[𝘀𝗽𝗿] changes introduced through rebaseupstream/users/ilovepi/spr/main.riscv-use-the-thread-local-stack-protector-for-android-targets

Created using spr 1.3.4

[skip ci]

262 files changed, 14163 insertions, 9155 deletions

diff --git a/clang-tools-extra/clang-tidy/hicpp/IgnoredRemoveResultCheck.cpp b/clang-tools-extra/clang-tidy/hicpp/IgnoredRemoveResultCheck.cpp
index 8020f8cd0625..b1a18485ce16 100644
--- a/clang-tools-extra/clang-tidy/hicpp/IgnoredRemoveResultCheck.cpp
+++ b/clang-tools-extra/clang-tidy/hicpp/IgnoredRemoveResultCheck.cpp
@@ -14,9 +14,9 @@ IgnoredRemoveResultCheck::IgnoredRemoveResultCheck(llvm::StringRef Name,
                                                    ClangTidyContext *Context)
     : UnusedReturnValueCheck(Name, Context,
                              {
-                                 "::std::remove",
-                                 "::std::remove_if",
-                                 "::std::unique",
+                                 "::std::remove$",
+                                 "::std::remove_if$",
+                                 "::std::unique$",
                              }) {
   // The constructor for ClangTidyCheck needs to have been called
   // before we can access options via Options.get().
diff --git a/clang-tools-extra/clangd/CompileCommands.cpp b/clang-tools-extra/clangd/CompileCommands.cpp
index 5b8128fca626..fddfffe7523d 100644
--- a/clang-tools-extra/clangd/CompileCommands.cpp
+++ b/clang-tools-extra/clangd/CompileCommands.cpp
@@ -466,7 +466,8 @@ llvm::ArrayRef<ArgStripper::Rule> ArgStripper::rulesFor(llvm::StringRef Arg) {
   static constexpr llvm::ArrayRef<llvm::StringLiteral> NAME(                   \
       NAME##_init, std::size(NAME##_init) - 1);
 #define OPTION(PREFIX, PREFIXED_NAME, ID, KIND, GROUP, ALIAS, ALIASARGS,       \
-               FLAGS, VISIBILITY, PARAM, HELP, METAVAR, VALUES)                \
+               FLAGS, VISIBILITY, PARAM, HELPTEXT, HELPTEXTSFORVARIANTS,       \
+               METAVAR, VALUES)                                                \
   Prefixes[DriverID::OPT_##ID] = PREFIX;
 #include "clang/Driver/Options.inc"
 #undef OPTION
@@ -478,7 +479,8 @@ llvm::ArrayRef<ArgStripper::Rule> ArgStripper::rulesFor(llvm::StringRef Arg) {
       const void *AliasArgs;
     } AliasTable[] = {
 #define OPTION(PREFIX, PREFIXED_NAME, ID, KIND, GROUP, ALIAS, ALIASARGS,       \
-               FLAGS, VISIBILITY, PARAM, HELP, METAVAR, VALUES)                \
+               FLAGS, VISIBILITY, PARAM, HELPTEXT, HELPTEXTSFORVARIANTS,       \
+               METAVAR, VALUES)                                                \
   {DriverID::OPT_##ID, DriverID::OPT_##ALIAS, ALIASARGS},
 #include "clang/Driver/Options.inc"
 #undef OPTION
diff --git a/clang-tools-extra/docs/ReleaseNotes.rst b/clang-tools-extra/docs/ReleaseNotes.rst
index 309b844615a1..456e09204fa2 100644
--- a/clang-tools-extra/docs/ReleaseNotes.rst
+++ b/clang-tools-extra/docs/ReleaseNotes.rst
@@ -210,6 +210,10 @@ Changes in existing checks
 - Improved :doc:`google-runtime-int <clang-tidy/checks/google/runtime-int>`
   check performance through optimizations.
 
+- Improved :doc:`hicpp-ignored-remove-result <clang-tidy/checks/hicpp/ignored-remove-result>`
+  check by ignoring other functions with same prefixes as the target specific
+  functions.
+
 - Improved :doc:`llvm-header-guard
   <clang-tidy/checks/llvm/header-guard>` check by replacing the local
   option `HeaderFileExtensions` by the global option of the same name.
diff --git a/clang-tools-extra/test/clang-tidy/checkers/hicpp/ignored-remove-result.cpp b/clang-tools-extra/test/clang-tidy/checkers/hicpp/ignored-remove-result.cpp
index b068f0859098..fc431024303a 100644
--- a/clang-tools-extra/test/clang-tidy/checkers/hicpp/ignored-remove-result.cpp
+++ b/clang-tools-extra/test/clang-tidy/checkers/hicpp/ignored-remove-result.cpp
@@ -15,6 +15,10 @@ ForwardIt unique(ForwardIt, ForwardIt);
 template <class InputIt, class T>
 InputIt find(InputIt, InputIt, const T&);
 
+struct unique_disposable {
+  void* release();
+};
+
 class error_code {
 };
 
@@ -63,4 +67,6 @@ void noWarning() {
   //   bugprone-unused-return-value's checked return types.
   errorFunc();
   (void) errorFunc();
+
+  std::unique_disposable{}.release();
 }
diff --git a/clang/docs/ReleaseNotes.rst b/clang/docs/ReleaseNotes.rst
index 8fc925350849..28e8ddb3c41c 100644
--- a/clang/docs/ReleaseNotes.rst
+++ b/clang/docs/ReleaseNotes.rst
@@ -334,6 +334,13 @@ Improvements to Clang's diagnostics
 - Clang now emits ``unused argument`` warning when the -fmodule-output flag is used
   with an input that is not of type c++-module.
 
+- Clang emits a ``-Wreturn-stack-address`` warning if a function returns a pointer or
+  reference to a struct literal. Fixes #GH8678
+
+- Clang emits a ``-Wunused-but-set-variable`` warning on C++ variables whose declaration
+  (with initializer) entirely consist the condition expression of a if/while/for construct
+  but are not actually used in the body of the if/while/for construct. Fixes #GH41447
+
 Improvements to Clang's time-trace
 ----------------------------------
 
@@ -473,6 +480,8 @@ Bug Fixes to C++ Support
   when one of the function had more specialized templates.
   Fixes (`#82509 <https://github.com/llvm/llvm-project/issues/82509>`_)
   and (`#74494 <https://github.com/llvm/llvm-project/issues/74494>`_)
+- Clang now supports direct lambda calls inside of a type alias template declarations.
+  This addresses (#GH70601), (#GH76674), (#GH79555), (#GH81145) and (#GH82104).
 - Allow access to a public template alias declaration that refers to friend's
   private nested type. (#GH25708).
 - Fixed a crash in constant evaluation when trying to access a
@@ -494,6 +503,7 @@ Bug Fixes to C++ Support
 
 - Fix crash when inheriting from a cv-qualified type. Fixes:
   (`#35603 <https://github.com/llvm/llvm-project/issues/35603>`_)
+- Fix a crash when the using enum declaration uses an anonymous enumeration. Fixes (#GH86790).
 
 Bug Fixes to AST Handling
 ^^^^^^^^^^^^^^^^^^^^^^^^^
diff --git a/clang/include/clang/AST/Decl.h b/clang/include/clang/AST/Decl.h
index a5879591f4c6..5f1f83bb0028 100644
--- a/clang/include/clang/AST/Decl.h
+++ b/clang/include/clang/AST/Decl.h
@@ -1100,6 +1100,9 @@ protected:
 
     LLVM_PREFERRED_TYPE(bool)
     unsigned EscapingByref : 1;
+
+    LLVM_PREFERRED_TYPE(bool)
+    unsigned IsCXXCondDecl : 1;
   };
 
   union {
@@ -1589,6 +1592,15 @@ public:
     NonParmVarDeclBits.EscapingByref = true;
   }
 
+  bool isCXXCondDecl() const {
+    return isa<ParmVarDecl>(this) ? false : NonParmVarDeclBits.IsCXXCondDecl;
+  }
+
+  void setCXXCondDecl() {
+    assert(!isa<ParmVarDecl>(this));
+    NonParmVarDeclBits.IsCXXCondDecl = true;
+  }
+
   /// Determines if this variable's alignment is dependent.
   bool hasDependentAlignment() const;
 
diff --git a/clang/include/clang/AST/DeclCXX.h b/clang/include/clang/AST/DeclCXX.h
index 9cebaff63bb0..7aed4d5cbc00 100644
--- a/clang/include/clang/AST/DeclCXX.h
+++ b/clang/include/clang/AST/DeclCXX.h
@@ -1869,6 +1869,10 @@ public:
     DL.MethodTyInfo = TS;
   }
 
+  void setLambdaDependencyKind(unsigned Kind) {
+    getLambdaData().DependencyKind = Kind;
+  }
+
   void setLambdaIsGeneric(bool IsGeneric) {
     assert(DefinitionData && DefinitionData->IsLambda &&
            "setting lambda property of non-lambda class");
diff --git a/clang/include/clang/Basic/Builtins.td b/clang/include/clang/Basic/Builtins.td
index f421223ff087..d6ceb450bd10 100644
--- a/clang/include/clang/Basic/Builtins.td
+++ b/clang/include/clang/Basic/Builtins.td
@@ -4587,6 +4587,12 @@ def GetDeviceSideMangledName : LangBuiltin<"CUDA_LANG"> {
 }
 
 // HLSL
+def HLSLAll : LangBuiltin<"HLSL_LANG"> {
+  let Spellings = ["__builtin_hlsl_elementwise_all"];
+  let Attributes = [NoThrow, Const];
+  let Prototype = "bool(...)";
+}
+
 def HLSLAny : LangBuiltin<"HLSL_LANG"> {
   let Spellings = ["__builtin_hlsl_elementwise_any"];
   let Attributes = [NoThrow, Const];
diff --git a/clang/include/clang/Basic/DiagnosticDriverKinds.td b/clang/include/clang/Basic/DiagnosticDriverKinds.td
index 3d86f7510bde..6186a88e1c80 100644
--- a/clang/include/clang/Basic/DiagnosticDriverKinds.td
+++ b/clang/include/clang/Basic/DiagnosticDriverKinds.td
@@ -142,6 +142,9 @@ def warn_drv_unsupported_diag_option_for_flang : Warning<
 def warn_drv_unsupported_option_for_processor : Warning<
   "ignoring '%0' option as it is not currently supported for processor '%1'">,
   InGroup<OptionIgnored>;
+def warn_drv_unsupported_openmp_library : Warning<
+  "The library '%0=%1' is not supported, openmp is not be enabled">,
+  InGroup<OptionIgnored>;
 
 def err_drv_invalid_thread_model_for_target : Error<
   "invalid thread model '%0' in '%1' for this target">;
diff --git a/clang/include/clang/Driver/Options.td b/clang/include/clang/Driver/Options.td
index 12e8dc7912c3..827d9d7c0c18 100644
--- a/clang/include/clang/Driver/Options.td
+++ b/clang/include/clang/Driver/Options.td
@@ -3409,17 +3409,24 @@ def fopenmp : Flag<["-"], "fopenmp">, Group<f_Group>,
   HelpText<"Parse OpenMP pragmas and generate parallel code.">;
 def fno_openmp : Flag<["-"], "fno-openmp">, Group<f_Group>,
   Flags<[NoArgumentUnused]>;
+class OpenMPVersionHelp<string program, string default> {
+  string str = !strconcat(
+    "Set OpenMP version (e.g. 45 for OpenMP 4.5, 51 for OpenMP 5.1). Default value is ",
+    default, " for ", program);
+}
 def fopenmp_version_EQ : Joined<["-"], "fopenmp-version=">, Group<f_Group>,
   Flags<[NoArgumentUnused]>,
   Visibility<[ClangOption, CC1Option, FlangOption, FC1Option]>,
-  HelpText<"Set OpenMP version (e.g. 45 for OpenMP 4.5, 51 for OpenMP 5.1). Default value is 51 for Clang">;
+  HelpText<OpenMPVersionHelp<"Clang", "51">.str>,
+  HelpTextForVariants<[FlangOption, FC1Option], OpenMPVersionHelp<"Flang", "11">.str>;
 defm openmp_extensions: BoolFOption<"openmp-extensions",
   LangOpts<"OpenMPExtensions">, DefaultTrue,
   PosFlag<SetTrue, [NoArgumentUnused], [ClangOption, CC1Option],
           "Enable all Clang extensions for OpenMP directives and clauses">,
   NegFlag<SetFalse, [NoArgumentUnused], [ClangOption, CC1Option],
           "Disable all Clang extensions for OpenMP directives and clauses">>;
-def fopenmp_EQ : Joined<["-"], "fopenmp=">, Group<f_Group>;
+def fopenmp_EQ : Joined<["-"], "fopenmp=">, Group<f_Group>,
+  Visibility<[ClangOption, CC1Option, FlangOption, FC1Option]>;
 def fopenmp_use_tls : Flag<["-"], "fopenmp-use-tls">, Group<f_Group>,
   Flags<[NoArgumentUnused, HelpHidden]>;
 def fnoopenmp_use_tls : Flag<["-"], "fnoopenmp-use-tls">, Group<f_Group>,
diff --git a/clang/include/clang/Lex/Token.h b/clang/include/clang/Lex/Token.h
index 36ec5ddaa29a..4f29fb7d1141 100644
--- a/clang/include/clang/Lex/Token.h
+++ b/clang/include/clang/Lex/Token.h
@@ -58,7 +58,7 @@ class Token {
   ///  Annotations (resolved type names, C++ scopes, etc): isAnnotation().
   ///    This is a pointer to sema-specific data for the annotation token.
   ///  Eof:
-  //     This is a pointer to a Decl.
+  ///    This is a pointer to a Decl.
   ///  Other:
   ///    This is null.
   void *PtrData;
diff --git a/clang/include/clang/Sema/Sema.h b/clang/include/clang/Sema/Sema.h
index 8c98d8c7fef7..149b268311bc 100644
--- a/clang/include/clang/Sema/Sema.h
+++ b/clang/include/clang/Sema/Sema.h
@@ -10170,6 +10170,9 @@ public:
 
       /// We are building deduction guides for a class.
       BuildingDeductionGuides,
+
+      /// We are instantiating a type alias template declaration.
+      TypeAliasTemplateInstantiation,
     } Kind;
 
     /// Was the enclosing context a non-instantiation SFINAE context?
@@ -10259,6 +10262,12 @@ public:
                           FunctionDecl *Entity, ExceptionSpecification,
                           SourceRange InstantiationRange = SourceRange());
 
+    /// Note that we are instantiating a type alias template declaration.
+    InstantiatingTemplate(Sema &SemaRef, SourceLocation PointOfInstantiation,
+                          TypeAliasTemplateDecl *Template,
+                          ArrayRef<TemplateArgument> TemplateArgs,
+                          SourceRange InstantiationRange = SourceRange());
+
     /// Note that we are instantiating a default argument in a
     /// template-id.
     InstantiatingTemplate(Sema &SemaRef, SourceLocation PointOfInstantiation,
diff --git a/clang/include/clang/StaticAnalyzer/Core/BugReporter/BugReporterVisitors.h b/clang/include/clang/StaticAnalyzer/Core/BugReporter/BugReporterVisitors.h
index d9b3d9352d32..cc3d93aabafd 100644
--- a/clang/include/clang/StaticAnalyzer/Core/BugReporter/BugReporterVisitors.h
+++ b/clang/include/clang/StaticAnalyzer/Core/BugReporter/BugReporterVisitors.h
@@ -374,6 +374,7 @@ bool trackExpressionValue(const ExplodedNode *N, const Expr *E,
 /// from.
 ///
 /// \param V We're searching for the store where \c R received this value.
+///        It may be either defined or undefined, but should not be unknown.
 /// \param R The region we're tracking.
 /// \param Opts Tracking options specifying how we want to track the value.
 /// \param Origin Only adds notes when the last store happened in a
@@ -383,7 +384,7 @@ bool trackExpressionValue(const ExplodedNode *N, const Expr *E,
 ///        changes to its value in a nested stackframe could be pruned, and
 ///        this visitor can prevent that without polluting the bugpath too
 ///        much.
-void trackStoredValue(KnownSVal V, const MemRegion *R,
+void trackStoredValue(SVal V, const MemRegion *R,
                       PathSensitiveBugReport &Report, TrackingOptions Opts = {},
                       const StackFrameContext *Origin = nullptr);
 
diff --git a/clang/include/clang/StaticAnalyzer/Core/PathSensitive/CallDescription.h b/clang/include/clang/StaticAnalyzer/Core/PathSensitive/CallDescription.h
index b4e1636130ca..ccfe8d47c290 100644
--- a/clang/include/clang/StaticAnalyzer/Core/PathSensitive/CallDescription.h
+++ b/clang/include/clang/StaticAnalyzer/Core/PathSensitive/CallDescription.h
@@ -32,19 +32,22 @@ namespace ento {
 class CallDescription {
 public:
   enum class Mode {
-    /// Match calls to functions from the C standard library. On some platforms
-    /// some functions may be implemented as macros that expand to calls to
-    /// built-in variants of the given functions, so in this mode we use some
-    /// heuristics to recognize these implementation-defined variants:
-    ///  - We also accept calls where the name is derived from the specified
-    ///    name by adding "__builtin" or similar prefixes/suffixes.
-    ///  - We also accept calls where the number of arguments or parameters is
-    ///    greater than the specified value.
+    /// Match calls to functions from the C standard library. This also
+    /// recognizes builtin variants whose name is derived by adding
+    /// "__builtin", "__inline" or similar prefixes or suffixes; but only
+    /// matches functions than are externally visible and are declared either
+    /// directly within a TU or in the namespace 'std'.
     /// For the exact heuristics, see CheckerContext::isCLibraryFunction().
-    /// (This mode only matches functions that are declared either directly
-    /// within a TU or in the namespace `std`.)
     CLibrary,
 
+    /// An extended version of the `CLibrary` mode that also matches the
+    /// hardened variants like __FOO_chk() and __builtin__FOO_chk() that take
+    /// additional arguments compared to the "regular" function FOO().
+    /// This is not the default behavior of `CLibrary` because in this case the
+    /// checker code must be prepared to handle the different parametrization.
+    /// For the exact heuristics, see CheckerContext::isHardenedVariantOf().
+    CLibraryMaybeHardened,
+
     /// Matches "simple" functions that are not methods. (Static methods are
     /// methods.)
     SimpleFunc,
@@ -187,6 +190,9 @@ public:
 private:
   bool matchesImpl(const FunctionDecl *Callee, size_t ArgCount,
                    size_t ParamCount) const;
+
+  bool matchNameOnly(const NamedDecl *ND) const;
+  bool matchQualifiedNameParts(const Decl *D) const;
 };
 
 /// An immutable map from CallDescriptions to arbitrary data. Provides a unified
diff --git a/clang/include/clang/StaticAnalyzer/Core/PathSensitive/CheckerContext.h b/clang/include/clang/StaticAnalyzer/Core/PathSensitive/CheckerContext.h
index 9923c41e6ad2..0365f9e41312 100644
--- a/clang/include/clang/StaticAnalyzer/Core/PathSensitive/CheckerContext.h
+++ b/clang/include/clang/StaticAnalyzer/Core/PathSensitive/CheckerContext.h
@@ -366,19 +366,31 @@ public:
     return getCalleeName(FunDecl);
   }
 
-  /// Returns true if the callee is an externally-visible function in the
-  /// top-level namespace, such as \c malloc.
+  /// Returns true if the given function is an externally-visible function in
+  /// the top-level namespace, such as \c malloc.
   ///
   /// If a name is provided, the function must additionally match the given
   /// name.
   ///
-  /// Note that this deliberately excludes C++ library functions in the \c std
-  /// namespace, but will include C library functions accessed through the
-  /// \c std namespace. This also does not check if the function is declared
-  /// as 'extern "C"', or if it uses C++ name mangling.
+  /// Note that this also accepts functions from the \c std namespace (because
+  /// headers like <cstdlib> declare them there) and does not check if the
+  /// function is declared as 'extern "C"' or if it uses C++ name mangling.
   static bool isCLibraryFunction(const FunctionDecl *FD,
                                  StringRef Name = StringRef());
 
+  /// In builds that use source hardening (-D_FORTIFY_SOURCE), many standard
+  /// functions are implemented as macros that expand to calls of hardened
+  /// functions that take additional arguments compared to the "usual"
+  /// variant and perform additional input validation. For example, a `memcpy`
+  /// call may expand to `__memcpy_chk()` or `__builtin___memcpy_chk()`.
+  ///
+  /// This method returns true if `FD` declares a fortified variant of the
+  /// standard library function `Name`.
+  ///
+  /// NOTE: This method relies on heuristics; extend it if you need to handle a
+  /// hardened variant that's not yet covered by it.
+  static bool isHardenedVariantOf(const FunctionDecl *FD, StringRef Name);
+
   /// Depending on wither the location corresponds to a macro, return
   /// either the macro name or the token spelling.
   ///
diff --git a/clang/include/clang/StaticAnalyzer/Core/PathSensitive/SVals.h b/clang/include/clang/StaticAnalyzer/Core/PathSensitive/SVals.h
index c60528b7685f..3a4b08725714 100644
--- a/clang/include/clang/StaticAnalyzer/Core/PathSensitive/SVals.h
+++ b/clang/include/clang/StaticAnalyzer/Core/PathSensitive/SVals.h
@@ -232,14 +232,6 @@ protected:
       : DefinedOrUnknownSVal(Kind, Data) {}
 };
 
-/// Represents an SVal that is guaranteed to not be UnknownVal.
-class KnownSVal : public SVal {
-public:
-  /*implicit*/ KnownSVal(DefinedSVal V) : SVal(V) {}
-  /*implicit*/ KnownSVal(UndefinedVal V) : SVal(V) {}
-  static bool classof(SVal V) { return !V.isUnknown(); }
-};
-
 class NonLoc : public DefinedSVal {
 protected:
   NonLoc(SValKind Kind, const void *Data) : DefinedSVal(Kind, Data) {}
diff --git a/clang/lib/AST/ASTImporter.cpp b/clang/lib/AST/ASTImporter.cpp
index 94a47a8f6190..45d4c9600537 100644
--- a/clang/lib/AST/ASTImporter.cpp
+++ b/clang/lib/AST/ASTImporter.cpp
@@ -4541,6 +4541,10 @@ ExpectedDecl ASTNodeImporter::VisitVarDecl(VarDecl *D) {
   ToVar->setQualifierInfo(ToQualifierLoc);
   ToVar->setAccess(D->getAccess());
   ToVar->setLexicalDeclContext(LexicalDC);
+  if (D->isInlineSpecified())
+    ToVar->setInlineSpecified();
+  if (D->isInline())
+    ToVar->setImplicitlyInline();
 
   if (FoundByLookup) {
     auto *Recent = const_cast<VarDecl *>(FoundByLookup->getMostRecentDecl());
diff --git a/clang/lib/CodeGen/BackendUtil.cpp b/clang/lib/CodeGen/BackendUtil.cpp
index e25a17658a34..6cc00b85664f 100644
--- a/clang/lib/CodeGen/BackendUtil.cpp
+++ b/clang/lib/CodeGen/BackendUtil.cpp
@@ -73,10 +73,10 @@
 #include "llvm/Transforms/Instrumentation/HWAddressSanitizer.h"
 #include "llvm/Transforms/Instrumentation/InstrProfiling.h"
 #include "llvm/Transforms/Instrumentation/KCFI.h"
+#include "llvm/Transforms/Instrumentation/LowerAllowCheckPass.h"
 #include "llvm/Transforms/Instrumentation/MemProfiler.h"
 #include "llvm/Transforms/Instrumentation/MemorySanitizer.h"
 #include "llvm/Transforms/Instrumentation/PGOInstrumentation.h"
-#include "llvm/Transforms/Instrumentation/RemoveTrapsPass.h"
 #include "llvm/Transforms/Instrumentation/SanitizerBinaryMetadata.h"
 #include "llvm/Transforms/Instrumentation/SanitizerCoverage.h"
 #include "llvm/Transforms/Instrumentation/ThreadSanitizer.h"
@@ -84,7 +84,6 @@
 #include "llvm/Transforms/Scalar/EarlyCSE.h"
 #include "llvm/Transforms/Scalar/GVN.h"
 #include "llvm/Transforms/Scalar/JumpThreading.h"
-#include "llvm/Transforms/Scalar/SimplifyCFG.h"
 #include "llvm/Transforms/Utils/Debugify.h"
 #include "llvm/Transforms/Utils/EntryExitInstrumenter.h"
 #include "llvm/Transforms/Utils/ModuleUtils.h"
@@ -747,18 +746,13 @@ static void addSanitizers(const Triple &TargetTriple,
     PB.registerOptimizerLastEPCallback(SanitizersCallback);
   }
 
-  if (RemoveTrapsPass::IsRequested()) {
+  if (LowerAllowCheckPass::IsRequested()) {
     // We can optimize after inliner, and PGO profile matching. The hook below
     // is called at the end `buildFunctionSimplificationPipeline`, which called
     // from `buildInlinerPipeline`, which called after profile matching.
     PB.registerScalarOptimizerLateEPCallback(
         [](FunctionPassManager &FPM, OptimizationLevel Level) {
-          // RemoveTrapsPass expects trap blocks preceded by conditional
-          // branches, which usually is not the case without SimplifyCFG.
-          // TODO: Remove `SimplifyCFGPass` after switching to dedicated
-          // intrinsic.
-          FPM.addPass(SimplifyCFGPass());
-          FPM.addPass(RemoveTrapsPass());
+          FPM.addPass(LowerAllowCheckPass());
         });
   }
 }
diff --git a/clang/lib/CodeGen/CGBuiltin.cpp b/clang/lib/CodeGen/CGBuiltin.cpp
index 2537e715b63e..df7502b8def5 100644
--- a/clang/lib/CodeGen/CGBuiltin.cpp
+++ b/clang/lib/CodeGen/CGBuiltin.cpp
@@ -13,6 +13,7 @@
 #include "ABIInfo.h"
 #include "CGCUDARuntime.h"
 #include "CGCXXABI.h"
+#include "CGHLSLRuntime.h"
 #include "CGObjCRuntime.h"
 #include "CGOpenCLRuntime.h"
 #include "CGRecordLayout.h"
@@ -18172,6 +18173,13 @@ Value *CodeGenFunction::EmitHLSLBuiltinExpr(unsigned BuiltinID,
     return nullptr;
 
   switch (BuiltinID) {
+  case Builtin::BI__builtin_hlsl_elementwise_all: {
+    Value *Op0 = EmitScalarExpr(E->getArg(0));
+    return Builder.CreateIntrinsic(
+        /*ReturnType=*/llvm::Type::getInt1Ty(getLLVMContext()),
+        CGM.getHLSLRuntime().getAllIntrinsic(), ArrayRef<Value *>{Op0}, nullptr,
+        "hlsl.all");
+  }
   case Builtin::BI__builtin_hlsl_elementwise_any: {
     Value *Op0 = EmitScalarExpr(E->getArg(0));
     return Builder.CreateIntrinsic(
diff --git a/clang/lib/CodeGen/CGExpr.cpp b/clang/lib/CodeGen/CGExpr.cpp
index 54432353e742..2480972f1432 100644
--- a/clang/lib/CodeGen/CGExpr.cpp
+++ b/clang/lib/CodeGen/CGExpr.cpp
@@ -56,7 +56,13 @@ using namespace CodeGen;
 // Experiment to make sanitizers easier to debug
 static llvm::cl::opt<bool> ClSanitizeDebugDeoptimization(
     "ubsan-unique-traps", llvm::cl::Optional,
-    llvm::cl::desc("Deoptimize traps for UBSAN so there is 1 trap per check"));
+    llvm::cl::desc("Deoptimize traps for UBSAN so there is 1 trap per check."));
+
+// TODO: Introduce frontend options to enabled per sanitizers, similar to
+// `fsanitize-trap`.
+static llvm::cl::opt<bool> ClSanitizeGuardChecks(
+    "ubsan-guard-checks", llvm::cl::Optional,
+    llvm::cl::desc("Guard UBSAN checks with `llvm.allow.ubsan.check()`."));
 
 //===--------------------------------------------------------------------===//
 //                        Miscellaneous Helper Methods
@@ -3522,6 +3528,17 @@ void CodeGenFunction::EmitCheck(
     Cond = Cond ? Builder.CreateAnd(Cond, Check) : Check;
   }
 
+  if (ClSanitizeGuardChecks) {
+    llvm::Value *Allow =
+        Builder.CreateCall(CGM.getIntrinsic(llvm::Intrinsic::allow_ubsan_check),
+                           llvm::ConstantInt::get(CGM.Int8Ty, CheckHandler));
+
+    for (llvm::Value **Cond : {&FatalCond, &RecoverableCond, &TrapCond}) {
+      if (*Cond)
+        *Cond = Builder.CreateOr(*Cond, Builder.CreateNot(Allow));
+    }
+  }
+
   if (TrapCond)
     EmitTrapCheck(TrapCond, CheckHandler);
   if (!FatalCond && !RecoverableCond)
diff --git a/clang/lib/CodeGen/CGHLSLRuntime.cpp b/clang/lib/CodeGen/CGHLSLRuntime.cpp
index 794d93358b0a..5e6a3dd4878f 100644
--- a/clang/lib/CodeGen/CGHLSLRuntime.cpp
+++ b/clang/lib/CodeGen/CGHLSLRuntime.cpp
@@ -17,8 +17,6 @@
 #include "CodeGenModule.h"
 #include "clang/AST/Decl.h"
 #include "clang/Basic/TargetOptions.h"
-#include "llvm/IR/IntrinsicsDirectX.h"
-#include "llvm/IR/IntrinsicsSPIRV.h"
 #include "llvm/IR/Metadata.h"
 #include "llvm/IR/Module.h"
 #include "llvm/Support/FormatVariadic.h"
@@ -117,6 +115,10 @@ GlobalVariable *replaceBuffer(CGHLSLRuntime::Buffer &Buf) {
 
 } // namespace
 
+llvm::Triple::ArchType CGHLSLRuntime::getArch() {
+  return CGM.getTarget().getTriple().getArch();
+}
+
 void CGHLSLRuntime::addConstant(VarDecl *D, Buffer &CB) {
   if (D->getStorageClass() == SC_Static) {
     // For static inside cbuffer, take as global static.
@@ -343,18 +345,8 @@ llvm::Value *CGHLSLRuntime::emitInputSemantic(IRBuilder<> &B,
     return B.CreateCall(FunctionCallee(DxGroupIndex));
   }
   if (D.hasAttr<HLSLSV_DispatchThreadIDAttr>()) {
-    llvm::Function *ThreadIDIntrinsic;
-    switch (CGM.getTarget().getTriple().getArch()) {
-    case llvm::Triple::dxil:
-      ThreadIDIntrinsic = CGM.getIntrinsic(Intrinsic::dx_thread_id);
-      break;
-    case llvm::Triple::spirv:
-      ThreadIDIntrinsic = CGM.getIntrinsic(Intrinsic::spv_thread_id);
-      break;
-    default:
-      llvm_unreachable("Input semantic not supported by target");
-      break;
-    }
+    llvm::Function *ThreadIDIntrinsic =
+        CGM.getIntrinsic(getThreadIdIntrinsic());
     return buildVectorInput(B, ThreadIDIntrinsic, Ty);
   }
   assert(false && "Unhandled parameter attribute");
diff --git a/clang/lib/CodeGen/CGHLSLRuntime.h b/clang/lib/CodeGen/CGHLSLRuntime.h
index bffefb66740a..2b8073aef973 100644
--- a/clang/lib/CodeGen/CGHLSLRuntime.h
+++ b/clang/lib/CodeGen/CGHLSLRuntime.h
@@ -16,7 +16,11 @@
 #define LLVM_CLANG_LIB_CODEGEN_CGHLSLRUNTIME_H
 
 #include "llvm/IR/IRBuilder.h"
+#include "llvm/IR/Intrinsics.h"
+#include "llvm/IR/IntrinsicsDirectX.h"
+#include "llvm/IR/IntrinsicsSPIRV.h"
 
+#include "clang/Basic/Builtins.h"
 #include "clang/Basic/HLSLRuntime.h"
 
 #include "llvm/ADT/SmallVector.h"
@@ -26,6 +30,22 @@
 #include <optional>
 #include <vector>
 
+// A function generator macro for picking the right intrinsic
+// for the target backend
+#define GENERATE_HLSL_INTRINSIC_FUNCTION(FunctionName, IntrinsicPostfix)       \
+  llvm::Intrinsic::ID get##FunctionName##Intrinsic() {                         \
+    llvm::Triple::ArchType Arch = getArch();                                   \
+    switch (Arch) {                                                            \
+    case llvm::Triple::dxil:                                                   \
+      return llvm::Intrinsic::dx_##IntrinsicPostfix;                           \
+    case llvm::Triple::spirv:                                                  \
+      return llvm::Intrinsic::spv_##IntrinsicPostfix;                          \
+    default:                                                                   \
+      llvm_unreachable("Intrinsic " #IntrinsicPostfix                          \
+                       " not supported by target architecture");               \
+    }                                                                          \
+  }
+
 namespace llvm {
 class GlobalVariable;
 class Function;
@@ -48,6 +68,17 @@ class CodeGenModule;
 
 class CGHLSLRuntime {
 public:
+  //===----------------------------------------------------------------------===//
+  // Start of reserved area for HLSL intrinsic getters.
+  //===----------------------------------------------------------------------===//
+
+  GENERATE_HLSL_INTRINSIC_FUNCTION(All, all)
+  GENERATE_HLSL_INTRINSIC_FUNCTION(ThreadId, thread_id)
+
+  //===----------------------------------------------------------------------===//
+  // End of reserved area for HLSL intrinsic getters.
+  //===----------------------------------------------------------------------===//
+
   struct BufferResBinding {
     // The ID like 2 in register(b2, space1).
     std::optional<unsigned> Reg;
@@ -96,6 +127,7 @@ private:
                                    BufferResBinding &Binding);
   void addConstant(VarDecl *D, Buffer &CB);
   void addBufferDecls(const DeclContext *DC, Buffer &CB);
+  llvm::Triple::ArchType getArch();
   llvm::SmallVector<Buffer> Buffers;
 };
 
diff --git a/clang/lib/CodeGen/CodeGenTBAA.cpp b/clang/lib/CodeGen/CodeGenTBAA.cpp
index a1e14c5f0a8c..837bf725da38 100644
--- a/clang/lib/CodeGen/CodeGenTBAA.cpp
+++ b/clang/lib/CodeGen/CodeGenTBAA.cpp
@@ -22,6 +22,7 @@
 #include "clang/AST/Mangle.h"
 #include "clang/AST/RecordLayout.h"
 #include "clang/Basic/CodeGenOptions.h"
+#include "clang/Basic/TargetInfo.h"
 #include "llvm/ADT/SmallSet.h"
 #include "llvm/IR/Constants.h"
 #include "llvm/IR/LLVMContext.h"
@@ -319,7 +320,13 @@ CodeGenTBAA::CollectFields(uint64_t BaseOffset,
       // base type.
       if ((*i)->isBitField()) {
         const CGBitFieldInfo &Info = CGRL.getBitFieldInfo(*i);
-        if (Info.Offset != 0)
+        // For big endian targets the first bitfield in the consecutive run is
+        // at the most-significant end; see CGRecordLowering::setBitFieldInfo
+        // for more information.
+        bool IsBE = Context.getTargetInfo().isBigEndian();
+        bool IsFirst = IsBE ? Info.StorageSize - (Info.Offset + Info.Size) == 0
+                            : Info.Offset == 0;
+        if (!IsFirst)
           continue;
         unsigned CurrentBitFieldSize = Info.StorageSize;
         uint64_t Size =
diff --git a/clang/lib/Driver/ToolChains/Flang.cpp b/clang/lib/Driver/ToolChains/Flang.cpp
index 70daa699e3a9..2c83f70eb788 100644
--- a/clang/lib/Driver/ToolChains/Flang.cpp
+++ b/clang/lib/Driver/ToolChains/Flang.cpp
@@ -35,27 +35,18 @@ static void addDashXForInput(const ArgList &Args, const InputInfo &Input,
 
 void Flang::addFortranDialectOptions(const ArgList &Args,
                                      ArgStringList &CmdArgs) const {
-  Args.addAllArgs(CmdArgs, {options::OPT_ffixed_form,
-                            options::OPT_ffree_form,
-                            options::OPT_ffixed_line_length_EQ,
-                            options::OPT_fopenmp,
-                            options::OPT_fopenmp_version_EQ,
-                            options::OPT_fopenacc,
-                            options::OPT_finput_charset_EQ,
-                            options::OPT_fimplicit_none,
-                            options::OPT_fno_implicit_none,
-                            options::OPT_fbackslash,
-                            options::OPT_fno_backslash,
-                            options::OPT_flogical_abbreviations,
-                            options::OPT_fno_logical_abbreviations,
-                            options::OPT_fxor_operator,
-                            options::OPT_fno_xor_operator,
-                            options::OPT_falternative_parameter_statement,
-                            options::OPT_fdefault_real_8,
-                            options::OPT_fdefault_integer_8,
-                            options::OPT_fdefault_double_8,
-                            options::OPT_flarge_sizes,
-                            options::OPT_fno_automatic});
+  Args.addAllArgs(
+      CmdArgs, {options::OPT_ffixed_form, options::OPT_ffree_form,
+                options::OPT_ffixed_line_length_EQ, options::OPT_fopenacc,
+                options::OPT_finput_charset_EQ, options::OPT_fimplicit_none,
+                options::OPT_fno_implicit_none, options::OPT_fbackslash,
+                options::OPT_fno_backslash, options::OPT_flogical_abbreviations,
+                options::OPT_fno_logical_abbreviations,
+                options::OPT_fxor_operator, options::OPT_fno_xor_operator,
+                options::OPT_falternative_parameter_statement,
+                options::OPT_fdefault_real_8, options::OPT_fdefault_integer_8,
+                options::OPT_fdefault_double_8, options::OPT_flarge_sizes,
+                options::OPT_fno_automatic});
 }
 
 void Flang::addPreprocessingOptions(const ArgList &Args,
@@ -763,6 +754,35 @@ void Flang::ConstructJob(Compilation &C, const JobAction &JA,
   // Add other compile options
   addOtherOptions(Args, CmdArgs);
 
+  // Forward flags for OpenMP. We don't do this if the current action is an
+  // device offloading action other than OpenMP.
+  if (Args.hasFlag(options::OPT_fopenmp, options::OPT_fopenmp_EQ,
+                   options::OPT_fno_openmp, false) &&
+      (JA.isDeviceOffloading(Action::OFK_None) ||
+       JA.isDeviceOffloading(Action::OFK_OpenMP))) {
+    switch (D.getOpenMPRuntime(Args)) {
+    case Driver::OMPRT_OMP:
+    case Driver::OMPRT_IOMP5:
+      // Clang can generate useful OpenMP code for these two runtime libraries.
+      CmdArgs.push_back("-fopenmp");
+      Args.AddAllArgs(CmdArgs, options::OPT_fopenmp_version_EQ);
+
+      // FIXME: Clang supports a whole bunch more flags here.
+      break;
+    default:
+      // By default, if Clang doesn't know how to generate useful OpenMP code
+      // for a specific runtime library, we just don't pass the '-fopenmp' flag
+      // down to the actual compilation.
+      // FIXME: It would be better to have a mode which *only* omits IR
+      // generation based on the OpenMP support so that we get consistent
+      // semantic analysis, etc.
+      const Arg *A = Args.getLastArg(options::OPT_fopenmp_EQ);
+      D.Diag(diag::warn_drv_unsupported_openmp_library)
+          << A->getSpelling() << A->getValue();
+      break;
+    }
+  }
+
   // Offloading related options
   addOffloadOptions(C, Inputs, JA, Args, CmdArgs);
 
diff --git a/clang/lib/Format/UnwrappedLineParser.cpp b/clang/lib/Format/UnwrappedLineParser.cpp
index 98ae1c8f62bb..55c627d87f08 100644
--- a/clang/lib/Format/UnwrappedLineParser.cpp
+++ b/clang/lib/Format/UnwrappedLineParser.cpp
@@ -495,20 +495,22 @@ void UnwrappedLineParser::calculateBraceTypes(bool ExpectClassBody) {
   };
   SmallVector<StackEntry, 8> LBraceStack;
   assert(Tok->is(tok::l_brace));
+
   do {
-    // Get next non-comment, non-preprocessor token.
     FormatToken *NextTok;
     do {
       NextTok = Tokens->getNextToken();
     } while (NextTok->is(tok::comment));
-    if (!Style.isTableGen()) {
-      // InTableGen, '#' is like binary operator. Not a preprocessor directive.
-      while (NextTok->is(tok::hash) && !Line->InMacroBody) {
-        NextTok = Tokens->getNextToken();
+
+    if (!Line->InMacroBody && !Style.isTableGen()) {
+      // Skip PPDirective lines and comments.
+      while (NextTok->is(tok::hash)) {
         do {
           NextTok = Tokens->getNextToken();
-        } while (NextTok->is(tok::comment) ||
-                 (NextTok->NewlinesBefore == 0 && NextTok->isNot(tok::eof)));
+        } while (NextTok->NewlinesBefore == 0 && NextTok->isNot(tok::eof));
+
+        while (NextTok->is(tok::comment))
+          NextTok = Tokens->getNextToken();
       }
     }
 
@@ -640,6 +642,7 @@ void UnwrappedLineParser::calculateBraceTypes(bool ExpectClassBody) {
     default:
       break;
     }
+
     PrevTok = Tok;
     Tok = NextTok;
   } while (Tok->isNot(tok::eof) && !LBraceStack.empty());
diff --git a/clang/lib/Frontend/CompilerInvocation.cpp b/clang/lib/Frontend/CompilerInvocation.cpp
index c5bfb8ef1b56..1f1f5440ddd7 100644
--- a/clang/lib/Frontend/CompilerInvocation.cpp
+++ b/clang/lib/Frontend/CompilerInvocation.cpp
@@ -533,10 +533,10 @@ static T extractMaskValue(T KeyPath) {
 
 #define PARSE_OPTION_WITH_MARSHALLING(                                         \
     ARGS, DIAGS, PREFIX_TYPE, SPELLING, ID, KIND, GROUP, ALIAS, ALIASARGS,     \
-    FLAGS, VISIBILITY, PARAM, HELPTEXT, METAVAR, VALUES, SHOULD_PARSE,         \
-    ALWAYS_EMIT, KEYPATH, DEFAULT_VALUE, IMPLIED_CHECK, IMPLIED_VALUE,         \
-    NORMALIZER, DENORMALIZER, MERGER, EXTRACTOR, TABLE_INDEX)                  \
-  if ((VISIBILITY)&options::CC1Option) {                                       \
+    FLAGS, VISIBILITY, PARAM, HELPTEXT, HELPTEXTSFORVARIANTS, METAVAR, VALUES, \
+    SHOULD_PARSE, ALWAYS_EMIT, KEYPATH, DEFAULT_VALUE, IMPLIED_CHECK,          \
+    IMPLIED_VALUE, NORMALIZER, DENORMALIZER, MERGER, EXTRACTOR, TABLE_INDEX)   \
+  if ((VISIBILITY) & options::CC1Option) {                                     \
     KEYPATH = MERGER(KEYPATH, DEFAULT_VALUE);                                  \
     if (IMPLIED_CHECK)                                                         \
       KEYPATH = MERGER(KEYPATH, IMPLIED_VALUE);                                \
@@ -550,10 +550,10 @@ static T extractMaskValue(T KeyPath) {
 // with lifetime extension of the reference.
 #define GENERATE_OPTION_WITH_MARSHALLING(                                      \
     CONSUMER, PREFIX_TYPE, SPELLING, ID, KIND, GROUP, ALIAS, ALIASARGS, FLAGS, \
-    VISIBILITY, PARAM, HELPTEXT, METAVAR, VALUES, SHOULD_PARSE, ALWAYS_EMIT,   \
-    KEYPATH, DEFAULT_VALUE, IMPLIED_CHECK, IMPLIED_VALUE, NORMALIZER,          \
-    DENORMALIZER, MERGER, EXTRACTOR, TABLE_INDEX)                              \
-  if ((VISIBILITY)&options::CC1Option) {                                       \
+    VISIBILITY, PARAM, HELPTEXT, HELPTEXTSFORVARIANTS, METAVAR, VALUES,        \
+    SHOULD_PARSE, ALWAYS_EMIT, KEYPATH, DEFAULT_VALUE, IMPLIED_CHECK,          \
+    IMPLIED_VALUE, NORMALIZER, DENORMALIZER, MERGER, EXTRACTOR, TABLE_INDEX)   \
+  if ((VISIBILITY) & options::CC1Option) {                                     \
     [&](const auto &Extracted) {                                               \
       if (ALWAYS_EMIT ||                                                       \
           (Extracted !=                                                        \
diff --git a/clang/lib/Frontend/FrontendActions.cpp b/clang/lib/Frontend/FrontendActions.cpp
index 0bc26b694cfc..7ee6ccf396a8 100644
--- a/clang/lib/Frontend/FrontendActions.cpp
+++ b/clang/lib/Frontend/FrontendActions.cpp
@@ -453,6 +453,8 @@ private:
       return "BuildingBuiltinDumpStructCall";
     case CodeSynthesisContext::BuildingDeductionGuides:
       return "BuildingDeductionGuides";
+    case Sema::CodeSynthesisContext::TypeAliasTemplateInstantiation:
+      return "TypeAliasTemplateInstantiation";
     }
     return "";
   }
diff --git a/clang/lib/Headers/hlsl/hlsl_intrinsics.h b/clang/lib/Headers/hlsl/hlsl_intrinsics.h
index 9fb6204f90c9..06409c6fc774 100644
--- a/clang/lib/Headers/hlsl/hlsl_intrinsics.h
+++ b/clang/lib/Headers/hlsl/hlsl_intrinsics.h
@@ -101,6 +101,118 @@ _HLSL_BUILTIN_ALIAS(__builtin_elementwise_abs)
 double4 abs(double4);
 
 //===----------------------------------------------------------------------===//
+// all builtins
+//===----------------------------------------------------------------------===//
+
+/// \fn bool all(T x)
+/// \brief Returns True if all components of the \a x parameter are non-zero;
+/// otherwise, false. \param x The input value.
+
+#ifdef __HLSL_ENABLE_16_BIT
+_HLSL_AVAILABILITY(shadermodel, 6.2)
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_all)
+bool all(int16_t);
+_HLSL_AVAILABILITY(shadermodel, 6.2)
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_all)
+bool all(int16_t2);
+_HLSL_AVAILABILITY(shadermodel, 6.2)
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_all)
+bool all(int16_t3);
+_HLSL_AVAILABILITY(shadermodel, 6.2)
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_all)
+bool all(int16_t4);
+_HLSL_AVAILABILITY(shadermodel, 6.2)
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_all)
+bool all(uint16_t);
+_HLSL_AVAILABILITY(shadermodel, 6.2)
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_all)
+bool all(uint16_t2);
+_HLSL_AVAILABILITY(shadermodel, 6.2)
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_all)
+bool all(uint16_t3);
+_HLSL_AVAILABILITY(shadermodel, 6.2)
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_all)
+bool all(uint16_t4);
+#endif
+
+_HLSL_16BIT_AVAILABILITY(shadermodel, 6.2)
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_all)
+bool all(half);
+_HLSL_16BIT_AVAILABILITY(shadermodel, 6.2)
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_all)
+bool all(half2);
+_HLSL_16BIT_AVAILABILITY(shadermodel, 6.2)
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_all)
+bool all(half3);
+_HLSL_16BIT_AVAILABILITY(shadermodel, 6.2)
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_all)
+bool all(half4);
+
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_all)
+bool all(bool);
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_all)
+bool all(bool2);
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_all)
+bool all(bool3);
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_all)
+bool all(bool4);
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_all)
+
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_all)
+bool all(int);
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_all)
+bool all(int2);
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_all)
+bool all(int3);
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_all)
+bool all(int4);
+
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_all)
+bool all(uint);
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_all)
+bool all(uint2);
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_all)
+bool all(uint3);
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_all)
+bool all(uint4);
+
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_all)
+bool all(float);
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_all)
+bool all(float2);
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_all)
+bool all(float3);
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_all)
+bool all(float4);
+
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_all)
+bool all(int64_t);
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_all)
+bool all(int64_t2);
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_all)
+bool all(int64_t3);
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_all)
+bool all(int64_t4);
+
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_all)
+bool all(uint64_t);
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_all)
+bool all(uint64_t2);
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_all)
+bool all(uint64_t3);
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_all)
+bool all(uint64_t4);
+
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_all)
+bool all(double);
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_all)
+bool all(double2);
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_all)
+bool all(double3);
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_all)
+bool all(double4);
+
+//===----------------------------------------------------------------------===//
 // any builtins
 //===----------------------------------------------------------------------===//
 
diff --git a/clang/lib/Sema/SemaChecking.cpp b/clang/lib/Sema/SemaChecking.cpp
index 3dcd18b3afc8..4c30cc3160de 100644
--- a/clang/lib/Sema/SemaChecking.cpp
+++ b/clang/lib/Sema/SemaChecking.cpp
@@ -5563,6 +5563,7 @@ void SetElementTypeAsReturnType(Sema *S, CallExpr *TheCall,
 // returning an ExprError
 bool Sema::CheckHLSLBuiltinFunctionCall(unsigned BuiltinID, CallExpr *TheCall) {
   switch (BuiltinID) {
+  case Builtin::BI__builtin_hlsl_elementwise_all:
   case Builtin::BI__builtin_hlsl_elementwise_any: {
     if (checkArgCount(*this, TheCall, 1))
       return true;
diff --git a/clang/lib/Sema/SemaConcept.cpp b/clang/lib/Sema/SemaConcept.cpp
index b2986c5012ea..e00c97260282 100644
--- a/clang/lib/Sema/SemaConcept.cpp
+++ b/clang/lib/Sema/SemaConcept.cpp
@@ -615,10 +615,12 @@ bool Sema::SetupConstraintScope(
     // reference the original primary template.
     // We walk up the instantiated template chain so that nested lambdas get
     // handled properly.
-    for (FunctionTemplateDecl *FromMemTempl =
-             PrimaryTemplate->getInstantiatedFromMemberTemplate();
-         FromMemTempl;
-         FromMemTempl = FromMemTempl->getInstantiatedFromMemberTemplate()) {
+    // We should only collect instantiated parameters from the primary template.
+    // Otherwise, we may have mismatched template parameter depth!
+    if (FunctionTemplateDecl *FromMemTempl =
+            PrimaryTemplate->getInstantiatedFromMemberTemplate()) {
+      while (FromMemTempl->getInstantiatedFromMemberTemplate())
+        FromMemTempl = FromMemTempl->getInstantiatedFromMemberTemplate();
       if (addInstantiatedParametersToScope(FD, FromMemTempl->getTemplatedDecl(),
                                            Scope, MLTAL))
         return true;
diff --git a/clang/lib/Sema/SemaDecl.cpp b/clang/lib/Sema/SemaDecl.cpp
index 5c1152896559..c790dab72dd7 100644
--- a/clang/lib/Sema/SemaDecl.cpp
+++ b/clang/lib/Sema/SemaDecl.cpp
@@ -1537,6 +1537,10 @@ void Sema::PushOnScopeChains(NamedDecl *D, Scope *S, bool AddToContext) {
       cast<FunctionDecl>(D)->isFunctionTemplateSpecialization())
     return;
 
+  if (isa<UsingEnumDecl>(D) && D->getDeclName().isEmpty()) {
+    S->AddDecl(D);
+    return;
+  }
   // If this replaces anything in the current scope,
   IdentifierResolver::iterator I = IdResolver.begin(D->getDeclName()),
                                IEnd = IdResolver.end();
@@ -2188,8 +2192,21 @@ void Sema::DiagnoseUnusedButSetDecl(const VarDecl *VD,
 
   assert(iter->getSecond() >= 0 &&
          "Found a negative number of references to a VarDecl");
-  if (iter->getSecond() != 0)
-    return;
+  if (int RefCnt = iter->getSecond(); RefCnt > 0) {
+    // Assume the given VarDecl is "used" if its ref count stored in
+    // `RefMinusAssignments` is positive, with one exception.
+    //
+    // For a C++ variable whose decl (with initializer) entirely consist the
+    // condition expression of a if/while/for construct,
+    // Clang creates a DeclRefExpr for the condition expression rather than a
+    // BinaryOperator of AssignmentOp. Thus, the C++ variable's ref
+    // count stored in `RefMinusAssignment` equals 1 when the variable is never
+    // used in the body of the if/while/for construct.
+    bool UnusedCXXCondDecl = VD->isCXXCondDecl() && (RefCnt == 1);
+    if (!UnusedCXXCondDecl)
+      return;
+  }
+
   unsigned DiagID = isa<ParmVarDecl>(VD) ? diag::warn_unused_but_set_parameter
                                          : diag::warn_unused_but_set_variable;
   DiagReceiver(VD->getLocation(), PDiag(DiagID) << VD);
diff --git a/clang/lib/Sema/SemaDeclCXX.cpp b/clang/lib/Sema/SemaDeclCXX.cpp
index f32ff396f8a5..068a2e4f04fa 100644
--- a/clang/lib/Sema/SemaDeclCXX.cpp
+++ b/clang/lib/Sema/SemaDeclCXX.cpp
@@ -18564,6 +18564,9 @@ DeclResult Sema::ActOnCXXConditionDeclaration(Scope *S, Declarator &D) {
     return true;
   }
 
+  if (auto *VD = dyn_cast<VarDecl>(Dcl))
+    VD->setCXXCondDecl();
+
   return Dcl;
 }
 
diff --git a/clang/lib/Sema/SemaTemplate.cpp b/clang/lib/Sema/SemaTemplate.cpp
index d3def13f495d..0b75f4fb401e 100644
--- a/clang/lib/Sema/SemaTemplate.cpp
+++ b/clang/lib/Sema/SemaTemplate.cpp
@@ -4383,6 +4383,11 @@ QualType Sema::CheckTemplateIdType(TemplateName Name,
     if (Inst.isInvalid())
       return QualType();
 
+    InstantiatingTemplate InstTemplate(
+        *this, /*PointOfInstantiation=*/AliasTemplate->getBeginLoc(),
+        /*Template=*/AliasTemplate,
+        /*TemplateArgs=*/TemplateArgLists.getInnermost());
+
     std::optional<ContextRAII> SavedContext;
     if (!AliasTemplate->getDeclContext()->isFileContext())
       SavedContext.emplace(*this, AliasTemplate->getDeclContext());
diff --git a/clang/lib/Sema/SemaTemplateInstantiate.cpp b/clang/lib/Sema/SemaTemplateInstantiate.cpp
index 5e05cb4f790b..d7b7291091ec 100644
--- a/clang/lib/Sema/SemaTemplateInstantiate.cpp
+++ b/clang/lib/Sema/SemaTemplateInstantiate.cpp
@@ -80,6 +80,81 @@ struct Response {
     return R;
   }
 };
+
+// Retrieve the primary template for a lambda call operator. It's
+// unfortunate that we only have the mappings of call operators rather
+// than lambda classes.
+const FunctionDecl *
+getPrimaryTemplateOfGenericLambda(const FunctionDecl *LambdaCallOperator) {
+  while (true) {
+    if (auto *FTD = dyn_cast_if_present<FunctionTemplateDecl>(
+            LambdaCallOperator->getDescribedTemplate());
+        FTD && FTD->getInstantiatedFromMemberTemplate()) {
+      LambdaCallOperator =
+          FTD->getInstantiatedFromMemberTemplate()->getTemplatedDecl();
+    } else if (auto *Prev = cast<CXXMethodDecl>(LambdaCallOperator)
+                                ->getInstantiatedFromMemberFunction())
+      LambdaCallOperator = Prev;
+    else
+      break;
+  }
+  return LambdaCallOperator;
+}
+
+struct EnclosingTypeAliasTemplateDetails {
+  TypeAliasTemplateDecl *Template = nullptr;
+  TypeAliasTemplateDecl *PrimaryTypeAliasDecl = nullptr;
+  ArrayRef<TemplateArgument> AssociatedTemplateArguments;
+
+  explicit operator bool() noexcept { return Template; }
+};
+
+// Find the enclosing type alias template Decl from CodeSynthesisContexts, as
+// well as its primary template and instantiating template arguments.
+EnclosingTypeAliasTemplateDetails
+getEnclosingTypeAliasTemplateDecl(Sema &SemaRef) {
+  for (auto &CSC : llvm::reverse(SemaRef.CodeSynthesisContexts)) {
+    if (CSC.Kind != Sema::CodeSynthesisContext::SynthesisKind::
+                        TypeAliasTemplateInstantiation)
+      continue;
+    EnclosingTypeAliasTemplateDetails Result;
+    auto *TATD = cast<TypeAliasTemplateDecl>(CSC.Entity),
+         *Next = TATD->getInstantiatedFromMemberTemplate();
+    Result = {
+        /*Template=*/TATD,
+        /*PrimaryTypeAliasDecl=*/TATD,
+        /*AssociatedTemplateArguments=*/CSC.template_arguments(),
+    };
+    while (Next) {
+      Result.PrimaryTypeAliasDecl = Next;
+      Next = Next->getInstantiatedFromMemberTemplate();
+    }
+    return Result;
+  }
+  return {};
+}
+
+// Check if we are currently inside of a lambda expression that is
+// surrounded by a using alias declaration. e.g.
+//   template <class> using type = decltype([](auto) { ^ }());
+// By checking if:
+//  1. The lambda expression and the using alias declaration share the
+//  same declaration context.
+//  2. They have the same template depth.
+// We have to do so since a TypeAliasTemplateDecl (or a TypeAliasDecl) is never
+// a DeclContext, nor does it have an associated specialization Decl from which
+// we could collect these template arguments.
+bool isLambdaEnclosedByTypeAliasDecl(
+    const FunctionDecl *PrimaryLambdaCallOperator,
+    const TypeAliasTemplateDecl *PrimaryTypeAliasDecl) {
+  return cast<CXXRecordDecl>(PrimaryLambdaCallOperator->getDeclContext())
+                 ->getTemplateDepth() ==
+             PrimaryTypeAliasDecl->getTemplateDepth() &&
+         getLambdaAwareParentOfDeclContext(
+             const_cast<FunctionDecl *>(PrimaryLambdaCallOperator)) ==
+             PrimaryTypeAliasDecl->getDeclContext();
+}
+
 // Add template arguments from a variable template instantiation.
 Response
 HandleVarTemplateSpec(const VarTemplateSpecializationDecl *VarTemplSpec,
@@ -176,7 +251,7 @@ HandleClassTemplateSpec(const ClassTemplateSpecializationDecl *ClassTemplSpec,
   return Response::UseNextDecl(ClassTemplSpec);
 }
 
-Response HandleFunction(const FunctionDecl *Function,
+Response HandleFunction(Sema &SemaRef, const FunctionDecl *Function,
                         MultiLevelTemplateArgumentList &Result,
                         const FunctionDecl *Pattern, bool RelativeToPrimary,
                         bool ForConstraintInstantiation) {
@@ -207,8 +282,23 @@ Response HandleFunction(const FunctionDecl *Function,
 
     // If this function is a generic lambda specialization, we are done.
     if (!ForConstraintInstantiation &&
-        isGenericLambdaCallOperatorOrStaticInvokerSpecialization(Function))
+        isGenericLambdaCallOperatorOrStaticInvokerSpecialization(Function)) {
+      // TypeAliasTemplateDecls should be taken into account, e.g.
+      // when we're deducing the return type of a lambda.
+      //
+      // template <class> int Value = 0;
+      // template <class T>
+      // using T = decltype([]<int U = 0>() { return Value<T>; }());
+      //
+      if (auto TypeAlias = getEnclosingTypeAliasTemplateDecl(SemaRef)) {
+        if (isLambdaEnclosedByTypeAliasDecl(
+                /*PrimaryLambdaCallOperator=*/getPrimaryTemplateOfGenericLambda(
+                    Function),
+                /*PrimaryTypeAliasDecl=*/TypeAlias.PrimaryTypeAliasDecl))
+          return Response::UseNextDecl(Function);
+      }
       return Response::Done();
+    }
 
   } else if (Function->getDescribedFunctionTemplate()) {
     assert(
@@ -283,7 +373,7 @@ Response HandleFunctionTemplateDecl(const FunctionTemplateDecl *FTD,
   return Response::ChangeDecl(FTD->getLexicalDeclContext());
 }
 
-Response HandleRecordDecl(const CXXRecordDecl *Rec,
+Response HandleRecordDecl(Sema &SemaRef, const CXXRecordDecl *Rec,
                           MultiLevelTemplateArgumentList &Result,
                           ASTContext &Context,
                           bool ForConstraintInstantiation) {
@@ -312,11 +402,39 @@ Response HandleRecordDecl(const CXXRecordDecl *Rec,
     return Response::ChangeDecl(Rec->getLexicalDeclContext());
   }
 
-  // This is to make sure we pick up the VarTemplateSpecializationDecl that this
-  // lambda is defined inside of.
-  if (Rec->isLambda())
+  // This is to make sure we pick up the VarTemplateSpecializationDecl or the
+  // TypeAliasTemplateDecl that this lambda is defined inside of.
+  if (Rec->isLambda()) {
     if (const Decl *LCD = Rec->getLambdaContextDecl())
       return Response::ChangeDecl(LCD);
+    // Retrieve the template arguments for a using alias declaration.
+    // This is necessary for constraint checking, since we always keep
+    // constraints relative to the primary template.
+    if (auto TypeAlias = getEnclosingTypeAliasTemplateDecl(SemaRef)) {
+      const FunctionDecl *PrimaryLambdaCallOperator =
+          getPrimaryTemplateOfGenericLambda(Rec->getLambdaCallOperator());
+      if (isLambdaEnclosedByTypeAliasDecl(PrimaryLambdaCallOperator,
+                                          TypeAlias.PrimaryTypeAliasDecl)) {
+        Result.addOuterTemplateArguments(TypeAlias.Template,
+                                         TypeAlias.AssociatedTemplateArguments,
+                                         /*Final=*/false);
+        // Visit the parent of the current type alias declaration rather than
+        // the lambda thereof.
+        // E.g., in the following example:
+        // struct S {
+        //  template <class> using T = decltype([]<Concept> {} ());
+        // };
+        // void foo() {
+        //   S::T var;
+        // }
+        // The instantiated lambda expression (which we're visiting at 'var')
+        // has a function DeclContext 'foo' rather than the Record DeclContext
+        // S. This seems to be an oversight to me that we may want to set a
+        // Sema Context from the CXXScopeSpec before substituting into T.
+        return Response::ChangeDecl(TypeAlias.Template->getDeclContext());
+      }
+    }
+  }
 
   return Response::UseNextDecl(Rec);
 }
@@ -410,10 +528,11 @@ MultiLevelTemplateArgumentList Sema::getTemplateInstantiationArgs(
       R = HandleClassTemplateSpec(ClassTemplSpec, Result,
                                   SkipForSpecialization);
     } else if (const auto *Function = dyn_cast<FunctionDecl>(CurDecl)) {
-      R = HandleFunction(Function, Result, Pattern, RelativeToPrimary,
+      R = HandleFunction(*this, Function, Result, Pattern, RelativeToPrimary,
                          ForConstraintInstantiation);
     } else if (const auto *Rec = dyn_cast<CXXRecordDecl>(CurDecl)) {
-      R = HandleRecordDecl(Rec, Result, Context, ForConstraintInstantiation);
+      R = HandleRecordDecl(*this, Rec, Result, Context,
+                           ForConstraintInstantiation);
     } else if (const auto *CSD =
                    dyn_cast<ImplicitConceptSpecializationDecl>(CurDecl)) {
       R = HandleImplicitConceptSpecializationDecl(CSD, Result);
@@ -470,6 +589,7 @@ bool Sema::CodeSynthesisContext::isInstantiationRecord() const {
   case BuildingBuiltinDumpStructCall:
   case LambdaExpressionSubstitution:
   case BuildingDeductionGuides:
+  case TypeAliasTemplateInstantiation:
     return false;
 
   // This function should never be called when Kind's value is Memoization.
@@ -616,6 +736,15 @@ Sema::InstantiatingTemplate::InstantiatingTemplate(
           TemplateArgs) {}
 
 Sema::InstantiatingTemplate::InstantiatingTemplate(
+    Sema &SemaRef, SourceLocation PointOfInstantiation,
+    TypeAliasTemplateDecl *Template, ArrayRef<TemplateArgument> TemplateArgs,
+    SourceRange InstantiationRange)
+    : InstantiatingTemplate(
+          SemaRef, Sema::CodeSynthesisContext::TypeAliasTemplateInstantiation,
+          PointOfInstantiation, InstantiationRange, /*Entity=*/Template,
+          /*Template=*/nullptr, TemplateArgs) {}
+
+Sema::InstantiatingTemplate::InstantiatingTemplate(
     Sema &SemaRef, SourceLocation PointOfInstantiation, TemplateDecl *Template,
     NamedDecl *Param, ArrayRef<TemplateArgument> TemplateArgs,
     SourceRange InstantiationRange)
@@ -1132,6 +1261,8 @@ void Sema::PrintInstantiationStack() {
       Diags.Report(Active->PointOfInstantiation,
                    diag::note_building_deduction_guide_here);
       break;
+    case CodeSynthesisContext::TypeAliasTemplateInstantiation:
+      break;
     }
   }
 }
@@ -1209,6 +1340,7 @@ std::optional<TemplateDeductionInfo *> Sema::isSFINAEContext() const {
       break;
 
     case CodeSynthesisContext::Memoization:
+    case CodeSynthesisContext::TypeAliasTemplateInstantiation:
       break;
     }
 
@@ -1534,6 +1666,18 @@ namespace {
                                            SubstTemplateTypeParmPackTypeLoc TL,
                                            bool SuppressObjCLifetime);
 
+    CXXRecordDecl::LambdaDependencyKind
+    ComputeLambdaDependency(LambdaScopeInfo *LSI) {
+      auto &CCS = SemaRef.CodeSynthesisContexts.back();
+      if (CCS.Kind ==
+          Sema::CodeSynthesisContext::TypeAliasTemplateInstantiation) {
+        unsigned TypeAliasDeclDepth = CCS.Entity->getTemplateDepth();
+        if (TypeAliasDeclDepth >= TemplateArgs.getNumSubstitutedLevels())
+          return CXXRecordDecl::LambdaDependencyKind::LDK_AlwaysDependent;
+      }
+      return inherited::ComputeLambdaDependency(LSI);
+    }
+
     ExprResult TransformLambdaExpr(LambdaExpr *E) {
       LocalInstantiationScope Scope(SemaRef, /*CombineWithOuterScope=*/true);
       Sema::ConstraintEvalRAII<TemplateInstantiator> RAII(*this);
diff --git a/clang/lib/Sema/SemaTemplateInstantiateDecl.cpp b/clang/lib/Sema/SemaTemplateInstantiateDecl.cpp
index dc972018e7b2..1cb071e4eb7d 100644
--- a/clang/lib/Sema/SemaTemplateInstantiateDecl.cpp
+++ b/clang/lib/Sema/SemaTemplateInstantiateDecl.cpp
@@ -1112,6 +1112,13 @@ TemplateDeclInstantiator::VisitTypeAliasTemplateDecl(TypeAliasTemplateDecl *D) {
     return nullptr;
 
   TypeAliasDecl *Pattern = D->getTemplatedDecl();
+  Sema::InstantiatingTemplate InstTemplate(
+      SemaRef, D->getBeginLoc(), D,
+      D->getTemplateDepth() >= TemplateArgs.getNumLevels()
+          ? ArrayRef<TemplateArgument>()
+          : (TemplateArgs.begin() + TemplateArgs.getNumLevels() - 1 -
+             D->getTemplateDepth())
+                ->Args);
 
   TypeAliasTemplateDecl *PrevAliasTemplate = nullptr;
   if (getPreviousDeclForInstantiation<TypedefNameDecl>(Pattern)) {
diff --git a/clang/lib/Sema/TreeTransform.h b/clang/lib/Sema/TreeTransform.h
index a2568ad0f82c..7df352c24e86 100644
--- a/clang/lib/Sema/TreeTransform.h
+++ b/clang/lib/Sema/TreeTransform.h
@@ -768,6 +768,12 @@ public:
   /// the body.
   StmtResult SkipLambdaBody(LambdaExpr *E, Stmt *Body);
 
+  CXXRecordDecl::LambdaDependencyKind
+  ComputeLambdaDependency(LambdaScopeInfo *LSI) {
+    return static_cast<CXXRecordDecl::LambdaDependencyKind>(
+        LSI->Lambda->getLambdaDependencyKind());
+  }
+
   QualType TransformReferenceType(TypeLocBuilder &TLB, ReferenceTypeLoc TL);
 
   StmtResult TransformCompoundStmt(CompoundStmt *S, bool IsStmtExpr);
@@ -13992,6 +13998,46 @@ TreeTransform<Derived>::TransformLambdaExpr(LambdaExpr *E) {
                                     /*IsInstantiation*/ true);
   SavedContext.pop();
 
+  // Recompute the dependency of the lambda so that we can defer the lambda call
+  // construction until after we have all the necessary template arguments. For
+  // example, given
+  //
+  // template <class> struct S {
+  //   template <class U>
+  //   using Type = decltype([](U){}(42.0));
+  // };
+  // void foo() {
+  //   using T = S<int>::Type<float>;
+  //             ^~~~~~
+  // }
+  //
+  // We would end up here from instantiating S<int> when ensuring its
+  // completeness. That would transform the lambda call expression regardless of
+  // the absence of the corresponding argument for U.
+  //
+  // Going ahead with unsubstituted type U makes things worse: we would soon
+  // compare the argument type (which is float) against the parameter U
+  // somewhere in Sema::BuildCallExpr. Then we would quickly run into a bogus
+  // error suggesting unmatched types 'U' and 'float'!
+  //
+  // That said, everything will be fine if we defer that semantic checking.
+  // Fortunately, we have such a mechanism that bypasses it if the CallExpr is
+  // dependent. Since the CallExpr's dependency boils down to the lambda's
+  // dependency in this case, we can harness that by recomputing the dependency
+  // from the instantiation arguments.
+  //
+  // FIXME: Creating the type of a lambda requires us to have a dependency
+  // value, which happens before its substitution. We update its dependency
+  // *after* the substitution in case we can't decide the dependency
+  // so early, e.g. because we want to see if any of the *substituted*
+  // parameters are dependent.
+  DependencyKind = getDerived().ComputeLambdaDependency(&LSICopy);
+  Class->setLambdaDependencyKind(DependencyKind);
+  // Clean up the type cache created previously. Then, we re-create a type for
+  // such Decl with the new DependencyKind.
+  Class->setTypeForDecl(nullptr);
+  getSema().Context.getTypeDeclType(Class);
+
   return getSema().BuildLambdaExpr(E->getBeginLoc(), Body.get()->getEndLoc(),
                                    &LSICopy);
 }
diff --git a/clang/lib/StaticAnalyzer/Checkers/CStringChecker.cpp b/clang/lib/StaticAnalyzer/Checkers/CStringChecker.cpp
index 59be236ca1c7..63844563de44 100644
--- a/clang/lib/StaticAnalyzer/Checkers/CStringChecker.cpp
+++ b/clang/lib/StaticAnalyzer/Checkers/CStringChecker.cpp
@@ -124,34 +124,45 @@ public:
                                      const CallEvent &)>;
 
   CallDescriptionMap<FnCheck> Callbacks = {
-      {{CDM::CLibrary, {"memcpy"}, 3},
+      {{CDM::CLibraryMaybeHardened, {"memcpy"}, 3},
        std::bind(&CStringChecker::evalMemcpy, _1, _2, _3, CK_Regular)},
-      {{CDM::CLibrary, {"wmemcpy"}, 3},
+      {{CDM::CLibraryMaybeHardened, {"wmemcpy"}, 3},
        std::bind(&CStringChecker::evalMemcpy, _1, _2, _3, CK_Wide)},
-      {{CDM::CLibrary, {"mempcpy"}, 3},
+      {{CDM::CLibraryMaybeHardened, {"mempcpy"}, 3},
        std::bind(&CStringChecker::evalMempcpy, _1, _2, _3, CK_Regular)},
-      {{CDM::Unspecified, {"wmempcpy"}, 3},
+      {{CDM::CLibraryMaybeHardened, {"wmempcpy"}, 3},
        std::bind(&CStringChecker::evalMempcpy, _1, _2, _3, CK_Wide)},
       {{CDM::CLibrary, {"memcmp"}, 3},
        std::bind(&CStringChecker::evalMemcmp, _1, _2, _3, CK_Regular)},
       {{CDM::CLibrary, {"wmemcmp"}, 3},
        std::bind(&CStringChecker::evalMemcmp, _1, _2, _3, CK_Wide)},
-      {{CDM::CLibrary, {"memmove"}, 3},
+      {{CDM::CLibraryMaybeHardened, {"memmove"}, 3},
        std::bind(&CStringChecker::evalMemmove, _1, _2, _3, CK_Regular)},
-      {{CDM::CLibrary, {"wmemmove"}, 3},
+      {{CDM::CLibraryMaybeHardened, {"wmemmove"}, 3},
        std::bind(&CStringChecker::evalMemmove, _1, _2, _3, CK_Wide)},
-      {{CDM::CLibrary, {"memset"}, 3}, &CStringChecker::evalMemset},
+      {{CDM::CLibraryMaybeHardened, {"memset"}, 3},
+       &CStringChecker::evalMemset},
       {{CDM::CLibrary, {"explicit_memset"}, 3}, &CStringChecker::evalMemset},
-      {{CDM::CLibrary, {"strcpy"}, 2}, &CStringChecker::evalStrcpy},
-      {{CDM::CLibrary, {"strncpy"}, 3}, &CStringChecker::evalStrncpy},
-      {{CDM::CLibrary, {"stpcpy"}, 2}, &CStringChecker::evalStpcpy},
-      {{CDM::CLibrary, {"strlcpy"}, 3}, &CStringChecker::evalStrlcpy},
-      {{CDM::CLibrary, {"strcat"}, 2}, &CStringChecker::evalStrcat},
-      {{CDM::CLibrary, {"strncat"}, 3}, &CStringChecker::evalStrncat},
-      {{CDM::CLibrary, {"strlcat"}, 3}, &CStringChecker::evalStrlcat},
-      {{CDM::CLibrary, {"strlen"}, 1}, &CStringChecker::evalstrLength},
+      // FIXME: C23 introduces 'memset_explicit', maybe also model that
+      {{CDM::CLibraryMaybeHardened, {"strcpy"}, 2},
+       &CStringChecker::evalStrcpy},
+      {{CDM::CLibraryMaybeHardened, {"strncpy"}, 3},
+       &CStringChecker::evalStrncpy},
+      {{CDM::CLibraryMaybeHardened, {"stpcpy"}, 2},
+       &CStringChecker::evalStpcpy},
+      {{CDM::CLibraryMaybeHardened, {"strlcpy"}, 3},
+       &CStringChecker::evalStrlcpy},
+      {{CDM::CLibraryMaybeHardened, {"strcat"}, 2},
+       &CStringChecker::evalStrcat},
+      {{CDM::CLibraryMaybeHardened, {"strncat"}, 3},
+       &CStringChecker::evalStrncat},
+      {{CDM::CLibraryMaybeHardened, {"strlcat"}, 3},
+       &CStringChecker::evalStrlcat},
+      {{CDM::CLibraryMaybeHardened, {"strlen"}, 1},
+       &CStringChecker::evalstrLength},
       {{CDM::CLibrary, {"wcslen"}, 1}, &CStringChecker::evalstrLength},
-      {{CDM::CLibrary, {"strnlen"}, 2}, &CStringChecker::evalstrnLength},
+      {{CDM::CLibraryMaybeHardened, {"strnlen"}, 2},
+       &CStringChecker::evalstrnLength},
       {{CDM::CLibrary, {"wcsnlen"}, 2}, &CStringChecker::evalstrnLength},
       {{CDM::CLibrary, {"strcmp"}, 2}, &CStringChecker::evalStrcmp},
       {{CDM::CLibrary, {"strncmp"}, 3}, &CStringChecker::evalStrncmp},
@@ -162,9 +173,19 @@ public:
       {{CDM::CLibrary, {"bcmp"}, 3},
        std::bind(&CStringChecker::evalMemcmp, _1, _2, _3, CK_Regular)},
       {{CDM::CLibrary, {"bzero"}, 2}, &CStringChecker::evalBzero},
-      {{CDM::CLibrary, {"explicit_bzero"}, 2}, &CStringChecker::evalBzero},
-      {{CDM::CLibrary, {"sprintf"}, 2}, &CStringChecker::evalSprintf},
-      {{CDM::CLibrary, {"snprintf"}, 2}, &CStringChecker::evalSnprintf},
+      {{CDM::CLibraryMaybeHardened, {"explicit_bzero"}, 2},
+       &CStringChecker::evalBzero},
+
+      // When recognizing calls to the following variadic functions, we accept
+      // any number of arguments in the call (std::nullopt = accept any
+      // number), but check that in the declaration there are 2 and 3
+      // parameters respectively. (Note that the parameter count does not
+      // include the "...". Calls where the number of arguments is too small
+      // will be discarded by the callback.)
+      {{CDM::CLibraryMaybeHardened, {"sprintf"}, std::nullopt, 2},
+       &CStringChecker::evalSprintf},
+      {{CDM::CLibraryMaybeHardened, {"snprintf"}, std::nullopt, 3},
+       &CStringChecker::evalSnprintf},
   };
 
   // These require a bit of special handling.
@@ -218,7 +239,7 @@ public:
   void evalSprintf(CheckerContext &C, const CallEvent &Call) const;
   void evalSnprintf(CheckerContext &C, const CallEvent &Call) const;
   void evalSprintfCommon(CheckerContext &C, const CallEvent &Call,
-                         bool IsBounded, bool IsBuiltin) const;
+                         bool IsBounded) const;
 
   // Utility methods
   std::pair<ProgramStateRef , ProgramStateRef >
@@ -2467,27 +2488,26 @@ void CStringChecker::evalBzero(CheckerContext &C, const CallEvent &Call) const {
 void CStringChecker::evalSprintf(CheckerContext &C,
                                  const CallEvent &Call) const {
   CurrentFunctionDescription = "'sprintf'";
-  const auto *CE = cast<CallExpr>(Call.getOriginExpr());
-  bool IsBI = CE->getBuiltinCallee() == Builtin::BI__builtin___sprintf_chk;
-  evalSprintfCommon(C, Call, /* IsBounded */ false, IsBI);
+  evalSprintfCommon(C, Call, /* IsBounded = */ false);
 }
 
 void CStringChecker::evalSnprintf(CheckerContext &C,
                                   const CallEvent &Call) const {
   CurrentFunctionDescription = "'snprintf'";
-  const auto *CE = cast<CallExpr>(Call.getOriginExpr());
-  bool IsBI = CE->getBuiltinCallee() == Builtin::BI__builtin___snprintf_chk;
-  evalSprintfCommon(C, Call, /* IsBounded */ true, IsBI);
+  evalSprintfCommon(C, Call, /* IsBounded = */ true);
 }
 
 void CStringChecker::evalSprintfCommon(CheckerContext &C, const CallEvent &Call,
-                                       bool IsBounded, bool IsBuiltin) const {
+                                       bool IsBounded) const {
   ProgramStateRef State = C.getState();
   const auto *CE = cast<CallExpr>(Call.getOriginExpr());
   DestinationArgExpr Dest = {{Call.getArgExpr(0), 0}};
 
   const auto NumParams = Call.parameters().size();
-  assert(CE->getNumArgs() >= NumParams);
+  if (CE->getNumArgs() < NumParams) {
+    // This is an invalid call, let's just ignore it.
+    return;
+  }
 
   const auto AllArguments =
       llvm::make_range(CE->getArgs(), CE->getArgs() + CE->getNumArgs());
diff --git a/clang/lib/StaticAnalyzer/Checkers/GenericTaintChecker.cpp b/clang/lib/StaticAnalyzer/Checkers/GenericTaintChecker.cpp
index 4c48fdf498f7..89054512d65a 100644
--- a/clang/lib/StaticAnalyzer/Checkers/GenericTaintChecker.cpp
+++ b/clang/lib/StaticAnalyzer/Checkers/GenericTaintChecker.cpp
@@ -718,20 +718,23 @@ void GenericTaintChecker::initTaintRules(CheckerContext &C) const {
       {{{"isupper"}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
       {{{"isxdigit"}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
 
-      {{CDM::CLibrary, {BI.getName(Builtin::BIstrncat)}},
+      {{CDM::CLibraryMaybeHardened, {BI.getName(Builtin::BIstrncat)}},
        TR::Prop({{1, 2}}, {{0, ReturnValueIndex}})},
-      {{CDM::CLibrary, {BI.getName(Builtin::BIstrlcpy)}},
+      {{CDM::CLibraryMaybeHardened, {BI.getName(Builtin::BIstrlcpy)}},
        TR::Prop({{1, 2}}, {{0}})},
-      {{CDM::CLibrary, {BI.getName(Builtin::BIstrlcat)}},
+      {{CDM::CLibraryMaybeHardened, {BI.getName(Builtin::BIstrlcat)}},
        TR::Prop({{1, 2}}, {{0}})},
-      {{CDM::CLibrary, {{"snprintf"}}},
+      {{CDM::CLibraryMaybeHardened, {{"snprintf"}}},
        TR::Prop({{1}, 3}, {{0, ReturnValueIndex}})},
-      {{CDM::CLibrary, {{"sprintf"}}},
+      {{CDM::CLibraryMaybeHardened, {{"sprintf"}}},
        TR::Prop({{1}, 2}, {{0, ReturnValueIndex}})},
-      {{CDM::CLibrary, {{"strcpy"}}}, TR::Prop({{1}}, {{0, ReturnValueIndex}})},
-      {{CDM::CLibrary, {{"stpcpy"}}}, TR::Prop({{1}}, {{0, ReturnValueIndex}})},
-      {{CDM::CLibrary, {{"strcat"}}}, TR::Prop({{1}}, {{0, ReturnValueIndex}})},
-      {{CDM::CLibrary, {{"wcsncat"}}},
+      {{CDM::CLibraryMaybeHardened, {{"strcpy"}}},
+       TR::Prop({{1}}, {{0, ReturnValueIndex}})},
+      {{CDM::CLibraryMaybeHardened, {{"stpcpy"}}},
+       TR::Prop({{1}}, {{0, ReturnValueIndex}})},
+      {{CDM::CLibraryMaybeHardened, {{"strcat"}}},
+       TR::Prop({{1}}, {{0, ReturnValueIndex}})},
+      {{CDM::CLibraryMaybeHardened, {{"wcsncat"}}},
        TR::Prop({{1}}, {{0, ReturnValueIndex}})},
       {{CDM::CLibrary, {{"strdup"}}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
       {{CDM::CLibrary, {{"strdupa"}}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
@@ -759,13 +762,13 @@ void GenericTaintChecker::initTaintRules(CheckerContext &C) const {
        TR::Sink({{0}, 1}, MsgUncontrolledFormatString)},
 
       // SinkProps
-      {{CDM::CLibrary, BI.getName(Builtin::BImemcpy)},
+      {{CDM::CLibraryMaybeHardened, BI.getName(Builtin::BImemcpy)},
        TR::SinkProp({{2}}, {{1, 2}}, {{0, ReturnValueIndex}},
                     MsgTaintedBufferSize)},
-      {{CDM::CLibrary, {BI.getName(Builtin::BImemmove)}},
+      {{CDM::CLibraryMaybeHardened, {BI.getName(Builtin::BImemmove)}},
        TR::SinkProp({{2}}, {{1, 2}}, {{0, ReturnValueIndex}},
                     MsgTaintedBufferSize)},
-      {{CDM::CLibrary, {BI.getName(Builtin::BIstrncpy)}},
+      {{CDM::CLibraryMaybeHardened, {BI.getName(Builtin::BIstrncpy)}},
        TR::SinkProp({{2}}, {{1, 2}}, {{0, ReturnValueIndex}},
                     MsgTaintedBufferSize)},
       {{CDM::CLibrary, {BI.getName(Builtin::BIstrndup)}},
diff --git a/clang/lib/StaticAnalyzer/Checkers/RetainCountChecker/RetainCountDiagnostics.cpp b/clang/lib/StaticAnalyzer/Checkers/RetainCountChecker/RetainCountDiagnostics.cpp
index c3acb73ba717..086c3e5e49b7 100644
--- a/clang/lib/StaticAnalyzer/Checkers/RetainCountChecker/RetainCountDiagnostics.cpp
+++ b/clang/lib/StaticAnalyzer/Checkers/RetainCountChecker/RetainCountDiagnostics.cpp
@@ -977,7 +977,7 @@ void RefLeakReport::findBindingToReport(CheckerContext &Ctx,
     //       something like derived regions if we want to construct SVal from
     //       Sym. Instead, we take the value that is definitely stored in that
     //       region, thus guaranteeing that trackStoredValue will work.
-    bugreporter::trackStoredValue(AllVarBindings[0].second.castAs<KnownSVal>(),
+    bugreporter::trackStoredValue(AllVarBindings[0].second,
                                   AllocBindingToReport, *this);
   } else {
     AllocBindingToReport = AllocFirstBinding;
diff --git a/clang/lib/StaticAnalyzer/Core/BugReporterVisitors.cpp b/clang/lib/StaticAnalyzer/Core/BugReporterVisitors.cpp
index a0822513a6d0..984755fa7e50 100644
--- a/clang/lib/StaticAnalyzer/Core/BugReporterVisitors.cpp
+++ b/clang/lib/StaticAnalyzer/Core/BugReporterVisitors.cpp
@@ -1238,7 +1238,7 @@ public:
   ///        changes to its value in a nested stackframe could be pruned, and
   ///        this visitor can prevent that without polluting the bugpath too
   ///        much.
-  StoreSiteFinder(bugreporter::TrackerRef ParentTracker, KnownSVal V,
+  StoreSiteFinder(bugreporter::TrackerRef ParentTracker, SVal V,
                   const MemRegion *R, TrackingOptions Options,
                   const StackFrameContext *OriginSFC = nullptr)
       : TrackingBugReporterVisitor(ParentTracker), R(R), V(V), Options(Options),
@@ -2539,9 +2539,9 @@ public:
         Report.addVisitor<UndefOrNullArgVisitor>(L->getRegion());
         Result.FoundSomethingToTrack = true;
 
-        if (auto KV = RVal.getAs<KnownSVal>())
+        if (!RVal.isUnknown())
           Result.combineWith(
-              getParentTracker().track(*KV, L->getRegion(), Opts, SFC));
+              getParentTracker().track(RVal, L->getRegion(), Opts, SFC));
       }
 
       const MemRegion *RegionRVal = RVal.getAsRegion();
@@ -2663,8 +2663,8 @@ Tracker::Result Tracker::track(const Expr *E, const ExplodedNode *N,
 
 Tracker::Result Tracker::track(SVal V, const MemRegion *R, TrackingOptions Opts,
                                const StackFrameContext *Origin) {
-  if (auto KV = V.getAs<KnownSVal>()) {
-    Report.addVisitor<StoreSiteFinder>(this, *KV, R, Opts, Origin);
+  if (!V.isUnknown()) {
+    Report.addVisitor<StoreSiteFinder>(this, V, R, Opts, Origin);
     return {true};
   }
   return {};
@@ -2692,7 +2692,7 @@ bool bugreporter::trackExpressionValue(const ExplodedNode *InputNode,
       .FoundSomethingToTrack;
 }
 
-void bugreporter::trackStoredValue(KnownSVal V, const MemRegion *R,
+void bugreporter::trackStoredValue(SVal V, const MemRegion *R,
                                    PathSensitiveBugReport &Report,
                                    TrackingOptions Opts,
                                    const StackFrameContext *Origin) {
diff --git a/clang/lib/StaticAnalyzer/Core/CallDescription.cpp b/clang/lib/StaticAnalyzer/Core/CallDescription.cpp
index 459e854cd44d..0bb0fe66e54f 100644
--- a/clang/lib/StaticAnalyzer/Core/CallDescription.cpp
+++ b/clang/lib/StaticAnalyzer/Core/CallDescription.cpp
@@ -75,6 +75,48 @@ bool ento::CallDescription::matchesAsWritten(const CallExpr &CE) const {
   return matchesImpl(FD, CE.getNumArgs(), FD->param_size());
 }
 
+bool ento::CallDescription::matchNameOnly(const NamedDecl *ND) const {
+  DeclarationName Name = ND->getDeclName();
+  if (const auto *NameII = Name.getAsIdentifierInfo()) {
+    if (!II)
+      II = &ND->getASTContext().Idents.get(getFunctionName());
+
+    return NameII == *II; // Fast case.
+  }
+
+  // Fallback to the slow stringification and comparison for:
+  // C++ overloaded operators, constructors, destructors, etc.
+  // FIXME This comparison is way SLOWER than comparing pointers.
+  // At some point in the future, we should compare FunctionDecl pointers.
+  return Name.getAsString() == getFunctionName();
+}
+
+bool ento::CallDescription::matchQualifiedNameParts(const Decl *D) const {
+  const auto FindNextNamespaceOrRecord =
+      [](const DeclContext *Ctx) -> const DeclContext * {
+    while (Ctx && !isa<NamespaceDecl, RecordDecl>(Ctx))
+      Ctx = Ctx->getParent();
+    return Ctx;
+  };
+
+  auto QualifierPartsIt = begin_qualified_name_parts();
+  const auto QualifierPartsEndIt = end_qualified_name_parts();
+
+  // Match namespace and record names. Skip unrelated names if they don't
+  // match.
+  const DeclContext *Ctx = FindNextNamespaceOrRecord(D->getDeclContext());
+  for (; Ctx && QualifierPartsIt != QualifierPartsEndIt;
+       Ctx = FindNextNamespaceOrRecord(Ctx->getParent())) {
+    // If not matched just continue and try matching for the next one.
+    if (cast<NamedDecl>(Ctx)->getName() != *QualifierPartsIt)
+      continue;
+    ++QualifierPartsIt;
+  }
+
+  // We matched if we consumed all expected qualifier segments.
+  return QualifierPartsIt == QualifierPartsEndIt;
+}
+
 bool ento::CallDescription::matchesImpl(const FunctionDecl *FD, size_t ArgCount,
                                         size_t ParamCount) const {
   if (!FD)
@@ -88,76 +130,34 @@ bool ento::CallDescription::matchesImpl(const FunctionDecl *FD, size_t ArgCount,
   if (MatchAs == Mode::CXXMethod && !isMethod)
     return false;
 
-  if (MatchAs == Mode::CLibrary) {
-    return CheckerContext::isCLibraryFunction(FD, getFunctionName()) &&
-           (!RequiredArgs || *RequiredArgs <= ArgCount) &&
-           (!RequiredParams || *RequiredParams <= ParamCount);
-  }
-
-  if (!II) {
-    II = &FD->getASTContext().Idents.get(getFunctionName());
-  }
-
-  const auto MatchNameOnly = [](const CallDescription &CD,
-                                const NamedDecl *ND) -> bool {
-    DeclarationName Name = ND->getDeclName();
-    if (const auto *II = Name.getAsIdentifierInfo())
-      return II == *CD.II; // Fast case.
-
-    // Fallback to the slow stringification and comparison for:
-    // C++ overloaded operators, constructors, destructors, etc.
-    // FIXME This comparison is way SLOWER than comparing pointers.
-    // At some point in the future, we should compare FunctionDecl pointers.
-    return Name.getAsString() == CD.getFunctionName();
-  };
-
-  const auto ExactMatchArgAndParamCounts =
-      [](size_t ArgCount, size_t ParamCount,
-         const CallDescription &CD) -> bool {
-    const bool ArgsMatch = !CD.RequiredArgs || *CD.RequiredArgs == ArgCount;
-    const bool ParamsMatch =
-        !CD.RequiredParams || *CD.RequiredParams == ParamCount;
-    return ArgsMatch && ParamsMatch;
-  };
-
-  const auto MatchQualifiedNameParts = [](const CallDescription &CD,
-                                          const Decl *D) -> bool {
-    const auto FindNextNamespaceOrRecord =
-        [](const DeclContext *Ctx) -> const DeclContext * {
-      while (Ctx && !isa<NamespaceDecl, RecordDecl>(Ctx))
-        Ctx = Ctx->getParent();
-      return Ctx;
-    };
-
-    auto QualifierPartsIt = CD.begin_qualified_name_parts();
-    const auto QualifierPartsEndIt = CD.end_qualified_name_parts();
-
-    // Match namespace and record names. Skip unrelated names if they don't
-    // match.
-    const DeclContext *Ctx = FindNextNamespaceOrRecord(D->getDeclContext());
-    for (; Ctx && QualifierPartsIt != QualifierPartsEndIt;
-         Ctx = FindNextNamespaceOrRecord(Ctx->getParent())) {
-      // If not matched just continue and try matching for the next one.
-      if (cast<NamedDecl>(Ctx)->getName() != *QualifierPartsIt)
-        continue;
-      ++QualifierPartsIt;
+  if (MatchAs == Mode::CLibraryMaybeHardened) {
+    // In addition to accepting FOO() with CLibrary rules, we also want to
+    // accept calls to __FOO_chk() and __builtin___FOO_chk().
+    if (CheckerContext::isCLibraryFunction(FD) &&
+        CheckerContext::isHardenedVariantOf(FD, getFunctionName())) {
+      // Check that the actual argument/parameter counts are greater or equal
+      // to the required counts. (Setting a requirement to std::nullopt matches
+      // anything, so in that case value_or ensures that the value is compared
+      // with itself.)
+      return (RequiredArgs.value_or(ArgCount) <= ArgCount &&
+              RequiredParams.value_or(ParamCount) <= ParamCount);
     }
+  }
 
-    // We matched if we consumed all expected qualifier segments.
-    return QualifierPartsIt == QualifierPartsEndIt;
-  };
-
-  // Let's start matching...
-  if (!ExactMatchArgAndParamCounts(ArgCount, ParamCount, *this))
+  if (RequiredArgs.value_or(ArgCount) != ArgCount ||
+      RequiredParams.value_or(ParamCount) != ParamCount)
     return false;
 
-  if (!MatchNameOnly(*this, FD))
+  if (MatchAs == Mode::CLibrary || MatchAs == Mode::CLibraryMaybeHardened)
+    return CheckerContext::isCLibraryFunction(FD, getFunctionName());
+
+  if (!matchNameOnly(FD))
     return false;
 
   if (!hasQualifiedNameParts())
     return true;
 
-  return MatchQualifiedNameParts(*this, FD);
+  return matchQualifiedNameParts(FD);
 }
 
 ento::CallDescriptionSet::CallDescriptionSet(
diff --git a/clang/lib/StaticAnalyzer/Core/CheckerContext.cpp b/clang/lib/StaticAnalyzer/Core/CheckerContext.cpp
index 1a9bff529e9b..113abcd4c2ab 100644
--- a/clang/lib/StaticAnalyzer/Core/CheckerContext.cpp
+++ b/clang/lib/StaticAnalyzer/Core/CheckerContext.cpp
@@ -110,13 +110,24 @@ bool CheckerContext::isCLibraryFunction(const FunctionDecl *FD,
   if (FName.starts_with("__inline") && FName.contains(Name))
     return true;
 
-  if (FName.starts_with("__") && FName.ends_with("_chk") &&
-      FName.contains(Name))
-    return true;
-
   return false;
 }
 
+bool CheckerContext::isHardenedVariantOf(const FunctionDecl *FD,
+                                         StringRef Name) {
+  const IdentifierInfo *II = FD->getIdentifier();
+  if (!II)
+    return false;
+
+  auto CompletelyMatchesParts = [II](auto... Parts) -> bool {
+    StringRef FName = II->getName();
+    return (FName.consume_front(Parts) && ...) && FName.empty();
+  };
+
+  return CompletelyMatchesParts("__", Name, "_chk") ||
+         CompletelyMatchesParts("__builtin_", "__", Name, "_chk");
+}
+
 StringRef CheckerContext::getMacroNameOrSpelling(SourceLocation &Loc) {
   if (Loc.isMacroID())
     return Lexer::getImmediateMacroName(Loc, getSourceManager(),
diff --git a/clang/test/CodeGen/allow-ubsan-check.c b/clang/test/CodeGen/allow-ubsan-check.c
index bc425230c8ec..5232d2408546 100644
--- a/clang/test/CodeGen/allow-ubsan-check.c
+++ b/clang/test/CodeGen/allow-ubsan-check.c
@@ -1,7 +1,7 @@
 // NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 4
-// RUN: %clang_cc1 -triple x86_64-pc-linux-gnu -emit-llvm -o - %s -fsanitize=signed-integer-overflow,integer-divide-by-zero,null | FileCheck %s
-// RUN: %clang_cc1 -triple x86_64-pc-linux-gnu -emit-llvm -o - %s -fsanitize=signed-integer-overflow,integer-divide-by-zero,null -fsanitize-trap=signed-integer-overflow,integer-divide-by-zero,null | FileCheck %s --check-prefixes=TRAP
-// RUN: %clang_cc1 -triple x86_64-pc-linux-gnu -emit-llvm -o - %s -fsanitize=signed-integer-overflow,integer-divide-by-zero,null -fsanitize-recover=signed-integer-overflow,integer-divide-by-zero,null | FileCheck %s --check-prefixes=RECOVER
+// RUN: %clang_cc1 -triple x86_64-pc-linux-gnu -emit-llvm -o - %s -fsanitize=signed-integer-overflow,integer-divide-by-zero,null -mllvm -ubsan-guard-checks | FileCheck %s
+// RUN: %clang_cc1 -triple x86_64-pc-linux-gnu -emit-llvm -o - %s -fsanitize=signed-integer-overflow,integer-divide-by-zero,null -mllvm -ubsan-guard-checks -fsanitize-trap=signed-integer-overflow,integer-divide-by-zero,null | FileCheck %s --check-prefixes=TRAP
+// RUN: %clang_cc1 -triple x86_64-pc-linux-gnu -emit-llvm -o - %s -fsanitize=signed-integer-overflow,integer-divide-by-zero,null -mllvm -ubsan-guard-checks -fsanitize-recover=signed-integer-overflow,integer-divide-by-zero,null | FileCheck %s --check-prefixes=RECOVER
 
 
 // CHECK-LABEL: define dso_local i32 @div(
@@ -18,11 +18,14 @@
 // CHECK-NEXT:    [[TMP4:%.*]] = icmp ne i32 [[TMP1]], -1, !nosanitize [[META2]]
 // CHECK-NEXT:    [[OR:%.*]] = or i1 [[TMP3]], [[TMP4]], !nosanitize [[META2]]
 // CHECK-NEXT:    [[TMP5:%.*]] = and i1 [[TMP2]], [[OR]], !nosanitize [[META2]]
-// CHECK-NEXT:    br i1 [[TMP5]], label [[CONT:%.*]], label [[HANDLER_DIVREM_OVERFLOW:%.*]], !prof [[PROF3:![0-9]+]], !nosanitize [[META2]]
+// CHECK-NEXT:    [[TMP6:%.*]] = call i1 @llvm.allow.ubsan.check(i8 3), !nosanitize [[META2]]
+// CHECK-NEXT:    [[TMP7:%.*]] = xor i1 [[TMP6]], true, !nosanitize [[META2]]
+// CHECK-NEXT:    [[TMP8:%.*]] = or i1 [[TMP5]], [[TMP7]], !nosanitize [[META2]]
+// CHECK-NEXT:    br i1 [[TMP8]], label [[CONT:%.*]], label [[HANDLER_DIVREM_OVERFLOW:%.*]], !prof [[PROF3:![0-9]+]], !nosanitize [[META2]]
 // CHECK:       handler.divrem_overflow:
-// CHECK-NEXT:    [[TMP6:%.*]] = zext i32 [[TMP0]] to i64, !nosanitize [[META2]]
-// CHECK-NEXT:    [[TMP7:%.*]] = zext i32 [[TMP1]] to i64, !nosanitize [[META2]]
-// CHECK-NEXT:    call void @__ubsan_handle_divrem_overflow_abort(ptr @[[GLOB1:[0-9]+]], i64 [[TMP6]], i64 [[TMP7]]) #[[ATTR3:[0-9]+]], !nosanitize [[META2]]
+// CHECK-NEXT:    [[TMP9:%.*]] = zext i32 [[TMP0]] to i64, !nosanitize [[META2]]
+// CHECK-NEXT:    [[TMP10:%.*]] = zext i32 [[TMP1]] to i64, !nosanitize [[META2]]
+// CHECK-NEXT:    call void @__ubsan_handle_divrem_overflow_abort(ptr @[[GLOB1:[0-9]+]], i64 [[TMP9]], i64 [[TMP10]]) #[[ATTR4:[0-9]+]], !nosanitize [[META2]]
 // CHECK-NEXT:    unreachable, !nosanitize [[META2]]
 // CHECK:       cont:
 // CHECK-NEXT:    [[DIV:%.*]] = sdiv i32 [[TMP0]], [[TMP1]]
@@ -42,9 +45,12 @@
 // TRAP-NEXT:    [[TMP4:%.*]] = icmp ne i32 [[TMP1]], -1, !nosanitize [[META2]]
 // TRAP-NEXT:    [[OR:%.*]] = or i1 [[TMP3]], [[TMP4]], !nosanitize [[META2]]
 // TRAP-NEXT:    [[TMP5:%.*]] = and i1 [[TMP2]], [[OR]], !nosanitize [[META2]]
-// TRAP-NEXT:    br i1 [[TMP5]], label [[CONT:%.*]], label [[TRAP:%.*]], !nosanitize [[META2]]
+// TRAP-NEXT:    [[TMP6:%.*]] = call i1 @llvm.allow.ubsan.check(i8 3), !nosanitize [[META2]]
+// TRAP-NEXT:    [[TMP7:%.*]] = xor i1 [[TMP6]], true, !nosanitize [[META2]]
+// TRAP-NEXT:    [[TMP8:%.*]] = or i1 [[TMP5]], [[TMP7]], !nosanitize [[META2]]
+// TRAP-NEXT:    br i1 [[TMP8]], label [[CONT:%.*]], label [[TRAP:%.*]], !nosanitize [[META2]]
 // TRAP:       trap:
-// TRAP-NEXT:    call void @llvm.ubsantrap(i8 3) #[[ATTR3:[0-9]+]], !nosanitize [[META2]]
+// TRAP-NEXT:    call void @llvm.ubsantrap(i8 3) #[[ATTR4:[0-9]+]], !nosanitize [[META2]]
 // TRAP-NEXT:    unreachable, !nosanitize [[META2]]
 // TRAP:       cont:
 // TRAP-NEXT:    [[DIV:%.*]] = sdiv i32 [[TMP0]], [[TMP1]]
@@ -64,11 +70,14 @@
 // RECOVER-NEXT:    [[TMP4:%.*]] = icmp ne i32 [[TMP1]], -1, !nosanitize [[META2]]
 // RECOVER-NEXT:    [[OR:%.*]] = or i1 [[TMP3]], [[TMP4]], !nosanitize [[META2]]
 // RECOVER-NEXT:    [[TMP5:%.*]] = and i1 [[TMP2]], [[OR]], !nosanitize [[META2]]
-// RECOVER-NEXT:    br i1 [[TMP5]], label [[CONT:%.*]], label [[HANDLER_DIVREM_OVERFLOW:%.*]], !prof [[PROF3:![0-9]+]], !nosanitize [[META2]]
+// RECOVER-NEXT:    [[TMP6:%.*]] = call i1 @llvm.allow.ubsan.check(i8 3), !nosanitize [[META2]]
+// RECOVER-NEXT:    [[TMP7:%.*]] = xor i1 [[TMP6]], true, !nosanitize [[META2]]
+// RECOVER-NEXT:    [[TMP8:%.*]] = or i1 [[TMP5]], [[TMP7]], !nosanitize [[META2]]
+// RECOVER-NEXT:    br i1 [[TMP8]], label [[CONT:%.*]], label [[HANDLER_DIVREM_OVERFLOW:%.*]], !prof [[PROF3:![0-9]+]], !nosanitize [[META2]]
 // RECOVER:       handler.divrem_overflow:
-// RECOVER-NEXT:    [[TMP6:%.*]] = zext i32 [[TMP0]] to i64, !nosanitize [[META2]]
-// RECOVER-NEXT:    [[TMP7:%.*]] = zext i32 [[TMP1]] to i64, !nosanitize [[META2]]
-// RECOVER-NEXT:    call void @__ubsan_handle_divrem_overflow(ptr @[[GLOB1:[0-9]+]], i64 [[TMP6]], i64 [[TMP7]]) #[[ATTR3:[0-9]+]], !nosanitize [[META2]]
+// RECOVER-NEXT:    [[TMP9:%.*]] = zext i32 [[TMP0]] to i64, !nosanitize [[META2]]
+// RECOVER-NEXT:    [[TMP10:%.*]] = zext i32 [[TMP1]] to i64, !nosanitize [[META2]]
+// RECOVER-NEXT:    call void @__ubsan_handle_divrem_overflow(ptr @[[GLOB1:[0-9]+]], i64 [[TMP9]], i64 [[TMP10]]) #[[ATTR4:[0-9]+]], !nosanitize [[META2]]
 // RECOVER-NEXT:    br label [[CONT]], !nosanitize [[META2]]
 // RECOVER:       cont:
 // RECOVER-NEXT:    [[DIV:%.*]] = sdiv i32 [[TMP0]], [[TMP1]]
@@ -85,14 +94,17 @@ int div(int x, int y) {
 // CHECK-NEXT:    store ptr [[X]], ptr [[X_ADDR]], align 8
 // CHECK-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[X_ADDR]], align 8
 // CHECK-NEXT:    [[TMP1:%.*]] = icmp ne ptr [[TMP0]], null, !nosanitize [[META2]]
-// CHECK-NEXT:    br i1 [[TMP1]], label [[CONT:%.*]], label [[HANDLER_TYPE_MISMATCH:%.*]], !prof [[PROF3]], !nosanitize [[META2]]
+// CHECK-NEXT:    [[TMP2:%.*]] = call i1 @llvm.allow.ubsan.check(i8 22), !nosanitize [[META2]]
+// CHECK-NEXT:    [[TMP3:%.*]] = xor i1 [[TMP2]], true, !nosanitize [[META2]]
+// CHECK-NEXT:    [[TMP4:%.*]] = or i1 [[TMP1]], [[TMP3]], !nosanitize [[META2]]
+// CHECK-NEXT:    br i1 [[TMP4]], label [[CONT:%.*]], label [[HANDLER_TYPE_MISMATCH:%.*]], !prof [[PROF3]], !nosanitize [[META2]]
 // CHECK:       handler.type_mismatch:
-// CHECK-NEXT:    [[TMP2:%.*]] = ptrtoint ptr [[TMP0]] to i64, !nosanitize [[META2]]
-// CHECK-NEXT:    call void @__ubsan_handle_type_mismatch_v1_abort(ptr @[[GLOB2:[0-9]+]], i64 [[TMP2]]) #[[ATTR3]], !nosanitize [[META2]]
+// CHECK-NEXT:    [[TMP5:%.*]] = ptrtoint ptr [[TMP0]] to i64, !nosanitize [[META2]]
+// CHECK-NEXT:    call void @__ubsan_handle_type_mismatch_v1_abort(ptr @[[GLOB2:[0-9]+]], i64 [[TMP5]]) #[[ATTR4]], !nosanitize [[META2]]
 // CHECK-NEXT:    unreachable, !nosanitize [[META2]]
 // CHECK:       cont:
-// CHECK-NEXT:    [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
-// CHECK-NEXT:    ret i32 [[TMP3]]
+// CHECK-NEXT:    [[TMP6:%.*]] = load i32, ptr [[TMP0]], align 4
+// CHECK-NEXT:    ret i32 [[TMP6]]
 //
 // TRAP-LABEL: define dso_local i32 @null(
 // TRAP-SAME: ptr noundef [[X:%.*]]) #[[ATTR0]] {
@@ -101,13 +113,16 @@ int div(int x, int y) {
 // TRAP-NEXT:    store ptr [[X]], ptr [[X_ADDR]], align 8
 // TRAP-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[X_ADDR]], align 8
 // TRAP-NEXT:    [[TMP1:%.*]] = icmp ne ptr [[TMP0]], null, !nosanitize [[META2]]
-// TRAP-NEXT:    br i1 [[TMP1]], label [[CONT:%.*]], label [[TRAP:%.*]], !nosanitize [[META2]]
+// TRAP-NEXT:    [[TMP2:%.*]] = call i1 @llvm.allow.ubsan.check(i8 22), !nosanitize [[META2]]
+// TRAP-NEXT:    [[TMP3:%.*]] = xor i1 [[TMP2]], true, !nosanitize [[META2]]
+// TRAP-NEXT:    [[TMP4:%.*]] = or i1 [[TMP1]], [[TMP3]], !nosanitize [[META2]]
+// TRAP-NEXT:    br i1 [[TMP4]], label [[CONT:%.*]], label [[TRAP:%.*]], !nosanitize [[META2]]
 // TRAP:       trap:
-// TRAP-NEXT:    call void @llvm.ubsantrap(i8 22) #[[ATTR3]], !nosanitize [[META2]]
+// TRAP-NEXT:    call void @llvm.ubsantrap(i8 22) #[[ATTR4]], !nosanitize [[META2]]
 // TRAP-NEXT:    unreachable, !nosanitize [[META2]]
 // TRAP:       cont:
-// TRAP-NEXT:    [[TMP2:%.*]] = load i32, ptr [[TMP0]], align 4
-// TRAP-NEXT:    ret i32 [[TMP2]]
+// TRAP-NEXT:    [[TMP5:%.*]] = load i32, ptr [[TMP0]], align 4
+// TRAP-NEXT:    ret i32 [[TMP5]]
 //
 // RECOVER-LABEL: define dso_local i32 @null(
 // RECOVER-SAME: ptr noundef [[X:%.*]]) #[[ATTR0]] {
@@ -116,14 +131,17 @@ int div(int x, int y) {
 // RECOVER-NEXT:    store ptr [[X]], ptr [[X_ADDR]], align 8
 // RECOVER-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[X_ADDR]], align 8
 // RECOVER-NEXT:    [[TMP1:%.*]] = icmp ne ptr [[TMP0]], null, !nosanitize [[META2]]
-// RECOVER-NEXT:    br i1 [[TMP1]], label [[CONT:%.*]], label [[HANDLER_TYPE_MISMATCH:%.*]], !prof [[PROF3]], !nosanitize [[META2]]
+// RECOVER-NEXT:    [[TMP2:%.*]] = call i1 @llvm.allow.ubsan.check(i8 22), !nosanitize [[META2]]
+// RECOVER-NEXT:    [[TMP3:%.*]] = xor i1 [[TMP2]], true, !nosanitize [[META2]]
+// RECOVER-NEXT:    [[TMP4:%.*]] = or i1 [[TMP1]], [[TMP3]], !nosanitize [[META2]]
+// RECOVER-NEXT:    br i1 [[TMP4]], label [[CONT:%.*]], label [[HANDLER_TYPE_MISMATCH:%.*]], !prof [[PROF3]], !nosanitize [[META2]]
 // RECOVER:       handler.type_mismatch:
-// RECOVER-NEXT:    [[TMP2:%.*]] = ptrtoint ptr [[TMP0]] to i64, !nosanitize [[META2]]
-// RECOVER-NEXT:    call void @__ubsan_handle_type_mismatch_v1(ptr @[[GLOB2:[0-9]+]], i64 [[TMP2]]) #[[ATTR3]], !nosanitize [[META2]]
+// RECOVER-NEXT:    [[TMP5:%.*]] = ptrtoint ptr [[TMP0]] to i64, !nosanitize [[META2]]
+// RECOVER-NEXT:    call void @__ubsan_handle_type_mismatch_v1(ptr @[[GLOB2:[0-9]+]], i64 [[TMP5]]) #[[ATTR4]], !nosanitize [[META2]]
 // RECOVER-NEXT:    br label [[CONT]], !nosanitize [[META2]]
 // RECOVER:       cont:
-// RECOVER-NEXT:    [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
-// RECOVER-NEXT:    ret i32 [[TMP3]]
+// RECOVER-NEXT:    [[TMP6:%.*]] = load i32, ptr [[TMP0]], align 4
+// RECOVER-NEXT:    ret i32 [[TMP6]]
 //
 int null(int* x) {
   return *x;
@@ -142,11 +160,14 @@ int null(int* x) {
 // CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { i32, i1 } [[TMP2]], 0, !nosanitize [[META2]]
 // CHECK-NEXT:    [[TMP4:%.*]] = extractvalue { i32, i1 } [[TMP2]], 1, !nosanitize [[META2]]
 // CHECK-NEXT:    [[TMP5:%.*]] = xor i1 [[TMP4]], true, !nosanitize [[META2]]
-// CHECK-NEXT:    br i1 [[TMP5]], label [[CONT:%.*]], label [[HANDLER_ADD_OVERFLOW:%.*]], !prof [[PROF3]], !nosanitize [[META2]]
+// CHECK-NEXT:    [[TMP6:%.*]] = call i1 @llvm.allow.ubsan.check(i8 0), !nosanitize [[META2]]
+// CHECK-NEXT:    [[TMP7:%.*]] = xor i1 [[TMP6]], true, !nosanitize [[META2]]
+// CHECK-NEXT:    [[TMP8:%.*]] = or i1 [[TMP5]], [[TMP7]], !nosanitize [[META2]]
+// CHECK-NEXT:    br i1 [[TMP8]], label [[CONT:%.*]], label [[HANDLER_ADD_OVERFLOW:%.*]], !prof [[PROF3]], !nosanitize [[META2]]
 // CHECK:       handler.add_overflow:
-// CHECK-NEXT:    [[TMP6:%.*]] = zext i32 [[TMP0]] to i64, !nosanitize [[META2]]
-// CHECK-NEXT:    [[TMP7:%.*]] = zext i32 [[TMP1]] to i64, !nosanitize [[META2]]
-// CHECK-NEXT:    call void @__ubsan_handle_add_overflow_abort(ptr @[[GLOB3:[0-9]+]], i64 [[TMP6]], i64 [[TMP7]]) #[[ATTR3]], !nosanitize [[META2]]
+// CHECK-NEXT:    [[TMP9:%.*]] = zext i32 [[TMP0]] to i64, !nosanitize [[META2]]
+// CHECK-NEXT:    [[TMP10:%.*]] = zext i32 [[TMP1]] to i64, !nosanitize [[META2]]
+// CHECK-NEXT:    call void @__ubsan_handle_add_overflow_abort(ptr @[[GLOB3:[0-9]+]], i64 [[TMP9]], i64 [[TMP10]]) #[[ATTR4]], !nosanitize [[META2]]
 // CHECK-NEXT:    unreachable, !nosanitize [[META2]]
 // CHECK:       cont:
 // CHECK-NEXT:    ret i32 [[TMP3]]
@@ -164,9 +185,12 @@ int null(int* x) {
 // TRAP-NEXT:    [[TMP3:%.*]] = extractvalue { i32, i1 } [[TMP2]], 0, !nosanitize [[META2]]
 // TRAP-NEXT:    [[TMP4:%.*]] = extractvalue { i32, i1 } [[TMP2]], 1, !nosanitize [[META2]]
 // TRAP-NEXT:    [[TMP5:%.*]] = xor i1 [[TMP4]], true, !nosanitize [[META2]]
-// TRAP-NEXT:    br i1 [[TMP5]], label [[CONT:%.*]], label [[TRAP:%.*]], !nosanitize [[META2]]
+// TRAP-NEXT:    [[TMP6:%.*]] = call i1 @llvm.allow.ubsan.check(i8 0), !nosanitize [[META2]]
+// TRAP-NEXT:    [[TMP7:%.*]] = xor i1 [[TMP6]], true, !nosanitize [[META2]]
+// TRAP-NEXT:    [[TMP8:%.*]] = or i1 [[TMP5]], [[TMP7]], !nosanitize [[META2]]
+// TRAP-NEXT:    br i1 [[TMP8]], label [[CONT:%.*]], label [[TRAP:%.*]], !nosanitize [[META2]]
 // TRAP:       trap:
-// TRAP-NEXT:    call void @llvm.ubsantrap(i8 0) #[[ATTR3]], !nosanitize [[META2]]
+// TRAP-NEXT:    call void @llvm.ubsantrap(i8 0) #[[ATTR4]], !nosanitize [[META2]]
 // TRAP-NEXT:    unreachable, !nosanitize [[META2]]
 // TRAP:       cont:
 // TRAP-NEXT:    ret i32 [[TMP3]]
@@ -184,11 +208,14 @@ int null(int* x) {
 // RECOVER-NEXT:    [[TMP3:%.*]] = extractvalue { i32, i1 } [[TMP2]], 0, !nosanitize [[META2]]
 // RECOVER-NEXT:    [[TMP4:%.*]] = extractvalue { i32, i1 } [[TMP2]], 1, !nosanitize [[META2]]
 // RECOVER-NEXT:    [[TMP5:%.*]] = xor i1 [[TMP4]], true, !nosanitize [[META2]]
-// RECOVER-NEXT:    br i1 [[TMP5]], label [[CONT:%.*]], label [[HANDLER_ADD_OVERFLOW:%.*]], !prof [[PROF3]], !nosanitize [[META2]]
+// RECOVER-NEXT:    [[TMP6:%.*]] = call i1 @llvm.allow.ubsan.check(i8 0), !nosanitize [[META2]]
+// RECOVER-NEXT:    [[TMP7:%.*]] = xor i1 [[TMP6]], true, !nosanitize [[META2]]
+// RECOVER-NEXT:    [[TMP8:%.*]] = or i1 [[TMP5]], [[TMP7]], !nosanitize [[META2]]
+// RECOVER-NEXT:    br i1 [[TMP8]], label [[CONT:%.*]], label [[HANDLER_ADD_OVERFLOW:%.*]], !prof [[PROF3]], !nosanitize [[META2]]
 // RECOVER:       handler.add_overflow:
-// RECOVER-NEXT:    [[TMP6:%.*]] = zext i32 [[TMP0]] to i64, !nosanitize [[META2]]
-// RECOVER-NEXT:    [[TMP7:%.*]] = zext i32 [[TMP1]] to i64, !nosanitize [[META2]]
-// RECOVER-NEXT:    call void @__ubsan_handle_add_overflow(ptr @[[GLOB3:[0-9]+]], i64 [[TMP6]], i64 [[TMP7]]) #[[ATTR3]], !nosanitize [[META2]]
+// RECOVER-NEXT:    [[TMP9:%.*]] = zext i32 [[TMP0]] to i64, !nosanitize [[META2]]
+// RECOVER-NEXT:    [[TMP10:%.*]] = zext i32 [[TMP1]] to i64, !nosanitize [[META2]]
+// RECOVER-NEXT:    call void @__ubsan_handle_add_overflow(ptr @[[GLOB3:[0-9]+]], i64 [[TMP9]], i64 [[TMP10]]) #[[ATTR4]], !nosanitize [[META2]]
 // RECOVER-NEXT:    br label [[CONT]], !nosanitize [[META2]]
 // RECOVER:       cont:
 // RECOVER-NEXT:    ret i32 [[TMP3]]
diff --git a/clang/test/CodeGen/remote-traps.c b/clang/test/CodeGen/remote-traps.c
deleted file mode 100644
index 6983ddbca892..000000000000
--- a/clang/test/CodeGen/remote-traps.c
+++ /dev/null
@@ -1,15 +0,0 @@
-// RUN: %clang_cc1 -O1 -emit-llvm -fsanitize=signed-integer-overflow -fsanitize-trap=signed-integer-overflow %s -o - | FileCheck %s 
-// RUN: %clang_cc1 -O1 -emit-llvm -fsanitize=signed-integer-overflow -fsanitize-trap=signed-integer-overflow -mllvm -remove-traps-random-rate=1 %s -o - | FileCheck %s --implicit-check-not="call void @llvm.ubsantrap" --check-prefixes=REMOVE
-
-int test(int x) {
-  return x + 123;
-}
-
-// CHECK-LABEL: define {{.*}}i32 @test(
-// CHECK: call { i32, i1 } @llvm.sadd.with.overflow.i32(
-// CHECK: trap:
-// CHECK-NEXT: call void @llvm.ubsantrap(i8 0)
-// CHECK-NEXT: unreachable
-
-// REMOVE-LABEL: define {{.*}}i32 @test(
-// REMOVE: call { i32, i1 } @llvm.sadd.with.overflow.i32(
diff --git a/clang/test/CodeGen/tbaa-struct-bitfield-endianness.cpp b/clang/test/CodeGen/tbaa-struct-bitfield-endianness.cpp
index 80884b49ddc6..e8bb46982537 100644
--- a/clang/test/CodeGen/tbaa-struct-bitfield-endianness.cpp
+++ b/clang/test/CodeGen/tbaa-struct-bitfield-endianness.cpp
@@ -1,13 +1,10 @@
 // RUN: %clang_cc1 -triple aarch64_be-apple-darwin -emit-llvm -o - -O1 %s | \
-// RUN:     FileCheck -check-prefixes=CHECK,CHECK-BE %s
+// RUN:     FileCheck -check-prefixes=CHECK %s
 // RUN: %clang_cc1 -triple aarch64-apple-darwin -emit-llvm -o - -O1 %s | \
-// RUN:     FileCheck -check-prefixes=CHECK,CHECK-LE %s
+// RUN:     FileCheck -check-prefixes=CHECK %s
 //
 // Check that TBAA metadata for structs containing bitfields is
 // consistent between big and little endian layouts.
-//
-// FIXME: The metadata below is invalid for the big endian layout: the
-// start offset of 2 is incorrect.
 
 struct NamedBitfields {
   int f1 : 8;
@@ -28,8 +25,7 @@ void copy(NamedBitfields *a1, NamedBitfields *a2) {
   *a1 = *a2;
 }
 
-// CHECK-BE: [[TBAA_STRUCT2]] = !{i64 2, i64 4, [[META3:![0-9]+]], i64 4, i64 4, [[META6:![0-9]+]], i64 8, i64 8, [[META8:![0-9]+]]}
-// CHECK-LE: [[TBAA_STRUCT2]] = !{i64 0, i64 4, [[META3:![0-9]+]], i64 4, i64 4, [[META6:![0-9]+]], i64 8, i64 8, [[META8:![0-9]+]]}
+// CHECK: [[TBAA_STRUCT2]] = !{i64 0, i64 4, [[META3:![0-9]+]], i64 4, i64 4, [[META6:![0-9]+]], i64 8, i64 8, [[META8:![0-9]+]]}
 // CHECK: [[META3]] = !{[[META4:![0-9]+]], [[META4]], i64 0}
 // CHECK: [[META4]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0}
 // CHECK: [[META5]] = !{!"Simple C++ TBAA"}
diff --git a/clang/test/CodeGenHLSL/builtins/all.hlsl b/clang/test/CodeGenHLSL/builtins/all.hlsl
new file mode 100644
index 000000000000..b48daa287480
--- /dev/null
+++ b/clang/test/CodeGenHLSL/builtins/all.hlsl
@@ -0,0 +1,277 @@
+// RUN: %clang_cc1 -finclude-default-header -x hlsl -triple \
+// RUN:   spirv-unknown-vulkan-compute %s -fnative-half-type \
+// RUN:   -emit-llvm -disable-llvm-passes -o - | FileCheck %s \ 
+// RUN:   --check-prefixes=CHECK,NATIVE_HALF,SPIR_NATIVE_HALF,SPIR_CHECK
+// RUN: %clang_cc1 -finclude-default-header -x hlsl -triple \
+// RUN:   spirv-unknown-vulkan-compute %s -emit-llvm -disable-llvm-passes \
+// RUN:   -o - | FileCheck %s --check-prefixes=CHECK,SPIR_NO_HALF,SPIR_CHECK
+// RUN: %clang_cc1 -finclude-default-header -x hlsl -triple \
+// RUN:   dxil-pc-shadermodel6.3-library %s -fnative-half-type \
+// RUN:   -emit-llvm -disable-llvm-passes -o - | FileCheck %s \ 
+// RUN:   --check-prefixes=CHECK,NATIVE_HALF,DXIL_NATIVE_HALF,DXIL_CHECK
+// RUN: %clang_cc1 -finclude-default-header -x hlsl -triple \
+// RUN:   dxil-pc-shadermodel6.3-library %s -emit-llvm -disable-llvm-passes \
+// RUN:   -o - | FileCheck %s --check-prefixes=CHECK,DXIL_NO_HALF,DXIL_CHECK
+
+#ifdef __HLSL_ENABLE_16_BIT
+// DXIL_NATIVE_HALF: define noundef i1 @
+// SPIR_NATIVE_HALF: define spir_func noundef i1 @
+// DXIL_NATIVE_HALF: %hlsl.all = call i1 @llvm.dx.all.i16
+// SPIR_NATIVE_HALF: %hlsl.all = call i1 @llvm.spv.all.i16
+// NATIVE_HALF: ret i1 %hlsl.all
+bool test_all_int16_t(int16_t p0) { return all(p0); }
+// DXIL_NATIVE_HALF: define noundef i1 @
+// SPIR_NATIVE_HALF: define spir_func noundef i1 @
+// DXIL_NATIVE_HALF: %hlsl.all = call i1 @llvm.dx.all.v2i16
+// SPIR_NATIVE_HALF: %hlsl.all = call i1 @llvm.spv.all.v2i16
+// NATIVE_HALF: ret i1 %hlsl.all
+bool test_all_int16_t2(int16_t2 p0) { return all(p0); }
+// DXIL_NATIVE_HALF: define noundef i1 @
+// SPIR_NATIVE_HALF: define spir_func noundef i1 @
+// DXIL_NATIVE_HALF: %hlsl.all = call i1 @llvm.dx.all.v3i16
+// SPIR_NATIVE_HALF: %hlsl.all = call i1 @llvm.spv.all.v3i16
+// NATIVE_HALF: ret i1 %hlsl.all
+bool test_all_int16_t3(int16_t3 p0) { return all(p0); }
+// DXIL_NATIVE_HALF: define noundef i1 @
+// SPIR_NATIVE_HALF: define spir_func noundef i1 @
+// DXIL_NATIVE_HALF: %hlsl.all = call i1 @llvm.dx.all.v4i16
+// SPIR_NATIVE_HALF: %hlsl.all = call i1 @llvm.spv.all.v4i16
+// NATIVE_HALF: ret i1 %hlsl.all
+bool test_all_int16_t4(int16_t4 p0) { return all(p0); }
+
+// DXIL_NATIVE_HALF: define noundef i1 @
+// SPIR_NATIVE_HALF: define spir_func noundef i1 @
+// DXIL_NATIVE_HALF: %hlsl.all = call i1 @llvm.dx.all.i16
+// SPIR_NATIVE_HALF: %hlsl.all = call i1 @llvm.spv.all.i16
+// NATIVE_HALF: ret i1 %hlsl.all
+bool test_all_uint16_t(uint16_t p0) { return all(p0); }
+// DXIL_NATIVE_HALF: define noundef i1 @
+// SPIR_NATIVE_HALF: define spir_func noundef i1 @
+// DXIL_NATIVE_HALF: %hlsl.all = call i1 @llvm.dx.all.v2i16
+// SPIR_NATIVE_HALF: %hlsl.all = call i1 @llvm.spv.all.v2i16
+// NATIVE_HALF: ret i1 %hlsl.all
+bool test_all_uint16_t2(uint16_t2 p0) { return all(p0); }
+// DXIL_NATIVE_HALF: define noundef i1 @
+// SPIR_NATIVE_HALF: define spir_func noundef i1 @
+// DXIL_NATIVE_HALF: %hlsl.all = call i1 @llvm.dx.all.v3i16
+// SPIR_NATIVE_HALF: %hlsl.all = call i1 @llvm.spv.all.v3i16
+// NATIVE_HALF: ret i1 %hlsl.all
+bool test_all_uint16_t3(uint16_t3 p0) { return all(p0); }
+// DXIL_NATIVE_HALF: define noundef i1 @
+// SPIR_NATIVE_HALF: define spir_func noundef i1 @
+// DXIL_NATIVE_HALF: %hlsl.all = call i1 @llvm.dx.all.v4i16
+// SPIR_NATIVE_HALF: %hlsl.all = call i1 @llvm.spv.all.v4i16
+// NATIVE_HALF: ret i1 %hlsl.all
+bool test_all_uint16_t4(uint16_t4 p0) { return all(p0); }
+#endif // __HLSL_ENABLE_16_BIT
+
+// DXIL_CHECK: define noundef i1 @
+// SPIR_CHECK: define spir_func noundef i1 @
+// DXIL_NATIVE_HALF: %hlsl.all = call i1 @llvm.dx.all.f16
+// SPIR_NATIVE_HALF: %hlsl.all = call i1 @llvm.spv.all.f16
+// DXIL_NO_HALF: %hlsl.all = call i1 @llvm.dx.all.f32
+// SPIR_NO_HALF: %hlsl.all = call i1 @llvm.spv.all.f32
+// CHECK: ret i1 %hlsl.all
+bool test_all_half(half p0) { return all(p0); }
+
+// DXIL_CHECK: define noundef i1 @
+// SPIR_CHECK: define spir_func noundef i1 @
+// DXIL_NATIVE_HALF: %hlsl.all = call i1 @llvm.dx.all.v2f16
+// SPIR_NATIVE_HALF: %hlsl.all = call i1 @llvm.spv.all.v2f16
+// DXIL_NO_HALF: %hlsl.all = call i1 @llvm.dx.all.v2f32
+// SPIR_NO_HALF: %hlsl.all = call i1 @llvm.spv.all.v2f32
+// CHECK: ret i1 %hlsl.all
+bool test_all_half2(half2 p0) { return all(p0); }
+
+// DXIL_CHECK: define noundef i1 @
+// SPIR_CHECK: define spir_func noundef i1 @
+// DXIL_NATIVE_HALF: %hlsl.all = call i1 @llvm.dx.all.v3f16
+// SPIR_NATIVE_HALF: %hlsl.all = call i1 @llvm.spv.all.v3f16
+// DXIL_NO_HALF: %hlsl.all = call i1 @llvm.dx.all.v3f32
+// SPIR_NO_HALF: %hlsl.all = call i1 @llvm.spv.all.v3f32
+// CHECK: ret i1 %hlsl.all
+bool test_all_half3(half3 p0) { return all(p0); }
+
+// DXIL_CHECK: define noundef i1 @
+// SPIR_CHECK: define spir_func noundef i1 @
+// DXIL_NATIVE_HALF: %hlsl.all = call i1 @llvm.dx.all.v4f16
+// SPIR_NATIVE_HALF: %hlsl.all = call i1 @llvm.spv.all.v4f16
+// DXIL_NO_HALF: %hlsl.all = call i1 @llvm.dx.all.v4f32
+// SPIR_NO_HALF: %hlsl.all = call i1 @llvm.spv.all.v4f32
+// CHECK: ret i1 %hlsl.all
+bool test_all_half4(half4 p0) { return all(p0); }
+
+// DXIL_CHECK: define noundef i1 @
+// SPIR_CHECK: define spir_func noundef i1 @
+// DXIL_CHECK: %hlsl.all = call i1 @llvm.dx.all.f32
+// SPIR_CHECK: %hlsl.all = call i1 @llvm.spv.all.f32
+// CHECK: ret i1 %hlsl.all
+bool test_all_float(float p0) { return all(p0); }
+// DXIL_CHECK: define noundef i1 @
+// SPIR_CHECK: define spir_func noundef i1 @
+// DXIL_CHECK: %hlsl.all = call i1 @llvm.dx.all.v2f32
+// SPIR_CHECK: %hlsl.all = call i1 @llvm.spv.all.v2f32
+// CHECK: ret i1 %hlsl.all
+bool test_all_float2(float2 p0) { return all(p0); }
+// DXIL_CHECK: define noundef i1 @
+// SPIR_CHECK: define spir_func noundef i1 @
+// DXIL_CHECK: %hlsl.all = call i1 @llvm.dx.all.v3f32
+// SPIR_CHECK: %hlsl.all = call i1 @llvm.spv.all.v3f32
+// CHECK: ret i1 %hlsl.all
+bool test_all_float3(float3 p0) { return all(p0); }
+// DXIL_CHECK: define noundef i1 @
+// SPIR_CHECK: define spir_func noundef i1 @
+// DXIL_CHECK: %hlsl.all = call i1 @llvm.dx.all.v4f32
+// SPIR_CHECK: %hlsl.all = call i1 @llvm.spv.all.v4f32
+// CHECK: ret i1 %hlsl.all
+bool test_all_float4(float4 p0) { return all(p0); }
+
+// DXIL_CHECK: define noundef i1 @
+// SPIR_CHECK: define spir_func noundef i1 @
+// DXIL_CHECK: %hlsl.all = call i1 @llvm.dx.all.f64
+// SPIR_CHECK: %hlsl.all = call i1 @llvm.spv.all.f64
+// CHECK: ret i1 %hlsl.all
+bool test_all_double(double p0) { return all(p0); }
+// DXIL_CHECK: define noundef i1 @
+// SPIR_CHECK: define spir_func noundef i1 @
+// DXIL_CHECK: %hlsl.all = call i1 @llvm.dx.all.v2f64
+// SPIR_CHECK: %hlsl.all = call i1 @llvm.spv.all.v2f64
+// CHECK: ret i1 %hlsl.all
+bool test_all_double2(double2 p0) { return all(p0); }
+// DXIL_CHECK: define noundef i1 @
+// SPIR_CHECK: define spir_func noundef i1 @
+// DXIL_CHECK: %hlsl.all = call i1 @llvm.dx.all.v3f64
+// SPIR_CHECK: %hlsl.all = call i1 @llvm.spv.all.v3f64
+// CHECK: ret i1 %hlsl.all
+bool test_all_double3(double3 p0) { return all(p0); }
+// DXIL_CHECK: define noundef i1 @
+// SPIR_CHECK: define spir_func noundef i1 @
+// DXIL_CHECK: %hlsl.all = call i1 @llvm.dx.all.v4f64
+// SPIR_CHECK: %hlsl.all = call i1 @llvm.spv.all.v4f64
+// CHECK: ret i1 %hlsl.all
+bool test_all_double4(double4 p0) { return all(p0); }
+
+// DXIL_CHECK: define noundef i1 @
+// SPIR_CHECK: define spir_func noundef i1 @
+// DXIL_CHECK: %hlsl.all = call i1 @llvm.dx.all.i32
+// SPIR_CHECK: %hlsl.all = call i1 @llvm.spv.all.i32
+// CHECK: ret i1 %hlsl.all
+bool test_all_int(int p0) { return all(p0); }
+// DXIL_CHECK: define noundef i1 @
+// SPIR_CHECK: define spir_func noundef i1 @
+// DXIL_CHECK: %hlsl.all = call i1 @llvm.dx.all.v2i32
+// SPIR_CHECK: %hlsl.all = call i1 @llvm.spv.all.v2i32
+// CHECK: ret i1 %hlsl.all
+bool test_all_int2(int2 p0) { return all(p0); }
+// DXIL_CHECK: define noundef i1 @
+// SPIR_CHECK: define spir_func noundef i1 @
+// DXIL_CHECK: %hlsl.all = call i1 @llvm.dx.all.v3i32
+// SPIR_CHECK: %hlsl.all = call i1 @llvm.spv.all.v3i32
+// CHECK: ret i1 %hlsl.all
+bool test_all_int3(int3 p0) { return all(p0); }
+// DXIL_CHECK: define noundef i1 @
+// SPIR_CHECK: define spir_func noundef i1 @
+// DXIL_CHECK: %hlsl.all = call i1 @llvm.dx.all.v4i32
+// SPIR_CHECK: %hlsl.all = call i1 @llvm.spv.all.v4i32
+// CHECK: ret i1 %hlsl.all
+bool test_all_int4(int4 p0) { return all(p0); }
+
+// DXIL_CHECK: define noundef i1 @
+// SPIR_CHECK: define spir_func noundef i1 @
+// DXIL_CHECK: %hlsl.all = call i1 @llvm.dx.all.i32
+// SPIR_CHECK: %hlsl.all = call i1 @llvm.spv.all.i32
+// CHECK: ret i1 %hlsl.all
+bool test_all_uint(uint p0) { return all(p0); }
+// DXIL_CHECK: define noundef i1 @
+// SPIR_CHECK: define spir_func noundef i1 @
+// DXIL_CHECK: %hlsl.all = call i1 @llvm.dx.all.v2i32
+// SPIR_CHECK: %hlsl.all = call i1 @llvm.spv.all.v2i32
+// CHECK: ret i1 %hlsl.all
+bool test_all_uint2(uint2 p0) { return all(p0); }
+// DXIL_CHECK: define noundef i1 @
+// SPIR_CHECK: define spir_func noundef i1 @
+// DXIL_CHECK: %hlsl.all = call i1 @llvm.dx.all.v3i32
+// SPIR_CHECK: %hlsl.all = call i1 @llvm.spv.all.v3i32
+// CHECK: ret i1 %hlsl.all
+bool test_all_uint3(uint3 p0) { return all(p0); }
+// DXIL_CHECK: define noundef i1 @
+// SPIR_CHECK: define spir_func noundef i1 @
+// DXIL_CHECK: %hlsl.all = call i1 @llvm.dx.all.v4i32
+// SPIR_CHECK: %hlsl.all = call i1 @llvm.spv.all.v4i32
+// CHECK: ret i1 %hlsl.all
+bool test_all_uint4(uint4 p0) { return all(p0); }
+
+// DXIL_CHECK: define noundef i1 @
+// SPIR_CHECK: define spir_func noundef i1 @
+// DXIL_CHECK: %hlsl.all = call i1 @llvm.dx.all.i64
+// SPIR_CHECK: %hlsl.all = call i1 @llvm.spv.all.i64
+// CHECK: ret i1 %hlsl.all
+bool test_all_int64_t(int64_t p0) { return all(p0); }
+// DXIL_CHECK: define noundef i1 @
+// SPIR_CHECK: define spir_func noundef i1 @
+// DXIL_CHECK: %hlsl.all = call i1 @llvm.dx.all.v2i64
+// SPIR_CHECK: %hlsl.all = call i1 @llvm.spv.all.v2i64
+// CHECK: ret i1 %hlsl.all
+bool test_all_int64_t2(int64_t2 p0) { return all(p0); }
+// DXIL_CHECK: define noundef i1 @
+// SPIR_CHECK: define spir_func noundef i1 @
+// DXIL_CHECK: %hlsl.all = call i1 @llvm.dx.all.v3i64
+// SPIR_CHECK: %hlsl.all = call i1 @llvm.spv.all.v3i64
+// CHECK: ret i1 %hlsl.all
+bool test_all_int64_t3(int64_t3 p0) { return all(p0); }
+// DXIL_CHECK: define noundef i1 @
+// SPIR_CHECK: define spir_func noundef i1 @
+// DXIL_CHECK: %hlsl.all = call i1 @llvm.dx.all.v4i64
+// SPIR_CHECK: %hlsl.all = call i1 @llvm.spv.all.v4i64
+// CHECK: ret i1 %hlsl.all
+bool test_all_int64_t4(int64_t4 p0) { return all(p0); }
+
+// DXIL_CHECK: define noundef i1 @
+// SPIR_CHECK: define spir_func noundef i1 @
+// DXIL_CHECK: %hlsl.all = call i1 @llvm.dx.all.i64
+// SPIR_CHECK: %hlsl.all = call i1 @llvm.spv.all.i64
+// CHECK: ret i1 %hlsl.all
+bool test_all_uint64_t(uint64_t p0) { return all(p0); }
+// DXIL_CHECK: define noundef i1 @
+// SPIR_CHECK: define spir_func noundef i1 @
+// DXIL_CHECK: %hlsl.all = call i1 @llvm.dx.all.v2i64
+// SPIR_CHECK: %hlsl.all = call i1 @llvm.spv.all.v2i64
+// CHECK: ret i1 %hlsl.all
+bool test_all_uint64_t2(uint64_t2 p0) { return all(p0); }
+// DXIL_CHECK: define noundef i1 @
+// SPIR_CHECK: define spir_func noundef i1 @
+// DXIL_CHECK: %hlsl.all = call i1 @llvm.dx.all.v3i64
+// SPIR_CHECK: %hlsl.all = call i1 @llvm.spv.all.v3i64
+// CHECK: ret i1 %hlsl.all
+bool test_all_uint64_t3(uint64_t3 p0) { return all(p0); }
+// DXIL_CHECK: define noundef i1 @
+// SPIR_CHECK: define spir_func noundef i1 @
+// DXIL_CHECK: %hlsl.all = call i1 @llvm.dx.all.v4i64
+// SPIR_CHECK: %hlsl.all = call i1 @llvm.spv.all.v4i64
+// CHECK: ret i1 %hlsl.all
+bool test_all_uint64_t4(uint64_t4 p0) { return all(p0); }
+
+// DXIL_CHECK: define noundef i1 @
+// SPIR_CHECK: define spir_func noundef i1 @
+// DXIL_CHECK: %hlsl.all = call i1 @llvm.dx.all.i1
+// SPIR_CHECK: %hlsl.all = call i1 @llvm.spv.all.i1
+// CHECK: ret i1 %hlsl.all
+bool test_all_bool(bool p0) { return all(p0); }
+// DXIL_CHECK: define noundef i1 @
+// SPIR_CHECK: define spir_func noundef i1 @
+// DXIL_CHECK: %hlsl.all = call i1 @llvm.dx.all.v2i1
+// SPIR_CHECK: %hlsl.all = call i1 @llvm.spv.all.v2i1
+// CHECK: ret i1 %hlsl.all
+bool test_all_bool2(bool2 p0) { return all(p0); }
+// DXIL_CHECK: define noundef i1 @
+// SPIR_CHECK: define spir_func noundef i1 @
+// DXIL_CHECK: %hlsl.all = call i1 @llvm.dx.all.v3i1
+// SPIR_CHECK: %hlsl.all = call i1 @llvm.spv.all.v3i1
+// CHECK: ret i1 %hlsl.all
+bool test_all_bool3(bool3 p0) { return all(p0); }
+// DXIL_CHECK: define noundef i1 @
+// SPIR_CHECK: define spir_func noundef i1 @
+// DXIL_CHECK: %hlsl.all = call i1 @llvm.dx.all.v4i1
+// SPIR_CHECK: %hlsl.all = call i1 @llvm.spv.all.v4i1
+// CHECK: ret i1 %hlsl.all
+bool test_all_bool4(bool4 p0) { return all(p0); }
diff --git a/clang/test/InstallAPI/Inputs/Foundation/Foundation.framework/Headers/Foundation.h b/clang/test/InstallAPI/Inputs/Foundation/Foundation.framework/Headers/Foundation.h
new file mode 100644
index 000000000000..e731b59ac530
--- /dev/null
+++ b/clang/test/InstallAPI/Inputs/Foundation/Foundation.framework/Headers/Foundation.h
@@ -0,0 +1,19 @@
+@interface NSObject 
+@end
+
+typedef unsigned char BOOL; 
+#ifndef NS_AVAILABLE
+#define NS_AVAILABLE(x,y) __attribute__((availability(macosx,introduced=x)))
+#endif 
+#ifndef NS_UNAVAILABLE
+#define NS_UNAVAILABLE  __attribute__((unavailable))
+#endif 
+#ifndef NS_DEPRECATED_MAC
+#define NS_DEPRECATED_MAC(x,y) __attribute__((availability(macosx,introduced=x,deprecated=y,message="" )));
+#endif 
+
+@interface NSManagedObject
+@end 
+
+@interface NSSet 
+@end 
diff --git a/clang/test/InstallAPI/extra-exclude-headers.test b/clang/test/InstallAPI/extra-exclude-headers.test
index 663ca1a5d500..addb81f5386f 100644
--- a/clang/test/InstallAPI/extra-exclude-headers.test
+++ b/clang/test/InstallAPI/extra-exclude-headers.test
@@ -2,6 +2,7 @@
 ; RUN: split-file %s %t
 ; RUN: mkdir -p %t/System/Library/Frameworks 
 ; RUN: cp -r %S/Inputs/Simple/Simple.framework %t/System/Library/Frameworks/
+; RUN: cp -r %S/Inputs/Foundation/Foundation.framework %t/System/Library/Frameworks/
 ; RUN: sed -e "s|DSTROOT|%/t|g" %t/inputs.json.in > %t/inputs.json
 ; RUN: yaml2obj %S/Inputs/Simple/Simple.yaml -o %t/Simple 
 
@@ -184,24 +185,3 @@
   ],
   "version": "3"
 }
-
-;--- System/Library/Frameworks/Foundation.framework/Headers/Foundation.h
-@interface NSObject 
-@end
-
-typedef unsigned char BOOL; 
-#ifndef NS_AVAILABLE
-#define NS_AVAILABLE(x,y) __attribute__((availability(macosx,introduced=x)))
-#endif 
-#ifndef NS_UNAVAILABLE
-#define NS_UNAVAILABLE  __attribute__((unavailable))
-#endif 
-#ifndef NS_DEPRECATED_MAC
-#define NS_DEPRECATED_MAC(x,y) __attribute__((availability(macosx,introduced=x,deprecated=y,message="" )));
-#endif 
-
-@interface NSManagedObject
-@end 
-
-@interface NSSet 
-@end 
diff --git a/clang/test/InstallAPI/forwarded-search-paths.test b/clang/test/InstallAPI/forwarded-search-paths.test
new file mode 100644
index 000000000000..dc1e9006060f
--- /dev/null
+++ b/clang/test/InstallAPI/forwarded-search-paths.test
@@ -0,0 +1,34 @@
+; RUN: rm -rf %t
+; RUN: split-file %s %t
+; RUN: sed -e "s|DSTROOT|%/t|g" %t/input.json.in > %t/input.json
+
+; RUN: mkdir -p %t/System/Library/Frameworks 
+; RUN: cp -r %S/Inputs/Foundation/Foundation.framework %t/System/Library/Frameworks/
+; RUN: cp -r %S/Inputs/Simple/Simple.framework %t/System/Library/Frameworks/
+; RUN: yaml2obj %S/Inputs/Simple/Simple.yaml -o %t/Simple 
+; RUN: mkdir -p %t/usr/include/after
+
+; RUN: clang-installapi -target x86_64-apple-macosx10.12 \
+; RUN: -install_name /System/Library/Frameworks/Simple.framework/Versions/A/Simple \
+; RUN: -current_version 1.2.3 -compatibility_version 1 -o %t/Simple.tbd \
+; RUN: -idirafter %t/usr/include/after \
+; RUN: -F %t/System/Library/Frameworks \
+; RUN: --verify-against=%t/Simple --verify-mode=ErrorsOnly \
+; RUN: %t/input.json  -v 2>&1 | FileCheck %s
+
+; CHECK: "-idirafter" {{.*}}/usr/include/after"
+; CHECK: #include "..." search starts here:
+; CHECK: #include <...> search starts here:
+; CHECK: usr/include/after 
+; CHECK-NEXT: End of search list.
+
+;--- input.json.in
+{
+  "version" : "3",
+  "headers" : [
+    {
+      "type" : "public",
+      "path" : "DSTROOT/System/Library/Frameworks/Simple.framework/Headers/Basic.h"
+    }
+  ]
+}
diff --git a/clang/test/OpenMP/thread_limit_amdgpu.c b/clang/test/OpenMP/thread_limit_amdgpu.c
new file mode 100644
index 000000000000..f884eeb73c3f
--- /dev/null
+++ b/clang/test/OpenMP/thread_limit_amdgpu.c
@@ -0,0 +1,34 @@
+// Test target codegen - host bc file has to be created first.
+// RUN: %clang_cc1 -verify -fopenmp -x c++ -triple x86_64-unknown-linux-gnu -fopenmp-targets=amdgcn-amd-amdhsa -emit-llvm-bc %s -o %t-x86-host.bc
+// RUN: %clang_cc1 -verify -fopenmp -x c++ -triple amdgcn-amd-amdhsa -fopenmp-targets=amdgcn-amd-amdhsa -emit-llvm %s -fopenmp-is-target-device -fopenmp-host-ir-file-path %t-x86-host.bc -o - | FileCheck %s
+// expected-no-diagnostics
+
+#ifndef HEADER
+#define HEADER
+
+void foo(int N) {
+#pragma omp target teams distribute parallel for simd
+  for (int i = 0; i < N; ++i)
+    ;
+#pragma omp target teams distribute parallel for simd thread_limit(4)
+  for (int i = 0; i < N; ++i)
+    ;
+#pragma omp target teams distribute parallel for simd ompx_attribute(__attribute__((launch_bounds(42, 42))))
+  for (int i = 0; i < N; ++i)
+    ;
+#pragma omp target teams distribute parallel for simd ompx_attribute(__attribute__((launch_bounds(42, 42)))) num_threads(22)
+  for (int i = 0; i < N; ++i)
+    ;
+}
+
+#endif
+
+// CHECK: define weak_odr protected amdgpu_kernel void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+__Z3fooi_}}l10({{.*}}) #[[ATTR1:.+]] {
+// CHECK: define weak_odr protected amdgpu_kernel void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+__Z3fooi_}}l13({{.*}}) #[[ATTR2:.+]] {
+// CHECK: define weak_odr protected amdgpu_kernel void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+__Z3fooi_}}l16({{.*}}) #[[ATTR3:.+]] {
+// CHECK: define weak_odr protected amdgpu_kernel void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+__Z3fooi_}}l19({{.*}}) #[[ATTR4:.+]] {
+
+// CHECK: attributes #[[ATTR1]] = { {{.*}} "amdgpu-flat-work-group-size"="1,256" {{.*}} }
+// CHECK: attributes #[[ATTR2]] = { {{.*}} "amdgpu-flat-work-group-size"="1,4" {{.*}} }
+// CHECK: attributes #[[ATTR3]] = { {{.*}} "amdgpu-flat-work-group-size"="1,42" "amdgpu-max-num-workgroups"="42,1,1"{{.*}} }
+// CHECK: attributes #[[ATTR4]] = { {{.*}} "amdgpu-flat-work-group-size"="1,22" "amdgpu-max-num-workgroups"="42,1,1"{{.*}} }
diff --git a/clang/test/SemaCXX/PR86790.cpp b/clang/test/SemaCXX/PR86790.cpp
new file mode 100644
index 000000000000..09e9bb3505e1
--- /dev/null
+++ b/clang/test/SemaCXX/PR86790.cpp
@@ -0,0 +1,32 @@
+// RUN: %clang_cc1 -verify -std=c++20 -fsyntax-only %s
+
+enum {A, S, D, F};
+int main() {
+    using asdf = decltype(A);
+    using enum asdf; // this line causes the crash
+    return 0;
+}
+
+namespace N1 {
+    enum {A, S, D, F};
+    constexpr struct T {
+    using asdf = decltype(A);
+    using enum asdf;
+    } t;
+
+    static_assert(t.D == D);
+    static_assert(T::S == S);
+}
+
+namespace N2 {
+    enum {A, S, D, F};
+    constexpr struct T {
+    struct {
+        using asdf = decltype(A);
+        using enum asdf;
+    } inner;
+    } t;
+
+    static_assert(t.inner.D == D);
+    static_assert(t.D == D); // expected-error {{no member named 'D' in 'N2::T'}}
+}
diff --git a/clang/test/SemaCXX/warn-unused-but-set-variables-cpp.cpp b/clang/test/SemaCXX/warn-unused-but-set-variables-cpp.cpp
index 418baa78aa96..eaedb53bf472 100644
--- a/clang/test/SemaCXX/warn-unused-but-set-variables-cpp.cpp
+++ b/clang/test/SemaCXX/warn-unused-but-set-variables-cpp.cpp
@@ -69,3 +69,11 @@ template <typename T> void f5() {
   SWarnUnused swu;
   ++swu;
 }
+
+void f6() {
+  if (int x = 123) {} // expected-warning{{variable 'x' set but not used}}
+
+  while (int x = 123) {} // expected-warning{{variable 'x' set but not used}}
+
+  for (; int x = 123;) {} // expected-warning{{variable 'x' set but not used}}
+}
diff --git a/clang/test/SemaTemplate/alias-template-with-lambdas.cpp b/clang/test/SemaTemplate/alias-template-with-lambdas.cpp
new file mode 100644
index 000000000000..ff94031e4d86
--- /dev/null
+++ b/clang/test/SemaTemplate/alias-template-with-lambdas.cpp
@@ -0,0 +1,105 @@
+// RUN: %clang_cc1 -std=c++2c -fsyntax-only -verify %s
+namespace lambda_calls {
+
+template <class>
+concept True = true;
+
+template <class>
+concept False = false; // #False
+
+template <class T> struct S {
+  template <class... U> using type = decltype([](U...) {}(U()...));
+  template <class U> using type2 = decltype([](auto) {}(1));
+  template <class U> using type3 = decltype([](True auto) {}(1));
+  template <class>
+  using type4 = decltype([](auto... pack) { return sizeof...(pack); }(1, 2));
+
+  template <class U> using type5 = decltype([](False auto...) {}(1)); // #Type5
+
+  template <class U>
+  using type6 = decltype([]<True> {}.template operator()<char>());
+  template <class U>
+  using type7 = decltype([]<False> {}.template operator()<char>()); // #Type7
+
+  template <class U>
+  using type8 = decltype([]() // #Type8
+                           requires(sizeof(U) == 32) // #Type8-requirement
+                         {}());
+
+  template <class... U>
+  using type9 = decltype([]<True>(U...) {}.template operator()<char>(U()...));
+  // https://github.com/llvm/llvm-project/issues/76674
+  template <class U>
+  using type10 = decltype([]<class V> { return V(); }.template operator()<U>());
+
+  template <class U> using type11 = decltype([] { return U{}; });
+};
+
+template <class> using Meow = decltype([]<True> {}.template operator()<int>());
+
+template <class... U>
+using MeowMeow = decltype([]<True>(U...) {}.template operator()<char>(U()...));
+
+// https://github.com/llvm/llvm-project/issues/70601
+template <class> using U = decltype([]<True> {}.template operator()<int>());
+
+U<int> foo();
+
+void bar() {
+  using T = S<int>::type<int, int, int>;
+  using T2 = S<int>::type2<int>;
+  using T3 = S<int>::type3<char>;
+  using T4 = S<int>::type4<void>;
+  using T5 = S<int>::type5<void>; // #T5
+  // expected-error@#Type5 {{no matching function for call}}
+  // expected-note@#T5 {{type alias 'type5' requested here}}
+  // expected-note@#Type5 {{constraints not satisfied [with auto:1 = <int>]}}
+  // expected-note@#Type5 {{because 'int' does not satisfy 'False'}}
+  // expected-note@#False {{because 'false' evaluated to false}}
+
+  using T6 = S<int>::type6<void>;
+  using T7 = S<int>::type7<void>; // #T7
+  // expected-error@#Type7 {{no matching member function for call}}
+  // expected-note@#T7 {{type alias 'type7' requested here}}
+  // expected-note@#Type7 {{constraints not satisfied [with $0 = char]}}
+  // expected-note@#Type7 {{because 'char' does not satisfy 'False'}}
+  // expected-note@#False {{because 'false' evaluated to false}}
+
+  using T8 = S<int>::type8<char>; // #T8
+  // expected-error@#Type8 {{no matching function for call}}
+  // expected-note@#T8 {{type alias 'type8' requested here}}
+  // expected-note@#Type8 {{constraints not satisfied}}
+  // expected-note@#Type8-requirement {{because 'sizeof(char) == 32' (1 == 32) evaluated to false}}
+
+  using T9 = S<int>::type9<long, long, char>;
+  using T10 = S<int>::type10<int>;
+  using T11 = S<int>::type11<int>;
+  int x = T11()();
+  using T12 = Meow<int>;
+  using T13 = MeowMeow<char, int, long, unsigned>;
+
+  static_assert(__is_same(T, void));
+  static_assert(__is_same(T2, void));
+  static_assert(__is_same(T3, void));
+  static_assert(__is_same(T4, decltype(sizeof(0))));
+  static_assert(__is_same(T6, void));
+  static_assert(__is_same(T9, void));
+  static_assert(__is_same(T10, int));
+  static_assert(__is_same(T12, void));
+  static_assert(__is_same(T13, void));
+}
+
+namespace GH82104 {
+
+template <typename, typename...> int Zero = 0;
+
+template <typename T, typename...U>
+using T14 = decltype([]<int V = 0>() { return Zero<T, U...>; }());
+
+template <typename T> using T15 = T14<T, T>;
+
+static_assert(__is_same(T15<char>, int));
+
+} // namespace GH82104
+
+} // namespace lambda_calls
diff --git a/clang/tools/clang-installapi/Options.cpp b/clang/tools/clang-installapi/Options.cpp
index c4f39b7c8417..04d5e3cbb9d6 100644
--- a/clang/tools/clang-installapi/Options.cpp
+++ b/clang/tools/clang-installapi/Options.cpp
@@ -45,9 +45,21 @@ static constexpr const ArrayRef<StringLiteral>
 /// Create table mapping all options defined in InstallAPIOpts.td.
 static constexpr OptTable::Info InfoTable[] = {
 #define OPTION(PREFIX, NAME, ID, KIND, GROUP, ALIAS, ALIASARGS, FLAGS,         \
-               VISIBILITY, PARAM, HELPTEXT, METAVAR, VALUES)                   \
-  {PREFIX, NAME,  HELPTEXT,   METAVAR,     OPT_##ID,    Option::KIND##Class,   \
-   PARAM,  FLAGS, VISIBILITY, OPT_##GROUP, OPT_##ALIAS, ALIASARGS,             \
+               VISIBILITY, PARAM, HELPTEXT, HELPTEXTSFORVARIANTS, METAVAR,     \
+               VALUES)                                                         \
+  {PREFIX,                                                                     \
+   NAME,                                                                       \
+   HELPTEXT,                                                                   \
+   HELPTEXTSFORVARIANTS,                                                       \
+   METAVAR,                                                                    \
+   OPT_##ID,                                                                   \
+   Option::KIND##Class,                                                        \
+   PARAM,                                                                      \
+   FLAGS,                                                                      \
+   VISIBILITY,                                                                 \
+   OPT_##GROUP,                                                                \
+   OPT_##ALIAS,                                                                \
+   ALIASARGS,                                                                  \
    VALUES},
 #include "InstallAPIOpts.inc"
 #undef OPTION
diff --git a/clang/tools/clang-installapi/Options.h b/clang/tools/clang-installapi/Options.h
index 82e04b49d125..cd74722c92c8 100644
--- a/clang/tools/clang-installapi/Options.h
+++ b/clang/tools/clang-installapi/Options.h
@@ -144,7 +144,8 @@ private:
 enum ID {
   OPT_INVALID = 0, // This is not an option ID.
 #define OPTION(PREFIX, NAME, ID, KIND, GROUP, ALIAS, ALIASARGS, FLAGS,         \
-               VISIBILITY, PARAM, HELPTEXT, METAVAR, VALUES)                   \
+               VISIBILITY, PARAM, HELPTEXT, HELPTEXTSFORVARIANTS, METAVAR,     \
+               VALUES)                                                         \
   OPT_##ID,
 #include "InstallAPIOpts.inc"
   LastOption
diff --git a/clang/unittests/AST/ASTImporterTest.cpp b/clang/unittests/AST/ASTImporterTest.cpp
index 35ab7e3b7fe3..acc596fef87b 100644
--- a/clang/unittests/AST/ASTImporterTest.cpp
+++ b/clang/unittests/AST/ASTImporterTest.cpp
@@ -5317,6 +5317,34 @@ TEST_P(ASTImporterOptionSpecificTestBase,
   EXPECT_FALSE(ToX);
 }
 
+TEST_P(ASTImporterOptionSpecificTestBase, VarTemplateDeclInlineWithCXX17) {
+  Decl *FromTU = getTuDecl(
+      R"(
+      struct S {
+        template <unsigned> static constexpr bool X = true;
+      };
+      )",
+      Lang_CXX17, "input1.cc");
+  Decl *FromTU2 = getTuDecl(
+      R"(
+      struct S {
+        template <unsigned> static constexpr bool X = true;
+        template <typename T> void get() { X<sizeof(T)>; }
+      };
+      template <typename U> U qvariant_cast(const S &v) { return v.get; }
+      )",
+      Lang_CXX17, "input2.cc");
+  auto *FromX = FirstDeclMatcher<VarTemplateDecl>().match(
+      FromTU, varTemplateDecl(hasName("X")));
+  auto *ToX = Import(FromX, Lang_CXX17);
+  ASSERT_TRUE(ToX);
+  auto *FromX2 = FirstDeclMatcher<VarTemplateDecl>().match(
+      FromTU2, varTemplateDecl(hasName("X")));
+  auto *ToX2 = Import(FromX2, Lang_CXX17);
+  EXPECT_TRUE(ToX2);
+  EXPECT_EQ(ToX, ToX2);
+}
+
 TEST_P(ASTImporterOptionSpecificTestBase, VarTemplateParameterDeclContext) {
   constexpr auto Code =
       R"(
diff --git a/clang/unittests/Format/FormatTest.cpp b/clang/unittests/Format/FormatTest.cpp
index ec0267da468a..91a8ff11889d 100644
--- a/clang/unittests/Format/FormatTest.cpp
+++ b/clang/unittests/Format/FormatTest.cpp
@@ -27272,6 +27272,18 @@ TEST_F(FormatTest, PPBranchesInBracedInit) {
                "};");
 }
 
+TEST_F(FormatTest, PPDirectivesAndCommentsInBracedInit) {
+  verifyFormat("{\n"
+               "  char *a[] = {\n"
+               "      /* abc */ \"abc\",\n"
+               "#if FOO\n"
+               "      /* xyz */ \"xyz\",\n"
+               "#endif\n"
+               "      /* last */ \"last\"};\n"
+               "}",
+               getLLVMStyleWithColumns(30));
+}
+
 TEST_F(FormatTest, StreamOutputOperator) {
   verifyFormat("std::cout << \"foo\" << \"bar\" << baz;");
   verifyFormat("std::cout << \"foo\\n\"\n"
diff --git a/clang/unittests/Format/TokenAnnotatorTest.cpp b/clang/unittests/Format/TokenAnnotatorTest.cpp
index 9425647d9998..5ba5e0fbd16f 100644
--- a/clang/unittests/Format/TokenAnnotatorTest.cpp
+++ b/clang/unittests/Format/TokenAnnotatorTest.cpp
@@ -2809,6 +2809,19 @@ TEST_F(TokenAnnotatorTest, BraceKind) {
   EXPECT_TOKEN(Tokens[13], tok::l_brace, TT_FunctionLBrace);
   EXPECT_BRACE_KIND(Tokens[13], BK_Block);
   EXPECT_BRACE_KIND(Tokens[14], BK_Block);
+
+  Tokens = annotate("{\n"
+                    "  char *a[] = {\n"
+                    "      /* abc */ \"abc\",\n"
+                    "#if FOO\n"
+                    "      /* xyz */ \"xyz\",\n"
+                    "#endif\n"
+                    "      /* last */ \"last\"};\n"
+                    "}");
+  ASSERT_EQ(Tokens.size(), 25u) << Tokens;
+  EXPECT_BRACE_KIND(Tokens[0], BK_Block);
+  EXPECT_BRACE_KIND(Tokens[7], BK_BracedInit);
+  EXPECT_BRACE_KIND(Tokens[21], BK_BracedInit);
 }
 
 TEST_F(TokenAnnotatorTest, StreamOperator) {
diff --git a/clang/unittests/StaticAnalyzer/CallDescriptionTest.cpp b/clang/unittests/StaticAnalyzer/CallDescriptionTest.cpp
index 3aac1f81f523..238f954d7133 100644
--- a/clang/unittests/StaticAnalyzer/CallDescriptionTest.cpp
+++ b/clang/unittests/StaticAnalyzer/CallDescriptionTest.cpp
@@ -488,16 +488,21 @@ TEST(CallDescription, NegativeMatchQualifiedNames) {
 }
 
 TEST(CallDescription, MatchBuiltins) {
-  // Test CDM::CLibrary - a flag that allows matching weird builtins.
-  EXPECT_TRUE(tooling::runToolOnCode(
-      std::unique_ptr<FrontendAction>(new CallDescriptionAction<>(
-          {{{{"memset"}, 3}, false}, {{CDM::CLibrary, {"memset"}, 3}, true}})),
-      "void foo() {"
-      "  int x;"
-      "  __builtin___memset_chk(&x, 0, sizeof(x),"
-      "                         __builtin_object_size(&x, 0));"
-      "}"));
-
+  // Test the matching modes CDM::CLibrary and CDM::CLibraryMaybeHardened,
+  // which can recognize builtin variants of C library functions.
+  {
+    SCOPED_TRACE("hardened variants of functions");
+    EXPECT_TRUE(tooling::runToolOnCode(
+        std::unique_ptr<FrontendAction>(new CallDescriptionAction<>(
+            {{{CDM::Unspecified, {"memset"}, 3}, false},
+             {{CDM::CLibrary, {"memset"}, 3}, false},
+             {{CDM::CLibraryMaybeHardened, {"memset"}, 3}, true}})),
+        "void foo() {"
+        "  int x;"
+        "  __builtin___memset_chk(&x, 0, sizeof(x),"
+        "                         __builtin_object_size(&x, 0));"
+        "}"));
+  }
   {
     SCOPED_TRACE("multiple similar builtins");
     EXPECT_TRUE(tooling::runToolOnCode(
@@ -519,6 +524,35 @@ TEST(CallDescription, MatchBuiltins) {
           })"));
   }
   {
+    SCOPED_TRACE("multiple similar builtins with hardened variant");
+    EXPECT_TRUE(tooling::runToolOnCode(
+        std::unique_ptr<FrontendAction>(new CallDescriptionAction<>(
+            {{{CDM::CLibraryMaybeHardened, {"memcpy"}, 3}, false},
+             {{CDM::CLibraryMaybeHardened, {"wmemcpy"}, 3}, true}})),
+        R"(typedef __typeof(sizeof(int)) size_t;
+          extern wchar_t *__wmemcpy_chk (wchar_t *__restrict __s1,
+                                          const wchar_t *__restrict __s2,
+                                          size_t __n, size_t __ns1);
+          void foo(wchar_t *x, wchar_t *y) {
+            __wmemcpy_chk(x, y, sizeof(wchar_t), 1234);
+          })"));
+  }
+  {
+    SCOPED_TRACE(
+        "multiple similar builtins with hardened variant reversed order");
+    EXPECT_TRUE(tooling::runToolOnCode(
+        std::unique_ptr<FrontendAction>(new CallDescriptionAction<>(
+            {{{CDM::CLibraryMaybeHardened, {"wmemcpy"}, 3}, true},
+             {{CDM::CLibraryMaybeHardened, {"memcpy"}, 3}, false}})),
+        R"(typedef __typeof(sizeof(int)) size_t;
+          extern wchar_t *__wmemcpy_chk (wchar_t *__restrict __s1,
+                                          const wchar_t *__restrict __s2,
+                                          size_t __n, size_t __ns1);
+          void foo(wchar_t *x, wchar_t *y) {
+            __wmemcpy_chk(x, y, sizeof(wchar_t), 1234);
+          })"));
+  }
+  {
     SCOPED_TRACE("lookbehind and lookahead mismatches");
     EXPECT_TRUE(tooling::runToolOnCode(
         std::unique_ptr<FrontendAction>(
diff --git a/clang/utils/TableGen/ClangOptionDocEmitter.cpp b/clang/utils/TableGen/ClangOptionDocEmitter.cpp
index 3fe989099407..86835611b842 100644
--- a/clang/utils/TableGen/ClangOptionDocEmitter.cpp
+++ b/clang/utils/TableGen/ClangOptionDocEmitter.cpp
@@ -359,8 +359,38 @@ void emitOption(const DocumentedOption &Option, const Record *DocInfo,
 
   // Emit the description, if we have one.
   const Record *R = Option.Option;
-  std::string Description =
-      getRSTStringWithTextFallback(R, "DocBrief", "HelpText");
+  std::string Description;
+
+  // Prefer a program specific help string.
+  // This is a list of (visibilities, string) pairs.
+  std::vector<Record *> VisibilitiesHelp =
+      R->getValueAsListOfDefs("HelpTextsForVariants");
+  for (Record *VisibilityHelp : VisibilitiesHelp) {
+    // This is a list of visibilities.
+    ArrayRef<Init *> Visibilities =
+        VisibilityHelp->getValueAsListInit("Visibilities")->getValues();
+
+    // See if any of the program's visibilities are in the list.
+    for (StringRef DocInfoMask :
+         DocInfo->getValueAsListOfStrings("VisibilityMask")) {
+      for (Init *Visibility : Visibilities) {
+        if (Visibility->getAsUnquotedString() == DocInfoMask) {
+          // Use the first one we find.
+          Description = escapeRST(VisibilityHelp->getValueAsString("Text"));
+          break;
+        }
+      }
+      if (!Description.empty())
+        break;
+    }
+
+    if (!Description.empty())
+      break;
+  }
+
+  // If there's not a program specific string, use the default one.
+  if (Description.empty())
+    Description = getRSTStringWithTextFallback(R, "DocBrief", "HelpText");
 
   if (!isa<UnsetInit>(R->getValueInit("Values"))) {
     if (!Description.empty() && Description.back() != '.')
diff --git a/flang/include/flang/Evaluate/tools.h b/flang/include/flang/Evaluate/tools.h
index 8c872a0579c8..29065c9f01ec 100644
--- a/flang/include/flang/Evaluate/tools.h
+++ b/flang/include/flang/Evaluate/tools.h
@@ -19,6 +19,7 @@
 #include "flang/Evaluate/type.h"
 #include "flang/Parser/message.h"
 #include "flang/Semantics/attr.h"
+#include "flang/Semantics/scope.h"
 #include "flang/Semantics/symbol.h"
 #include <array>
 #include <optional>
@@ -1240,6 +1241,35 @@ inline bool HasCUDAAttrs(const Expr<SomeType> &expr) {
   return false;
 }
 
+/// Check if the expression is a mix of host and device variables that require
+/// implicit data transfer.
+inline bool HasCUDAImplicitTransfer(const Expr<SomeType> &expr) {
+  unsigned hostSymbols{0};
+  unsigned deviceSymbols{0};
+  for (const Symbol &sym : CollectSymbols(expr)) {
+    if (const auto *details =
+            sym.GetUltimate().detailsIf<semantics::ObjectEntityDetails>()) {
+      if (details->cudaDataAttr()) {
+        ++deviceSymbols;
+      } else {
+        if (sym.owner().IsDerivedType()) {
+          if (const auto *details =
+                  sym.owner()
+                      .GetSymbol()
+                      ->GetUltimate()
+                      .detailsIf<semantics::ObjectEntityDetails>()) {
+            if (details->cudaDataAttr()) {
+              ++deviceSymbols;
+            }
+          }
+        }
+        ++hostSymbols;
+      }
+    }
+  }
+  return hostSymbols > 0 && deviceSymbols > 0;
+}
+
 } // namespace Fortran::evaluate
 
 namespace Fortran::semantics {
diff --git a/flang/include/flang/Optimizer/Dialect/FIROpsSupport.h b/flang/include/flang/Optimizer/Dialect/FIROpsSupport.h
index f29e44504acb..3266ea3aa7fd 100644
--- a/flang/include/flang/Optimizer/Dialect/FIROpsSupport.h
+++ b/flang/include/flang/Optimizer/Dialect/FIROpsSupport.h
@@ -155,13 +155,7 @@ bool valueMayHaveFirAttributes(mlir::Value value,
 bool anyFuncArgsHaveAttr(mlir::func::FuncOp func, llvm::StringRef attr);
 
 /// Unwrap integer constant from an mlir::Value.
-inline std::optional<std::int64_t> getIntIfConstant(mlir::Value value) {
-  if (auto *definingOp = value.getDefiningOp())
-    if (auto cst = mlir::dyn_cast<mlir::arith::ConstantOp>(definingOp))
-      if (auto intAttr = cst.getValue().dyn_cast<mlir::IntegerAttr>())
-        return intAttr.getInt();
-  return {};
-}
+std::optional<std::int64_t> getIntIfConstant(mlir::Value value);
 
 static constexpr llvm::StringRef getAdaptToByRefAttrName() {
   return "adapt.valuebyref";
diff --git a/flang/include/flang/Runtime/io-api.h b/flang/include/flang/Runtime/io-api.h
index 1b6c4f5d6a65..328afc715a3f 100644
--- a/flang/include/flang/Runtime/io-api.h
+++ b/flang/include/flang/Runtime/io-api.h
@@ -92,18 +92,18 @@ constexpr std::size_t RecommendedInternalIoScratchAreaBytes(
 
 // Internal I/O to/from character arrays &/or non-default-kind character
 // requires a descriptor, which is copied.
-Cookie IONAME(BeginInternalArrayListOutput)(const Descriptor &,
+Cookie IODECL(BeginInternalArrayListOutput)(const Descriptor &,
     void **scratchArea = nullptr, std::size_t scratchBytes = 0,
     const char *sourceFile = nullptr, int sourceLine = 0);
-Cookie IONAME(BeginInternalArrayListInput)(const Descriptor &,
+Cookie IODECL(BeginInternalArrayListInput)(const Descriptor &,
     void **scratchArea = nullptr, std::size_t scratchBytes = 0,
     const char *sourceFile = nullptr, int sourceLine = 0);
-Cookie IONAME(BeginInternalArrayFormattedOutput)(const Descriptor &,
+Cookie IODECL(BeginInternalArrayFormattedOutput)(const Descriptor &,
     const char *format, std::size_t formatLength,
     const Descriptor *formatDescriptor = nullptr, void **scratchArea = nullptr,
     std::size_t scratchBytes = 0, const char *sourceFile = nullptr,
     int sourceLine = 0);
-Cookie IONAME(BeginInternalArrayFormattedInput)(const Descriptor &,
+Cookie IODECL(BeginInternalArrayFormattedInput)(const Descriptor &,
     const char *format, std::size_t formatLength,
     const Descriptor *formatDescriptor = nullptr, void **scratchArea = nullptr,
     std::size_t scratchBytes = 0, const char *sourceFile = nullptr,
@@ -111,20 +111,20 @@ Cookie IONAME(BeginInternalArrayFormattedInput)(const Descriptor &,
 
 // Internal I/O to/from a default-kind character scalar can avoid a
 // descriptor.
-Cookie IONAME(BeginInternalListOutput)(char *internal,
+Cookie IODECL(BeginInternalListOutput)(char *internal,
     std::size_t internalLength, void **scratchArea = nullptr,
     std::size_t scratchBytes = 0, const char *sourceFile = nullptr,
     int sourceLine = 0);
-Cookie IONAME(BeginInternalListInput)(const char *internal,
+Cookie IODECL(BeginInternalListInput)(const char *internal,
     std::size_t internalLength, void **scratchArea = nullptr,
     std::size_t scratchBytes = 0, const char *sourceFile = nullptr,
     int sourceLine = 0);
-Cookie IONAME(BeginInternalFormattedOutput)(char *internal,
+Cookie IODECL(BeginInternalFormattedOutput)(char *internal,
     std::size_t internalLength, const char *format, std::size_t formatLength,
     const Descriptor *formatDescriptor = nullptr, void **scratchArea = nullptr,
     std::size_t scratchBytes = 0, const char *sourceFile = nullptr,
     int sourceLine = 0);
-Cookie IONAME(BeginInternalFormattedInput)(const char *internal,
+Cookie IODECL(BeginInternalFormattedInput)(const char *internal,
     std::size_t internalLength, const char *format, std::size_t formatLength,
     const Descriptor *formatDescriptor = nullptr, void **scratchArea = nullptr,
     std::size_t scratchBytes = 0, const char *sourceFile = nullptr,
@@ -139,63 +139,63 @@ Cookie IONAME(BeginInternalFormattedInput)(const char *internal,
 // If handleError is false, and the unit number is out of range, the program
 // will be terminated. Otherwise, if unit is out of range, a nonzero Iostat
 // code is returned and ioMsg is set if it is not a nullptr.
-enum Iostat IONAME(CheckUnitNumberInRange64)(std::int64_t unit,
+enum Iostat IODECL(CheckUnitNumberInRange64)(std::int64_t unit,
     bool handleError, char *ioMsg = nullptr, std::size_t ioMsgLength = 0,
     const char *sourceFile = nullptr, int sourceLine = 0);
-enum Iostat IONAME(CheckUnitNumberInRange128)(common::int128_t unit,
+enum Iostat IODECL(CheckUnitNumberInRange128)(common::int128_t unit,
     bool handleError, char *ioMsg = nullptr, std::size_t ioMsgLength = 0,
     const char *sourceFile = nullptr, int sourceLine = 0);
 
 // External synchronous I/O initiation
 Cookie IODECL(BeginExternalListOutput)(ExternalUnit = DefaultOutputUnit,
     const char *sourceFile = nullptr, int sourceLine = 0);
-Cookie IONAME(BeginExternalListInput)(ExternalUnit = DefaultInputUnit,
+Cookie IODECL(BeginExternalListInput)(ExternalUnit = DefaultInputUnit,
     const char *sourceFile = nullptr, int sourceLine = 0);
-Cookie IONAME(BeginExternalFormattedOutput)(const char *format, std::size_t,
+Cookie IODECL(BeginExternalFormattedOutput)(const char *format, std::size_t,
     const Descriptor *formatDescriptor = nullptr,
     ExternalUnit = DefaultOutputUnit, const char *sourceFile = nullptr,
     int sourceLine = 0);
-Cookie IONAME(BeginExternalFormattedInput)(const char *format, std::size_t,
+Cookie IODECL(BeginExternalFormattedInput)(const char *format, std::size_t,
     const Descriptor *formatDescriptor = nullptr,
     ExternalUnit = DefaultInputUnit, const char *sourceFile = nullptr,
     int sourceLine = 0);
-Cookie IONAME(BeginUnformattedOutput)(ExternalUnit = DefaultOutputUnit,
+Cookie IODECL(BeginUnformattedOutput)(ExternalUnit = DefaultOutputUnit,
     const char *sourceFile = nullptr, int sourceLine = 0);
-Cookie IONAME(BeginUnformattedInput)(ExternalUnit = DefaultInputUnit,
+Cookie IODECL(BeginUnformattedInput)(ExternalUnit = DefaultInputUnit,
     const char *sourceFile = nullptr, int sourceLine = 0);
 
 // WAIT(ID=)
-Cookie IONAME(BeginWait)(ExternalUnit, AsynchronousId,
+Cookie IODECL(BeginWait)(ExternalUnit, AsynchronousId,
     const char *sourceFile = nullptr, int sourceLine = 0);
 // WAIT(no ID=)
-Cookie IONAME(BeginWaitAll)(
+Cookie IODECL(BeginWaitAll)(
     ExternalUnit, const char *sourceFile = nullptr, int sourceLine = 0);
 
 // Other I/O statements
-Cookie IONAME(BeginClose)(
+Cookie IODECL(BeginClose)(
     ExternalUnit, const char *sourceFile = nullptr, int sourceLine = 0);
-Cookie IONAME(BeginFlush)(
+Cookie IODECL(BeginFlush)(
     ExternalUnit, const char *sourceFile = nullptr, int sourceLine = 0);
-Cookie IONAME(BeginBackspace)(
+Cookie IODECL(BeginBackspace)(
     ExternalUnit, const char *sourceFile = nullptr, int sourceLine = 0);
-Cookie IONAME(BeginEndfile)(
+Cookie IODECL(BeginEndfile)(
     ExternalUnit, const char *sourceFile = nullptr, int sourceLine = 0);
-Cookie IONAME(BeginRewind)(
+Cookie IODECL(BeginRewind)(
     ExternalUnit, const char *sourceFile = nullptr, int sourceLine = 0);
 
 // OPEN(UNIT=) and OPEN(NEWUNIT=) have distinct interfaces.
-Cookie IONAME(BeginOpenUnit)(
+Cookie IODECL(BeginOpenUnit)(
     ExternalUnit, const char *sourceFile = nullptr, int sourceLine = 0);
-Cookie IONAME(BeginOpenNewUnit)(
+Cookie IODECL(BeginOpenNewUnit)(
     const char *sourceFile = nullptr, int sourceLine = 0);
 
 // The variant forms of INQUIRE() statements have distinct interfaces.
 // BeginInquireIoLength() is basically a no-op output statement.
-Cookie IONAME(BeginInquireUnit)(
+Cookie IODECL(BeginInquireUnit)(
     ExternalUnit, const char *sourceFile = nullptr, int sourceLine = 0);
-Cookie IONAME(BeginInquireFile)(const char *, std::size_t,
+Cookie IODECL(BeginInquireFile)(const char *, std::size_t,
     const char *sourceFile = nullptr, int sourceLine = 0);
-Cookie IONAME(BeginInquireIoLength)(
+Cookie IODECL(BeginInquireIoLength)(
     const char *sourceFile = nullptr, int sourceLine = 0);
 
 // If an I/O statement has any IOSTAT=, ERR=, END=, or EOR= specifiers,
@@ -214,33 +214,33 @@ Cookie IONAME(BeginInquireIoLength)(
 //     }
 //   }
 //   if (EndIoStatement(cookie) == FORTRAN_RUTIME_IOSTAT_END) goto label666;
-void IONAME(EnableHandlers)(Cookie, bool hasIoStat = false, bool hasErr = false,
+void IODECL(EnableHandlers)(Cookie, bool hasIoStat = false, bool hasErr = false,
     bool hasEnd = false, bool hasEor = false, bool hasIoMsg = false);
 
 // ASYNCHRONOUS='YES' or 'NO' on READ/WRITE/OPEN
 // Use GetAsynchronousId() to handle ID=.
-bool IONAME(SetAsynchronous)(Cookie, const char *, std::size_t);
+bool IODECL(SetAsynchronous)(Cookie, const char *, std::size_t);
 
 // Control list options.  These return false on a error that the
 // Begin...() call has specified will be handled by the caller.
 // The interfaces that pass a default-kind CHARACTER argument
 // are limited to passing specific case-insensitive keyword values.
 // ADVANCE=YES, NO
-bool IONAME(SetAdvance)(Cookie, const char *, std::size_t);
+bool IODECL(SetAdvance)(Cookie, const char *, std::size_t);
 // BLANK=NULL, ZERO
-bool IONAME(SetBlank)(Cookie, const char *, std::size_t);
+bool IODECL(SetBlank)(Cookie, const char *, std::size_t);
 // DECIMAL=COMMA, POINT
-bool IONAME(SetDecimal)(Cookie, const char *, std::size_t);
+bool IODECL(SetDecimal)(Cookie, const char *, std::size_t);
 // DELIM=APOSTROPHE, QUOTE, NONE
-bool IONAME(SetDelim)(Cookie, const char *, std::size_t);
+bool IODECL(SetDelim)(Cookie, const char *, std::size_t);
 // PAD=YES, NO
-bool IONAME(SetPad)(Cookie, const char *, std::size_t);
-bool IONAME(SetPos)(Cookie, std::int64_t);
-bool IONAME(SetRec)(Cookie, std::int64_t);
+bool IODECL(SetPad)(Cookie, const char *, std::size_t);
+bool IODECL(SetPos)(Cookie, std::int64_t);
+bool IODECL(SetRec)(Cookie, std::int64_t);
 // ROUND=UP, DOWN, ZERO, NEAREST, COMPATIBLE, PROCESSOR_DEFINED
-bool IONAME(SetRound)(Cookie, const char *, std::size_t);
+bool IODECL(SetRound)(Cookie, const char *, std::size_t);
 // SIGN=PLUS, SUPPRESS, PROCESSOR_DEFINED
-bool IONAME(SetSign)(Cookie, const char *, std::size_t);
+bool IODECL(SetSign)(Cookie, const char *, std::size_t);
 
 // Data item transfer for modes other than NAMELIST:
 // Any data object that can be passed as an actual argument without the
@@ -256,34 +256,34 @@ bool IONAME(SetSign)(Cookie, const char *, std::size_t);
 // Once the statement has encountered an error, all following items will be
 // ignored and also return false; but compiled code should check for errors
 // and avoid the following items when they might crash.
-bool IONAME(OutputDescriptor)(Cookie, const Descriptor &);
-bool IONAME(InputDescriptor)(Cookie, const Descriptor &);
+bool IODECL(OutputDescriptor)(Cookie, const Descriptor &);
+bool IODECL(InputDescriptor)(Cookie, const Descriptor &);
 // Formatted (including list directed) I/O data items
-bool IONAME(OutputInteger8)(Cookie, std::int8_t);
-bool IONAME(OutputInteger16)(Cookie, std::int16_t);
+bool IODECL(OutputInteger8)(Cookie, std::int8_t);
+bool IODECL(OutputInteger16)(Cookie, std::int16_t);
 bool IODECL(OutputInteger32)(Cookie, std::int32_t);
-bool IONAME(OutputInteger64)(Cookie, std::int64_t);
-bool IONAME(OutputInteger128)(Cookie, common::int128_t);
-bool IONAME(InputInteger)(Cookie, std::int64_t &, int kind = 8);
-bool IONAME(OutputReal32)(Cookie, float);
-bool IONAME(InputReal32)(Cookie, float &);
-bool IONAME(OutputReal64)(Cookie, double);
-bool IONAME(InputReal64)(Cookie, double &);
-bool IONAME(OutputComplex32)(Cookie, float, float);
-bool IONAME(InputComplex32)(Cookie, float[2]);
-bool IONAME(OutputComplex64)(Cookie, double, double);
-bool IONAME(InputComplex64)(Cookie, double[2]);
-bool IONAME(OutputCharacter)(Cookie, const char *, std::size_t, int kind = 1);
-bool IONAME(OutputAscii)(Cookie, const char *, std::size_t);
-bool IONAME(InputCharacter)(Cookie, char *, std::size_t, int kind = 1);
-bool IONAME(InputAscii)(Cookie, char *, std::size_t);
-bool IONAME(OutputLogical)(Cookie, bool);
-bool IONAME(InputLogical)(Cookie, bool &);
+bool IODECL(OutputInteger64)(Cookie, std::int64_t);
+bool IODECL(OutputInteger128)(Cookie, common::int128_t);
+bool IODECL(InputInteger)(Cookie, std::int64_t &, int kind = 8);
+bool IODECL(OutputReal32)(Cookie, float);
+bool IODECL(InputReal32)(Cookie, float &);
+bool IODECL(OutputReal64)(Cookie, double);
+bool IODECL(InputReal64)(Cookie, double &);
+bool IODECL(OutputComplex32)(Cookie, float, float);
+bool IODECL(InputComplex32)(Cookie, float[2]);
+bool IODECL(OutputComplex64)(Cookie, double, double);
+bool IODECL(InputComplex64)(Cookie, double[2]);
+bool IODECL(OutputCharacter)(Cookie, const char *, std::size_t, int kind = 1);
+bool IODECL(OutputAscii)(Cookie, const char *, std::size_t);
+bool IODECL(InputCharacter)(Cookie, char *, std::size_t, int kind = 1);
+bool IODECL(InputAscii)(Cookie, char *, std::size_t);
+bool IODECL(OutputLogical)(Cookie, bool);
+bool IODECL(InputLogical)(Cookie, bool &);
 
 // NAMELIST I/O must be the only data item in an (otherwise)
 // list-directed I/O statement.
-bool IONAME(OutputNamelist)(Cookie, const NamelistGroup &);
-bool IONAME(InputNamelist)(Cookie, const NamelistGroup &);
+bool IODECL(OutputNamelist)(Cookie, const NamelistGroup &);
+bool IODECL(InputNamelist)(Cookie, const NamelistGroup &);
 
 // When an I/O list item has a derived type with a specific defined
 // I/O subroutine of the appropriate generic kind for the active
@@ -294,9 +294,9 @@ bool IONAME(InputNamelist)(Cookie, const NamelistGroup &);
 // made such a generic interface inaccessible), these data item transfer
 // APIs enable the I/O runtime to make the right calls to defined I/O
 // subroutines.
-bool IONAME(OutputDerivedType)(
+bool IODECL(OutputDerivedType)(
     Cookie, const Descriptor &, const NonTbpDefinedIoTable *);
-bool IONAME(InputDerivedType)(
+bool IODECL(InputDerivedType)(
     Cookie, const Descriptor &, const NonTbpDefinedIoTable *);
 
 // Additional specifier interfaces for the connection-list of
@@ -304,56 +304,56 @@ bool IONAME(InputDerivedType)(
 // SetDelim(), GetIoMsg(), SetPad(), SetRound(), SetSign(),
 // & SetAsynchronous() are also acceptable for OPEN.
 // ACCESS=SEQUENTIAL, DIRECT, STREAM
-bool IONAME(SetAccess)(Cookie, const char *, std::size_t);
+bool IODECL(SetAccess)(Cookie, const char *, std::size_t);
 // ACTION=READ, WRITE, or READWRITE
-bool IONAME(SetAction)(Cookie, const char *, std::size_t);
+bool IODECL(SetAction)(Cookie, const char *, std::size_t);
 // CARRIAGECONTROL=LIST, FORTRAN, NONE
-bool IONAME(SetCarriagecontrol)(Cookie, const char *, std::size_t);
+bool IODECL(SetCarriagecontrol)(Cookie, const char *, std::size_t);
 // CONVERT=NATIVE, LITTLE_ENDIAN, BIG_ENDIAN, or SWAP
-bool IONAME(SetConvert)(Cookie, const char *, std::size_t);
+bool IODECL(SetConvert)(Cookie, const char *, std::size_t);
 // ENCODING=UTF-8, DEFAULT
-bool IONAME(SetEncoding)(Cookie, const char *, std::size_t);
+bool IODECL(SetEncoding)(Cookie, const char *, std::size_t);
 // FORM=FORMATTED, UNFORMATTED
-bool IONAME(SetForm)(Cookie, const char *, std::size_t);
+bool IODECL(SetForm)(Cookie, const char *, std::size_t);
 // POSITION=ASIS, REWIND, APPEND
-bool IONAME(SetPosition)(Cookie, const char *, std::size_t);
-bool IONAME(SetRecl)(Cookie, std::size_t); // RECL=
+bool IODECL(SetPosition)(Cookie, const char *, std::size_t);
+bool IODECL(SetRecl)(Cookie, std::size_t); // RECL=
 
 // STATUS can be set during an OPEN or CLOSE statement.
 // For OPEN: STATUS=OLD, NEW, SCRATCH, REPLACE, UNKNOWN
 // For CLOSE: STATUS=KEEP, DELETE
-bool IONAME(SetStatus)(Cookie, const char *, std::size_t);
+bool IODECL(SetStatus)(Cookie, const char *, std::size_t);
 
-bool IONAME(SetFile)(Cookie, const char *, std::size_t chars);
+bool IODECL(SetFile)(Cookie, const char *, std::size_t chars);
 
 // Acquires the runtime-created unit number for OPEN(NEWUNIT=)
-bool IONAME(GetNewUnit)(Cookie, int &, int kind = 4);
+bool IODECL(GetNewUnit)(Cookie, int &, int kind = 4);
 
 // READ(SIZE=), after all input items
-std::size_t IONAME(GetSize)(Cookie);
+std::size_t IODECL(GetSize)(Cookie);
 
 // INQUIRE(IOLENGTH=), after all output items
-std::size_t IONAME(GetIoLength)(Cookie);
+std::size_t IODECL(GetIoLength)(Cookie);
 
 // GetIoMsg() does not modify its argument unless an error or
 // end-of-record/file condition is present.
-void IONAME(GetIoMsg)(Cookie, char *, std::size_t); // IOMSG=
+void IODECL(GetIoMsg)(Cookie, char *, std::size_t); // IOMSG=
 
 // Defines ID= on READ/WRITE(ASYNCHRONOUS='YES')
-AsynchronousId IONAME(GetAsynchronousId)(Cookie);
+AsynchronousId IODECL(GetAsynchronousId)(Cookie);
 
 // INQUIRE() specifiers are mostly identified by their NUL-terminated
 // case-insensitive names.
 // ACCESS, ACTION, ASYNCHRONOUS, BLANK, CONVERT, DECIMAL, DELIM, DIRECT,
 // ENCODING, FORM, FORMATTED, NAME, PAD, POSITION, READ, READWRITE, ROUND,
 // SEQUENTIAL, SIGN, STREAM, UNFORMATTED, WRITE:
-bool IONAME(InquireCharacter)(Cookie, InquiryKeywordHash, char *, std::size_t);
+bool IODECL(InquireCharacter)(Cookie, InquiryKeywordHash, char *, std::size_t);
 // EXIST, NAMED, OPENED, and PENDING (without ID):
-bool IONAME(InquireLogical)(Cookie, InquiryKeywordHash, bool &);
+bool IODECL(InquireLogical)(Cookie, InquiryKeywordHash, bool &);
 // PENDING with ID
-bool IONAME(InquirePendingId)(Cookie, AsynchronousId, bool &);
+bool IODECL(InquirePendingId)(Cookie, AsynchronousId, bool &);
 // NEXTREC, NUMBER, POS, RECL, SIZE
-bool IONAME(InquireInteger64)(
+bool IODECL(InquireInteger64)(
     Cookie, InquiryKeywordHash, std::int64_t &, int kind = 8);
 
 // This function must be called to end an I/O statement, and its
diff --git a/flang/lib/Lower/Bridge.cpp b/flang/lib/Lower/Bridge.cpp
index 5bba0978617c..478c8f4c17ec 100644
--- a/flang/lib/Lower/Bridge.cpp
+++ b/flang/lib/Lower/Bridge.cpp
@@ -3710,16 +3710,18 @@ private:
     return false;
   }
 
-  static void genCUDADataTransfer(fir::FirOpBuilder &builder,
-                                  mlir::Location loc, bool lhsIsDevice,
-                                  hlfir::Entity &lhs, bool rhsIsDevice,
-                                  hlfir::Entity &rhs) {
+  void genCUDADataTransfer(fir::FirOpBuilder &builder, mlir::Location loc,
+                           const Fortran::evaluate::Assignment &assign,
+                           hlfir::Entity &lhs, hlfir::Entity &rhs) {
+    bool lhsIsDevice = Fortran::evaluate::HasCUDAAttrs(assign.lhs);
+    bool rhsIsDevice = Fortran::evaluate::HasCUDAAttrs(assign.rhs);
     if (rhs.isBoxAddressOrValue() || lhs.isBoxAddressOrValue())
       TODO(loc, "CUDA data transfler with descriptors");
+
+    // device = host
     if (lhsIsDevice && !rhsIsDevice) {
       auto transferKindAttr = fir::CUDADataTransferKindAttr::get(
           builder.getContext(), fir::CUDADataTransferKind::HostDevice);
-      // device = host
       if (!rhs.isVariable()) {
         auto associate = hlfir::genAssociateExpr(
             loc, builder, rhs, rhs.getType(), ".cuf_host_tmp");
@@ -3732,7 +3734,73 @@ private:
       }
       return;
     }
-    TODO(loc, "Assignement with CUDA Fortran variables");
+
+    // host = device
+    if (!lhsIsDevice && rhsIsDevice) {
+      auto transferKindAttr = fir::CUDADataTransferKindAttr::get(
+          builder.getContext(), fir::CUDADataTransferKind::DeviceHost);
+      if (!rhs.isVariable()) {
+        // evaluateRhs loads scalar. Look for the memory reference to be used in
+        // the transfer.
+        if (mlir::isa_and_nonnull<fir::LoadOp>(rhs.getDefiningOp())) {
+          auto loadOp = mlir::dyn_cast<fir::LoadOp>(rhs.getDefiningOp());
+          builder.create<fir::CUDADataTransferOp>(loc, loadOp.getMemref(), lhs,
+                                                  transferKindAttr);
+          return;
+        }
+      } else {
+        builder.create<fir::CUDADataTransferOp>(loc, rhs, lhs,
+                                                transferKindAttr);
+      }
+      return;
+    }
+
+    if (lhsIsDevice && rhsIsDevice) {
+      assert(rhs.isVariable() && "CUDA Fortran assignment rhs is not legal");
+      auto transferKindAttr = fir::CUDADataTransferKindAttr::get(
+          builder.getContext(), fir::CUDADataTransferKind::DeviceDevice);
+      builder.create<fir::CUDADataTransferOp>(loc, rhs, lhs, transferKindAttr);
+      return;
+    }
+    llvm_unreachable("Unhandled CUDA data transfer");
+  }
+
+  llvm::SmallVector<mlir::Value>
+  genCUDAImplicitDataTransfer(fir::FirOpBuilder &builder, mlir::Location loc,
+                              const Fortran::evaluate::Assignment &assign) {
+    llvm::SmallVector<mlir::Value> temps;
+    localSymbols.pushScope();
+    auto transferKindAttr = fir::CUDADataTransferKindAttr::get(
+        builder.getContext(), fir::CUDADataTransferKind::DeviceHost);
+    unsigned nbDeviceResidentObject = 0;
+    for (const Fortran::semantics::Symbol &sym :
+         Fortran::evaluate::CollectSymbols(assign.rhs)) {
+      if (const auto *details =
+              sym.GetUltimate()
+                  .detailsIf<Fortran::semantics::ObjectEntityDetails>()) {
+        if (details->cudaDataAttr()) {
+          if (sym.owner().IsDerivedType() && IsAllocatable(sym.GetUltimate()))
+            TODO(loc, "Device resident allocatable derived-type component");
+          // TODO: This should probably being checked in semantic and give a
+          // proper error.
+          assert(
+              nbDeviceResidentObject <= 1 &&
+              "Only one reference to the device resident object is supported");
+          auto addr = getSymbolAddress(sym);
+          hlfir::Entity entity{addr};
+          auto [temp, cleanup] =
+              hlfir::createTempFromMold(loc, builder, entity);
+          auto needCleanup = fir::getIntIfConstant(cleanup);
+          if (needCleanup && *needCleanup)
+            temps.push_back(temp);
+          addSymbol(sym, temp, /*forced=*/true);
+          builder.create<fir::CUDADataTransferOp>(loc, addr, temp,
+                                                  transferKindAttr);
+          ++nbDeviceResidentObject;
+        }
+      }
+    }
+    return temps;
   }
 
   void genDataAssignment(
@@ -3741,8 +3809,13 @@ private:
     mlir::Location loc = getCurrentLocation();
     fir::FirOpBuilder &builder = getFirOpBuilder();
 
-    bool lhsIsDevice = Fortran::evaluate::HasCUDAAttrs(assign.lhs);
-    bool rhsIsDevice = Fortran::evaluate::HasCUDAAttrs(assign.rhs);
+    bool isCUDATransfer = Fortran::evaluate::HasCUDAAttrs(assign.lhs) ||
+                          Fortran::evaluate::HasCUDAAttrs(assign.rhs);
+    bool hasCUDAImplicitTransfer =
+        Fortran::evaluate::HasCUDAImplicitTransfer(assign.rhs);
+    llvm::SmallVector<mlir::Value> implicitTemps;
+    if (hasCUDAImplicitTransfer)
+      implicitTemps = genCUDAImplicitDataTransfer(builder, loc, assign);
 
     // Gather some information about the assignment that will impact how it is
     // lowered.
@@ -3800,12 +3873,16 @@ private:
       Fortran::lower::StatementContext localStmtCtx;
       hlfir::Entity rhs = evaluateRhs(localStmtCtx);
       hlfir::Entity lhs = evaluateLhs(localStmtCtx);
-      if (lhsIsDevice || rhsIsDevice) {
-        genCUDADataTransfer(builder, loc, lhsIsDevice, lhs, rhsIsDevice, rhs);
-      } else {
+      if (isCUDATransfer && !hasCUDAImplicitTransfer)
+        genCUDADataTransfer(builder, loc, assign, lhs, rhs);
+      else
         builder.create<hlfir::AssignOp>(loc, rhs, lhs,
                                         isWholeAllocatableAssignment,
                                         keepLhsLengthInAllocatableAssignment);
+      if (hasCUDAImplicitTransfer) {
+        localSymbols.popScope();
+        for (mlir::Value temp : implicitTemps)
+          builder.create<fir::FreeMemOp>(loc, temp);
       }
       return;
     }
diff --git a/flang/lib/Optimizer/CodeGen/CodeGen.cpp b/flang/lib/Optimizer/CodeGen/CodeGen.cpp
index 06ce84f1543a..d909bda89cde 100644
--- a/flang/lib/Optimizer/CodeGen/CodeGen.cpp
+++ b/flang/lib/Optimizer/CodeGen/CodeGen.cpp
@@ -92,27 +92,10 @@ static mlir::Block *createBlock(mlir::ConversionPatternRewriter &rewriter,
                               mlir::Region::iterator(insertBefore));
 }
 
-/// Extract constant from a value if it is a result of one of the
-/// ConstantOp operations, otherwise, return std::nullopt.
-static std::optional<int64_t> getIfConstantIntValue(mlir::Value val) {
-  if (!val || !val.dyn_cast<mlir::OpResult>())
-    return {};
-
-  mlir::Operation *defop = val.getDefiningOp();
-
-  if (auto constOp = mlir::dyn_cast<mlir::arith::ConstantIntOp>(defop))
-    return constOp.value();
-  if (auto llConstOp = mlir::dyn_cast<mlir::LLVM::ConstantOp>(defop))
-    if (auto attr = llConstOp.getValue().dyn_cast<mlir::IntegerAttr>())
-      return attr.getValue().getSExtValue();
-
-  return {};
-}
-
 /// Extract constant from a value that must be the result of one of the
 /// ConstantOp operations.
 static int64_t getConstantIntValue(mlir::Value val) {
-  if (auto constVal = getIfConstantIntValue(val))
+  if (auto constVal = fir::getIntIfConstant(val))
     return *constVal;
   fir::emitFatalError(val.getLoc(), "must be a constant");
 }
@@ -664,7 +647,7 @@ struct ConvertOpConversion : public fir::FIROpConversion<fir::ConvertOp> {
                << " -> " << toTy;
 
       // Do folding for constant inputs.
-      if (auto constVal = getIfConstantIntValue(op0)) {
+      if (auto constVal = fir::getIntIfConstant(op0)) {
         mlir::Value normVal =
             genConstantIndex(loc, toTy, rewriter, *constVal ? 1 : 0);
         rewriter.replaceOp(convert, normVal);
diff --git a/flang/lib/Optimizer/Dialect/FIROps.cpp b/flang/lib/Optimizer/Dialect/FIROps.cpp
index dba2c30d1851..8ab74103cb6a 100644
--- a/flang/lib/Optimizer/Dialect/FIROps.cpp
+++ b/flang/lib/Optimizer/Dialect/FIROps.cpp
@@ -3834,6 +3834,18 @@ bool fir::anyFuncArgsHaveAttr(mlir::func::FuncOp func, llvm::StringRef attr) {
   return false;
 }
 
+std::optional<std::int64_t> fir::getIntIfConstant(mlir::Value value) {
+  if (auto *definingOp = value.getDefiningOp()) {
+    if (auto cst = mlir::dyn_cast<mlir::arith::ConstantOp>(definingOp))
+      if (auto intAttr = cst.getValue().dyn_cast<mlir::IntegerAttr>())
+        return intAttr.getInt();
+    if (auto llConstOp = mlir::dyn_cast<mlir::LLVM::ConstantOp>(definingOp))
+      if (auto attr = llConstOp.getValue().dyn_cast<mlir::IntegerAttr>())
+        return attr.getValue().getSExtValue();
+  }
+  return {};
+}
+
 mlir::Type fir::applyPathToType(mlir::Type eleTy, mlir::ValueRange path) {
   for (auto i = path.begin(), end = path.end(); eleTy && i < end;) {
     eleTy = llvm::TypeSwitch<mlir::Type, mlir::Type>(eleTy)
diff --git a/flang/runtime/environment.cpp b/flang/runtime/environment.cpp
index b74067a37777..b2c9665a28df 100644
--- a/flang/runtime/environment.cpp
+++ b/flang/runtime/environment.cpp
@@ -49,6 +49,7 @@ static void SetEnvironmentDefaults(const EnvironmentDefaultList *envDefaults) {
   }
 }
 
+RT_OFFLOAD_API_GROUP_BEGIN
 Fortran::common::optional<Convert> GetConvertFromString(
     const char *x, std::size_t n) {
   static const char *keywords[]{
@@ -68,6 +69,7 @@ Fortran::common::optional<Convert> GetConvertFromString(
     return Fortran::common::nullopt;
   }
 }
+RT_OFFLOAD_API_GROUP_END
 
 void ExecutionEnvironment::Configure(int ac, const char *av[],
     const char *env[], const EnvironmentDefaultList *envDefaults) {
diff --git a/flang/runtime/environment.h b/flang/runtime/environment.h
index 6c56993fb1d6..b8b9f10e4e57 100644
--- a/flang/runtime/environment.h
+++ b/flang/runtime/environment.h
@@ -31,7 +31,7 @@ RT_OFFLOAD_VAR_GROUP_END
 // External unformatted I/O data conversions
 enum class Convert { Unknown, Native, LittleEndian, BigEndian, Swap };
 
-Fortran::common::optional<Convert> GetConvertFromString(
+RT_API_ATTRS Fortran::common::optional<Convert> GetConvertFromString(
     const char *, std::size_t);
 
 struct ExecutionEnvironment {
diff --git a/flang/runtime/freestanding-tools.h b/flang/runtime/freestanding-tools.h
index 451bf13b9fa6..9089dc6bcf53 100644
--- a/flang/runtime/freestanding-tools.h
+++ b/flang/runtime/freestanding-tools.h
@@ -52,6 +52,11 @@
 #define STD_STRCPY_UNSUPPORTED 1
 #endif
 
+#if !defined(STD_STRCMP_UNSUPPORTED) && \
+    (defined(__CUDACC__) || defined(__CUDA__)) && defined(__CUDA_ARCH__)
+#define STD_STRCMP_UNSUPPORTED 1
+#endif
+
 namespace Fortran::runtime {
 
 #if STD_FILL_N_UNSUPPORTED
@@ -176,5 +181,19 @@ static inline RT_API_ATTRS char *strcpy(char *dest, const char *src) {
 using std::strcpy;
 #endif // !STD_STRCPY_UNSUPPORTED
 
+#if STD_STRCMP_UNSUPPORTED
+// Provides alternative implementation for std::strcmp(), if
+// it is not supported.
+static inline RT_API_ATTRS int strcmp(const char *lhs, const char *rhs) {
+  while (*lhs != '\0' && *lhs == *rhs) {
+    ++lhs;
+    ++rhs;
+  }
+  return static_cast<unsigned char>(*lhs) - static_cast<unsigned char>(*rhs);
+}
+#else // !STD_STRCMP_UNSUPPORTED
+using std::strcmp;
+#endif // !STD_STRCMP_UNSUPPORTED
+
 } // namespace Fortran::runtime
 #endif // FORTRAN_RUNTIME_FREESTANDING_TOOLS_H_
diff --git a/flang/runtime/io-api.cpp b/flang/runtime/io-api.cpp
index 3a86c9fa7375..ccb5b576451d 100644
--- a/flang/runtime/io-api.cpp
+++ b/flang/runtime/io-api.cpp
@@ -25,8 +25,9 @@
 #include <memory>
 
 namespace Fortran::runtime::io {
+RT_EXT_API_GROUP_BEGIN
 
-const char *InquiryKeywordHashDecode(
+RT_API_ATTRS const char *InquiryKeywordHashDecode(
     char *buffer, std::size_t n, InquiryKeywordHash hash) {
   if (n < 1) {
     return nullptr;
@@ -44,7 +45,7 @@ const char *InquiryKeywordHashDecode(
 }
 
 template <Direction DIR>
-Cookie BeginInternalArrayListIO(const Descriptor &descriptor,
+RT_API_ATTRS Cookie BeginInternalArrayListIO(const Descriptor &descriptor,
     void ** /*scratchArea*/, std::size_t /*scratchBytes*/,
     const char *sourceFile, int sourceLine) {
   Terminator oom{sourceFile, sourceLine};
@@ -54,14 +55,14 @@ Cookie BeginInternalArrayListIO(const Descriptor &descriptor,
               ->ioStatementState();
 }
 
-Cookie IONAME(BeginInternalArrayListOutput)(const Descriptor &descriptor,
+Cookie IODEF(BeginInternalArrayListOutput)(const Descriptor &descriptor,
     void **scratchArea, std::size_t scratchBytes, const char *sourceFile,
     int sourceLine) {
   return BeginInternalArrayListIO<Direction::Output>(
       descriptor, scratchArea, scratchBytes, sourceFile, sourceLine);
 }
 
-Cookie IONAME(BeginInternalArrayListInput)(const Descriptor &descriptor,
+Cookie IODEF(BeginInternalArrayListInput)(const Descriptor &descriptor,
     void **scratchArea, std::size_t scratchBytes, const char *sourceFile,
     int sourceLine) {
   return BeginInternalArrayListIO<Direction::Input>(
@@ -69,7 +70,7 @@ Cookie IONAME(BeginInternalArrayListInput)(const Descriptor &descriptor,
 }
 
 template <Direction DIR>
-Cookie BeginInternalArrayFormattedIO(const Descriptor &descriptor,
+RT_API_ATTRS Cookie BeginInternalArrayFormattedIO(const Descriptor &descriptor,
     const char *format, std::size_t formatLength,
     const Descriptor *formatDescriptor, void ** /*scratchArea*/,
     std::size_t /*scratchBytes*/, const char *sourceFile, int sourceLine) {
@@ -80,7 +81,7 @@ Cookie BeginInternalArrayFormattedIO(const Descriptor &descriptor,
               ->ioStatementState();
 }
 
-Cookie IONAME(BeginInternalArrayFormattedOutput)(const Descriptor &descriptor,
+Cookie IODEF(BeginInternalArrayFormattedOutput)(const Descriptor &descriptor,
     const char *format, std::size_t formatLength,
     const Descriptor *formatDescriptor, void **scratchArea,
     std::size_t scratchBytes, const char *sourceFile, int sourceLine) {
@@ -89,7 +90,7 @@ Cookie IONAME(BeginInternalArrayFormattedOutput)(const Descriptor &descriptor,
       sourceLine);
 }
 
-Cookie IONAME(BeginInternalArrayFormattedInput)(const Descriptor &descriptor,
+Cookie IODEF(BeginInternalArrayFormattedInput)(const Descriptor &descriptor,
     const char *format, std::size_t formatLength,
     const Descriptor *formatDescriptor, void **scratchArea,
     std::size_t scratchBytes, const char *sourceFile, int sourceLine) {
@@ -110,14 +111,14 @@ RT_API_ATTRS Cookie BeginInternalListIO(
               ->ioStatementState();
 }
 
-Cookie IONAME(BeginInternalListOutput)(char *internal,
+Cookie IODEF(BeginInternalListOutput)(char *internal,
     std::size_t internalLength, void **scratchArea, std::size_t scratchBytes,
     const char *sourceFile, int sourceLine) {
   return BeginInternalListIO<Direction::Output>(internal, internalLength,
       scratchArea, scratchBytes, sourceFile, sourceLine);
 }
 
-Cookie IONAME(BeginInternalListInput)(const char *internal,
+Cookie IODEF(BeginInternalListInput)(const char *internal,
     std::size_t internalLength, void **scratchArea, std::size_t scratchBytes,
     const char *sourceFile, int sourceLine) {
   return BeginInternalListIO<Direction::Input>(internal, internalLength,
@@ -125,7 +126,7 @@ Cookie IONAME(BeginInternalListInput)(const char *internal,
 }
 
 template <Direction DIR>
-Cookie BeginInternalFormattedIO(
+RT_API_ATTRS Cookie BeginInternalFormattedIO(
     std::conditional_t<DIR == Direction::Input, const char, char> *internal,
     std::size_t internalLength, const char *format, std::size_t formatLength,
     const Descriptor *formatDescriptor, void ** /*scratchArea*/,
@@ -138,7 +139,7 @@ Cookie BeginInternalFormattedIO(
               ->ioStatementState();
 }
 
-Cookie IONAME(BeginInternalFormattedOutput)(char *internal,
+Cookie IODEF(BeginInternalFormattedOutput)(char *internal,
     std::size_t internalLength, const char *format, std::size_t formatLength,
     const Descriptor *formatDescriptor, void **scratchArea,
     std::size_t scratchBytes, const char *sourceFile, int sourceLine) {
@@ -147,7 +148,7 @@ Cookie IONAME(BeginInternalFormattedOutput)(char *internal,
       sourceFile, sourceLine);
 }
 
-Cookie IONAME(BeginInternalFormattedInput)(const char *internal,
+Cookie IODEF(BeginInternalFormattedInput)(const char *internal,
     std::size_t internalLength, const char *format, std::size_t formatLength,
     const Descriptor *formatDescriptor, void **scratchArea,
     std::size_t scratchBytes, const char *sourceFile, int sourceLine) {
@@ -227,24 +228,22 @@ RT_API_ATTRS Cookie BeginExternalListIO(
   }
 }
 
-RT_EXT_API_GROUP_BEGIN
 Cookie IODEF(BeginExternalListOutput)(
     ExternalUnit unitNumber, const char *sourceFile, int sourceLine) {
   return BeginExternalListIO<Direction::Output, ExternalListIoStatementState>(
       unitNumber, sourceFile, sourceLine);
 }
-RT_EXT_API_GROUP_END
 
-Cookie IONAME(BeginExternalListInput)(
+Cookie IODEF(BeginExternalListInput)(
     ExternalUnit unitNumber, const char *sourceFile, int sourceLine) {
   return BeginExternalListIO<Direction::Input, ExternalListIoStatementState>(
       unitNumber, sourceFile, sourceLine);
 }
 
 template <Direction DIR>
-Cookie BeginExternalFormattedIO(const char *format, std::size_t formatLength,
-    const Descriptor *formatDescriptor, ExternalUnit unitNumber,
-    const char *sourceFile, int sourceLine) {
+RT_API_ATTRS Cookie BeginExternalFormattedIO(const char *format,
+    std::size_t formatLength, const Descriptor *formatDescriptor,
+    ExternalUnit unitNumber, const char *sourceFile, int sourceLine) {
   Terminator terminator{sourceFile, sourceLine};
   Cookie errorCookie{nullptr};
   ExternalFileUnit *unit{GetOrCreateUnit(
@@ -286,14 +285,14 @@ Cookie BeginExternalFormattedIO(const char *format, std::size_t formatLength,
   }
 }
 
-Cookie IONAME(BeginExternalFormattedOutput)(const char *format,
+Cookie IODEF(BeginExternalFormattedOutput)(const char *format,
     std::size_t formatLength, const Descriptor *formatDescriptor,
     ExternalUnit unitNumber, const char *sourceFile, int sourceLine) {
   return BeginExternalFormattedIO<Direction::Output>(format, formatLength,
       formatDescriptor, unitNumber, sourceFile, sourceLine);
 }
 
-Cookie IONAME(BeginExternalFormattedInput)(const char *format,
+Cookie IODEF(BeginExternalFormattedInput)(const char *format,
     std::size_t formatLength, const Descriptor *formatDescriptor,
     ExternalUnit unitNumber, const char *sourceFile, int sourceLine) {
   return BeginExternalFormattedIO<Direction::Input>(format, formatLength,
@@ -301,7 +300,7 @@ Cookie IONAME(BeginExternalFormattedInput)(const char *format,
 }
 
 template <Direction DIR>
-Cookie BeginUnformattedIO(
+RT_API_ATTRS Cookie BeginUnformattedIO(
     ExternalUnit unitNumber, const char *sourceFile, int sourceLine) {
   Terminator terminator{sourceFile, sourceLine};
   Cookie errorCookie{nullptr};
@@ -352,19 +351,19 @@ Cookie BeginUnformattedIO(
   }
 }
 
-Cookie IONAME(BeginUnformattedOutput)(
+Cookie IODEF(BeginUnformattedOutput)(
     ExternalUnit unitNumber, const char *sourceFile, int sourceLine) {
   return BeginUnformattedIO<Direction::Output>(
       unitNumber, sourceFile, sourceLine);
 }
 
-Cookie IONAME(BeginUnformattedInput)(
+Cookie IODEF(BeginUnformattedInput)(
     ExternalUnit unitNumber, const char *sourceFile, int sourceLine) {
   return BeginUnformattedIO<Direction::Input>(
       unitNumber, sourceFile, sourceLine);
 }
 
-Cookie IONAME(BeginOpenUnit)( // OPEN(without NEWUNIT=)
+Cookie IODEF(BeginOpenUnit)( // OPEN(without NEWUNIT=)
     ExternalUnit unitNumber, const char *sourceFile, int sourceLine) {
   Terminator terminator{sourceFile, sourceLine};
   bool wasExtant{false};
@@ -384,7 +383,7 @@ Cookie IONAME(BeginOpenUnit)( // OPEN(without NEWUNIT=)
   }
 }
 
-Cookie IONAME(BeginOpenNewUnit)( // OPEN(NEWUNIT=j)
+Cookie IODEF(BeginOpenNewUnit)( // OPEN(NEWUNIT=j)
     const char *sourceFile, int sourceLine) {
   Terminator terminator{sourceFile, sourceLine};
   ExternalFileUnit &unit{
@@ -394,7 +393,7 @@ Cookie IONAME(BeginOpenNewUnit)( // OPEN(NEWUNIT=j)
       sourceLine);
 }
 
-Cookie IONAME(BeginWait)(ExternalUnit unitNumber, AsynchronousId id,
+Cookie IODEF(BeginWait)(ExternalUnit unitNumber, AsynchronousId id,
     const char *sourceFile, int sourceLine) {
   Terminator terminator{sourceFile, sourceLine};
   if (ExternalFileUnit * unit{ExternalFileUnit::LookUp(unitNumber)}) {
@@ -410,12 +409,12 @@ Cookie IONAME(BeginWait)(ExternalUnit unitNumber, AsynchronousId id,
         terminator, unitNumber, id == 0 ? IostatOk : IostatBadWaitUnit);
   }
 }
-Cookie IONAME(BeginWaitAll)(
+Cookie IODEF(BeginWaitAll)(
     ExternalUnit unitNumber, const char *sourceFile, int sourceLine) {
   return IONAME(BeginWait)(unitNumber, 0 /*no ID=*/, sourceFile, sourceLine);
 }
 
-Cookie IONAME(BeginClose)(
+Cookie IODEF(BeginClose)(
     ExternalUnit unitNumber, const char *sourceFile, int sourceLine) {
   Terminator terminator{sourceFile, sourceLine};
   if (ExternalFileUnit * unit{ExternalFileUnit::LookUp(unitNumber)}) {
@@ -434,7 +433,7 @@ Cookie IONAME(BeginClose)(
   }
 }
 
-Cookie IONAME(BeginFlush)(
+Cookie IODEF(BeginFlush)(
     ExternalUnit unitNumber, const char *sourceFile, int sourceLine) {
   Terminator terminator{sourceFile, sourceLine};
   if (ExternalFileUnit * unit{ExternalFileUnit::LookUp(unitNumber)}) {
@@ -452,7 +451,7 @@ Cookie IONAME(BeginFlush)(
   }
 }
 
-Cookie IONAME(BeginBackspace)(
+Cookie IODEF(BeginBackspace)(
     ExternalUnit unitNumber, const char *sourceFile, int sourceLine) {
   Terminator terminator{sourceFile, sourceLine};
   if (ExternalFileUnit * unit{ExternalFileUnit::LookUp(unitNumber)}) {
@@ -470,7 +469,7 @@ Cookie IONAME(BeginBackspace)(
   }
 }
 
-Cookie IONAME(BeginEndfile)(
+Cookie IODEF(BeginEndfile)(
     ExternalUnit unitNumber, const char *sourceFile, int sourceLine) {
   Terminator terminator{sourceFile, sourceLine};
   Cookie errorCookie{nullptr};
@@ -490,7 +489,7 @@ Cookie IONAME(BeginEndfile)(
   }
 }
 
-Cookie IONAME(BeginRewind)(
+Cookie IODEF(BeginRewind)(
     ExternalUnit unitNumber, const char *sourceFile, int sourceLine) {
   Terminator terminator{sourceFile, sourceLine};
   Cookie errorCookie{nullptr};
@@ -510,7 +509,7 @@ Cookie IONAME(BeginRewind)(
   }
 }
 
-Cookie IONAME(BeginInquireUnit)(
+Cookie IODEF(BeginInquireUnit)(
     ExternalUnit unitNumber, const char *sourceFile, int sourceLine) {
   Terminator terminator{sourceFile, sourceLine};
   if (ExternalFileUnit * unit{ExternalFileUnit::LookUp(unitNumber)}) {
@@ -530,14 +529,14 @@ Cookie IONAME(BeginInquireUnit)(
   }
 }
 
-Cookie IONAME(BeginInquireFile)(const char *path, std::size_t pathLength,
+Cookie IODEF(BeginInquireFile)(const char *path, std::size_t pathLength,
     const char *sourceFile, int sourceLine) {
   Terminator terminator{sourceFile, sourceLine};
   auto trimmed{SaveDefaultCharacter(
       path, TrimTrailingSpaces(path, pathLength), terminator)};
   if (ExternalFileUnit *
       unit{ExternalFileUnit::LookUp(
-          trimmed.get(), std::strlen(trimmed.get()))}) {
+          trimmed.get(), Fortran::runtime::strlen(trimmed.get()))}) {
     // INQUIRE(FILE=) to a connected unit
     if (ChildIo * child{unit->GetChildIo()}) {
       return &child->BeginIoStatement<InquireUnitState>(
@@ -554,7 +553,7 @@ Cookie IONAME(BeginInquireFile)(const char *path, std::size_t pathLength,
   }
 }
 
-Cookie IONAME(BeginInquireIoLength)(const char *sourceFile, int sourceLine) {
+Cookie IODEF(BeginInquireIoLength)(const char *sourceFile, int sourceLine) {
   Terminator oom{sourceFile, sourceLine};
   return &New<InquireIOLengthState>{oom}(sourceFile, sourceLine)
               .release()
@@ -563,7 +562,7 @@ Cookie IONAME(BeginInquireIoLength)(const char *sourceFile, int sourceLine) {
 
 // Control list items
 
-void IONAME(EnableHandlers)(Cookie cookie, bool hasIoStat, bool hasErr,
+void IODEF(EnableHandlers)(Cookie cookie, bool hasIoStat, bool hasErr,
     bool hasEnd, bool hasEor, bool hasIoMsg) {
   IoErrorHandler &handler{cookie->GetIoErrorHandler()};
   if (hasIoStat) {
@@ -583,8 +582,8 @@ void IONAME(EnableHandlers)(Cookie cookie, bool hasIoStat, bool hasErr,
   }
 }
 
-static bool YesOrNo(const char *keyword, std::size_t length, const char *what,
-    IoErrorHandler &handler) {
+static RT_API_ATTRS bool YesOrNo(const char *keyword, std::size_t length,
+    const char *what, IoErrorHandler &handler) {
   static const char *keywords[]{"YES", "NO", nullptr};
   switch (IdentifyValue(keyword, length, keywords)) {
   case 0:
@@ -598,8 +597,7 @@ static bool YesOrNo(const char *keyword, std::size_t length, const char *what,
   }
 }
 
-bool IONAME(SetAdvance)(
-    Cookie cookie, const char *keyword, std::size_t length) {
+bool IODEF(SetAdvance)(Cookie cookie, const char *keyword, std::size_t length) {
   IoStatementState &io{*cookie};
   IoErrorHandler &handler{io.GetIoErrorHandler()};
   bool nonAdvancing{!YesOrNo(keyword, length, "ADVANCE", handler)};
@@ -616,7 +614,7 @@ bool IONAME(SetAdvance)(
   return !handler.InError();
 }
 
-bool IONAME(SetBlank)(Cookie cookie, const char *keyword, std::size_t length) {
+bool IODEF(SetBlank)(Cookie cookie, const char *keyword, std::size_t length) {
   IoStatementState &io{*cookie};
   static const char *keywords[]{"NULL", "ZERO", nullptr};
   switch (IdentifyValue(keyword, length, keywords)) {
@@ -633,8 +631,7 @@ bool IONAME(SetBlank)(Cookie cookie, const char *keyword, std::size_t length) {
   }
 }
 
-bool IONAME(SetDecimal)(
-    Cookie cookie, const char *keyword, std::size_t length) {
+bool IODEF(SetDecimal)(Cookie cookie, const char *keyword, std::size_t length) {
   IoStatementState &io{*cookie};
   static const char *keywords[]{"COMMA", "POINT", nullptr};
   switch (IdentifyValue(keyword, length, keywords)) {
@@ -651,7 +648,7 @@ bool IONAME(SetDecimal)(
   }
 }
 
-bool IONAME(SetDelim)(Cookie cookie, const char *keyword, std::size_t length) {
+bool IODEF(SetDelim)(Cookie cookie, const char *keyword, std::size_t length) {
   IoStatementState &io{*cookie};
   static const char *keywords[]{"APOSTROPHE", "QUOTE", "NONE", nullptr};
   switch (IdentifyValue(keyword, length, keywords)) {
@@ -671,14 +668,14 @@ bool IONAME(SetDelim)(Cookie cookie, const char *keyword, std::size_t length) {
   }
 }
 
-bool IONAME(SetPad)(Cookie cookie, const char *keyword, std::size_t length) {
+bool IODEF(SetPad)(Cookie cookie, const char *keyword, std::size_t length) {
   IoStatementState &io{*cookie};
   IoErrorHandler &handler{io.GetIoErrorHandler()};
   io.mutableModes().pad = YesOrNo(keyword, length, "PAD", handler);
   return !handler.InError();
 }
 
-bool IONAME(SetPos)(Cookie cookie, std::int64_t pos) {
+bool IODEF(SetPos)(Cookie cookie, std::int64_t pos) {
   IoStatementState &io{*cookie};
   IoErrorHandler &handler{io.GetIoErrorHandler()};
   if (auto *unit{io.GetExternalFileUnit()}) {
@@ -689,7 +686,7 @@ bool IONAME(SetPos)(Cookie cookie, std::int64_t pos) {
   return false;
 }
 
-bool IONAME(SetRec)(Cookie cookie, std::int64_t rec) {
+bool IODEF(SetRec)(Cookie cookie, std::int64_t rec) {
   IoStatementState &io{*cookie};
   IoErrorHandler &handler{io.GetIoErrorHandler()};
   if (auto *unit{io.GetExternalFileUnit()}) {
@@ -705,7 +702,7 @@ bool IONAME(SetRec)(Cookie cookie, std::int64_t rec) {
   return true;
 }
 
-bool IONAME(SetRound)(Cookie cookie, const char *keyword, std::size_t length) {
+bool IODEF(SetRound)(Cookie cookie, const char *keyword, std::size_t length) {
   IoStatementState &io{*cookie};
   static const char *keywords[]{"UP", "DOWN", "ZERO", "NEAREST", "COMPATIBLE",
       "PROCESSOR_DEFINED", nullptr};
@@ -735,7 +732,7 @@ bool IONAME(SetRound)(Cookie cookie, const char *keyword, std::size_t length) {
   }
 }
 
-bool IONAME(SetSign)(Cookie cookie, const char *keyword, std::size_t length) {
+bool IODEF(SetSign)(Cookie cookie, const char *keyword, std::size_t length) {
   IoStatementState &io{*cookie};
   static const char *keywords[]{
       "PLUS", "SUPPRESS", "PROCESSOR_DEFINED", nullptr};
@@ -754,7 +751,7 @@ bool IONAME(SetSign)(Cookie cookie, const char *keyword, std::size_t length) {
   }
 }
 
-bool IONAME(SetAccess)(Cookie cookie, const char *keyword, std::size_t length) {
+bool IODEF(SetAccess)(Cookie cookie, const char *keyword, std::size_t length) {
   IoStatementState &io{*cookie};
   auto *open{io.get_if<OpenStatementState>()};
   if (!open) {
@@ -790,7 +787,7 @@ bool IONAME(SetAccess)(Cookie cookie, const char *keyword, std::size_t length) {
   return true;
 }
 
-bool IONAME(SetAction)(Cookie cookie, const char *keyword, std::size_t length) {
+bool IODEF(SetAction)(Cookie cookie, const char *keyword, std::size_t length) {
   IoStatementState &io{*cookie};
   auto *open{io.get_if<OpenStatementState>()};
   if (!open) {
@@ -832,7 +829,7 @@ bool IONAME(SetAction)(Cookie cookie, const char *keyword, std::size_t length) {
   return true;
 }
 
-bool IONAME(SetAsynchronous)(
+bool IODEF(SetAsynchronous)(
     Cookie cookie, const char *keyword, std::size_t length) {
   IoStatementState &io{*cookie};
   IoErrorHandler &handler{io.GetIoErrorHandler()};
@@ -859,7 +856,7 @@ bool IONAME(SetAsynchronous)(
   return !handler.InError();
 }
 
-bool IONAME(SetCarriagecontrol)(
+bool IODEF(SetCarriagecontrol)(
     Cookie cookie, const char *keyword, std::size_t length) {
   IoStatementState &io{*cookie};
   auto *open{io.get_if<OpenStatementState>()};
@@ -891,8 +888,7 @@ bool IONAME(SetCarriagecontrol)(
   }
 }
 
-bool IONAME(SetConvert)(
-    Cookie cookie, const char *keyword, std::size_t length) {
+bool IODEF(SetConvert)(Cookie cookie, const char *keyword, std::size_t length) {
   IoStatementState &io{*cookie};
   auto *open{io.get_if<OpenStatementState>()};
   if (!open) {
@@ -916,7 +912,7 @@ bool IONAME(SetConvert)(
   }
 }
 
-bool IONAME(SetEncoding)(
+bool IODEF(SetEncoding)(
     Cookie cookie, const char *keyword, std::size_t length) {
   IoStatementState &io{*cookie};
   auto *open{io.get_if<OpenStatementState>()};
@@ -948,7 +944,7 @@ bool IONAME(SetEncoding)(
   return true;
 }
 
-bool IONAME(SetForm)(Cookie cookie, const char *keyword, std::size_t length) {
+bool IODEF(SetForm)(Cookie cookie, const char *keyword, std::size_t length) {
   IoStatementState &io{*cookie};
   auto *open{io.get_if<OpenStatementState>()};
   if (!open) {
@@ -976,7 +972,7 @@ bool IONAME(SetForm)(Cookie cookie, const char *keyword, std::size_t length) {
   return true;
 }
 
-bool IONAME(SetPosition)(
+bool IODEF(SetPosition)(
     Cookie cookie, const char *keyword, std::size_t length) {
   IoStatementState &io{*cookie};
   auto *open{io.get_if<OpenStatementState>()};
@@ -1009,7 +1005,7 @@ bool IONAME(SetPosition)(
   return true;
 }
 
-bool IONAME(SetRecl)(Cookie cookie, std::size_t n) {
+bool IODEF(SetRecl)(Cookie cookie, std::size_t n) {
   IoStatementState &io{*cookie};
   auto *open{io.get_if<OpenStatementState>()};
   if (!open) {
@@ -1036,7 +1032,7 @@ bool IONAME(SetRecl)(Cookie cookie, std::size_t n) {
   }
 }
 
-bool IONAME(SetStatus)(Cookie cookie, const char *keyword, std::size_t length) {
+bool IODEF(SetStatus)(Cookie cookie, const char *keyword, std::size_t length) {
   IoStatementState &io{*cookie};
   if (auto *open{io.get_if<OpenStatementState>()}) {
     if (open->completedOperation()) {
@@ -1090,7 +1086,7 @@ bool IONAME(SetStatus)(Cookie cookie, const char *keyword, std::size_t length) {
       "SetStatus() called when not in an OPEN or CLOSE statement");
 }
 
-bool IONAME(SetFile)(Cookie cookie, const char *path, std::size_t chars) {
+bool IODEF(SetFile)(Cookie cookie, const char *path, std::size_t chars) {
   IoStatementState &io{*cookie};
   if (auto *open{io.get_if<OpenStatementState>()}) {
     if (open->completedOperation()) {
@@ -1107,7 +1103,7 @@ bool IONAME(SetFile)(Cookie cookie, const char *path, std::size_t chars) {
   return false;
 }
 
-bool IONAME(GetNewUnit)(Cookie cookie, int &unit, int kind) {
+bool IODEF(GetNewUnit)(Cookie cookie, int &unit, int kind) {
   IoStatementState &io{*cookie};
   auto *open{io.get_if<OpenStatementState>()};
   if (!open) {
@@ -1135,15 +1131,15 @@ bool IONAME(GetNewUnit)(Cookie cookie, int &unit, int kind) {
 
 // Data transfers
 
-bool IONAME(OutputDescriptor)(Cookie cookie, const Descriptor &descriptor) {
+bool IODEF(OutputDescriptor)(Cookie cookie, const Descriptor &descriptor) {
   return descr::DescriptorIO<Direction::Output>(*cookie, descriptor);
 }
 
-bool IONAME(InputDescriptor)(Cookie cookie, const Descriptor &descriptor) {
+bool IODEF(InputDescriptor)(Cookie cookie, const Descriptor &descriptor) {
   return descr::DescriptorIO<Direction::Input>(*cookie, descriptor);
 }
 
-bool IONAME(OutputInteger8)(Cookie cookie, std::int8_t n) {
+bool IODEF(OutputInteger8)(Cookie cookie, std::int8_t n) {
   if (!cookie->CheckFormattedStmtType<Direction::Output>("OutputInteger8")) {
     return false;
   }
@@ -1154,7 +1150,7 @@ bool IONAME(OutputInteger8)(Cookie cookie, std::int8_t n) {
   return descr::DescriptorIO<Direction::Output>(*cookie, descriptor);
 }
 
-bool IONAME(OutputInteger16)(Cookie cookie, std::int16_t n) {
+bool IODEF(OutputInteger16)(Cookie cookie, std::int16_t n) {
   if (!cookie->CheckFormattedStmtType<Direction::Output>("OutputInteger16")) {
     return false;
   }
@@ -1165,7 +1161,6 @@ bool IONAME(OutputInteger16)(Cookie cookie, std::int16_t n) {
   return descr::DescriptorIO<Direction::Output>(*cookie, descriptor);
 }
 
-RT_EXT_API_GROUP_BEGIN
 bool IODEF(OutputInteger32)(Cookie cookie, std::int32_t n) {
   if (!cookie->CheckFormattedStmtType<Direction::Output>("OutputInteger32")) {
     return false;
@@ -1176,9 +1171,8 @@ bool IODEF(OutputInteger32)(Cookie cookie, std::int32_t n) {
       TypeCategory::Integer, 4, reinterpret_cast<void *>(&n), 0);
   return descr::DescriptorIO<Direction::Output>(*cookie, descriptor);
 }
-RT_EXT_API_GROUP_END
 
-bool IONAME(OutputInteger64)(Cookie cookie, std::int64_t n) {
+bool IODEF(OutputInteger64)(Cookie cookie, std::int64_t n) {
   if (!cookie->CheckFormattedStmtType<Direction::Output>("OutputInteger64")) {
     return false;
   }
@@ -1190,7 +1184,7 @@ bool IONAME(OutputInteger64)(Cookie cookie, std::int64_t n) {
 }
 
 #ifdef __SIZEOF_INT128__
-bool IONAME(OutputInteger128)(Cookie cookie, common::int128_t n) {
+bool IODEF(OutputInteger128)(Cookie cookie, common::int128_t n) {
   if (!cookie->CheckFormattedStmtType<Direction::Output>("OutputInteger128")) {
     return false;
   }
@@ -1202,7 +1196,7 @@ bool IONAME(OutputInteger128)(Cookie cookie, common::int128_t n) {
 }
 #endif
 
-bool IONAME(InputInteger)(Cookie cookie, std::int64_t &n, int kind) {
+bool IODEF(InputInteger)(Cookie cookie, std::int64_t &n, int kind) {
   if (!cookie->CheckFormattedStmtType<Direction::Input>("InputInteger")) {
     return false;
   }
@@ -1213,7 +1207,7 @@ bool IONAME(InputInteger)(Cookie cookie, std::int64_t &n, int kind) {
   return descr::DescriptorIO<Direction::Input>(*cookie, descriptor);
 }
 
-bool IONAME(OutputReal32)(Cookie cookie, float x) {
+bool IODEF(OutputReal32)(Cookie cookie, float x) {
   if (!cookie->CheckFormattedStmtType<Direction::Output>("OutputReal32")) {
     return false;
   }
@@ -1223,7 +1217,7 @@ bool IONAME(OutputReal32)(Cookie cookie, float x) {
   return descr::DescriptorIO<Direction::Output>(*cookie, descriptor);
 }
 
-bool IONAME(OutputReal64)(Cookie cookie, double x) {
+bool IODEF(OutputReal64)(Cookie cookie, double x) {
   if (!cookie->CheckFormattedStmtType<Direction::Output>("OutputReal64")) {
     return false;
   }
@@ -1233,7 +1227,7 @@ bool IONAME(OutputReal64)(Cookie cookie, double x) {
   return descr::DescriptorIO<Direction::Output>(*cookie, descriptor);
 }
 
-bool IONAME(InputReal32)(Cookie cookie, float &x) {
+bool IODEF(InputReal32)(Cookie cookie, float &x) {
   if (!cookie->CheckFormattedStmtType<Direction::Input>("InputReal32")) {
     return false;
   }
@@ -1243,7 +1237,7 @@ bool IONAME(InputReal32)(Cookie cookie, float &x) {
   return descr::DescriptorIO<Direction::Input>(*cookie, descriptor);
 }
 
-bool IONAME(InputReal64)(Cookie cookie, double &x) {
+bool IODEF(InputReal64)(Cookie cookie, double &x) {
   if (!cookie->CheckFormattedStmtType<Direction::Input>("InputReal64")) {
     return false;
   }
@@ -1253,7 +1247,7 @@ bool IONAME(InputReal64)(Cookie cookie, double &x) {
   return descr::DescriptorIO<Direction::Input>(*cookie, descriptor);
 }
 
-bool IONAME(OutputComplex32)(Cookie cookie, float r, float i) {
+bool IODEF(OutputComplex32)(Cookie cookie, float r, float i) {
   if (!cookie->CheckFormattedStmtType<Direction::Output>("OutputComplex32")) {
     return false;
   }
@@ -1265,7 +1259,7 @@ bool IONAME(OutputComplex32)(Cookie cookie, float r, float i) {
   return descr::DescriptorIO<Direction::Output>(*cookie, descriptor);
 }
 
-bool IONAME(OutputComplex64)(Cookie cookie, double r, double i) {
+bool IODEF(OutputComplex64)(Cookie cookie, double r, double i) {
   if (!cookie->CheckFormattedStmtType<Direction::Output>("OutputComplex64")) {
     return false;
   }
@@ -1277,7 +1271,7 @@ bool IONAME(OutputComplex64)(Cookie cookie, double r, double i) {
   return descr::DescriptorIO<Direction::Output>(*cookie, descriptor);
 }
 
-bool IONAME(InputComplex32)(Cookie cookie, float z[2]) {
+bool IODEF(InputComplex32)(Cookie cookie, float z[2]) {
   if (!cookie->CheckFormattedStmtType<Direction::Input>("InputComplex32")) {
     return false;
   }
@@ -1288,7 +1282,7 @@ bool IONAME(InputComplex32)(Cookie cookie, float z[2]) {
   return descr::DescriptorIO<Direction::Input>(*cookie, descriptor);
 }
 
-bool IONAME(InputComplex64)(Cookie cookie, double z[2]) {
+bool IODEF(InputComplex64)(Cookie cookie, double z[2]) {
   if (!cookie->CheckFormattedStmtType<Direction::Input>("InputComplex64")) {
     return false;
   }
@@ -1299,7 +1293,7 @@ bool IONAME(InputComplex64)(Cookie cookie, double z[2]) {
   return descr::DescriptorIO<Direction::Input>(*cookie, descriptor);
 }
 
-bool IONAME(OutputCharacter)(
+bool IODEF(OutputCharacter)(
     Cookie cookie, const char *x, std::size_t length, int kind) {
   if (!cookie->CheckFormattedStmtType<Direction::Output>("OutputCharacter")) {
     return false;
@@ -1311,11 +1305,11 @@ bool IONAME(OutputCharacter)(
   return descr::DescriptorIO<Direction::Output>(*cookie, descriptor);
 }
 
-bool IONAME(OutputAscii)(Cookie cookie, const char *x, std::size_t length) {
+bool IODEF(OutputAscii)(Cookie cookie, const char *x, std::size_t length) {
   return IONAME(OutputCharacter(cookie, x, length, 1));
 }
 
-bool IONAME(InputCharacter)(
+bool IODEF(InputCharacter)(
     Cookie cookie, char *x, std::size_t length, int kind) {
   if (!cookie->CheckFormattedStmtType<Direction::Input>("InputCharacter")) {
     return false;
@@ -1326,11 +1320,11 @@ bool IONAME(InputCharacter)(
   return descr::DescriptorIO<Direction::Input>(*cookie, descriptor);
 }
 
-bool IONAME(InputAscii)(Cookie cookie, char *x, std::size_t length) {
+bool IODEF(InputAscii)(Cookie cookie, char *x, std::size_t length) {
   return IONAME(InputCharacter)(cookie, x, length, 1);
 }
 
-bool IONAME(OutputLogical)(Cookie cookie, bool truth) {
+bool IODEF(OutputLogical)(Cookie cookie, bool truth) {
   if (!cookie->CheckFormattedStmtType<Direction::Output>("OutputLogical")) {
     return false;
   }
@@ -1341,7 +1335,7 @@ bool IONAME(OutputLogical)(Cookie cookie, bool truth) {
   return descr::DescriptorIO<Direction::Output>(*cookie, descriptor);
 }
 
-bool IONAME(InputLogical)(Cookie cookie, bool &truth) {
+bool IODEF(InputLogical)(Cookie cookie, bool &truth) {
   if (!cookie->CheckFormattedStmtType<Direction::Input>("InputLogical")) {
     return false;
   }
@@ -1352,17 +1346,17 @@ bool IONAME(InputLogical)(Cookie cookie, bool &truth) {
   return descr::DescriptorIO<Direction::Input>(*cookie, descriptor);
 }
 
-bool IONAME(OutputDerivedType)(Cookie cookie, const Descriptor &descriptor,
+bool IODEF(OutputDerivedType)(Cookie cookie, const Descriptor &descriptor,
     const NonTbpDefinedIoTable *table) {
   return descr::DescriptorIO<Direction::Output>(*cookie, descriptor, table);
 }
 
-bool IONAME(InputDerivedType)(Cookie cookie, const Descriptor &descriptor,
+bool IODEF(InputDerivedType)(Cookie cookie, const Descriptor &descriptor,
     const NonTbpDefinedIoTable *table) {
   return descr::DescriptorIO<Direction::Input>(*cookie, descriptor, table);
 }
 
-std::size_t IONAME(GetSize)(Cookie cookie) {
+std::size_t IODEF(GetSize)(Cookie cookie) {
   IoStatementState &io{*cookie};
   IoErrorHandler &handler{io.GetIoErrorHandler()};
   if (!handler.InError()) {
@@ -1379,7 +1373,7 @@ std::size_t IONAME(GetSize)(Cookie cookie) {
   return 0;
 }
 
-std::size_t IONAME(GetIoLength)(Cookie cookie) {
+std::size_t IODEF(GetIoLength)(Cookie cookie) {
   IoStatementState &io{*cookie};
   IoErrorHandler &handler{io.GetIoErrorHandler()};
   if (!handler.InError()) {
@@ -1395,7 +1389,7 @@ std::size_t IONAME(GetIoLength)(Cookie cookie) {
   return 0;
 }
 
-void IONAME(GetIoMsg)(Cookie cookie, char *msg, std::size_t length) {
+void IODEF(GetIoMsg)(Cookie cookie, char *msg, std::size_t length) {
   IoStatementState &io{*cookie};
   IoErrorHandler &handler{io.GetIoErrorHandler()};
   if (!handler.InError()) {
@@ -1406,7 +1400,7 @@ void IONAME(GetIoMsg)(Cookie cookie, char *msg, std::size_t length) {
   }
 }
 
-AsynchronousId IONAME(GetAsynchronousId)(Cookie cookie) {
+AsynchronousId IODEF(GetAsynchronousId)(Cookie cookie) {
   IoStatementState &io{*cookie};
   IoErrorHandler &handler{io.GetIoErrorHandler()};
   if (auto *ext{io.get_if<ExternalIoStatementBase>()}) {
@@ -1419,24 +1413,24 @@ AsynchronousId IONAME(GetAsynchronousId)(Cookie cookie) {
   return 0;
 }
 
-bool IONAME(InquireCharacter)(Cookie cookie, InquiryKeywordHash inquiry,
+bool IODEF(InquireCharacter)(Cookie cookie, InquiryKeywordHash inquiry,
     char *result, std::size_t length) {
   IoStatementState &io{*cookie};
   return io.Inquire(inquiry, result, length);
 }
 
-bool IONAME(InquireLogical)(
+bool IODEF(InquireLogical)(
     Cookie cookie, InquiryKeywordHash inquiry, bool &result) {
   IoStatementState &io{*cookie};
   return io.Inquire(inquiry, result);
 }
 
-bool IONAME(InquirePendingId)(Cookie cookie, AsynchronousId id, bool &result) {
+bool IODEF(InquirePendingId)(Cookie cookie, AsynchronousId id, bool &result) {
   IoStatementState &io{*cookie};
   return io.Inquire(HashInquiryKeyword("PENDING"), id, result);
 }
 
-bool IONAME(InquireInteger64)(
+bool IODEF(InquireInteger64)(
     Cookie cookie, InquiryKeywordHash inquiry, std::int64_t &result, int kind) {
   IoStatementState &io{*cookie};
   std::int64_t n{0}; // safe "undefined" value
@@ -1452,17 +1446,15 @@ bool IONAME(InquireInteger64)(
   return false;
 }
 
-RT_EXT_API_GROUP_BEGIN
 enum Iostat IODEF(EndIoStatement)(Cookie cookie) {
   IoStatementState &io{*cookie};
   return static_cast<enum Iostat>(io.EndIoStatement());
 }
-RT_EXT_API_GROUP_END
 
 template <typename INT>
-static enum Iostat CheckUnitNumberInRangeImpl(INT unit, bool handleError,
-    char *ioMsg, std::size_t ioMsgLength, const char *sourceFile,
-    int sourceLine) {
+static RT_API_ATTRS enum Iostat CheckUnitNumberInRangeImpl(INT unit,
+    bool handleError, char *ioMsg, std::size_t ioMsgLength,
+    const char *sourceFile, int sourceLine) {
   static_assert(sizeof(INT) >= sizeof(ExternalUnit),
       "only intended to be used when the INT to ExternalUnit conversion is "
       "narrowing");
@@ -1494,15 +1486,15 @@ static enum Iostat CheckUnitNumberInRangeImpl(INT unit, bool handleError,
   return IostatOk;
 }
 
-enum Iostat IONAME(CheckUnitNumberInRange64)(std::int64_t unit,
-    bool handleError, char *ioMsg, std::size_t ioMsgLength,
-    const char *sourceFile, int sourceLine) {
+enum Iostat IODEF(CheckUnitNumberInRange64)(std::int64_t unit, bool handleError,
+    char *ioMsg, std::size_t ioMsgLength, const char *sourceFile,
+    int sourceLine) {
   return CheckUnitNumberInRangeImpl(
       unit, handleError, ioMsg, ioMsgLength, sourceFile, sourceLine);
 }
 
 #ifdef __SIZEOF_INT128__
-enum Iostat IONAME(CheckUnitNumberInRange128)(common::int128_t unit,
+enum Iostat IODEF(CheckUnitNumberInRange128)(common::int128_t unit,
     bool handleError, char *ioMsg, std::size_t ioMsgLength,
     const char *sourceFile, int sourceLine) {
   return CheckUnitNumberInRangeImpl(
@@ -1525,3 +1517,5 @@ void std::__libcpp_verbose_abort(char const *format, ...) {
   std::abort();
 }
 #endif
+
+RT_EXT_API_GROUP_END
diff --git a/flang/runtime/io-error.cpp b/flang/runtime/io-error.cpp
index b006b82f6224..7a90966f8104 100644
--- a/flang/runtime/io-error.cpp
+++ b/flang/runtime/io-error.cpp
@@ -109,8 +109,6 @@ void IoErrorHandler::SignalPendingError() {
   SignalError(error);
 }
 
-RT_OFFLOAD_API_GROUP_END
-
 void IoErrorHandler::SignalErrno() { SignalError(errno); }
 
 bool IoErrorHandler::GetIoMsg(char *buffer, std::size_t bufferLength) {
@@ -127,7 +125,10 @@ bool IoErrorHandler::GetIoMsg(char *buffer, std::size_t bufferLength) {
   // in LLVM v9.0.1 with inadequate modification for Fortran,
   // since rectified.
   bool ok{false};
-#if HAVE_STRERROR_R
+#if defined(RT_DEVICE_COMPILATION)
+  // strerror_r is not available on device.
+  msg = "errno description is not available on device";
+#elif HAVE_STRERROR_R
   // strerror_r is thread-safe.
 #if defined(__GLIBC__) && defined(_GNU_SOURCE)
   // glibc defines its own incompatible version of strerror_r
@@ -157,4 +158,6 @@ bool IoErrorHandler::GetIoMsg(char *buffer, std::size_t bufferLength) {
     return false;
   }
 }
+
+RT_OFFLOAD_API_GROUP_END
 } // namespace Fortran::runtime::io
diff --git a/flang/runtime/io-error.h b/flang/runtime/io-error.h
index 0fe11c9185c0..426573e2faf0 100644
--- a/flang/runtime/io-error.h
+++ b/flang/runtime/io-error.h
@@ -61,7 +61,7 @@ public:
   RT_API_ATTRS void SignalPendingError();
 
   RT_API_ATTRS int GetIoStat() const { return ioStat_; }
-  bool GetIoMsg(char *, std::size_t);
+  RT_API_ATTRS bool GetIoMsg(char *, std::size_t);
 
 private:
   enum Flag : std::uint8_t {
diff --git a/flang/runtime/namelist.cpp b/flang/runtime/namelist.cpp
index b502d41a8d5c..b9eed2101ecf 100644
--- a/flang/runtime/namelist.cpp
+++ b/flang/runtime/namelist.cpp
@@ -17,16 +17,20 @@
 
 namespace Fortran::runtime::io {
 
+RT_VAR_GROUP_BEGIN
 // Max size of a group, symbol or component identifier that can appear in
 // NAMELIST input, plus a byte for NUL termination.
-static constexpr std::size_t nameBufferSize{201};
+static constexpr RT_CONST_VAR_ATTRS std::size_t nameBufferSize{201};
+RT_VAR_GROUP_END
 
-static inline char32_t GetComma(IoStatementState &io) {
+RT_OFFLOAD_API_GROUP_BEGIN
+
+static inline RT_API_ATTRS char32_t GetComma(IoStatementState &io) {
   return io.mutableModes().editingFlags & decimalComma ? char32_t{';'}
                                                        : char32_t{','};
 }
 
-bool IONAME(OutputNamelist)(Cookie cookie, const NamelistGroup &group) {
+bool IODEF(OutputNamelist)(Cookie cookie, const NamelistGroup &group) {
   IoStatementState &io{*cookie};
   io.CheckFormattedStmtType<Direction::Output>("OutputNamelist");
   io.mutableModes().inNamelist = true;
@@ -40,7 +44,8 @@ bool IONAME(OutputNamelist)(Cookie cookie, const NamelistGroup &group) {
     if ((connection.NeedAdvance(prefixLen) &&
             !(io.AdvanceRecord() && EmitAscii(io, " ", 1))) ||
         !EmitAscii(io, prefix, prefixLen) ||
-        (connection.NeedAdvance(std::strlen(str) + (suffix != ' ')) &&
+        (connection.NeedAdvance(
+             Fortran::runtime::strlen(str) + (suffix != ' ')) &&
             !(io.AdvanceRecord() && EmitAscii(io, " ", 1)))) {
       return false;
     }
@@ -84,20 +89,20 @@ bool IONAME(OutputNamelist)(Cookie cookie, const NamelistGroup &group) {
   return EmitUpperCase("/", 1, "", ' ');
 }
 
-static constexpr bool IsLegalIdStart(char32_t ch) {
+static constexpr RT_API_ATTRS bool IsLegalIdStart(char32_t ch) {
   return (ch >= 'A' && ch <= 'Z') || (ch >= 'a' && ch <= 'z') || ch == '_' ||
       ch == '@';
 }
 
-static constexpr bool IsLegalIdChar(char32_t ch) {
+static constexpr RT_API_ATTRS bool IsLegalIdChar(char32_t ch) {
   return IsLegalIdStart(ch) || (ch >= '0' && ch <= '9');
 }
 
-static constexpr char NormalizeIdChar(char32_t ch) {
+static constexpr RT_API_ATTRS char NormalizeIdChar(char32_t ch) {
   return static_cast<char>(ch >= 'A' && ch <= 'Z' ? ch - 'A' + 'a' : ch);
 }
 
-static bool GetLowerCaseName(
+static RT_API_ATTRS bool GetLowerCaseName(
     IoStatementState &io, char buffer[], std::size_t maxLength) {
   std::size_t byteLength{0};
   if (auto ch{io.GetNextNonBlank(byteLength)}) {
@@ -119,7 +124,7 @@ static bool GetLowerCaseName(
   return false;
 }
 
-static Fortran::common::optional<SubscriptValue> GetSubscriptValue(
+static RT_API_ATTRS Fortran::common::optional<SubscriptValue> GetSubscriptValue(
     IoStatementState &io) {
   Fortran::common::optional<SubscriptValue> value;
   std::size_t byteCount{0};
@@ -152,8 +157,8 @@ static Fortran::common::optional<SubscriptValue> GetSubscriptValue(
   return value;
 }
 
-static bool HandleSubscripts(IoStatementState &io, Descriptor &desc,
-    const Descriptor &source, const char *name) {
+static RT_API_ATTRS bool HandleSubscripts(IoStatementState &io,
+    Descriptor &desc, const Descriptor &source, const char *name) {
   IoErrorHandler &handler{io.GetIoErrorHandler()};
   // Allow for blanks in subscripts; they're nonstandard, but not
   // ambiguous within the parentheses.
@@ -252,7 +257,7 @@ static bool HandleSubscripts(IoStatementState &io, Descriptor &desc,
   return false;
 }
 
-static void StorageSequenceExtension(
+static RT_API_ATTRS void StorageSequenceExtension(
     Descriptor &desc, const Descriptor &source) {
   // Support the near-universal extension of NAMELIST input into a
   // designatable storage sequence identified by its initial scalar array
@@ -274,7 +279,7 @@ static void StorageSequenceExtension(
   }
 }
 
-static bool HandleSubstring(
+static RT_API_ATTRS bool HandleSubstring(
     IoStatementState &io, Descriptor &desc, const char *name) {
   IoErrorHandler &handler{io.GetIoErrorHandler()};
   auto pair{desc.type().GetCategoryAndKind()};
@@ -335,7 +340,7 @@ static bool HandleSubstring(
   return false;
 }
 
-static bool HandleComponent(IoStatementState &io, Descriptor &desc,
+static RT_API_ATTRS bool HandleComponent(IoStatementState &io, Descriptor &desc,
     const Descriptor &source, const char *name) {
   IoErrorHandler &handler{io.GetIoErrorHandler()};
   char compName[nameBufferSize];
@@ -344,7 +349,8 @@ static bool HandleComponent(IoStatementState &io, Descriptor &desc,
     if (const typeInfo::DerivedType *
         type{addendum ? addendum->derivedType() : nullptr}) {
       if (const typeInfo::Component *
-          comp{type->FindDataComponent(compName, std::strlen(compName))}) {
+          comp{type->FindDataComponent(
+              compName, Fortran::runtime::strlen(compName))}) {
         bool createdDesc{false};
         if (comp->rank() > 0 && source.rank() > 0) {
           // If base and component are both arrays, the component name
@@ -408,7 +414,7 @@ static bool HandleComponent(IoStatementState &io, Descriptor &desc,
 
 // Advance to the terminal '/' of a namelist group or leading '&'/'$'
 // of the next.
-static void SkipNamelistGroup(IoStatementState &io) {
+static RT_API_ATTRS void SkipNamelistGroup(IoStatementState &io) {
   std::size_t byteCount{0};
   while (auto ch{io.GetNextNonBlank(byteCount)}) {
     io.HandleRelativePosition(byteCount);
@@ -431,7 +437,7 @@ static void SkipNamelistGroup(IoStatementState &io) {
   }
 }
 
-bool IONAME(InputNamelist)(Cookie cookie, const NamelistGroup &group) {
+bool IODEF(InputNamelist)(Cookie cookie, const NamelistGroup &group) {
   IoStatementState &io{*cookie};
   io.CheckFormattedStmtType<Direction::Input>("InputNamelist");
   io.mutableModes().inNamelist = true;
@@ -470,7 +476,7 @@ bool IONAME(InputNamelist)(Cookie cookie, const NamelistGroup &group) {
       handler.SignalError("NAMELIST input group has no name");
       return false;
     }
-    if (std::strcmp(group.groupName, name) == 0) {
+    if (Fortran::runtime::strcmp(group.groupName, name) == 0) {
       break; // found it
     }
     SkipNamelistGroup(io);
@@ -489,7 +495,7 @@ bool IONAME(InputNamelist)(Cookie cookie, const NamelistGroup &group) {
     }
     std::size_t itemIndex{0};
     for (; itemIndex < group.items; ++itemIndex) {
-      if (std::strcmp(name, group.item[itemIndex].name) == 0) {
+      if (Fortran::runtime::strcmp(name, group.item[itemIndex].name) == 0) {
         break;
       }
     }
@@ -590,8 +596,6 @@ bool IONAME(InputNamelist)(Cookie cookie, const NamelistGroup &group) {
   return true;
 }
 
-RT_OFFLOAD_API_GROUP_BEGIN
-
 bool IsNamelistNameOrSlash(IoStatementState &io) {
   if (auto *listInput{
           io.get_if<ListDirectedStatementState<Direction::Input>>()}) {
diff --git a/flang/test/Driver/driver-help-hidden.f90 b/flang/test/Driver/driver-help-hidden.f90
index 4405b64a8d00..48f48f5384fd 100644
--- a/flang/test/Driver/driver-help-hidden.f90
+++ b/flang/test/Driver/driver-help-hidden.f90
@@ -81,7 +81,7 @@
 ! CHECK-NEXT: -fopenmp-targets=<value>
 ! CHECK-NEXT:                         Specify comma-separated list of triples OpenMP offloading targets to be supported
 ! CHECK-NEXT: -fopenmp-version=<value>
-! CHECK-NEXT:                         Set OpenMP version (e.g. 45 for OpenMP 4.5, 51 for OpenMP 5.1). Default value is 51 for Clang
+! CHECK-NEXT:                         Set OpenMP version (e.g. 45 for OpenMP 4.5, 51 for OpenMP 5.1). Default value is 11 for Flang
 ! CHECK-NEXT: -fopenmp                Parse OpenMP pragmas and generate parallel code.
 ! CHECK-NEXT: -foptimization-record-file=<file>
 ! CHECK-NEXT:                         Specify the output name of the file containing the optimization remarks. Implies -fsave-optimization-record. On Darwin platforms, this cannot be used with multiple -arch <arch> options.
diff --git a/flang/test/Driver/driver-help.f90 b/flang/test/Driver/driver-help.f90
index c80453f0ef21..38f74395a678 100644
--- a/flang/test/Driver/driver-help.f90
+++ b/flang/test/Driver/driver-help.f90
@@ -69,7 +69,7 @@
 ! HELP-NEXT: -fopenmp-targets=<value>
 ! HELP-NEXT:                         Specify comma-separated list of triples OpenMP offloading targets to be supported
 ! HELP-NEXT: -fopenmp-version=<value>
-! HELP-NEXT:                         Set OpenMP version (e.g. 45 for OpenMP 4.5, 51 for OpenMP 5.1). Default value is 51 for Clang
+! HELP-NEXT:                         Set OpenMP version (e.g. 45 for OpenMP 4.5, 51 for OpenMP 5.1). Default value is 11 for Flang
 ! HELP-NEXT: -fopenmp                Parse OpenMP pragmas and generate parallel code.
 ! HELP-NEXT: -foptimization-record-file=<file>
 ! HELP-NEXT:                         Specify the output name of the file containing the optimization remarks. Implies -fsave-optimization-record. On Darwin platforms, this cannot be used with multiple -arch <arch> options.
@@ -226,7 +226,7 @@
 ! HELP-FC1-NEXT:                         Generate code only for an OpenMP target device.
 ! HELP-FC1-NEXT: -fopenmp-target-debug   Enable debugging in the OpenMP offloading device RTL
 ! HELP-FC1-NEXT: -fopenmp-version=<value>
-! HELP-FC1-NEXT:                         Set OpenMP version (e.g. 45 for OpenMP 4.5, 51 for OpenMP 5.1). Default value is 51 for Clang
+! HELP-FC1-NEXT:                         Set OpenMP version (e.g. 45 for OpenMP 4.5, 51 for OpenMP 5.1). Default value is 11 for Flang
 ! HELP-FC1-NEXT: -fopenmp                Parse OpenMP pragmas and generate parallel code.
 ! HELP-FC1-NEXT: -fpass-plugin=<dsopath> Load pass plugin from a dynamic shared object file (only with new pass manager).
 ! HELP-FC1-NEXT: -fppc-native-vector-element-order
diff --git a/flang/test/Driver/fopenmp.f90 b/flang/test/Driver/fopenmp.f90
new file mode 100644
index 000000000000..c71d34dc9e7e
--- /dev/null
+++ b/flang/test/Driver/fopenmp.f90
@@ -0,0 +1,61 @@
+! RUN: %flang -target x86_64-linux-gnu -fopenmp=libomp -c %s -### 2>&1 | FileCheck %s --check-prefix=CHECK-FC1-OPENMP
+! RUN: %flang -target x86_64-linux-gnu -fopenmp=libgomp -c %s -### 2>&1 | FileCheck %s --check-prefix=CHECK-FC1-NO-OPENMP
+! RUN: %flang -target x86_64-linux-gnu -fopenmp=libiomp5 -c %s -### 2>&1 | FileCheck %s --check-prefix=CHECK-FC1-OPENMP
+! RUN: %flang -target x86_64-apple-darwin -fopenmp=libomp -c %s -### 2>&1 | FileCheck %s --check-prefix=CHECK-FC1-OPENMP
+! RUN: %flang -target x86_64-apple-darwin -fopenmp=libgomp -c %s -### 2>&1 | FileCheck %s --check-prefix=CHECK-FC1-NO-OPENMP
+! RUN: %flang -target x86_64-apple-darwin -fopenmp=libiomp5 -c %s -### 2>&1 | FileCheck %s --check-prefix=CHECK-FC1-OPENMP
+! RUN: %flang -target x86_64-freebsd -fopenmp=libomp -c %s -### 2>&1 | FileCheck %s --check-prefix=CHECK-FC1-OPENMP
+! RUN: %flang -target x86_64-freebsd -fopenmp=libgomp -c %s -### 2>&1 | FileCheck %s --check-prefix=CHECK-FC1-NO-OPENMP
+! RUN: %flang -target x86_64-freebsd -fopenmp=libiomp5 -c %s -### 2>&1 | FileCheck %s --check-prefix=CHECK-FC1-OPENMP
+! RUN: %flang -target x86_64-windows-gnu -fopenmp=libomp -c %s -### 2>&1 | FileCheck %s --check-prefix=CHECK-FC1-OPENMP
+! RUN: %flang -target x86_64-windows-gnu -fopenmp=libgomp -c %s -### 2>&1 | FileCheck %s --check-prefix=CHECK-FC1-NO-OPENMP --check-prefix=CHECK-WARNING
+! RUN: %flang -target x86_64-windows-gnu -fopenmp=libiomp5 -c %s -### 2>&1 | FileCheck %s --check-prefix=CHECK-FC1-OPENMP
+
+! CHECK-FC1-OPENMP: "-fc1"
+! CHECK-FC1-OPENMP: "-fopenmp"
+!
+! CHECK-WARNING: warning: The library '-fopenmp=={{.*}}' is not supported, openmp is not be enabled
+! CHECK-FC1-NO-OPENMP: "-fc1"
+! CHECK-FC1-NO-OPENMP-NOT: "-fopenmp"
+!
+! RUN: %flang -target x86_64-linux-gnu -fopenmp=libomp %s -o %t -### 2>&1 | FileCheck %s --check-prefix=CHECK-LD-OMP
+! RUN: %flang -target x86_64-linux-gnu -fopenmp=libgomp %s -o %t -### 2>&1 | FileCheck %s --check-prefix=CHECK-LD-GOMP --check-prefix=CHECK-LD-GOMP-RT
+! RUN: %flang -target x86_64-linux-gnu -fopenmp=libiomp5 %s -o %t -### 2>&1 | FileCheck %s --check-prefix=CHECK-LD-IOMP5
+!
+! RUN: %flang -target x86_64-darwin -fopenmp=libomp %s -o %t -### 2>&1 | FileCheck %s --check-prefix=CHECK-LD-OMP
+! RUN: %flang -target x86_64-darwin -fopenmp=libgomp %s -o %t -### 2>&1 | FileCheck %s --check-prefix=CHECK-LD-GOMP --check-prefix=CHECK-LD-GOMP-NO-RT
+! RUN: %flang -target x86_64-darwin -fopenmp=libiomp5 %s -o %t -### 2>&1 | FileCheck %s --check-prefix=CHECK-LD-IOMP5
+!
+! RUN: %flang -target x86_64-freebsd -fopenmp=libomp %s -o %t -### 2>&1 | FileCheck %s --check-prefix=CHECK-LD-OMP
+! RUN: %flang -target x86_64-freebsd -fopenmp=libgomp %s -o %t -### 2>&1 | FileCheck %s --check-prefix=CHECK-LD-GOMP --check-prefix=CHECK-LD-GOMP-NO-RT
+! RUN: %flang -target x86_64-freebsd -fopenmp=libiomp5 %s -o %t -### 2>&1 | FileCheck %s --check-prefix=CHECK-LD-IOMP5
+!
+! RUN: %flang -target x86_64-windows-gnu -fopenmp=libomp %s -o %t -### 2>&1 | FileCheck %s --check-prefix=CHECK-LD-OMP
+! RUN: %flang -target x86_64-windows-gnu -fopenmp=libgomp %s -o %t -### 2>&1 | FileCheck %s --check-prefix=CHECK-LD-GOMP --check-prefix=CHECK-LD-GOMP-NO-RT
+! RUN: %flang -target x86_64-windows-gnu -fopenmp=libiomp5 %s -o %t -### 2>&1 | FileCheck %s --check-prefix=CHECK-LD-IOMP5MD
+!
+! CHECK-LD-OMP: "{{.*}}ld{{(.exe)?}}"
+! CHECK-LD-OMP: "-lomp"
+!
+! CHECK-LD-GOMP: "{{.*}}ld{{(.exe)?}}"
+! CHECK-LD-GOMP: "-lgomp"
+! CHECK-LD-GOMP-RT: "-lrt"
+! CHECK-LD-GOMP-NO-RT-NOT: "-lrt"
+!
+! CHECK-LD-IOMP5: "{{.*}}ld{{(.exe)?}}"
+! CHECK-LD-IOMP5: "-liomp5"
+!
+! CHECK-LD-IOMP5MD: "{{.*}}ld{{(.exe)?}}"
+! CHECK-LD-IOMP5MD: "-liomp5md"
+!
+! We'd like to check that the default is sane, but until we have the ability
+! to *always* semantically analyze OpenMP without always generating runtime
+! calls (in the event of an unsupported runtime), we don't have a good way to
+! test the CC1 invocation. Instead, just ensure we do eventually link *some*
+! OpenMP runtime.
+!
+! CHECK-LD-ANY: "{{.*}}ld{{(.exe)?}}"
+! CHECK-LD-ANY: "-l{{(omp|gomp|iomp5)}}"
+!
+! CHECK-LD-ANYMD: "{{.*}}ld{{(.exe)?}}"
+! CHECK-LD-ANYMD: "-l{{(omp|gomp|iomp5md)}}"
diff --git a/flang/test/Driver/omp-driver-offload.f90 b/flang/test/Driver/omp-driver-offload.f90
index 9b62699030c6..7e9a73627cd7 100644
--- a/flang/test/Driver/omp-driver-offload.f90
+++ b/flang/test/Driver/omp-driver-offload.f90
@@ -57,9 +57,14 @@
 ! RUN: --target=aarch64-unknown-linux-gnu \
 ! RUN:   | FileCheck %s --check-prefix=OPENMP-OFFLOAD-ARGS
 ! OPENMP-OFFLOAD-ARGS: "{{[^"]*}}flang-new" "-fc1" "-triple" "aarch64-unknown-linux-gnu" {{.*}} "-fopenmp" {{.*}}.f90"
-! OPENMP-OFFLOAD-ARGS-NEXT: "{{[^"]*}}flang-new" "-fc1" "-triple" "amdgcn-amd-amdhsa" {{.*}} "-fopenmp" {{.*}} "-fopenmp-host-ir-file-path" "{{.*}}.bc" "-fopenmp-is-target-device" {{.*}}.f90"
+! OPENMP-OFFLOAD-ARGS-NEXT: "{{[^"]*}}flang-new" "-fc1" "-triple" "amdgcn-amd-amdhsa"
+! OPENMP-OFFLOAD-ARGS-SAME:  "-fopenmp"
+! OPENMP-OFFLOAD-ARGS-SAME:  "-fopenmp-host-ir-file-path" "{{.*}}.bc" "-fopenmp-is-target-device"
+! OPENMP-OFFLOAD-ARGS-SAME:  {{.*}}.f90"
 ! OPENMP-OFFLOAD-ARGS: "{{[^"]*}}clang-offload-packager{{.*}}" {{.*}} "--image=file={{.*}}.bc,triple=amdgcn-amd-amdhsa,arch=gfx90a,kind=openmp"
-! OPENMP-OFFLOAD-ARGS-NEXT: "{{[^"]*}}flang-new" "-fc1" "-triple" "aarch64-unknown-linux-gnu" {{.*}} "-fopenmp" {{.*}} "-fembed-offload-object={{.*}}.out" {{.*}}.bc"
+! OPENMP-OFFLOAD-ARGS-NEXT: "{{[^"]*}}flang-new" "-fc1" "-triple" "aarch64-unknown-linux-gnu"
+! OPENMP-OFFLOAD-ARGS-SAME:  "-fopenmp"
+! OPENMP-OFFLOAD-ARGS-SAME:  "-fembed-offload-object={{.*}}.out" {{.*}}.bc"
 
 ! Test -fopenmp with offload for RTL Flag Options
 ! RUN: %flang -### %s -o %t 2>&1 \
diff --git a/flang/test/Lower/CUDA/cuda-data-transfer.cuf b/flang/test/Lower/CUDA/cuda-data-transfer.cuf
index 54226b8623e6..4ebd736315bc 100644
--- a/flang/test/Lower/CUDA/cuda-data-transfer.cuf
+++ b/flang/test/Lower/CUDA/cuda-data-transfer.cuf
@@ -2,6 +2,12 @@
 
 ! Test CUDA Fortran data transfer using assignment statements.
 
+module mod1
+  type :: t1
+    integer :: i
+  end type
+end
+
 subroutine sub1()
   integer, device :: m
   integer, device :: adev(10)
@@ -55,3 +61,61 @@ end
 ! CHECK: %[[ASSOC:.*]]:3 = hlfir.associate %[[ELEMENTAL]](%{{.*}}) {uniq_name = ".cuf_host_tmp"} : (!hlfir.expr<10xi32>, !fir.shape<1>) -> (!fir.ref<!fir.array<10xi32>>, !fir.ref<!fir.array<10xi32>>, i1)
 ! CHECK: fir.cuda_data_transfer %[[ASSOC]]#0 to %[[ADEV]]#0 {transfer_kind = #fir.cuda_transfer<host_device>} : !fir.ref<!fir.array<10xi32>>, !fir.ref<!fir.array<10xi32>>
 ! CHECK: hlfir.end_associate %[[ASSOC]]#1, %[[ASSOC]]#2 : !fir.ref<!fir.array<10xi32>>, i1
+
+subroutine sub2()
+  integer, device :: m
+  integer, device :: adev(10), bdev(10)
+  integer :: i, ahost(10), bhost(10)
+
+  ahost = adev
+
+  i = m
+
+  ahost(1:5) = adev(1:5)
+
+  bdev = adev
+
+  ! Implicit data transfer of adev before evaluation.
+  bhost = ahost + adev
+
+end
+
+! CHECK-LABEL: func.func @_QPsub2()
+! CHECK: %[[ADEV:.*]]:2 = hlfir.declare %{{.*}}(%{{.*}}) {cuda_attr = #fir.cuda<device>, uniq_name = "_QFsub2Eadev"} : (!fir.ref<!fir.array<10xi32>>, !fir.shape<1>) -> (!fir.ref<!fir.array<10xi32>>, !fir.ref<!fir.array<10xi32>>)
+! CHECK: %[[AHOST:.*]]:2 = hlfir.declare %{{.*}}(%{{.*}}) {uniq_name = "_QFsub2Eahost"} : (!fir.ref<!fir.array<10xi32>>, !fir.shape<1>) -> (!fir.ref<!fir.array<10xi32>>, !fir.ref<!fir.array<10xi32>>)
+! CHECK: %[[BDEV:.*]]:2 = hlfir.declare %{{.*}}(%{{.*}}) {cuda_attr = #fir.cuda<device>, uniq_name = "_QFsub2Ebdev"} : (!fir.ref<!fir.array<10xi32>>, !fir.shape<1>) -> (!fir.ref<!fir.array<10xi32>>, !fir.ref<!fir.array<10xi32>>)
+! CHECK: %[[BHOST:.*]]:2 = hlfir.declare %{{.*}}(%{{.*}}) {uniq_name = "_QFsub2Ebhost"} : (!fir.ref<!fir.array<10xi32>>, !fir.shape<1>) -> (!fir.ref<!fir.array<10xi32>>, !fir.ref<!fir.array<10xi32>>)
+! CHECK: %[[I:.*]]:2 = hlfir.declare %{{.*}} {uniq_name = "_QFsub2Ei"} : (!fir.ref<i32>) -> (!fir.ref<i32>, !fir.ref<i32>)
+! CHECK: %[[M:.*]]:2 = hlfir.declare %{{.*}} {cuda_attr = #fir.cuda<device>, uniq_name = "_QFsub2Em"} : (!fir.ref<i32>) -> (!fir.ref<i32>, !fir.ref<i32>)
+! CHECK: fir.cuda_data_transfer %[[ADEV]]#0 to %[[AHOST]]#0 {transfer_kind = #fir.cuda_transfer<device_host>} : !fir.ref<!fir.array<10xi32>>, !fir.ref<!fir.array<10xi32>>
+! CHECK: fir.cuda_data_transfer %[[M]]#0 to %[[I]]#0 {transfer_kind = #fir.cuda_transfer<device_host>} : !fir.ref<i32>, !fir.ref<i32>
+
+! CHECK: %[[DES_ADEV:.*]] = hlfir.designate %[[ADEV]]#0 (%{{.*}}:%{{.*}}:%{{.*}})  shape %{{.*}} : (!fir.ref<!fir.array<10xi32>>, index, index, index, !fir.shape<1>) -> !fir.ref<!fir.array<5xi32>>
+! CHECK: %[[DES_AHOST:.*]] = hlfir.designate %[[AHOST]]#0 (%{{.*}}:%{{.*}}:%{{.*}})  shape %{{.*}} : (!fir.ref<!fir.array<10xi32>>, index, index, index, !fir.shape<1>) -> !fir.ref<!fir.array<5xi32>>
+! CHECK: fir.cuda_data_transfer %[[DES_ADEV]] to %[[DES_AHOST]] {transfer_kind = #fir.cuda_transfer<device_host>} : !fir.ref<!fir.array<5xi32>>, !fir.ref<!fir.array<5xi32>>
+
+! CHECK: fir.cuda_data_transfer %[[ADEV]]#0 to %[[BDEV]]#0 {transfer_kind = #fir.cuda_transfer<device_device>} : !fir.ref<!fir.array<10xi32>>, !fir.ref<!fir.array<10xi32>>
+
+! CHECK: %[[TEMP:.*]] = fir.allocmem !fir.array<10xi32> {bindc_name = ".tmp", uniq_name = ""}
+! CHECK: %[[DECL_TEMP:.*]]:2 = hlfir.declare %[[TEMP]](%{{.*}}) {uniq_name = ".tmp"} : (!fir.heap<!fir.array<10xi32>>, !fir.shape<1>) -> (!fir.heap<!fir.array<10xi32>>, !fir.heap<!fir.array<10xi32>>)
+! CHECK: %[[ADEV_TEMP:.*]]:2 = hlfir.declare %21#0 {cuda_attr = #fir.cuda<device>, uniq_name = "_QFsub2Eadev"} : (!fir.heap<!fir.array<10xi32>>) -> (!fir.heap<!fir.array<10xi32>>, !fir.heap<!fir.array<10xi32>>)
+! CHECK: fir.cuda_data_transfer %[[ADEV]]#1 to %[[DECL_TEMP]]#0 {transfer_kind = #fir.cuda_transfer<device_host>} : !fir.ref<!fir.array<10xi32>>, !fir.heap<!fir.array<10xi32>>
+! CHECK: %[[ELEMENTAL:.*]] = hlfir.elemental %{{.*}} unordered : (!fir.shape<1>) -> !hlfir.expr<10xi32>
+! CHECK: hlfir.assign %[[ELEMENTAL]] to %[[BHOST]]#0 : !hlfir.expr<10xi32>, !fir.ref<!fir.array<10xi32>>
+! CHECK: fir.freemem %[[DECL_TEMP]]#0 : !fir.heap<!fir.array<10xi32>>
+
+subroutine sub3()
+  use mod1
+  type(t1), device :: t
+  integer :: ahost(10), bhost(10)
+
+  bhost = ahost + t%i
+end
+
+! CHECK-LABEL: func.func @_QPsub3()
+! CHECK: %[[TMP:.*]] = fir.alloca !fir.type<_QMmod1Tt1{i:i32}> {bindc_name = ".tmp"}
+! CHECK: %[[AHOST:.*]]:2 = hlfir.declare %{{.*}}(%{{.*}}) {uniq_name = "_QFsub3Eahost"} : (!fir.ref<!fir.array<10xi32>>, !fir.shape<1>) -> (!fir.ref<!fir.array<10xi32>>, !fir.ref<!fir.array<10xi32>>)
+! CHECK: %[[BHOST:.*]]:2 = hlfir.declare %{{.*}}(%{{.*}}) {uniq_name = "_QFsub3Ebhost"} : (!fir.ref<!fir.array<10xi32>>, !fir.shape<1>) -> (!fir.ref<!fir.array<10xi32>>, !fir.ref<!fir.array<10xi32>>)
+! CHECK: %[[T:.*]]:2 = hlfir.declare %7 {cuda_attr = #fir.cuda<device>, uniq_name = "_QFsub3Et"} : (!fir.ref<!fir.type<_QMmod1Tt1{i:i32}>>) -> (!fir.ref<!fir.type<_QMmod1Tt1{i:i32}>>, !fir.ref<!fir.type<_QMmod1Tt1{i:i32}>>)
+! CHECK: %[[TMP_DECL:.*]]:2 = hlfir.declare %0 {uniq_name = ".tmp"} : (!fir.ref<!fir.type<_QMmod1Tt1{i:i32}>>) -> (!fir.ref<!fir.type<_QMmod1Tt1{i:i32}>>, !fir.ref<!fir.type<_QMmod1Tt1{i:i32}>>)
+! CHECK: fir.cuda_data_transfer %[[T]]#1 to %[[TMP_DECL]]#0 {transfer_kind = #fir.cuda_transfer<device_host>} : !fir.ref<!fir.type<_QMmod1Tt1{i:i32}>>, !fir.ref<!fir.type<_QMmod1Tt1{i:i32}>>
diff --git a/libc/config/linux/x86_64/entrypoints.txt b/libc/config/linux/x86_64/entrypoints.txt
index 2742c33ae478..b522e7f03015 100644
--- a/libc/config/linux/x86_64/entrypoints.txt
+++ b/libc/config/linux/x86_64/entrypoints.txt
@@ -494,6 +494,9 @@ set(TARGET_LIBM_ENTRYPOINTS
     libc.src.math.round
     libc.src.math.roundf
     libc.src.math.roundl
+    libc.src.math.roundeven
+    libc.src.math.roundevenf
+    libc.src.math.roundevenl
     libc.src.math.scalbn
     libc.src.math.scalbnf
     libc.src.math.scalbnl
@@ -555,6 +558,7 @@ if(LIBC_TYPES_HAS_FLOAT128)
     libc.src.math.nextdownf128
     libc.src.math.nextupf128
     libc.src.math.rintf128
+    libc.src.math.roundevenf128
     libc.src.math.roundf128
     libc.src.math.sqrtf128
     libc.src.math.truncf128
diff --git a/libc/docs/c23.rst b/libc/docs/c23.rst
index 3f64722bc8e6..152178eaf751 100644
--- a/libc/docs/c23.rst
+++ b/libc/docs/c23.rst
@@ -54,7 +54,7 @@ Additions:
   * pown*
   * powr*
   * rootn*
-  * roundeven*
+  * roundeven* |check|
   * fromfp*
   * ufromfp*
   * fromfpx*
diff --git a/libc/docs/math/index.rst b/libc/docs/math/index.rst
index 970a43ca87c9..7a7b6c9c8db5 100644
--- a/libc/docs/math/index.rst
+++ b/libc/docs/math/index.rst
@@ -206,7 +206,7 @@ Basic Operations
 +------------------+------------------+-----------------+------------------------+----------------------+------------------------+------------------------+----------------------------+
 | round            | |check|          | |check|         | |check|                |                      | |check|                | 7.12.9.6               | F.10.6.6                   |
 +------------------+------------------+-----------------+------------------------+----------------------+------------------------+------------------------+----------------------------+
-| roundeven        |                  |                 |                        |                      |                        | 7.12.9.8               | F.10.6.8                   |
+| roundeven        | |check|          | |check|         | |check|                |                      | |check|                | 7.12.9.8               | F.10.6.8                   |
 +------------------+------------------+-----------------+------------------------+----------------------+------------------------+------------------------+----------------------------+
 | scalbn           | |check|          | |check|         | |check|                |                      |                        | 7.12.6.19              | F.10.3.19                  |
 +------------------+------------------+-----------------+------------------------+----------------------+------------------------+------------------------+----------------------------+
diff --git a/libc/src/math/CMakeLists.txt b/libc/src/math/CMakeLists.txt
index c89792b8ac7b..e8f699fabe36 100644
--- a/libc/src/math/CMakeLists.txt
+++ b/libc/src/math/CMakeLists.txt
@@ -293,6 +293,11 @@ add_math_entrypoint_object(roundf)
 add_math_entrypoint_object(roundl)
 add_math_entrypoint_object(roundf128)
 
+add_math_entrypoint_object(roundeven)
+add_math_entrypoint_object(roundevenf)
+add_math_entrypoint_object(roundevenl)
+add_math_entrypoint_object(roundevenf128)
+
 add_math_entrypoint_object(scalbn)
 add_math_entrypoint_object(scalbnf)
 add_math_entrypoint_object(scalbnl)
diff --git a/libc/src/math/generic/CMakeLists.txt b/libc/src/math/generic/CMakeLists.txt
index dc77f8b5ddba..5bc9066c6263 100644
--- a/libc/src/math/generic/CMakeLists.txt
+++ b/libc/src/math/generic/CMakeLists.txt
@@ -418,6 +418,55 @@ add_entrypoint_object(
 )
 
 add_entrypoint_object(
+  roundeven
+  SRCS
+    roundeven.cpp
+  HDRS
+    ../roundeven.h
+  COMPILE_OPTIONS
+    -O3
+  DEPENDS
+    libc.src.__support.FPUtil.nearest_integer_operations
+)
+
+add_entrypoint_object(
+  roundevenf
+  SRCS
+    roundevenf.cpp
+  HDRS
+    ../roundevenf.h
+  COMPILE_OPTIONS
+    -O3
+  DEPENDS
+    libc.src.__support.FPUtil.nearest_integer_operations
+)
+
+add_entrypoint_object(
+  roundevenl
+  SRCS
+    roundevenl.cpp
+  HDRS
+    ../roundevenl.h
+  COMPILE_OPTIONS
+    -O3
+  DEPENDS
+    libc.src.__support.FPUtil.nearest_integer_operations
+)
+
+add_entrypoint_object(
+  roundevenf128
+  SRCS
+    roundevenf128.cpp
+  HDRS
+    ../roundevenf128.h
+  COMPILE_OPTIONS
+    -O3
+  DEPENDS
+    libc.src.__support.macros.properties.types
+    libc.src.__support.FPUtil.nearest_integer_operations
+)
+
+add_entrypoint_object(
   lround
   SRCS
     lround.cpp
diff --git a/libc/src/math/generic/roundeven.cpp b/libc/src/math/generic/roundeven.cpp
new file mode 100644
index 000000000000..5f2adf9b5fce
--- /dev/null
+++ b/libc/src/math/generic/roundeven.cpp
@@ -0,0 +1,19 @@
+//===-- Implementation of roundeven function ------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#include "src/math/roundeven.h"
+#include "src/__support/FPUtil/NearestIntegerOperations.h"
+#include "src/__support/common.h"
+
+namespace LIBC_NAMESPACE {
+
+LLVM_LIBC_FUNCTION(double, roundeven, (double x)) {
+  return fputil::round_using_specific_rounding_mode(x, FP_INT_TONEAREST);
+}
+
+} // namespace LIBC_NAMESPACE
diff --git a/libc/src/math/generic/roundevenf.cpp b/libc/src/math/generic/roundevenf.cpp
new file mode 100644
index 000000000000..353bec74ecf0
--- /dev/null
+++ b/libc/src/math/generic/roundevenf.cpp
@@ -0,0 +1,19 @@
+//===-- Implementation of roundevenf function -----------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#include "src/math/roundevenf.h"
+#include "src/__support/FPUtil/NearestIntegerOperations.h"
+#include "src/__support/common.h"
+
+namespace LIBC_NAMESPACE {
+
+LLVM_LIBC_FUNCTION(float, roundevenf, (float x)) {
+  return fputil::round_using_specific_rounding_mode(x, FP_INT_TONEAREST);
+}
+
+} // namespace LIBC_NAMESPACE
diff --git a/libc/src/math/generic/roundevenf128.cpp b/libc/src/math/generic/roundevenf128.cpp
new file mode 100644
index 000000000000..259388c86fd3
--- /dev/null
+++ b/libc/src/math/generic/roundevenf128.cpp
@@ -0,0 +1,19 @@
+//===-- Implementation of roundevenf128 function --------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#include "src/math/roundevenf128.h"
+#include "src/__support/FPUtil/NearestIntegerOperations.h"
+#include "src/__support/common.h"
+
+namespace LIBC_NAMESPACE {
+
+LLVM_LIBC_FUNCTION(float128, roundevenf128, (float128 x)) {
+  return fputil::round_using_specific_rounding_mode(x, FP_INT_TONEAREST);
+}
+
+} // namespace LIBC_NAMESPACE
diff --git a/libc/src/math/generic/roundevenl.cpp b/libc/src/math/generic/roundevenl.cpp
new file mode 100644
index 000000000000..f8f429faeec8
--- /dev/null
+++ b/libc/src/math/generic/roundevenl.cpp
@@ -0,0 +1,19 @@
+//===-- Implementation of roundevenl function -----------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#include "src/math/roundevenl.h"
+#include "src/__support/FPUtil/NearestIntegerOperations.h"
+#include "src/__support/common.h"
+
+namespace LIBC_NAMESPACE {
+
+LLVM_LIBC_FUNCTION(long double, roundevenl, (long double x)) {
+  return fputil::round_using_specific_rounding_mode(x, FP_INT_TONEAREST);
+}
+
+} // namespace LIBC_NAMESPACE
diff --git a/libc/src/math/roundeven.h b/libc/src/math/roundeven.h
new file mode 100644
index 000000000000..9c76b1fe334a
--- /dev/null
+++ b/libc/src/math/roundeven.h
@@ -0,0 +1,18 @@
+//===-- Implementation header for roundeven ---------------------*- C++ -*-===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_LIBC_SRC_MATH_ROUNDEVEN_H
+#define LLVM_LIBC_SRC_MATH_ROUNDEVEN_H
+
+namespace LIBC_NAMESPACE {
+
+double roundeven(double x);
+
+} // namespace LIBC_NAMESPACE
+
+#endif // LLVM_LIBC_SRC_MATH_ROUNDEVEN_H
diff --git a/libc/src/math/roundevenf.h b/libc/src/math/roundevenf.h
new file mode 100644
index 000000000000..447e7fd940c1
--- /dev/null
+++ b/libc/src/math/roundevenf.h
@@ -0,0 +1,18 @@
+//===-- Implementation header for roundevenf --------------------*- C++ -*-===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_LIBC_SRC_MATH_ROUNDEVENF_H
+#define LLVM_LIBC_SRC_MATH_ROUNDEVENF_H
+
+namespace LIBC_NAMESPACE {
+
+float roundevenf(float x);
+
+} // namespace LIBC_NAMESPACE
+
+#endif // LLVM_LIBC_SRC_MATH_ROUNDEVENF_H
diff --git a/libc/src/math/roundevenf128.h b/libc/src/math/roundevenf128.h
new file mode 100644
index 000000000000..589839d09075
--- /dev/null
+++ b/libc/src/math/roundevenf128.h
@@ -0,0 +1,20 @@
+//===-- Implementation header for roundevenf128 -----------------*- C++ -*-===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_LIBC_SRC_MATH_ROUNDEVENF128_H
+#define LLVM_LIBC_SRC_MATH_ROUNDEVENF128_H
+
+#include "src/__support/macros/properties/types.h"
+
+namespace LIBC_NAMESPACE {
+
+float128 roundevenf128(float128 x);
+
+} // namespace LIBC_NAMESPACE
+
+#endif // LLVM_LIBC_SRC_MATH_ROUNDEVENF128_H
diff --git a/libc/src/math/roundevenl.h b/libc/src/math/roundevenl.h
new file mode 100644
index 000000000000..a2f3397e4479
--- /dev/null
+++ b/libc/src/math/roundevenl.h
@@ -0,0 +1,18 @@
+//===-- Implementation header for roundevenl --------------------*- C++ -*-===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_LIBC_SRC_MATH_ROUNDEVENL_H
+#define LLVM_LIBC_SRC_MATH_ROUNDEVENL_H
+
+namespace LIBC_NAMESPACE {
+
+long double roundevenl(long double x);
+
+} // namespace LIBC_NAMESPACE
+
+#endif // LLVM_LIBC_SRC_MATH_ROUNDEVENL_H
diff --git a/libc/test/src/math/CMakeLists.txt b/libc/test/src/math/CMakeLists.txt
index bbf8f071e1e0..d928cb5c6475 100644
--- a/libc/test/src/math/CMakeLists.txt
+++ b/libc/test/src/math/CMakeLists.txt
@@ -324,6 +324,51 @@ add_fp_unittest(
 )
 
 add_fp_unittest(
+  roundeven_test
+  NEED_MPFR
+  SUITE
+    libc-math-unittests
+  SRCS
+    roundeven_test.cpp
+  HDRS
+  RoundEvenTest.h
+  DEPENDS
+    libc.include.math
+    libc.src.math.roundeven
+    libc.src.__support.FPUtil.fp_bits
+)
+
+add_fp_unittest(
+  roundevenf_test
+  NEED_MPFR
+  SUITE
+    libc-math-unittests
+  SRCS
+    roundevenf_test.cpp
+  HDRS
+  RoundEvenTest.h
+  DEPENDS
+    libc.include.math
+    libc.src.math.roundevenf
+    libc.src.__support.FPUtil.fp_bits
+)
+
+add_fp_unittest(
+  roundevenl_test
+  NEED_MPFR
+  SUITE
+    libc-math-unittests
+  SRCS
+    roundevenl_test.cpp
+  HDRS
+  RoundEvenTest.h
+  DEPENDS
+    libc.include.math
+    libc.src.math.roundevenl
+    libc.src.__support.FPUtil.fp_bits
+)
+
+add_fp_unittest(
   lround_test
   NEED_MPFR
   SUITE
diff --git a/libc/test/src/math/RoundEvenTest.h b/libc/test/src/math/RoundEvenTest.h
new file mode 100644
index 000000000000..db7263a39c0f
--- /dev/null
+++ b/libc/test/src/math/RoundEvenTest.h
@@ -0,0 +1,92 @@
+//===-- Utility class to test roundeven[f|l] --------------------*- C++ -*-===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_LIBC_TEST_SRC_MATH_ROUNDEVENTEST_H
+#define LLVM_LIBC_TEST_SRC_MATH_ROUNDEVENTEST_H
+
+#include "test/UnitTest/FPMatcher.h"
+#include "test/UnitTest/Test.h"
+#include "utils/MPFRWrapper/MPFRUtils.h"
+
+#include "include/llvm-libc-macros/math-macros.h"
+
+namespace mpfr = LIBC_NAMESPACE::testing::mpfr;
+
+template <typename T>
+class RoundEvenTest : public LIBC_NAMESPACE::testing::Test {
+
+  DECLARE_SPECIAL_CONSTANTS(T)
+
+public:
+  typedef T (*RoundEvenFunc)(T);
+
+  void testSpecialNumbers(RoundEvenFunc func) {
+    EXPECT_FP_EQ(zero, func(zero));
+    EXPECT_FP_EQ(neg_zero, func(neg_zero));
+
+    EXPECT_FP_EQ(inf, func(inf));
+    EXPECT_FP_EQ(neg_inf, func(neg_inf));
+
+    EXPECT_FP_EQ(aNaN, func(aNaN));
+  }
+
+  void testRoundedNumbers(RoundEvenFunc func) {
+    EXPECT_FP_EQ(T(1.0), func(T(1.0)));
+    EXPECT_FP_EQ(T(-1.0), func(T(-1.0)));
+    EXPECT_FP_EQ(T(10.0), func(T(10.0)));
+    EXPECT_FP_EQ(T(-10.0), func(T(-10.0)));
+    EXPECT_FP_EQ(T(1234.0), func(T(1234.0)));
+    EXPECT_FP_EQ(T(-1234.0), func(T(-1234.0)));
+  }
+
+  void testFractions(RoundEvenFunc func) {
+    EXPECT_FP_EQ(T(0.0), func(T(0.5)));
+    EXPECT_FP_EQ(T(-0.0), func(T(-0.5)));
+    EXPECT_FP_EQ(T(0.0), func(T(0.115)));
+    EXPECT_FP_EQ(T(-0.0), func(T(-0.115)));
+    EXPECT_FP_EQ(T(1.0), func(T(0.715)));
+    EXPECT_FP_EQ(T(-1.0), func(T(-0.715)));
+    EXPECT_FP_EQ(T(1.0), func(T(1.3)));
+    EXPECT_FP_EQ(T(-1.0), func(T(-1.3)));
+    EXPECT_FP_EQ(T(2.0), func(T(1.5)));
+    EXPECT_FP_EQ(T(-2.0), func(T(-1.5)));
+    EXPECT_FP_EQ(T(2.0), func(T(1.75)));
+    EXPECT_FP_EQ(T(-2.0), func(T(-1.75)));
+    EXPECT_FP_EQ(T(11.0), func(T(10.65)));
+    EXPECT_FP_EQ(T(-11.0), func(T(-10.65)));
+    EXPECT_FP_EQ(T(1233.0), func(T(1233.25)));
+    EXPECT_FP_EQ(T(1234.0), func(T(1233.50)));
+    EXPECT_FP_EQ(T(1234.0), func(T(1233.75)));
+    EXPECT_FP_EQ(T(-1233.0), func(T(-1233.25)));
+    EXPECT_FP_EQ(T(-1234.0), func(T(-1233.50)));
+    EXPECT_FP_EQ(T(-1234.0), func(T(-1233.75)));
+    EXPECT_FP_EQ(T(1234.0), func(T(1234.50)));
+    EXPECT_FP_EQ(T(-1234.0), func(T(-1234.50)));
+  }
+
+  void testRange(RoundEvenFunc func) {
+    constexpr StorageType COUNT = 100'000;
+    constexpr StorageType STEP = STORAGE_MAX / COUNT;
+    for (StorageType i = 0, v = 0; i <= COUNT; ++i, v += STEP) {
+      T x = FPBits(v).get_val();
+      if (isnan(x) || isinf(x))
+        continue;
+
+      ASSERT_MPFR_MATCH(mpfr::Operation::RoundEven, x, func(x), 0.0);
+    }
+  }
+};
+
+#define LIST_ROUNDEVEN_TESTS(T, func)                                          \
+  using LlvmLibcRoundEvenTest = RoundEvenTest<T>;                              \
+  TEST_F(LlvmLibcRoundEvenTest, SpecialNumbers) { testSpecialNumbers(&func); } \
+  TEST_F(LlvmLibcRoundEvenTest, RoundedNubmers) { testRoundedNumbers(&func); } \
+  TEST_F(LlvmLibcRoundEvenTest, Fractions) { testFractions(&func); }           \
+  TEST_F(LlvmLibcRoundEvenTest, Range) { testRange(&func); }
+
+#endif // LLVM_LIBC_TEST_SRC_MATH_ROUNDEVENTEST_H
diff --git a/libc/test/src/math/roundeven_test.cpp b/libc/test/src/math/roundeven_test.cpp
new file mode 100644
index 000000000000..cd1a7bf2429f
--- /dev/null
+++ b/libc/test/src/math/roundeven_test.cpp
@@ -0,0 +1,13 @@
+//===-- Unittests for roundeven -------------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#include "RoundEvenTest.h"
+
+#include "src/math/roundeven.h"
+
+LIST_ROUNDEVEN_TESTS(double, LIBC_NAMESPACE::roundeven)
diff --git a/libc/test/src/math/roundevenf_test.cpp b/libc/test/src/math/roundevenf_test.cpp
new file mode 100644
index 000000000000..68dff9b3eca9
--- /dev/null
+++ b/libc/test/src/math/roundevenf_test.cpp
@@ -0,0 +1,13 @@
+//===-- Unittests for roundevenf ------------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#include "RoundEvenTest.h"
+
+#include "src/math/roundevenf.h"
+
+LIST_ROUNDEVEN_TESTS(float, LIBC_NAMESPACE::roundevenf)
diff --git a/libc/test/src/math/roundevenl_test.cpp b/libc/test/src/math/roundevenl_test.cpp
new file mode 100644
index 000000000000..f4031bd65ec2
--- /dev/null
+++ b/libc/test/src/math/roundevenl_test.cpp
@@ -0,0 +1,13 @@
+//===-- Unittests for roundevenl ------------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#include "RoundEvenTest.h"
+
+#include "src/math/roundevenl.h"
+
+LIST_ROUNDEVEN_TESTS(long double, LIBC_NAMESPACE::roundevenl)
diff --git a/libc/test/src/math/smoke/CMakeLists.txt b/libc/test/src/math/smoke/CMakeLists.txt
index 4ac1842cf5fa..e9ff35577b95 100644
--- a/libc/test/src/math/smoke/CMakeLists.txt
+++ b/libc/test/src/math/smoke/CMakeLists.txt
@@ -398,6 +398,62 @@ add_fp_unittest(
 )
 
 add_fp_unittest(
+  roundeven_test
+  SUITE
+    libc-math-smoke-tests
+  SRCS
+    roundeven_test.cpp
+  HDRS
+  RoundEvenTest.h
+  DEPENDS
+    libc.include.math
+    libc.src.math.roundeven
+    libc.src.__support.FPUtil.fp_bits
+)
+
+add_fp_unittest(
+  roundevenf_test
+  SUITE
+    libc-math-smoke-tests
+  SRCS
+    roundevenf_test.cpp
+  HDRS
+  RoundEvenTest.h
+  DEPENDS
+    libc.include.math
+    libc.src.math.roundevenf
+    libc.src.__support.FPUtil.fp_bits
+)
+
+add_fp_unittest(
+  roundevenl_test
+  SUITE
+    libc-math-smoke-tests
+  SRCS
+    roundevenl_test.cpp
+  HDRS
+  RoundEvenTest.h
+  DEPENDS
+    libc.include.math
+    libc.src.math.roundevenl
+    libc.src.__support.FPUtil.fp_bits
+)
+
+add_fp_unittest(
+  roundevenf128_test
+  SUITE
+    libc-math-smoke-tests
+  SRCS
+    roundevenf128_test.cpp
+  HDRS
+  RoundEvenTest.h
+  DEPENDS
+    libc.include.math
+    libc.src.math.roundevenf128
+    libc.src.__support.FPUtil.fp_bits
+)
+
+add_fp_unittest(
   lround_test
   SUITE
     libc-math-smoke-tests
diff --git a/libc/test/src/math/smoke/RoundEvenTest.h b/libc/test/src/math/smoke/RoundEvenTest.h
new file mode 100644
index 000000000000..107052fa0e28
--- /dev/null
+++ b/libc/test/src/math/smoke/RoundEvenTest.h
@@ -0,0 +1,72 @@
+//===-- Utility class to test roundeven[f|l] --------------------*- C++ -*-===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_LIBC_TEST_SRC_MATH_SMOKE_ROUNDEVENTEST_H
+#define LLVM_LIBC_TEST_SRC_MATH_SMOKE_ROUNDEVENTEST_H
+
+#include "test/UnitTest/FPMatcher.h"
+#include "test/UnitTest/Test.h"
+
+#include "include/llvm-libc-macros/math-macros.h"
+
+template <typename T>
+class RoundEvenTest : public LIBC_NAMESPACE::testing::Test {
+
+  DECLARE_SPECIAL_CONSTANTS(T)
+
+public:
+  typedef T (*RoundEvenFunc)(T);
+
+  void testSpecialNumbers(RoundEvenFunc func) {
+    EXPECT_FP_EQ(zero, func(zero));
+    EXPECT_FP_EQ(neg_zero, func(neg_zero));
+
+    EXPECT_FP_EQ(inf, func(inf));
+    EXPECT_FP_EQ(neg_inf, func(neg_inf));
+
+    EXPECT_FP_EQ(aNaN, func(aNaN));
+  }
+
+  void testRoundedNumbers(RoundEvenFunc func) {
+    EXPECT_FP_EQ(T(1.0), func(T(1.0)));
+    EXPECT_FP_EQ(T(-1.0), func(T(-1.0)));
+    EXPECT_FP_EQ(T(10.0), func(T(10.0)));
+    EXPECT_FP_EQ(T(-10.0), func(T(-10.0)));
+    EXPECT_FP_EQ(T(1234.0), func(T(1234.0)));
+    EXPECT_FP_EQ(T(-1234.0), func(T(-1234.0)));
+  }
+
+  void testFractions(RoundEvenFunc func) {
+    EXPECT_FP_EQ(T(0.0), func(T(0.5)));
+    EXPECT_FP_EQ(T(-0.0), func(T(-0.5)));
+    EXPECT_FP_EQ(T(0.0), func(T(0.115)));
+    EXPECT_FP_EQ(T(-0.0), func(T(-0.115)));
+    EXPECT_FP_EQ(T(1.0), func(T(0.715)));
+    EXPECT_FP_EQ(T(-1.0), func(T(-0.715)));
+    EXPECT_FP_EQ(T(2.0), func(T(1.5)));
+    EXPECT_FP_EQ(T(-2.0), func(T(-1.5)));
+    EXPECT_FP_EQ(T(2.0), func(T(1.75)));
+    EXPECT_FP_EQ(T(-2.0), func(T(-1.75)));
+    EXPECT_FP_EQ(T(10.0), func(T(10.50)));
+    EXPECT_FP_EQ(T(-10.0), func(T(-10.50)));
+    EXPECT_FP_EQ(T(11.0), func(T(10.65)));
+    EXPECT_FP_EQ(T(-11.0), func(T(-10.65)));
+    EXPECT_FP_EQ(T(1234.0), func(T(1234.50)));
+    EXPECT_FP_EQ(T(-1234.0), func(T(-1234.50)));
+    EXPECT_FP_EQ(T(1236.0), func(T(1235.50)));
+    EXPECT_FP_EQ(T(-1236.0), func(T(-1235.50)));
+  }
+};
+
+#define LIST_ROUNDEVEN_TESTS(T, func)                                          \
+  using LlvmLibcRoundEvenTest = RoundEvenTest<T>;                              \
+  TEST_F(LlvmLibcRoundEvenTest, SpecialNumbers) { testSpecialNumbers(&func); } \
+  TEST_F(LlvmLibcRoundEvenTest, RoundedNubmers) { testRoundedNumbers(&func); } \
+  TEST_F(LlvmLibcRoundEvenTest, Fractions) { testFractions(&func); }
+
+#endif // LLVM_LIBC_TEST_SRC_MATH_SMOKE_ROUNDEVENTEST_H
diff --git a/libc/test/src/math/smoke/roundeven_test.cpp b/libc/test/src/math/smoke/roundeven_test.cpp
new file mode 100644
index 000000000000..e2d625fb0d31
--- /dev/null
+++ b/libc/test/src/math/smoke/roundeven_test.cpp
@@ -0,0 +1,12 @@
+//===-- Unittests for roundeven -------------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#include "RoundEvenTest.h"
+#include "src/math/roundeven.h"
+
+LIST_ROUNDEVEN_TESTS(double, LIBC_NAMESPACE::roundeven)
diff --git a/libc/test/src/math/smoke/roundevenf128_test.cpp b/libc/test/src/math/smoke/roundevenf128_test.cpp
new file mode 100644
index 000000000000..a1fdc40d577e
--- /dev/null
+++ b/libc/test/src/math/smoke/roundevenf128_test.cpp
@@ -0,0 +1,12 @@
+//===-- Unittests for roundevenf128 ---------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#include "RoundEvenTest.h"
+#include "src/math/roundevenf128.h"
+
+LIST_ROUNDEVEN_TESTS(float128, LIBC_NAMESPACE::roundevenf128)
diff --git a/libc/test/src/math/smoke/roundevenf_test.cpp b/libc/test/src/math/smoke/roundevenf_test.cpp
new file mode 100644
index 000000000000..f033e26988fa
--- /dev/null
+++ b/libc/test/src/math/smoke/roundevenf_test.cpp
@@ -0,0 +1,12 @@
+//===-- Unittests for roundevenf ------------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#include "RoundEvenTest.h"
+#include "src/math/roundevenf.h"
+
+LIST_ROUNDEVEN_TESTS(float, LIBC_NAMESPACE::roundevenf)
diff --git a/libc/test/src/math/smoke/roundevenl_test.cpp b/libc/test/src/math/smoke/roundevenl_test.cpp
new file mode 100644
index 000000000000..be09f1283aa7
--- /dev/null
+++ b/libc/test/src/math/smoke/roundevenl_test.cpp
@@ -0,0 +1,12 @@
+//===-- Unittests for roundevenf ------------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#include "RoundEvenTest.h"
+#include "src/math/roundevenl.h"
+
+LIST_ROUNDEVEN_TESTS(long double, LIBC_NAMESPACE::roundevenl)
diff --git a/libc/utils/MPFRWrapper/MPFRUtils.cpp b/libc/utils/MPFRWrapper/MPFRUtils.cpp
index eaa47da6bda2..938e2c3c0edf 100644
--- a/libc/utils/MPFRWrapper/MPFRUtils.cpp
+++ b/libc/utils/MPFRWrapper/MPFRUtils.cpp
@@ -351,6 +351,16 @@ public:
     return result;
   }
 
+  MPFRNumber roundeven() const {
+    MPFRNumber result(*this);
+#if MPFR_VERSION_MAJOR >= 4
+    mpfr_roundeven(result.value, value);
+#else
+    mpfr_rint(result.value, value, MPFR_RNDN);
+#endif
+    return result;
+  }
+
   bool round_to_long(long &result) const {
     // We first calculate the rounded value. This way, when converting
     // to long using mpfr_get_si, the rounding direction of MPFR_RNDN
@@ -634,6 +644,8 @@ unary_operation(Operation op, InputType input, unsigned int precision,
     return mpfrInput.mod_pi_over_4();
   case Operation::Round:
     return mpfrInput.round();
+  case Operation::RoundEven:
+    return mpfrInput.roundeven();
   case Operation::Sin:
     return mpfrInput.sin();
   case Operation::Sinh:
diff --git a/libc/utils/MPFRWrapper/MPFRUtils.h b/libc/utils/MPFRWrapper/MPFRUtils.h
index 0a41ac639798..d2f73e2628e1 100644
--- a/libc/utils/MPFRWrapper/MPFRUtils.h
+++ b/libc/utils/MPFRWrapper/MPFRUtils.h
@@ -49,6 +49,7 @@ enum class Operation : int {
   ModPIOver2,
   ModPIOver4,
   Round,
+  RoundEven,
   Sin,
   Sinh,
   Sqrt,
diff --git a/libcxx/include/__format/formatter_floating_point.h b/libcxx/include/__format/formatter_floating_point.h
index 1d94cc349c0d..fa42ba203b0b 100644
--- a/libcxx/include/__format/formatter_floating_point.h
+++ b/libcxx/include/__format/formatter_floating_point.h
@@ -690,7 +690,7 @@ __format_floating_point(_Tp __value, _FormatContext& __ctx, __format_spec::__par
       // Let P equal the precision if nonzero, 6 if the precision is not
       // specified, or 1 if the precision is 0. Then, if a conversion with
       // style E would have an exponent of X:
-      int __p = std::max(1, (__specs.__has_precision() ? __specs.__precision_ : 6));
+      int __p = std::max<int>(1, (__specs.__has_precision() ? __specs.__precision_ : 6));
       if (__result.__exponent == __result.__last)
         // if P > X >= -4, the conversion is with style f or F and precision P - 1 - X.
         // By including the radix point it calculates P - (1 + X)
diff --git a/libcxx/include/__memory/construct_at.h b/libcxx/include/__memory/construct_at.h
index 91d17134db01..eb0213248006 100644
--- a/libcxx/include/__memory/construct_at.h
+++ b/libcxx/include/__memory/construct_at.h
@@ -44,7 +44,7 @@ _LIBCPP_HIDE_FROM_ABI constexpr _Tp* construct_at(_Tp* __location, _Args&&... __
 #endif
 
 template <class _Tp, class... _Args, class = decltype(::new(std::declval<void*>()) _Tp(std::declval<_Args>()...))>
-_LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR _Tp* __construct_at(_Tp* __location, _Args&&... __args) {
+_LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX20 _Tp* __construct_at(_Tp* __location, _Args&&... __args) {
 #if _LIBCPP_STD_VER >= 20
   return std::construct_at(__location, std::forward<_Args>(__args)...);
 #else
diff --git a/lld/COFF/PDB.cpp b/lld/COFF/PDB.cpp
index 35e4c68dcda6..c0739b37aeb0 100644
--- a/lld/COFF/PDB.cpp
+++ b/lld/COFF/PDB.cpp
@@ -1726,15 +1726,15 @@ void PDBLinker::commit(codeview::GUID *guid) {
   }
 }
 
-static uint32_t getSecrelReloc(llvm::COFF::MachineTypes machine) {
-  switch (machine) {
-  case AMD64:
+static uint32_t getSecrelReloc(Triple::ArchType arch) {
+  switch (arch) {
+  case Triple::x86_64:
     return COFF::IMAGE_REL_AMD64_SECREL;
-  case I386:
+  case Triple::x86:
     return COFF::IMAGE_REL_I386_SECREL;
-  case ARMNT:
+  case Triple::thumb:
     return COFF::IMAGE_REL_ARM_SECREL;
-  case ARM64:
+  case Triple::aarch64:
     return COFF::IMAGE_REL_ARM64_SECREL;
   default:
     llvm_unreachable("unknown machine type");
@@ -1752,7 +1752,7 @@ static bool findLineTable(const SectionChunk *c, uint32_t addr,
                           DebugLinesSubsectionRef &lines,
                           uint32_t &offsetInLinetable) {
   ExitOnError exitOnErr;
-  const uint32_t secrelReloc = getSecrelReloc(c->file->ctx.config.machine);
+  const uint32_t secrelReloc = getSecrelReloc(c->getArch());
 
   for (SectionChunk *dbgC : c->file->getDebugChunks()) {
     if (dbgC->getSectionName() != ".debug$S")
diff --git a/lld/MachO/DriverUtils.cpp b/lld/MachO/DriverUtils.cpp
index d6f18ecb85b8..077a639bf7ab 100644
--- a/lld/MachO/DriverUtils.cpp
+++ b/lld/MachO/DriverUtils.cpp
@@ -45,11 +45,21 @@ using namespace lld::macho;
 // Create table mapping all options defined in Options.td
 static constexpr OptTable::Info optInfo[] = {
 #define OPTION(PREFIX, NAME, ID, KIND, GROUP, ALIAS, ALIASARGS, FLAGS,         \
-               VISIBILITY, PARAM, HELPTEXT, METAVAR, VALUES)                   \
-  {PREFIX,      NAME,        HELPTEXT,                                         \
-   METAVAR,     OPT_##ID,    opt::Option::KIND##Class,                         \
-   PARAM,       FLAGS,       VISIBILITY,                                       \
-   OPT_##GROUP, OPT_##ALIAS, ALIASARGS,                                        \
+               VISIBILITY, PARAM, HELPTEXT, HELPTEXTSFORVARIANTS, METAVAR,     \
+               VALUES)                                                         \
+  {PREFIX,                                                                     \
+   NAME,                                                                       \
+   HELPTEXT,                                                                   \
+   HELPTEXTSFORVARIANTS,                                                       \
+   METAVAR,                                                                    \
+   OPT_##ID,                                                                   \
+   opt::Option::KIND##Class,                                                   \
+   PARAM,                                                                      \
+   FLAGS,                                                                      \
+   VISIBILITY,                                                                 \
+   OPT_##GROUP,                                                                \
+   OPT_##ALIAS,                                                                \
+   ALIASARGS,                                                                  \
    VALUES},
 #include "Options.inc"
 #undef OPTION
diff --git a/lld/MinGW/Driver.cpp b/lld/MinGW/Driver.cpp
index bb08c77b2e11..0d55d5b3672a 100644
--- a/lld/MinGW/Driver.cpp
+++ b/lld/MinGW/Driver.cpp
@@ -69,11 +69,21 @@ enum {
 // Create table mapping all options defined in Options.td
 static constexpr opt::OptTable::Info infoTable[] = {
 #define OPTION(PREFIX, NAME, ID, KIND, GROUP, ALIAS, ALIASARGS, FLAGS,         \
-               VISIBILITY, PARAM, HELPTEXT, METAVAR, VALUES)                   \
-  {PREFIX,      NAME,        HELPTEXT,                                         \
-   METAVAR,     OPT_##ID,    opt::Option::KIND##Class,                         \
-   PARAM,       FLAGS,       VISIBILITY,                                       \
-   OPT_##GROUP, OPT_##ALIAS, ALIASARGS,                                        \
+               VISIBILITY, PARAM, HELPTEXT, HELPTEXTSFORVARIANTS, METAVAR,     \
+               VALUES)                                                         \
+  {PREFIX,                                                                     \
+   NAME,                                                                       \
+   HELPTEXT,                                                                   \
+   HELPTEXTSFORVARIANTS,                                                       \
+   METAVAR,                                                                    \
+   OPT_##ID,                                                                   \
+   opt::Option::KIND##Class,                                                   \
+   PARAM,                                                                      \
+   FLAGS,                                                                      \
+   VISIBILITY,                                                                 \
+   OPT_##GROUP,                                                                \
+   OPT_##ALIAS,                                                                \
+   ALIASARGS,                                                                  \
    VALUES},
 #include "Options.inc"
 #undef OPTION
diff --git a/lld/test/COFF/undefined-symbol-cv.s b/lld/test/COFF/undefined-symbol-cv.s
index 08a85826b9f9..502f28044c17 100644
--- a/lld/test/COFF/undefined-symbol-cv.s
+++ b/lld/test/COFF/undefined-symbol-cv.s
@@ -1,6 +1,14 @@
-# REQUIRES: x86
-# RUN: llvm-mc -triple=x86_64-windows-msvc -filetype=obj -o %t.obj %s
-# RUN: not lld-link /out:%t.exe %t.obj 2>&1 | FileCheck %s
+# REQUIRES: aarch64, x86
+# RUN: split-file %s %t.dir && cd %t.dir
+
+# RUN: llvm-mc -triple=x86_64-windows-msvc -filetype=obj -o test-x86_64.obj test-x86_64.s
+# RUN: llvm-mc -triple=aarch64-windows-msvc -filetype=obj -o test-aarch64.obj test-aarch64.s
+# RUN: llvm-mc -triple=arm64ec-windows-msvc -filetype=obj -o test-arm64ec.obj test-aarch64.s
+
+# RUN: not lld-link -out:test-x86_64.exe test-x86_64.obj 2>&1 | FileCheck %s
+# RUN: not lld-link -out:test-aarch64.exe test-aarch64.obj 2>&1 | FileCheck %s
+# RUN: not lld-link -out:test-arm64ec.exe -machine:arm64ec test-arm64ec.obj 2>&1 | FileCheck %s
+# RUN: not lld-link -out:test-arm64ec2.exe -machine:arm64ec test-x86_64.obj 2>&1 | FileCheck %s
 
 # CHECK: error: undefined symbol: int __cdecl foo(void)
 # CHECK-NEXT: >>> referenced by file1.cpp:1
@@ -18,6 +26,7 @@
 # CHECK-NEXT: >>> referenced by file1.cpp:5
 # CHECK-NEXT: >>>               {{.*}}.obj:(f2)
 
+#--- test-x86_64.s
 	.cv_file	1 "file1.cpp" "EDA15C78BB573E49E685D8549286F33C" 1
 	.cv_file	2 "file2.cpp" "EDA15C78BB573E49E685D8549286F33D" 1
 
@@ -60,3 +69,47 @@ f2:
 	.long	4
 	.cv_filechecksums
 	.cv_stringtable
+
+#--- test-aarch64.s
+	.cv_file	1 "file1.cpp" "EDA15C78BB573E49E685D8549286F33C" 1
+	.cv_file	2 "file2.cpp" "EDA15C78BB573E49E685D8549286F33D" 1
+
+        .section        .text,"xr",one_only,main
+.globl main
+main:
+	.cv_func_id 0
+	.cv_loc	0 1 1 0 is_stmt 0
+	bl	"?foo@@YAHXZ"
+	.cv_loc	0 1 2 0
+	bl	"?foo@@YAHXZ"
+	.cv_loc	0 2 3 0
+	b	"?bar@@YAHXZ"
+.Lfunc_end0:
+
+f1:
+	.cv_func_id 1
+	.cv_loc	1 1 4 0 is_stmt 0
+	bl	"?bar@@YAHXZ"
+.Lfunc_end1:
+
+        .section        .text,"xr",one_only,f2
+.globl f2
+f2:
+	.cv_func_id 2
+	.cv_loc	2 1 5 0 is_stmt 0
+	bl	"?baz@@YAHXZ"
+.Lfunc_end2:
+
+	.section	.debug$S,"dr",associative,main
+	.long	4
+	.cv_linetable	0, main, .Lfunc_end0
+	.cv_linetable	1, f1, .Lfunc_end1
+
+	.section	.debug$S,"dr",associative,f2
+	.long	4
+	.cv_linetable	2, f2, .Lfunc_end2
+
+	.section	.debug$S,"dr"
+	.long	4
+	.cv_filechecksums
+	.cv_stringtable
diff --git a/lld/wasm/Driver.cpp b/lld/wasm/Driver.cpp
index df7d4d1cc3d6..d5d763b0a4ae 100644
--- a/lld/wasm/Driver.cpp
+++ b/lld/wasm/Driver.cpp
@@ -132,11 +132,21 @@ bool link(ArrayRef<const char *> args, llvm::raw_ostream &stdoutOS,
 // Create table mapping all options defined in Options.td
 static constexpr opt::OptTable::Info optInfo[] = {
 #define OPTION(PREFIX, NAME, ID, KIND, GROUP, ALIAS, ALIASARGS, FLAGS,         \
-               VISIBILITY, PARAM, HELPTEXT, METAVAR, VALUES)                   \
-  {PREFIX,      NAME,        HELPTEXT,                                         \
-   METAVAR,     OPT_##ID,    opt::Option::KIND##Class,                         \
-   PARAM,       FLAGS,       VISIBILITY,                                       \
-   OPT_##GROUP, OPT_##ALIAS, ALIASARGS,                                        \
+               VISIBILITY, PARAM, HELPTEXT, HELPTEXTSFORVARIANTS, METAVAR,     \
+               VALUES)                                                         \
+  {PREFIX,                                                                     \
+   NAME,                                                                       \
+   HELPTEXT,                                                                   \
+   HELPTEXTSFORVARIANTS,                                                       \
+   METAVAR,                                                                    \
+   OPT_##ID,                                                                   \
+   opt::Option::KIND##Class,                                                   \
+   PARAM,                                                                      \
+   FLAGS,                                                                      \
+   VISIBILITY,                                                                 \
+   OPT_##GROUP,                                                                \
+   OPT_##ALIAS,                                                                \
+   ALIASARGS,                                                                  \
    VALUES},
 #include "Options.inc"
 #undef OPTION
diff --git a/lldb/packages/Python/lldbsuite/test/make/Makefile.rules b/lldb/packages/Python/lldbsuite/test/make/Makefile.rules
index ee8793fa1b30..bfd249ccd43f 100644
--- a/lldb/packages/Python/lldbsuite/test/make/Makefile.rules
+++ b/lldb/packages/Python/lldbsuite/test/make/Makefile.rules
@@ -51,7 +51,7 @@ LLDB_BASE_DIR := $(THIS_FILE_DIR)/../../../../../
 #
 # GNUWin32 uname gives "windows32" or "server version windows32" while
 # some versions of MSYS uname return "MSYS_NT*", but most environments
-# standardize on "Windows_NT", so we'll make it consistent here.
+# standardize on "Windows_NT", so we'll make it consistent here. 
 # When running tests from Visual Studio, the environment variable isn't
 # inherited all the way down to the process spawned for make.
 #----------------------------------------------------------------------
@@ -210,12 +210,6 @@ else
 	ifeq "$(SPLIT_DEBUG_SYMBOLS)" "YES"
 		DSYM = $(EXE).debug
 	endif
-
-	ifeq "$(MAKE_DWP)" "YES"
-		MAKE_DWO := YES
-		DWP_NAME = $(EXE).dwp
-		DYLIB_DWP_NAME = $(DYLIB_NAME).dwp
-	endif
 endif
 
 LIMIT_DEBUG_INFO_FLAGS =
@@ -363,7 +357,6 @@ ifneq "$(OS)" "Darwin"
 
 	OBJCOPY ?= $(call replace_cc_with,objcopy)
 	ARCHIVER ?= $(call replace_cc_with,ar)
-	DWP ?= $(call replace_cc_with,dwp)
 	override AR = $(ARCHIVER)
 endif
 
@@ -534,10 +527,6 @@ ifneq "$(CXX)" ""
 	endif
 endif
 
-ifeq "$(GEN_GNU_BUILD_ID)" "YES"
-	LDFLAGS += -Wl,--build-id
-endif
-
 #----------------------------------------------------------------------
 # DYLIB_ONLY variable can be used to skip the building of a.out.
 # See the sections below regarding dSYM file as well as the building of
@@ -576,17 +565,10 @@ else
 endif
 else
 ifeq "$(SPLIT_DEBUG_SYMBOLS)" "YES"
-ifeq "$(SAVE_FULL_DEBUG_BINARY)" "YES"
-	cp "$(EXE)" "$(EXE).unstripped"
-endif
 	$(OBJCOPY) --only-keep-debug "$(EXE)" "$(DSYM)"
 	$(OBJCOPY) --strip-debug --add-gnu-debuglink="$(DSYM)" "$(EXE)" "$(EXE)"
 endif
-ifeq "$(MAKE_DWP)" "YES"
-	$(DWP) -o "$(DWP_NAME)" $(DWOS)
 endif
-endif
-
 
 #----------------------------------------------------------------------
 # Make the dylib
@@ -628,15 +610,9 @@ endif
 else
 	$(LD) $(DYLIB_OBJECTS) $(LDFLAGS) -shared -o "$(DYLIB_FILENAME)"
 ifeq "$(SPLIT_DEBUG_SYMBOLS)" "YES"
-	ifeq "$(SAVE_FULL_DEBUG_BINARY)" "YES"
-	cp "$(DYLIB_FILENAME)" "$(DYLIB_FILENAME).unstripped"
-	endif
 	$(OBJCOPY) --only-keep-debug "$(DYLIB_FILENAME)" "$(DYLIB_FILENAME).debug"
 	$(OBJCOPY) --strip-debug --add-gnu-debuglink="$(DYLIB_FILENAME).debug" "$(DYLIB_FILENAME)" "$(DYLIB_FILENAME)"
 endif
-ifeq "$(MAKE_DWP)" "YES"
-	$(DWP) -o $(DYLIB_DWP_FILE) $(DYLIB_DWOS)
-endif
 endif
 
 #----------------------------------------------------------------------
diff --git a/lldb/source/Plugins/SymbolFile/DWARF/SymbolFileDWARF.cpp b/lldb/source/Plugins/SymbolFile/DWARF/SymbolFileDWARF.cpp
index dafdf241f9db..49f13d2c89e3 100644
--- a/lldb/source/Plugins/SymbolFile/DWARF/SymbolFileDWARF.cpp
+++ b/lldb/source/Plugins/SymbolFile/DWARF/SymbolFileDWARF.cpp
@@ -4378,38 +4378,26 @@ const std::shared_ptr<SymbolFileDWARFDwo> &SymbolFileDWARF::GetDwpSymbolFile() {
     FileSpecList search_paths = Target::GetDefaultDebugFileSearchPaths();
     ModuleSpec module_spec;
     module_spec.GetFileSpec() = m_objfile_sp->GetFileSpec();
-    FileSpec dwp_filespec;
     for (const auto &symfile : symfiles.files()) {
       module_spec.GetSymbolFileSpec() =
           FileSpec(symfile.GetPath() + ".dwp", symfile.GetPathStyle());
       LLDB_LOG(log, "Searching for DWP using: \"{0}\"",
                module_spec.GetSymbolFileSpec());
-      dwp_filespec =
+      FileSpec dwp_filespec =
           PluginManager::LocateExecutableSymbolFile(module_spec, search_paths);
       if (FileSystem::Instance().Exists(dwp_filespec)) {
-        break;
-      }
-    }
-    if (!FileSystem::Instance().Exists(dwp_filespec)) {
-      LLDB_LOG(log, "No DWP file found locally");
-      // Fill in the UUID for the module we're trying to match for, so we can
-      // find the correct DWP file, as the Debuginfod plugin uses *only* this
-      // data to correctly match the DWP file with the binary.
-      module_spec.GetUUID() = m_objfile_sp->GetUUID();
-      dwp_filespec =
-          PluginManager::LocateExecutableSymbolFile(module_spec, search_paths);
-    }
-    if (FileSystem::Instance().Exists(dwp_filespec)) {
-      LLDB_LOG(log, "Found DWP file: \"{0}\"", dwp_filespec);
-      DataBufferSP dwp_file_data_sp;
-      lldb::offset_t dwp_file_data_offset = 0;
-      ObjectFileSP dwp_obj_file = ObjectFile::FindPlugin(
-          GetObjectFile()->GetModule(), &dwp_filespec, 0,
-          FileSystem::Instance().GetByteSize(dwp_filespec), dwp_file_data_sp,
-          dwp_file_data_offset);
-      if (dwp_obj_file) {
-        m_dwp_symfile = std::make_shared<SymbolFileDWARFDwo>(
-            *this, dwp_obj_file, DIERef::k_file_index_mask);
+        LLDB_LOG(log, "Found DWP file: \"{0}\"", dwp_filespec);
+        DataBufferSP dwp_file_data_sp;
+        lldb::offset_t dwp_file_data_offset = 0;
+        ObjectFileSP dwp_obj_file = ObjectFile::FindPlugin(
+            GetObjectFile()->GetModule(), &dwp_filespec, 0,
+            FileSystem::Instance().GetByteSize(dwp_filespec), dwp_file_data_sp,
+            dwp_file_data_offset);
+        if (dwp_obj_file) {
+          m_dwp_symfile = std::make_shared<SymbolFileDWARFDwo>(
+              *this, dwp_obj_file, DIERef::k_file_index_mask);
+          break;
+        }
       }
     }
     if (!m_dwp_symfile) {
diff --git a/lldb/source/Plugins/SymbolLocator/CMakeLists.txt b/lldb/source/Plugins/SymbolLocator/CMakeLists.txt
index 3367022639ab..ca969626f4ff 100644
--- a/lldb/source/Plugins/SymbolLocator/CMakeLists.txt
+++ b/lldb/source/Plugins/SymbolLocator/CMakeLists.txt
@@ -1,10 +1,5 @@
-# Order matters here: the first symbol locator prevents further searching.
-# For DWARF binaries that are both stripped and split, the Default plugin
-# will return the stripped binary when asked for the ObjectFile, which then
-# prevents an unstripped binary from being requested from the Debuginfod
-# provider.
-add_subdirectory(Debuginfod)
 add_subdirectory(Default)
 if (CMAKE_SYSTEM_NAME MATCHES "Darwin")
   add_subdirectory(DebugSymbols)
 endif()
+add_subdirectory(Debuginfod)
diff --git a/lldb/source/Plugins/SymbolVendor/ELF/SymbolVendorELF.cpp b/lldb/source/Plugins/SymbolVendor/ELF/SymbolVendorELF.cpp
index f296e655cc46..b5fe35d71032 100644
--- a/lldb/source/Plugins/SymbolVendor/ELF/SymbolVendorELF.cpp
+++ b/lldb/source/Plugins/SymbolVendor/ELF/SymbolVendorELF.cpp
@@ -44,24 +44,6 @@ llvm::StringRef SymbolVendorELF::GetPluginDescriptionStatic() {
          "executables.";
 }
 
-// If this is needed elsewhere, it can be exported/moved.
-static bool IsDwpSymbolFile(const lldb::ModuleSP &module_sp,
-                            const FileSpec &file_spec) {
-  DataBufferSP dwp_file_data_sp;
-  lldb::offset_t dwp_file_data_offset = 0;
-  // Try to create an ObjectFile from the file_spec.
-  ObjectFileSP dwp_obj_file = ObjectFile::FindPlugin(
-      module_sp, &file_spec, 0, FileSystem::Instance().GetByteSize(file_spec),
-      dwp_file_data_sp, dwp_file_data_offset);
-  // The presence of a debug_cu_index section is the key identifying feature of
-  // a DWP file. Make sure we don't fill in the section list on dwp_obj_file
-  // (by calling GetSectionList(false)) as this function could be called before
-  // we may have all the symbol files collected and available.
-  return dwp_obj_file && ObjectFileELF::classof(dwp_obj_file.get()) &&
-         dwp_obj_file->GetSectionList(false)->FindSectionByType(
-             eSectionTypeDWARFDebugCuIndex, false);
-}
-
 // CreateInstance
 //
 // Platforms can register a callback to use when creating symbol vendors to
@@ -105,15 +87,8 @@ SymbolVendorELF::CreateInstance(const lldb::ModuleSP &module_sp,
   FileSpecList search_paths = Target::GetDefaultDebugFileSearchPaths();
   FileSpec dsym_fspec =
       PluginManager::LocateExecutableSymbolFile(module_spec, search_paths);
-  if (!dsym_fspec || IsDwpSymbolFile(module_sp, dsym_fspec)) {
-    // If we have a stripped binary or if we got a DWP file, we should prefer
-    // symbols in the executable acquired through a plugin.
-    ModuleSpec unstripped_spec =
-        PluginManager::LocateExecutableObjectFile(module_spec);
-    if (!unstripped_spec)
-      return nullptr;
-    dsym_fspec = unstripped_spec.GetFileSpec();
-  }
+  if (!dsym_fspec)
+    return nullptr;
 
   DataBufferSP dsym_file_data_sp;
   lldb::offset_t dsym_file_data_offset = 0;
diff --git a/lldb/test/API/debuginfod/Normal/Makefile b/lldb/test/API/debuginfod/Normal/Makefile
deleted file mode 100644
index 54bd7adae241..000000000000
--- a/lldb/test/API/debuginfod/Normal/Makefile
+++ /dev/null
@@ -1,19 +0,0 @@
-C_SOURCES := main.c
-
-# For normal (non DWP) Debuginfod tests, we need:
-
-# * The full binary: a.out.unstripped
-#   Produced by Makefile.rules with SAVE_FULL_DEBUG_BINARY set to YES and
-#   SPLIT_DEBUG_SYMBOLS set to YES
-
-# * The stripped binary (a.out)
-#   Produced by Makefile.rules with SPLIT_DEBUG_SYMBOLS set to YES
-
-# * The 'only-keep-debug' binary (a.out.debug)
-#   Produced below
-
-SPLIT_DEBUG_SYMBOLS := YES
-SAVE_FULL_DEBUG_BINARY := YES
-GEN_GNU_BUILD_ID := YES
-
-include Makefile.rules
diff --git a/lldb/test/API/debuginfod/Normal/TestDebuginfod.py b/lldb/test/API/debuginfod/Normal/TestDebuginfod.py
deleted file mode 100644
index f1be1e705ee0..000000000000
--- a/lldb/test/API/debuginfod/Normal/TestDebuginfod.py
+++ /dev/null
@@ -1,183 +0,0 @@
-import os
-import shutil
-import tempfile
-
-import lldb
-from lldbsuite.test.decorators import *
-import lldbsuite.test.lldbutil as lldbutil
-from lldbsuite.test.lldbtest import *
-
-
-"""
-Test support for the DebugInfoD network symbol acquisition protocol.
-This one is for simple / no split-dwarf scenarios.
-
-For no-split-dwarf scenarios, there are 2 variations:
-1 - A stripped binary with it's corresponding unstripped binary:
-2 - A stripped binary with a corresponding --only-keep-debug symbols file
-"""
-
-
-# It looks like Linux-AArch64 doesn't support build-id's on the LLDB builtbots
-class DebugInfodTests(TestBase):
-    # No need to try every flavor of debug inf.
-    NO_DEBUG_INFO_TESTCASE = True
-
-    @skipIf(oslist=no_match(["linux"]), archs=no_match(["i386", "x86_64"]))
-    def test_normal_no_symbols(self):
-        """
-        Validate behavior with no symbols or symbol locator.
-        ('baseline negative' behavior)
-        """
-        test_root = self.config_test(["a.out"])
-        self.try_breakpoint(False)
-
-    @skipIf(oslist=no_match(["linux"]), archs=no_match(["i386", "x86_64"]))
-    def test_normal_default(self):
-        """
-        Validate behavior with symbols, but no symbol locator.
-        ('baseline positive' behavior)
-        """
-        test_root = self.config_test(["a.out", "a.out.debug"])
-        self.try_breakpoint(True)
-
-    @skipIf(oslist=no_match(["linux"]), archs=no_match(["i386", "x86_64"]))
-    def test_debuginfod_symbols(self):
-        """
-        Test behavior with the full binary available from Debuginfod as
-        'debuginfo' from the plug-in.
-        """
-        test_root = self.config_test(["a.out"], "a.out.unstripped")
-        self.try_breakpoint(True)
-
-    @skipIf(oslist=no_match(["linux"]), archs=no_match(["i386", "x86_64"]))
-    def test_debuginfod_executable(self):
-        """
-        Test behavior with the full binary available from Debuginfod as
-        'executable' from the plug-in.
-        """
-        test_root = self.config_test(["a.out"], None, "a.out.unstripped")
-        self.try_breakpoint(True)
-
-    @skipIf(oslist=no_match(["linux"]), archs=no_match(["i386", "x86_64"]))
-    def test_debuginfod_okd_symbols(self):
-        """
-        Test behavior with the 'only-keep-debug' symbols available from Debuginfod.
-        """
-        test_root = self.config_test(["a.out"], "a.out.debug")
-        self.try_breakpoint(True)
-
-    def try_breakpoint(self, should_have_loc):
-        """
-        This function creates a target from self.aout, sets a function-name
-        breakpoint, and checks to see if we have a file/line location,
-        as a way to validate that the symbols have been loaded.
-        should_have_loc specifies if we're testing that symbols have or
-        haven't been loaded.
-        """
-        target = self.dbg.CreateTarget(self.aout)
-        self.assertTrue(target and target.IsValid(), "Target is valid")
-
-        bp = target.BreakpointCreateByName("func")
-        self.assertTrue(bp and bp.IsValid(), "Breakpoint is valid")
-        self.assertEqual(bp.GetNumLocations(), 1)
-
-        loc = bp.GetLocationAtIndex(0)
-        self.assertTrue(loc and loc.IsValid(), "Location is valid")
-        addr = loc.GetAddress()
-        self.assertTrue(addr and addr.IsValid(), "Loc address is valid")
-        line_entry = addr.GetLineEntry()
-        self.assertEqual(
-            should_have_loc,
-            line_entry != None and line_entry.IsValid(),
-            "Loc line entry is valid",
-        )
-        if should_have_loc:
-            self.assertEqual(line_entry.GetLine(), 4)
-            self.assertEqual(
-                line_entry.GetFileSpec().GetFilename(),
-                self.main_source_file.GetFilename(),
-            )
-        self.dbg.DeleteTarget(target)
-        shutil.rmtree(self.tmp_dir)
-
-    def config_test(self, local_files, debuginfo=None, executable=None):
-        """
-        Set up a test with local_files[] copied to a different location
-        so that we control which files are, or are not, found in the file system.
-        Also, create a stand-alone file-system 'hosted' debuginfod server with the
-        provided debuginfo and executable files (if they exist)
-
-        Make the filesystem look like:
-
-        /tmp/<tmpdir>/test/[local_files]
-
-        /tmp/<tmpdir>/cache (for lldb to use as a temp cache)
-
-        /tmp/<tmpdir>/buildid/<uuid>/executable -> <executable>
-        /tmp/<tmpdir>/buildid/<uuid>/debuginfo -> <debuginfo>
-        Returns the /tmp/<tmpdir> path
-        """
-
-        self.build()
-
-        uuid = self.getUUID("a.out")
-        if not uuid:
-            self.fail("Could not get UUID for a.out")
-            return
-        self.main_source_file = lldb.SBFileSpec("main.c")
-        self.tmp_dir = tempfile.mkdtemp()
-        test_dir = os.path.join(self.tmp_dir, "test")
-        os.makedirs(test_dir)
-
-        self.aout = ""
-        # Copy the files used by the test:
-        for f in local_files:
-            shutil.copy(self.getBuildArtifact(f), test_dir)
-            # The first item is the binary to be used for the test
-            if self.aout == "":
-                self.aout = os.path.join(test_dir, f)
-
-        use_debuginfod = debuginfo != None or executable != None
-
-        # Populated the 'file://... mocked' Debuginfod server:
-        if use_debuginfod:
-            os.makedirs(os.path.join(self.tmp_dir, "cache"))
-            uuid_dir = os.path.join(self.tmp_dir, "buildid", uuid)
-            os.makedirs(uuid_dir)
-            if debuginfo:
-                shutil.copy(
-                    self.getBuildArtifact(debuginfo),
-                    os.path.join(uuid_dir, "debuginfo"),
-                )
-            if executable:
-                shutil.copy(
-                    self.getBuildArtifact(executable),
-                    os.path.join(uuid_dir, "executable"),
-                )
-
-        # Configure LLDB for the test:
-        self.runCmd(
-            "settings set symbols.enable-external-lookup %s"
-            % str(use_debuginfod).lower()
-        )
-        self.runCmd("settings clear plugin.symbol-locator.debuginfod.server-urls")
-        if use_debuginfod:
-            self.runCmd(
-                "settings set plugin.symbol-locator.debuginfod.cache-path %s/cache"
-                % self.tmp_dir
-            )
-            self.runCmd(
-                "settings insert-before plugin.symbol-locator.debuginfod.server-urls 0 file://%s"
-                % self.tmp_dir
-            )
-
-    def getUUID(self, filename):
-        try:
-            target = self.dbg.CreateTarget(self.getBuildArtifact(filename))
-            module = target.GetModuleAtIndex(0)
-            uuid = module.GetUUIDString().replace("-", "").lower()
-            self.dbg.DeleteTarget(target)
-            return uuid if len(uuid) == 40 else None
-        except:
-            return None
diff --git a/lldb/test/API/debuginfod/Normal/main.c b/lldb/test/API/debuginfod/Normal/main.c
deleted file mode 100644
index 4c7184609b45..000000000000
--- a/lldb/test/API/debuginfod/Normal/main.c
+++ /dev/null
@@ -1,7 +0,0 @@
-// This is a dump little pair of test files
-
-int func(int argc, const char *argv[]) {
-  return (argc + 1) * (argv[argc][0] + 2);
-}
-
-int main(int argc, const char *argv[]) { return func(0, argv); }
diff --git a/lldb/test/API/debuginfod/SplitDWARF/Makefile b/lldb/test/API/debuginfod/SplitDWARF/Makefile
deleted file mode 100644
index 3ab9a969e5a4..000000000000
--- a/lldb/test/API/debuginfod/SplitDWARF/Makefile
+++ /dev/null
@@ -1,23 +0,0 @@
-C_SOURCES := main.c
-
-# For split-dwarf Debuginfod tests, we need:
-
-# * A .DWP file (a.out.dwp)
-#   Produced by Makefile.rules with MAKE_DWP set to YES
-
-# * The "full" binary (missing things that live in .dwo's) (a.out.unstripped)
-#   Produced by Makefile.rules with SAVE_FULL_DEBUG_BINARY set to YES and
-#   SPLIT_DEBUG_SYMBOLS set to YES
-
-# * The stripped binary (a.out)
-#   Produced by Makefile.rules
-
-# * The 'only-keep-debug' binary (a.out.debug)
-#   Produced below
-
-MAKE_DWP := YES
-SPLIT_DEBUG_SYMBOLS := YES
-SAVE_FULL_DEBUG_BINARY := YES
-GEN_GNU_BUILD_ID := YES
-
-include Makefile.rules
diff --git a/lldb/test/API/debuginfod/SplitDWARF/TestDebuginfodDWP.py b/lldb/test/API/debuginfod/SplitDWARF/TestDebuginfodDWP.py
deleted file mode 100644
index fec2fa1132d7..000000000000
--- a/lldb/test/API/debuginfod/SplitDWARF/TestDebuginfodDWP.py
+++ /dev/null
@@ -1,193 +0,0 @@
-"""
-Test support for the DebugInfoD network symbol acquisition protocol.
-"""
-import os
-import shutil
-import tempfile
-
-import lldb
-from lldbsuite.test.decorators import *
-import lldbsuite.test.lldbutil as lldbutil
-from lldbsuite.test.lldbtest import *
-
-
-"""
-Test support for the DebugInfoD network symbol acquisition protocol.
-This file is for split-dwarf (dwp) scenarios.
-
-1 - A split binary target with it's corresponding DWP file
-2 - A stripped, split binary target with an unstripped binary and a DWP file
-3 - A stripped, split binary target with an --only-keep-debug symbols file and a DWP file
-"""
-
-
-# It looks like Linux-AArch64 doesn't support build-id's on the LLDB builtbots
-class DebugInfodDWPTests(TestBase):
-    # No need to try every flavor of debug inf.
-    NO_DEBUG_INFO_TESTCASE = True
-
-    @skipIf(oslist=no_match(["linux"]), archs=no_match(["i386", "x86_64"]))
-    def test_normal_stripped(self):
-        """
-        Validate behavior with a stripped binary, no symbols or symbol locator.
-        """
-        self.config_test(["a.out"])
-        self.try_breakpoint(False)
-
-    @skipIf(oslist=no_match(["linux"]), archs=no_match(["i386", "x86_64"]))
-    def test_normal_stripped_split_with_dwp(self):
-        """
-        Validate behavior with symbols, but no symbol locator.
-        """
-        self.config_test(["a.out", "a.out.debug", "a.out.dwp"])
-        self.try_breakpoint(True)
-
-    @skipIf(oslist=no_match(["linux"]), archs=no_match(["i386", "x86_64"]))
-    def test_normal_stripped_only_dwp(self):
-        """
-        Validate behavior *with* dwp symbols only, but missing other symbols,
-        but no symbol locator. This shouldn't work: without the other symbols
-        DWO's appear mostly useless.
-        """
-        self.config_test(["a.out", "a.out.dwp"])
-        self.try_breakpoint(False)
-
-    @skipIf(oslist=no_match(["linux"]), archs=no_match(["i386", "x86_64"]))
-    def test_debuginfod_dwp_from_service(self):
-        """
-        Test behavior with the unstripped binary, and DWP from the service.
-        """
-        self.config_test(["a.out.debug"], "a.out.dwp")
-        self.try_breakpoint(True)
-
-    @skipIf(oslist=no_match(["linux"]), archs=no_match(["i386", "x86_64"]))
-    def test_debuginfod_both_symfiles_from_service(self):
-        """
-        Test behavior with a stripped binary, with the unstripped binary and
-        dwp symbols from Debuginfod.
-        """
-        self.config_test(["a.out"], "a.out.dwp", "a.out.unstripped")
-        self.try_breakpoint(True)
-
-    @skipIf(oslist=no_match(["linux"]), archs=no_match(["i386", "x86_64"]))
-    def test_debuginfod_both_okd_symfiles_from_service(self):
-        """
-        Test behavior with both the only-keep-debug symbols and the dwp symbols
-        from Debuginfod.
-        """
-        self.config_test(["a.out"], "a.out.dwp", "a.out.debug")
-        self.try_breakpoint(True)
-
-    def try_breakpoint(self, should_have_loc):
-        """
-        This function creates a target from self.aout, sets a function-name
-        breakpoint, and checks to see if we have a file/line location,
-        as a way to validate that the symbols have been loaded.
-        should_have_loc specifies if we're testing that symbols have or
-        haven't been loaded.
-        """
-        target = self.dbg.CreateTarget(self.aout)
-        self.assertTrue(target and target.IsValid(), "Target is valid")
-
-        bp = target.BreakpointCreateByName("func")
-        self.assertTrue(bp and bp.IsValid(), "Breakpoint is valid")
-        self.assertEqual(bp.GetNumLocations(), 1)
-
-        loc = bp.GetLocationAtIndex(0)
-        self.assertTrue(loc and loc.IsValid(), "Location is valid")
-        addr = loc.GetAddress()
-        self.assertTrue(addr and addr.IsValid(), "Loc address is valid")
-        line_entry = addr.GetLineEntry()
-        self.assertEqual(
-            should_have_loc,
-            line_entry != None and line_entry.IsValid(),
-            "Loc line entry is valid",
-        )
-        if should_have_loc:
-            self.assertEqual(line_entry.GetLine(), 4)
-            self.assertEqual(
-                line_entry.GetFileSpec().GetFilename(),
-                self.main_source_file.GetFilename(),
-            )
-        self.dbg.DeleteTarget(target)
-        shutil.rmtree(self.tmp_dir)
-
-    def config_test(self, local_files, debuginfo=None, executable=None):
-        """
-        Set up a test with local_files[] copied to a different location
-        so that we control which files are, or are not, found in the file system.
-        Also, create a stand-alone file-system 'hosted' debuginfod server with the
-        provided debuginfo and executable files (if they exist)
-
-        Make the filesystem look like:
-
-        /tmp/<tmpdir>/test/[local_files]
-
-        /tmp/<tmpdir>/cache (for lldb to use as a temp cache)
-
-        /tmp/<tmpdir>/buildid/<uuid>/executable -> <executable>
-        /tmp/<tmpdir>/buildid/<uuid>/debuginfo -> <debuginfo>
-        Returns the /tmp/<tmpdir> path
-        """
-
-        self.build()
-
-        uuid = self.getUUID("a.out")
-        if not uuid:
-            self.fail("Could not get UUID for a.out")
-            return
-        self.main_source_file = lldb.SBFileSpec("main.c")
-        self.tmp_dir = tempfile.mkdtemp()
-        self.test_dir = os.path.join(self.tmp_dir, "test")
-        os.makedirs(self.test_dir)
-
-        self.aout = ""
-        # Copy the files used by the test:
-        for f in local_files:
-            shutil.copy(self.getBuildArtifact(f), self.test_dir)
-            if self.aout == "":
-                self.aout = os.path.join(self.test_dir, f)
-
-        use_debuginfod = debuginfo != None or executable != None
-
-        # Populated the 'file://... mocked' Debuginfod server:
-        if use_debuginfod:
-            os.makedirs(os.path.join(self.tmp_dir, "cache"))
-            uuid_dir = os.path.join(self.tmp_dir, "buildid", uuid)
-            os.makedirs(uuid_dir)
-            if debuginfo:
-                shutil.copy(
-                    self.getBuildArtifact(debuginfo),
-                    os.path.join(uuid_dir, "debuginfo"),
-                )
-            if executable:
-                shutil.copy(
-                    self.getBuildArtifact(executable),
-                    os.path.join(uuid_dir, "executable"),
-                )
-        os.remove(self.getBuildArtifact("main.dwo"))
-        # Configure LLDB for the test:
-        self.runCmd(
-            "settings set symbols.enable-external-lookup %s"
-            % str(use_debuginfod).lower()
-        )
-        self.runCmd("settings clear plugin.symbol-locator.debuginfod.server-urls")
-        if use_debuginfod:
-            self.runCmd(
-                "settings set plugin.symbol-locator.debuginfod.cache-path %s/cache"
-                % self.tmp_dir
-            )
-            self.runCmd(
-                "settings insert-before plugin.symbol-locator.debuginfod.server-urls 0 file://%s"
-                % self.tmp_dir
-            )
-
-    def getUUID(self, filename):
-        try:
-            target = self.dbg.CreateTarget(self.getBuildArtifact(filename))
-            module = target.GetModuleAtIndex(0)
-            uuid = module.GetUUIDString().replace("-", "").lower()
-            self.dbg.DeleteTarget(target)
-            return uuid if len(uuid) == 40 else None
-        except:
-            return None
diff --git a/lldb/test/API/debuginfod/SplitDWARF/main.c b/lldb/test/API/debuginfod/SplitDWARF/main.c
deleted file mode 100644
index 4c7184609b45..000000000000
--- a/lldb/test/API/debuginfod/SplitDWARF/main.c
+++ /dev/null
@@ -1,7 +0,0 @@
-// This is a dump little pair of test files
-
-int func(int argc, const char *argv[]) {
-  return (argc + 1) * (argv[argc][0] + 2);
-}
-
-int main(int argc, const char *argv[]) { return func(0, argv); }
diff --git a/llvm/docs/CommandGuide/llvm-objcopy.rst b/llvm/docs/CommandGuide/llvm-objcopy.rst
index 57d6280d57c8..985d16eb11cf 100644
--- a/llvm/docs/CommandGuide/llvm-objcopy.rst
+++ b/llvm/docs/CommandGuide/llvm-objcopy.rst
@@ -309,14 +309,6 @@ them.
  Compress DWARF debug sections in the output, using the specified format.
  Supported formats are ``zlib`` and ``zstd``. Use ``zlib`` if ``<format>`` is omitted.
 
-.. option:: --compress-sections <section>=<format>
-
- Compress or decompress sections matched by ``<section>`` using the specified
- format. Supported formats are ``zlib`` and ``zstd``. Specify ``none`` for
- decompression. When a section is matched by multiple options, the last one
- wins. A wildcard ``<section>`` starting with '!' is disallowed.
- Sections within a segment cannot be (de)compressed.
-
 .. option:: --decompress-debug-sections
 
  Decompress any compressed DWARF debug sections in the output.
diff --git a/llvm/docs/ReleaseNotes.rst b/llvm/docs/ReleaseNotes.rst
index ff7fed9ad590..d2d542752b55 100644
--- a/llvm/docs/ReleaseNotes.rst
+++ b/llvm/docs/ReleaseNotes.rst
@@ -174,6 +174,10 @@ Changes to the LLVM tools
 * llvm-ar now allows specifying COFF archive format with ``--format`` argument
   and uses it by default for COFF targets.
 
+* llvm-ranlib now supports ``-V`` as an alias for ``--version``.
+  ``-v`` (``--verbose`` in llvm-ar) has been removed.
+  (`#87661 <https://github.com/llvm/llvm-project/pull/87661>`_)
+
 * llvm-objcopy now supports ``--set-symbol-visibility`` and
   ``--set-symbols-visibility`` options for ELF input to change the
   visibility of symbols.
@@ -182,10 +186,6 @@ Changes to the LLVM tools
   for ELF input to skip the specified symbols when executing other options
   that can change a symbol's name, binding or visibility.
 
-* llvm-objcopy now supports ``--compress-sections`` to compress or decompress
-  arbitrary sections not within a segment.
-  (`#85036 <https://github.com/llvm/llvm-project/pull/85036>`_.)
-
 * llvm-profgen now supports COFF+DWARF binaries. This enables Sample-based PGO
   on Windows using Intel VTune's SEP. For details on usage, see the `end-user
   documentation for SPGO
diff --git a/llvm/include/llvm/ADT/STLExtras.h b/llvm/include/llvm/ADT/STLExtras.h
index 02a3074ae1f0..b5eb319ccf34 100644
--- a/llvm/include/llvm/ADT/STLExtras.h
+++ b/llvm/include/llvm/ADT/STLExtras.h
@@ -2151,7 +2151,7 @@ template <typename Container, typename UnaryFunctor, typename NullaryFunctor,
               !std::is_constructible<StringRef, NullaryFunctor>::value>>
 inline void interleave(const Container &c, UnaryFunctor each_fn,
                        NullaryFunctor between_fn) {
-  interleave(c.begin(), c.end(), each_fn, between_fn);
+  interleave(adl_begin(c), adl_end(c), each_fn, between_fn);
 }
 
 /// Overload of interleave for the common case of string separator.
@@ -2159,7 +2159,7 @@ template <typename Container, typename UnaryFunctor, typename StreamT,
           typename T = detail::ValueOfRange<Container>>
 inline void interleave(const Container &c, StreamT &os, UnaryFunctor each_fn,
                        const StringRef &separator) {
-  interleave(c.begin(), c.end(), each_fn, [&] { os << separator; });
+  interleave(adl_begin(c), adl_end(c), each_fn, [&] { os << separator; });
 }
 template <typename Container, typename StreamT,
           typename T = detail::ValueOfRange<Container>>
diff --git a/llvm/include/llvm/Analysis/TargetTransformInfo.h b/llvm/include/llvm/Analysis/TargetTransformInfo.h
index bad0a77b0f2d..fa9392b86c15 100644
--- a/llvm/include/llvm/Analysis/TargetTransformInfo.h
+++ b/llvm/include/llvm/Analysis/TargetTransformInfo.h
@@ -190,7 +190,10 @@ enum class TailFoldingStyle {
   /// Use predicate to control both data and control flow, but modify
   /// the trip count so that a runtime overflow check can be avoided
   /// and such that the scalar epilogue loop can always be removed.
-  DataAndControlFlowWithoutRuntimeCheck
+  DataAndControlFlowWithoutRuntimeCheck,
+  /// Use predicated EVL instructions for tail-folding.
+  /// Indicates that VP intrinsics should be used.
+  DataWithEVL,
 };
 
 struct TailFoldingInfo {
diff --git a/llvm/include/llvm/AsmParser/LLParser.h b/llvm/include/llvm/AsmParser/LLParser.h
index 8ebd0d3409c8..b4e971fea1a1 100644
--- a/llvm/include/llvm/AsmParser/LLParser.h
+++ b/llvm/include/llvm/AsmParser/LLParser.h
@@ -335,6 +335,7 @@ namespace llvm {
 
     // Top-Level Entities
     bool parseTopLevelEntities();
+    bool finalizeDebugInfoFormat(Module *M);
     void dropUnknownMetadataReferences();
     bool validateEndOfModule(bool UpgradeDebugInfo);
     bool validateEndOfIndex();
diff --git a/llvm/include/llvm/CodeGen/BasicTTIImpl.h b/llvm/include/llvm/CodeGen/BasicTTIImpl.h
index 92fa726c31df..42d8f74fd427 100644
--- a/llvm/include/llvm/CodeGen/BasicTTIImpl.h
+++ b/llvm/include/llvm/CodeGen/BasicTTIImpl.h
@@ -209,28 +209,31 @@ private:
                                               Align Alignment,
                                               bool VariableMask,
                                               bool IsGatherScatter,
-                                              TTI::TargetCostKind CostKind) {
+                                              TTI::TargetCostKind CostKind,
+                                              unsigned AddressSpace = 0) {
     // We cannot scalarize scalable vectors, so return Invalid.
     if (isa<ScalableVectorType>(DataTy))
       return InstructionCost::getInvalid();
 
     auto *VT = cast<FixedVectorType>(DataTy);
+    unsigned VF = VT->getNumElements();
+
     // Assume the target does not have support for gather/scatter operations
     // and provide a rough estimate.
     //
     // First, compute the cost of the individual memory operations.
     InstructionCost AddrExtractCost =
         IsGatherScatter
-            ? getVectorInstrCost(Instruction::ExtractElement,
-                                 FixedVectorType::get(
-                                     PointerType::get(VT->getElementType(), 0),
-                                     VT->getNumElements()),
-                                 CostKind, -1, nullptr, nullptr)
+            ? getVectorInstrCost(
+                  Instruction::ExtractElement,
+                  FixedVectorType::get(
+                      PointerType::get(VT->getElementType(), 0), VF),
+                  CostKind, -1, nullptr, nullptr)
             : 0;
     InstructionCost LoadCost =
-        VT->getNumElements() *
-        (AddrExtractCost +
-         getMemoryOpCost(Opcode, VT->getElementType(), Alignment, 0, CostKind));
+        VF *
+        (AddrExtractCost + getMemoryOpCost(Opcode, VT->getElementType(),
+                                           Alignment, AddressSpace, CostKind));
 
     // Next, compute the cost of packing the result in a vector.
     InstructionCost PackingCost =
@@ -246,11 +249,10 @@ private:
       // operations accurately is quite difficult and the current solution
       // provides a very rough estimate only.
       ConditionalCost =
-          VT->getNumElements() *
+          VF *
           (getVectorInstrCost(
                Instruction::ExtractElement,
-               FixedVectorType::get(Type::getInt1Ty(DataTy->getContext()),
-                                    VT->getNumElements()),
+               FixedVectorType::get(Type::getInt1Ty(DataTy->getContext()), VF),
                CostKind, -1, nullptr, nullptr) +
            getCFInstrCost(Instruction::Br, CostKind) +
            getCFInstrCost(Instruction::PHI, CostKind));
@@ -1369,6 +1371,7 @@ public:
   InstructionCost getMaskedMemoryOpCost(unsigned Opcode, Type *DataTy,
                                         Align Alignment, unsigned AddressSpace,
                                         TTI::TargetCostKind CostKind) {
+    // TODO: Pass on AddressSpace when we have test coverage.
     return getCommonMaskedMemoryOpCost(Opcode, DataTy, Alignment, true, false,
                                        CostKind);
   }
diff --git a/llvm/include/llvm/IR/BasicBlock.h b/llvm/include/llvm/IR/BasicBlock.h
index 0c5a07bde4ec..e1220966e7e6 100644
--- a/llvm/include/llvm/IR/BasicBlock.h
+++ b/llvm/include/llvm/IR/BasicBlock.h
@@ -92,6 +92,7 @@ public:
   /// in the new format (\p NewFlag == true), converting to the desired format
   /// if necessary.
   void setIsNewDbgInfoFormat(bool NewFlag);
+  void setNewDbgInfoFormatFlag(bool NewFlag);
 
   /// Record that the collection of DbgRecords in \p M "trails" after the last
   /// instruction of this block. These are equivalent to dbg.value intrinsics
diff --git a/llvm/include/llvm/IR/Function.h b/llvm/include/llvm/IR/Function.h
index d96d506a9b05..60f41b30e91c 100644
--- a/llvm/include/llvm/IR/Function.h
+++ b/llvm/include/llvm/IR/Function.h
@@ -120,6 +120,7 @@ public:
   void convertFromNewDbgValues();
 
   void setIsNewDbgInfoFormat(bool NewVal);
+  void setNewDbgInfoFormatFlag(bool NewVal);
 
 private:
   friend class TargetLibraryInfoImpl;
diff --git a/llvm/include/llvm/IR/IntrinsicsDirectX.td b/llvm/include/llvm/IR/IntrinsicsDirectX.td
index a871fac46b9f..a7f212da2f5b 100644
--- a/llvm/include/llvm/IR/IntrinsicsDirectX.td
+++ b/llvm/include/llvm/IR/IntrinsicsDirectX.td
@@ -20,7 +20,8 @@ def int_dx_flattened_thread_id_in_group : Intrinsic<[llvm_i32_ty], [], [IntrNoMe
 def int_dx_create_handle : ClangBuiltin<"__builtin_hlsl_create_handle">,
     Intrinsic<[ llvm_ptr_ty ], [llvm_i8_ty], [IntrWillReturn]>;
 
-def int_dx_any  : DefaultAttrsIntrinsic<[llvm_i1_ty], [llvm_any_ty]>;
+def int_dx_all : DefaultAttrsIntrinsic<[llvm_i1_ty], [llvm_any_ty]>;
+def int_dx_any : DefaultAttrsIntrinsic<[llvm_i1_ty], [llvm_any_ty]>;
 def int_dx_clamp : DefaultAttrsIntrinsic<[llvm_any_ty], [LLVMMatchType<0>, LLVMMatchType<0>, LLVMMatchType<0>]>;
 def int_dx_uclamp : DefaultAttrsIntrinsic<[llvm_anyint_ty], [LLVMMatchType<0>, LLVMMatchType<0>, LLVMMatchType<0>]>; 
 
diff --git a/llvm/include/llvm/IR/IntrinsicsSPIRV.td b/llvm/include/llvm/IR/IntrinsicsSPIRV.td
index 0eb09b1699af..f843383f0b00 100644
--- a/llvm/include/llvm/IR/IntrinsicsSPIRV.td
+++ b/llvm/include/llvm/IR/IntrinsicsSPIRV.td
@@ -56,4 +56,5 @@ let TargetPrefix = "spv" in {
   def int_spv_thread_id : Intrinsic<[llvm_i32_ty], [llvm_i32_ty], [IntrNoMem, IntrWillReturn]>;
   def int_spv_create_handle : ClangBuiltin<"__builtin_hlsl_create_handle">,
       Intrinsic<[ llvm_ptr_ty ], [llvm_i8_ty], [IntrWillReturn]>;
+  def int_spv_all : DefaultAttrsIntrinsic<[llvm_i1_ty], [llvm_any_ty]>;
 }
diff --git a/llvm/include/llvm/IR/Module.h b/llvm/include/llvm/IR/Module.h
index bb2e667ef6f4..6135e15fd030 100644
--- a/llvm/include/llvm/IR/Module.h
+++ b/llvm/include/llvm/IR/Module.h
@@ -245,6 +245,12 @@ public:
     else if (!UseNewFormat && IsNewDbgInfoFormat)
       convertFromNewDbgValues();
   }
+  void setNewDbgInfoFormatFlag(bool NewFlag) {
+    for (auto &F : *this) {
+      F.setNewDbgInfoFormatFlag(NewFlag);
+    }
+    IsNewDbgInfoFormat = NewFlag;
+  }
 
   /// The Module constructor. Note that there is no default constructor. You
   /// must provide a name for the module upon construction.
diff --git a/llvm/include/llvm/ObjCopy/CommonConfig.h b/llvm/include/llvm/ObjCopy/CommonConfig.h
index ae08d4032736..9d6d5fb23b18 100644
--- a/llvm/include/llvm/ObjCopy/CommonConfig.h
+++ b/llvm/include/llvm/ObjCopy/CommonConfig.h
@@ -262,9 +262,6 @@ struct CommonConfig {
   bool DecompressDebugSections = false;
 
   DebugCompressionType CompressionType = DebugCompressionType::None;
-
-  SmallVector<std::pair<NameMatcher, llvm::DebugCompressionType>, 0>
-      compressSections;
 };
 
 } // namespace objcopy
diff --git a/llvm/include/llvm/Option/OptParser.td b/llvm/include/llvm/Option/OptParser.td
index 7bbee1da643b..9fd606b0d6fc 100644
--- a/llvm/include/llvm/Option/OptParser.td
+++ b/llvm/include/llvm/Option/OptParser.td
@@ -93,6 +93,11 @@ class OptionGroup<string name> {
 
 // Define the option class.
 
+class HelpTextVariant<list<OptionVisibility> visibilities, string text> {
+  list<OptionVisibility> Visibilities = visibilities;
+  string Text = text;
+}
+
 class Option<list<string> prefixes, string name, OptionKind kind> {
   string EnumName = ?; // Uses the def name if undefined.
   list<string> Prefixes = prefixes;
@@ -101,6 +106,7 @@ class Option<list<string> prefixes, string name, OptionKind kind> {
   // Used by MultiArg option kind.
   int NumArgs = 0;
   string HelpText = ?;
+  list<HelpTextVariant> HelpTextsForVariants = [];
   string MetaVarName = ?;
   string Values = ?;
   code ValuesCode = ?;
@@ -155,6 +161,12 @@ class Visibility<list<OptionVisibility> visibility> {
 }
 class Group<OptionGroup group> { OptionGroup Group = group; }
 class HelpText<string text> { string HelpText = text; }
+class HelpTextForVariants<list<OptionVisibility> Visibilities, string text> {
+  list<HelpTextVariant> HelpTextsForVariants = [
+    HelpTextVariant<Visibilities, text>
+  ];
+}
+
 class MetaVarName<string name> { string MetaVarName = name; }
 class Values<string value> { string Values = value; }
 class ValuesCode<code valuecode> { code ValuesCode = valuecode; }
diff --git a/llvm/include/llvm/Option/OptTable.h b/llvm/include/llvm/Option/OptTable.h
index bb3b665a1631..d8bf292bac21 100644
--- a/llvm/include/llvm/Option/OptTable.h
+++ b/llvm/include/llvm/Option/OptTable.h
@@ -58,6 +58,17 @@ public:
     ArrayRef<StringLiteral> Prefixes;
     StringLiteral PrefixedName;
     const char *HelpText;
+    // Help text for specific visibilities. A list of pairs, where each pair
+    // is a list of visibilities and a specific help string for those
+    // visibilities. If no help text is found in this list for the visibility of
+    // the program, HelpText is used instead. This cannot use std::vector
+    // because OptTable is used in constexpr contexts. Increase the array sizes
+    // here if you need more entries and adjust the constants in
+    // OptParserEmitter::EmitHelpTextsForVariants.
+    std::array<std::pair<std::array<unsigned int, 2 /*MaxVisibilityPerHelp*/>,
+                         const char *>,
+               1 /*MaxVisibilityHelp*/>
+        HelpTextsForVariants;
     const char *MetaVar;
     unsigned ID;
     unsigned char Kind;
@@ -145,7 +156,20 @@ public:
 
   /// Get the help text to use to describe this option.
   const char *getOptionHelpText(OptSpecifier id) const {
-    return getInfo(id).HelpText;
+    return getOptionHelpText(id, Visibility(0));
+  }
+
+  // Get the help text to use to describe this option.
+  // If it has visibility specific help text and that visibility is in the
+  // visibility mask, use that text instead of the generic text.
+  const char *getOptionHelpText(OptSpecifier id,
+                                Visibility VisibilityMask) const {
+    auto Info = getInfo(id);
+    for (auto [Visibilities, Text] : Info.HelpTextsForVariants)
+      for (auto Visibility : Visibilities)
+        if (VisibilityMask & Visibility)
+          return Text;
+    return Info.HelpText;
   }
 
   /// Get the meta-variable name to use when describing
@@ -323,7 +347,8 @@ public:
 private:
   void internalPrintHelp(raw_ostream &OS, const char *Usage, const char *Title,
                          bool ShowHidden, bool ShowAllAliases,
-                         std::function<bool(const Info &)> ExcludeOption) const;
+                         std::function<bool(const Info &)> ExcludeOption,
+                         Visibility VisibilityMask) const;
 };
 
 /// Specialization of OptTable
@@ -358,30 +383,30 @@ protected:
 
 #define LLVM_MAKE_OPT_ID_WITH_ID_PREFIX(                                       \
     ID_PREFIX, PREFIX, PREFIXED_NAME, ID, KIND, GROUP, ALIAS, ALIASARGS,       \
-    FLAGS, VISIBILITY, PARAM, HELPTEXT, METAVAR, VALUES)                       \
+    FLAGS, VISIBILITY, PARAM, HELPTEXT, HELPTEXTSFORVARIANTS, METAVAR, VALUES) \
   ID_PREFIX##ID
 
 #define LLVM_MAKE_OPT_ID(PREFIX, PREFIXED_NAME, ID, KIND, GROUP, ALIAS,        \
                          ALIASARGS, FLAGS, VISIBILITY, PARAM, HELPTEXT,        \
-                         METAVAR, VALUES)                                      \
-  LLVM_MAKE_OPT_ID_WITH_ID_PREFIX(OPT_, PREFIX, PREFIXED_NAME, ID, KIND,       \
-                                  GROUP, ALIAS, ALIASARGS, FLAGS, VISIBILITY,  \
-                                  PARAM, HELPTEXT, METAVAR, VALUE)
+                         HELPTEXTSFORVARIANTS, METAVAR, VALUES)                \
+  LLVM_MAKE_OPT_ID_WITH_ID_PREFIX(                                             \
+      OPT_, PREFIX, PREFIXED_NAME, ID, KIND, GROUP, ALIAS, ALIASARGS, FLAGS,   \
+      VISIBILITY, PARAM, HELPTEXT, HELPTEXTSFORVARIANTS, METAVAR, VALUE)
 
 #define LLVM_CONSTRUCT_OPT_INFO_WITH_ID_PREFIX(                                \
     ID_PREFIX, PREFIX, PREFIXED_NAME, ID, KIND, GROUP, ALIAS, ALIASARGS,       \
-    FLAGS, VISIBILITY, PARAM, HELPTEXT, METAVAR, VALUES)                       \
+    FLAGS, VISIBILITY, PARAM, HELPTEXT, HELPTEXTSFORVARIANTS, METAVAR, VALUES) \
   llvm::opt::OptTable::Info {                                                  \
-    PREFIX, PREFIXED_NAME, HELPTEXT, METAVAR, ID_PREFIX##ID,                   \
-        llvm::opt::Option::KIND##Class, PARAM, FLAGS, VISIBILITY,              \
-        ID_PREFIX##GROUP, ID_PREFIX##ALIAS, ALIASARGS, VALUES                  \
+    PREFIX, PREFIXED_NAME, HELPTEXT, HELPTEXTSFORVARIANTS, METAVAR,            \
+        ID_PREFIX##ID, llvm::opt::Option::KIND##Class, PARAM, FLAGS,           \
+        VISIBILITY, ID_PREFIX##GROUP, ID_PREFIX##ALIAS, ALIASARGS, VALUES      \
   }
 
 #define LLVM_CONSTRUCT_OPT_INFO(PREFIX, PREFIXED_NAME, ID, KIND, GROUP, ALIAS, \
                                 ALIASARGS, FLAGS, VISIBILITY, PARAM, HELPTEXT, \
-                                METAVAR, VALUES)                               \
+                                HELPTEXTSFORVARIANTS, METAVAR, VALUES)         \
   LLVM_CONSTRUCT_OPT_INFO_WITH_ID_PREFIX(                                      \
       OPT_, PREFIX, PREFIXED_NAME, ID, KIND, GROUP, ALIAS, ALIASARGS, FLAGS,   \
-      VISIBILITY, PARAM, HELPTEXT, METAVAR, VALUES)
+      VISIBILITY, PARAM, HELPTEXT, HELPTEXTSFORVARIANTS, METAVAR, VALUES)
 
 #endif // LLVM_OPTION_OPTTABLE_H
diff --git a/llvm/include/llvm/Transforms/Instrumentation/RemoveTrapsPass.h b/llvm/include/llvm/Transforms/Instrumentation/LowerAllowCheckPass.h
index bae15840f992..af974818fec5 100644
--- a/llvm/include/llvm/Transforms/Instrumentation/RemoveTrapsPass.h
+++ b/llvm/include/llvm/Transforms/Instrumentation/LowerAllowCheckPass.h
@@ -1,4 +1,4 @@
-//===- RemoveTrapsPass.h ----------------------------------------*- C++ -*-===//
+//===- LowerAllowCheckPass.h ------------------------------------*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
@@ -11,8 +11,8 @@
 ///
 //===----------------------------------------------------------------------===//
 
-#ifndef LLVM_TRANSFORMS_INSTRUMENTATION_UBSANOPTIMIZATIONPASS_H
-#define LLVM_TRANSFORMS_INSTRUMENTATION_UBSANOPTIMIZATIONPASS_H
+#ifndef LLVM_TRANSFORMS_INSTRUMENTATION_LOWERALLOWCHECKPASS_H
+#define LLVM_TRANSFORMS_INSTRUMENTATION_LOWERALLOWCHECKPASS_H
 
 #include "llvm/IR/Function.h"
 #include "llvm/IR/PassManager.h"
@@ -22,7 +22,7 @@ namespace llvm {
 
 // This pass is responsible for removing optional traps, like llvm.ubsantrap
 // from the hot code.
-class RemoveTrapsPass : public PassInfoMixin<RemoveTrapsPass> {
+class LowerAllowCheckPass : public PassInfoMixin<LowerAllowCheckPass> {
 public:
   PreservedAnalyses run(Function &F, FunctionAnalysisManager &AM);
 
diff --git a/llvm/include/llvm/Transforms/Utils/LoopUtils.h b/llvm/include/llvm/Transforms/Utils/LoopUtils.h
index 187ace3a0cbe..345e09dce0b2 100644
--- a/llvm/include/llvm/Transforms/Utils/LoopUtils.h
+++ b/llvm/include/llvm/Transforms/Utils/LoopUtils.h
@@ -372,15 +372,6 @@ RecurKind getMinMaxReductionRecurKind(Intrinsic::ID RdxID);
 /// Returns the comparison predicate used when expanding a min/max reduction.
 CmpInst::Predicate getMinMaxReductionPredicate(RecurKind RK);
 
-/// See RecurrenceDescriptor::isAnyOfPattern for a description of the pattern we
-/// are trying to match. In this pattern, we are only ever selecting between two
-/// values: 1) an initial start value \p StartVal of the reduction PHI, and 2) a
-/// loop invariant value. If any of lane value in \p Left, \p Right is not equal
-/// to \p StartVal, select the loop invariant value. This is done by selecting
-/// \p Right iff \p Left is equal to \p StartVal.
-Value *createAnyOfOp(IRBuilderBase &Builder, Value *StartVal, RecurKind RK,
-                     Value *Left, Value *Right);
-
 /// Returns a Min/Max operation corresponding to MinMaxRecurrenceKind.
 /// The Builder's fast-math-flags must be set to propagate the expected values.
 Value *createMinMaxOp(IRBuilderBase &Builder, RecurKind RK, Value *Left,
diff --git a/llvm/lib/AsmParser/LLParser.cpp b/llvm/lib/AsmParser/LLParser.cpp
index fe49e52ae428..67ec4a7c2dd0 100644
--- a/llvm/lib/AsmParser/LLParser.cpp
+++ b/llvm/lib/AsmParser/LLParser.cpp
@@ -63,6 +63,9 @@ static cl::opt<bool> AllowIncompleteIR(
         "metadata will be dropped)"));
 
 extern llvm::cl::opt<bool> UseNewDbgInfoFormat;
+extern cl::opt<cl::boolOrDefault> PreserveInputDbgFormat;
+extern bool WriteNewDbgInfoFormatToBitcode;
+extern cl::opt<bool> WriteNewDbgInfoFormat;
 
 static std::string getTypeString(Type *T) {
   std::string Result;
@@ -71,12 +74,20 @@ static std::string getTypeString(Type *T) {
   return Tmp.str();
 }
 
-// Currently, we should always process modules in the old debug info format by
-// default regardless of the module's format in IR; convert it to the old format
-// here.
-bool finalizeDebugInfoFormat(Module *M) {
-  if (M)
+// Whatever debug info format we parsed, we should convert to the expected debug
+// info format immediately afterwards.
+bool LLParser::finalizeDebugInfoFormat(Module *M) {
+  // We should have already returned an error if we observed both intrinsics and
+  // records in this IR.
+  assert(!(SeenNewDbgInfoFormat && SeenOldDbgInfoFormat) &&
+         "Mixed debug intrinsics/records seen without a parsing error?");
+  if (PreserveInputDbgFormat == cl::boolOrDefault::BOU_TRUE) {
+    UseNewDbgInfoFormat = SeenNewDbgInfoFormat;
+    WriteNewDbgInfoFormatToBitcode = SeenNewDbgInfoFormat;
+    WriteNewDbgInfoFormat = SeenNewDbgInfoFormat;
+  } else if (M) {
     M->setIsNewDbgInfoFormat(false);
+  }
   return false;
 }
 
@@ -6511,10 +6522,10 @@ bool LLParser::parseBasicBlock(PerFunctionState &PFS) {
       if (SeenOldDbgInfoFormat)
         return error(Lex.getLoc(), "debug record should not appear in a module "
                                    "containing debug info intrinsics");
+      if (!SeenNewDbgInfoFormat)
+        M->setNewDbgInfoFormatFlag(true);
       SeenNewDbgInfoFormat = true;
       Lex.Lex();
-      if (!M->IsNewDbgInfoFormat)
-        M->convertToNewDbgValues();
 
       DbgRecord *DR;
       if (parseDebugRecord(DR, PFS))
@@ -7928,6 +7939,8 @@ bool LLParser::parseCall(Instruction *&Inst, PerFunctionState &PFS,
       return error(CallLoc, "llvm.dbg intrinsic should not appear in a module "
                             "using non-intrinsic debug info");
     }
+    if (!SeenOldDbgInfoFormat)
+      M->setNewDbgInfoFormatFlag(false);
     SeenOldDbgInfoFormat = true;
   }
   CI->setAttributes(PAL);
diff --git a/llvm/lib/Bitcode/Reader/BitcodeReader.cpp b/llvm/lib/Bitcode/Reader/BitcodeReader.cpp
index aa6c9c95ca24..5ea6cc2c6b52 100644
--- a/llvm/lib/Bitcode/Reader/BitcodeReader.cpp
+++ b/llvm/lib/Bitcode/Reader/BitcodeReader.cpp
@@ -108,6 +108,9 @@ cl::opt<cl::boolOrDefault> LoadBitcodeIntoNewDbgInfoFormat(
     "load-bitcode-into-experimental-debuginfo-iterators", cl::Hidden,
     cl::desc("Load bitcode directly into the new debug info format (regardless "
              "of input format)"));
+extern cl::opt<cl::boolOrDefault> PreserveInputDbgFormat;
+extern bool WriteNewDbgInfoFormatToBitcode;
+extern cl::opt<bool> WriteNewDbgInfoFormat;
 
 namespace {
 
@@ -682,6 +685,11 @@ class BitcodeReader : public BitcodeReaderBase, public GVMaterializer {
   /// (e.g.) blockaddress forward references.
   bool WillMaterializeAllForwardRefs = false;
 
+  /// Tracks whether we have seen debug intrinsics or records in this bitcode;
+  /// seeing both in a single module is currently a fatal error.
+  bool SeenDebugIntrinsic = false;
+  bool SeenDebugRecord = false;
+
   bool StripDebugInfo = false;
   TBAAVerifier TBAAVerifyHelper;
 
@@ -3774,7 +3782,11 @@ Error BitcodeReader::globalCleanup() {
   for (Function &F : *TheModule) {
     MDLoader->upgradeDebugIntrinsics(F);
     Function *NewFn;
-    if (UpgradeIntrinsicFunction(&F, NewFn))
+    // If PreserveInputDbgFormat=true, then we don't know whether we want
+    // intrinsics or records, and we won't perform any conversions in either
+    // case, so don't upgrade intrinsics to records.
+    if (UpgradeIntrinsicFunction(
+            &F, NewFn, PreserveInputDbgFormat != cl::boolOrDefault::BOU_TRUE))
       UpgradedIntrinsics[&F] = NewFn;
     // Look for functions that rely on old function attribute behavior.
     UpgradeFunctionAttributes(F);
@@ -4301,10 +4313,13 @@ Error BitcodeReader::parseModule(uint64_t ResumeBit,
                                  bool ShouldLazyLoadMetadata,
                                  ParserCallbacks Callbacks) {
   // Load directly into RemoveDIs format if LoadBitcodeIntoNewDbgInfoFormat
-  // has been set to true (default action: load into the old debug format).
-  TheModule->IsNewDbgInfoFormat =
-      UseNewDbgInfoFormat &&
-      LoadBitcodeIntoNewDbgInfoFormat == cl::boolOrDefault::BOU_TRUE;
+  // has been set to true and we aren't attempting to preserve the existing
+  // format in the bitcode (default action: load into the old debug format).
+  if (PreserveInputDbgFormat != cl::boolOrDefault::BOU_TRUE) {
+    TheModule->IsNewDbgInfoFormat =
+        UseNewDbgInfoFormat &&
+        LoadBitcodeIntoNewDbgInfoFormat == cl::boolOrDefault::BOU_TRUE;
+  }
 
   this->ValueTypeCallback = std::move(Callbacks.ValueType);
   if (ResumeBit) {
@@ -6453,6 +6468,7 @@ Error BitcodeReader::parseFunctionBody(Function *F) {
     case bitc::FUNC_CODE_DEBUG_RECORD_ASSIGN: {
       // DbgVariableRecords are placed after the Instructions that they are
       // attached to.
+      SeenDebugRecord = true;
       Instruction *Inst = getLastInstruction();
       if (!Inst)
         return error("Invalid dbg record: missing instruction");
@@ -6613,6 +6629,8 @@ Error BitcodeReader::parseFunctionBody(Function *F) {
         TCK = CallInst::TCK_NoTail;
       cast<CallInst>(I)->setTailCallKind(TCK);
       cast<CallInst>(I)->setAttributes(PAL);
+      if (isa<DbgInfoIntrinsic>(I))
+        SeenDebugIntrinsic = true;
       if (Error Err = propagateAttributeTypes(cast<CallBase>(I), ArgTyIDs)) {
         I->deleteValue();
         return Err;
@@ -6801,20 +6819,48 @@ Error BitcodeReader::materialize(GlobalValue *GV) {
   if (Error JumpFailed = Stream.JumpToBit(DFII->second))
     return JumpFailed;
 
-  // Set the debug info mode to "new", possibly creating a mismatch between
-  // module and function debug modes. This is okay because we'll convert
-  // everything back to the old mode after parsing if needed.
-  // FIXME: Remove this once all tools support RemoveDIs.
+  // Regardless of the debug info format we want to end up in, we need
+  // IsNewDbgInfoFormat=true to construct any debug records seen in the bitcode.
   F->IsNewDbgInfoFormat = true;
 
   if (Error Err = parseFunctionBody(F))
     return Err;
   F->setIsMaterializable(false);
 
-  // Convert new debug info records into intrinsics.
-  // FIXME: Remove this once all tools support RemoveDIs.
-  if (!F->getParent()->IsNewDbgInfoFormat)
-    F->convertFromNewDbgValues();
+  // All parsed Functions should load into the debug info format dictated by the
+  // Module, unless we're attempting to preserve the input debug info format.
+  if (SeenDebugIntrinsic && SeenDebugRecord)
+    return error("Mixed debug intrinsics and debug records in bitcode module!");
+  if (PreserveInputDbgFormat == cl::boolOrDefault::BOU_TRUE) {
+    bool SeenAnyDebugInfo = SeenDebugIntrinsic || SeenDebugRecord;
+    bool NewDbgInfoFormatDesired =
+        SeenAnyDebugInfo ? SeenDebugRecord : F->getParent()->IsNewDbgInfoFormat;
+    if (SeenAnyDebugInfo) {
+      UseNewDbgInfoFormat = SeenDebugRecord;
+      WriteNewDbgInfoFormatToBitcode = SeenDebugRecord;
+      WriteNewDbgInfoFormat = SeenDebugRecord;
+    }
+    // If the module's debug info format doesn't match the observed input
+    // format, then set its format now; we don't need to call the conversion
+    // function because there must be no existing intrinsics to convert.
+    // Otherwise, just set the format on this function now.
+    if (NewDbgInfoFormatDesired != F->getParent()->IsNewDbgInfoFormat)
+      F->getParent()->setNewDbgInfoFormatFlag(NewDbgInfoFormatDesired);
+    else
+      F->setNewDbgInfoFormatFlag(NewDbgInfoFormatDesired);
+  } else {
+    // If we aren't preserving formats, we use the Module flag to get our
+    // desired format instead of reading flags, in case we are lazy-loading and
+    // the format of the module has been changed since it was set by the flags.
+    // We only need to convert debug info here if we have debug records but
+    // desire the intrinsic format; everything else is a no-op or handled by the
+    // autoupgrader.
+    bool ModuleIsNewDbgInfoFormat = F->getParent()->IsNewDbgInfoFormat;
+    if (ModuleIsNewDbgInfoFormat || !SeenDebugRecord)
+      F->setNewDbgInfoFormatFlag(ModuleIsNewDbgInfoFormat);
+    else
+      F->setIsNewDbgInfoFormat(ModuleIsNewDbgInfoFormat);
+  }
 
   if (StripDebugInfo)
     stripDebugInfo(*F);
diff --git a/llvm/lib/CodeGen/GlobalISel/CombinerHelper.cpp b/llvm/lib/CodeGen/GlobalISel/CombinerHelper.cpp
index e53e35d98cec..719209e0edd5 100644
--- a/llvm/lib/CodeGen/GlobalISel/CombinerHelper.cpp
+++ b/llvm/lib/CodeGen/GlobalISel/CombinerHelper.cpp
@@ -6273,14 +6273,15 @@ bool CombinerHelper::matchShiftsTooBig(MachineInstr &MI) {
 bool CombinerHelper::matchCommuteConstantToRHS(MachineInstr &MI) {
   Register LHS = MI.getOperand(1).getReg();
   Register RHS = MI.getOperand(2).getReg();
-  auto *LHSDef = MRI.getVRegDef(LHS);
-  if (getIConstantVRegVal(LHS, MRI).has_value())
-    return true;
-
-  // LHS may be a G_CONSTANT_FOLD_BARRIER. If so we commute
-  // as long as we don't already have a constant on the RHS.
-  if (LHSDef->getOpcode() != TargetOpcode::G_CONSTANT_FOLD_BARRIER)
-    return false;
+  if (!getIConstantVRegVal(LHS, MRI)) {
+    // Skip commuting if LHS is not a constant. But, LHS may be a
+    // G_CONSTANT_FOLD_BARRIER. If so we commute as long as we don't already
+    // have a constant on the RHS.
+    if (MRI.getVRegDef(LHS)->getOpcode() !=
+        TargetOpcode::G_CONSTANT_FOLD_BARRIER)
+      return false;
+  }
+  // Commute as long as RHS is not a constant or G_CONSTANT_FOLD_BARRIER.
   return MRI.getVRegDef(RHS)->getOpcode() !=
              TargetOpcode::G_CONSTANT_FOLD_BARRIER &&
          !getIConstantVRegVal(RHS, MRI);
diff --git a/llvm/lib/ExecutionEngine/Orc/Core.cpp b/llvm/lib/ExecutionEngine/Orc/Core.cpp
index 37c32f86e8d8..60265c4f72ff 100644
--- a/llvm/lib/ExecutionEngine/Orc/Core.cpp
+++ b/llvm/lib/ExecutionEngine/Orc/Core.cpp
@@ -3379,12 +3379,17 @@ ExecutionSession::IL_failSymbols(JITDylib &JD,
       for (auto &DependantEDU : MI.DependantEDUs) {
 
         // Remove DependantEDU from all of its users DependantEDUs lists.
-        for (auto &[JD, Syms] : DependantEDU->Dependencies) {
-          for (auto Sym : Syms) {
-            assert(JD->Symbols.count(SymbolStringPtr(Sym)) && "Sym not in JD?");
-            assert(JD->MaterializingInfos.count(SymbolStringPtr(Sym)) &&
+        for (auto &[DepJD, DepSyms] : DependantEDU->Dependencies) {
+          for (auto DepSym : DepSyms) {
+            // Skip self-reference to avoid invalidating the MI.DependantEDUs
+            // map. We'll clear this later.
+            if (DepJD == &JD && DepSym == Name)
+              continue;
+            assert(DepJD->Symbols.count(SymbolStringPtr(DepSym)) &&
+                   "DepSym not in DepJD?");
+            assert(DepJD->MaterializingInfos.count(SymbolStringPtr(DepSym)) &&
                    "DependantEDU not registered with symbol it depends on");
-            auto SymMI = JD->MaterializingInfos[SymbolStringPtr(Sym)];
+            auto &SymMI = DepJD->MaterializingInfos[SymbolStringPtr(DepSym)];
             assert(SymMI.DependantEDUs.count(DependantEDU) &&
                    "DependantEDU missing from DependantEDUs list");
             SymMI.DependantEDUs.erase(DependantEDU);
diff --git a/llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp b/llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp
index 16507a69ea85..7fd8474c2ec8 100644
--- a/llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp
+++ b/llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp
@@ -4791,6 +4791,9 @@ void OpenMPIRBuilder::writeTeamsForKernel(const Triple &T, Function &Kernel,
       updateNVPTXMetadata(Kernel, "maxclusterrank", UB, true);
     updateNVPTXMetadata(Kernel, "minctasm", LB, false);
   }
+  if (T.isAMDGPU())
+    Kernel.addFnAttr("amdgpu-max-num-workgroups", llvm::utostr(LB) + ",1,1");
+
   Kernel.addFnAttr("omp_target_num_teams", std::to_string(LB));
 }
 
diff --git a/llvm/lib/IR/BasicBlock.cpp b/llvm/lib/IR/BasicBlock.cpp
index ae99267f5ba8..6e62767c99e2 100644
--- a/llvm/lib/IR/BasicBlock.cpp
+++ b/llvm/lib/IR/BasicBlock.cpp
@@ -30,11 +30,19 @@ using namespace llvm;
 #define DEBUG_TYPE "ir"
 STATISTIC(NumInstrRenumberings, "Number of renumberings across all blocks");
 
-cl::opt<bool>
-    UseNewDbgInfoFormat("experimental-debuginfo-iterators",
-                        cl::desc("Enable communicating debuginfo positions "
-                                 "through iterators, eliminating intrinsics"),
-                        cl::init(true));
+cl::opt<bool> UseNewDbgInfoFormat(
+    "experimental-debuginfo-iterators",
+    cl::desc("Enable communicating debuginfo positions through iterators, "
+             "eliminating intrinsics. Has no effect if "
+             "--preserve-input-debuginfo-format=true."),
+    cl::init(true));
+cl::opt<cl::boolOrDefault> PreserveInputDbgFormat(
+    "preserve-input-debuginfo-format", cl::Hidden,
+    cl::desc("When set to true, IR files will be processed and printed in "
+             "their current debug info format, regardless of default behaviour "
+             "or other flags passed. Has no effect if input IR does not "
+             "contain debug records or intrinsics. Ignored in llvm-link, "
+             "llvm-lto, and llvm-lto2."));
 
 bool WriteNewDbgInfoFormatToBitcode /*set default value in cl::init() below*/;
 cl::opt<bool, true> WriteNewDbgInfoFormatToBitcode2(
@@ -147,6 +155,9 @@ void BasicBlock::setIsNewDbgInfoFormat(bool NewFlag) {
   else if (!NewFlag && IsNewDbgInfoFormat)
     convertFromNewDbgValues();
 }
+void BasicBlock::setNewDbgInfoFormatFlag(bool NewFlag) {
+  IsNewDbgInfoFormat = NewFlag;
+}
 
 ValueSymbolTable *BasicBlock::getValueSymbolTable() {
   if (Function *F = getParent())
diff --git a/llvm/lib/IR/Function.cpp b/llvm/lib/IR/Function.cpp
index eb126f182ead..b5fda9bb3d12 100644
--- a/llvm/lib/IR/Function.cpp
+++ b/llvm/lib/IR/Function.cpp
@@ -103,6 +103,12 @@ void Function::setIsNewDbgInfoFormat(bool NewFlag) {
   else if (!NewFlag && IsNewDbgInfoFormat)
     convertFromNewDbgValues();
 }
+void Function::setNewDbgInfoFormatFlag(bool NewFlag) {
+  for (auto &BB : *this) {
+    BB.setNewDbgInfoFormatFlag(NewFlag);
+  }
+  IsNewDbgInfoFormat = NewFlag;
+}
 
 //===----------------------------------------------------------------------===//
 // Argument Implementation
diff --git a/llvm/lib/IR/IRPrintingPasses.cpp b/llvm/lib/IR/IRPrintingPasses.cpp
index 84fb8e6c66b8..43252c57afca 100644
--- a/llvm/lib/IR/IRPrintingPasses.cpp
+++ b/llvm/lib/IR/IRPrintingPasses.cpp
@@ -25,7 +25,8 @@ using namespace llvm;
 
 cl::opt<bool> WriteNewDbgInfoFormat(
     "write-experimental-debuginfo",
-    cl::desc("Write debug info in the new non-intrinsic format"),
+    cl::desc("Write debug info in the new non-intrinsic format. Has no effect "
+             "if --preserve-input-debuginfo-format=true."),
     cl::init(false));
 
 namespace {
diff --git a/llvm/lib/ObjCopy/ELF/ELFObjcopy.cpp b/llvm/lib/ObjCopy/ELF/ELFObjcopy.cpp
index f343d1447e05..205bc1ef5b1a 100644
--- a/llvm/lib/ObjCopy/ELF/ELFObjcopy.cpp
+++ b/llvm/lib/ObjCopy/ELF/ELFObjcopy.cpp
@@ -215,41 +215,23 @@ static Error dumpSectionToFile(StringRef SecName, StringRef Filename,
 }
 
 Error Object::compressOrDecompressSections(const CommonConfig &Config) {
-  // Build a list of sections we are going to replace.
-  // We can't call `addSection` while iterating over sections,
+  // Build a list of the debug sections we are going to replace.
+  // We can't call `AddSection` while iterating over sections,
   // because it would mutate the sections array.
   SmallVector<std::pair<SectionBase *, std::function<SectionBase *()>>, 0>
       ToReplace;
   for (SectionBase &Sec : sections()) {
-    std::optional<DebugCompressionType> CType;
-    for (auto &[Matcher, T] : Config.compressSections)
-      if (Matcher.matches(Sec.Name))
-        CType = T;
-    // Handle --compress-debug-sections and --decompress-debug-sections, which
-    // apply to non-ALLOC debug sections.
-    if (!(Sec.Flags & SHF_ALLOC) && StringRef(Sec.Name).starts_with(".debug")) {
-      if (Config.CompressionType != DebugCompressionType::None)
-        CType = Config.CompressionType;
-      else if (Config.DecompressDebugSections)
-        CType = DebugCompressionType::None;
-    }
-    if (!CType)
+    if ((Sec.Flags & SHF_ALLOC) || !StringRef(Sec.Name).starts_with(".debug"))
       continue;
-
-    if (Sec.ParentSegment)
-      return createStringError(
-          errc::invalid_argument,
-          "section '" + Sec.Name +
-              "' within a segment cannot be (de)compressed");
-
     if (auto *CS = dyn_cast<CompressedSection>(&Sec)) {
-      if (*CType == DebugCompressionType::None)
+      if (Config.DecompressDebugSections) {
         ToReplace.emplace_back(
             &Sec, [=] { return &addSection<DecompressedSection>(*CS); });
-    } else if (*CType != DebugCompressionType::None) {
-      ToReplace.emplace_back(&Sec, [=, S = &Sec] {
+      }
+    } else if (Config.CompressionType != DebugCompressionType::None) {
+      ToReplace.emplace_back(&Sec, [&, S = &Sec] {
         return &addSection<CompressedSection>(
-            CompressedSection(*S, *CType, Is64Bits));
+            CompressedSection(*S, Config.CompressionType, Is64Bits));
       });
     }
   }
diff --git a/llvm/lib/Option/OptTable.cpp b/llvm/lib/Option/OptTable.cpp
index cf69f6173b6d..b8b6b90c253f 100644
--- a/llvm/lib/Option/OptTable.cpp
+++ b/llvm/lib/Option/OptTable.cpp
@@ -710,7 +710,8 @@ void OptTable::printHelp(raw_ostream &OS, const char *Usage, const char *Title,
       OS, Usage, Title, ShowHidden, ShowAllAliases,
       [VisibilityMask](const Info &CandidateInfo) -> bool {
         return (CandidateInfo.Visibility & VisibilityMask) == 0;
-      });
+      },
+      VisibilityMask);
 }
 
 void OptTable::printHelp(raw_ostream &OS, const char *Usage, const char *Title,
@@ -726,13 +727,14 @@ void OptTable::printHelp(raw_ostream &OS, const char *Usage, const char *Title,
         if (CandidateInfo.Flags & FlagsToExclude)
           return true;
         return false;
-      });
+      },
+      Visibility(0));
 }
 
 void OptTable::internalPrintHelp(
     raw_ostream &OS, const char *Usage, const char *Title, bool ShowHidden,
-    bool ShowAllAliases,
-    std::function<bool(const Info &)> ExcludeOption) const {
+    bool ShowAllAliases, std::function<bool(const Info &)> ExcludeOption,
+    Visibility VisibilityMask) const {
   OS << "OVERVIEW: " << Title << "\n\n";
   OS << "USAGE: " << Usage << "\n\n";
 
@@ -754,11 +756,11 @@ void OptTable::internalPrintHelp(
 
     // If an alias doesn't have a help text, show a help text for the aliased
     // option instead.
-    const char *HelpText = getOptionHelpText(Id);
+    const char *HelpText = getOptionHelpText(Id, VisibilityMask);
     if (!HelpText && ShowAllAliases) {
       const Option Alias = getOption(Id).getAlias();
       if (Alias.isValid())
-        HelpText = getOptionHelpText(Alias.getID());
+        HelpText = getOptionHelpText(Alias.getID(), VisibilityMask);
     }
 
     if (HelpText && (strlen(HelpText) != 0)) {
diff --git a/llvm/lib/Passes/PassBuilder.cpp b/llvm/lib/Passes/PassBuilder.cpp
index 57975e34d426..832ee352205a 100644
--- a/llvm/lib/Passes/PassBuilder.cpp
+++ b/llvm/lib/Passes/PassBuilder.cpp
@@ -172,12 +172,12 @@
 #include "llvm/Transforms/Instrumentation/InstrOrderFile.h"
 #include "llvm/Transforms/Instrumentation/InstrProfiling.h"
 #include "llvm/Transforms/Instrumentation/KCFI.h"
+#include "llvm/Transforms/Instrumentation/LowerAllowCheckPass.h"
 #include "llvm/Transforms/Instrumentation/MemProfiler.h"
 #include "llvm/Transforms/Instrumentation/MemorySanitizer.h"
 #include "llvm/Transforms/Instrumentation/PGOForceFunctionAttrs.h"
 #include "llvm/Transforms/Instrumentation/PGOInstrumentation.h"
 #include "llvm/Transforms/Instrumentation/PoisonChecking.h"
-#include "llvm/Transforms/Instrumentation/RemoveTrapsPass.h"
 #include "llvm/Transforms/Instrumentation/SanitizerBinaryMetadata.h"
 #include "llvm/Transforms/Instrumentation/SanitizerCoverage.h"
 #include "llvm/Transforms/Instrumentation/ThreadSanitizer.h"
diff --git a/llvm/lib/Passes/PassRegistry.def b/llvm/lib/Passes/PassRegistry.def
index 41f16d0915bf..d15f58d7adfa 100644
--- a/llvm/lib/Passes/PassRegistry.def
+++ b/llvm/lib/Passes/PassRegistry.def
@@ -363,6 +363,7 @@ FUNCTION_PASS("loop-load-elim", LoopLoadEliminationPass())
 FUNCTION_PASS("loop-simplify", LoopSimplifyPass())
 FUNCTION_PASS("loop-sink", LoopSinkPass())
 FUNCTION_PASS("loop-versioning", LoopVersioningPass())
+FUNCTION_PASS("lower-allow-check", LowerAllowCheckPass())
 FUNCTION_PASS("lower-atomic", LowerAtomicPass())
 FUNCTION_PASS("lower-constant-intrinsics", LowerConstantIntrinsicsPass())
 FUNCTION_PASS("lower-expect", LowerExpectIntrinsicPass())
@@ -422,7 +423,6 @@ FUNCTION_PASS("print<uniformity>", UniformityInfoPrinterPass(dbgs()))
 FUNCTION_PASS("reassociate", ReassociatePass())
 FUNCTION_PASS("redundant-dbg-inst-elim", RedundantDbgInstEliminationPass())
 FUNCTION_PASS("reg2mem", RegToMemPass())
-FUNCTION_PASS("remove-traps", RemoveTrapsPass())
 FUNCTION_PASS("safe-stack", SafeStackPass(TM))
 FUNCTION_PASS("scalarize-masked-mem-intrin", ScalarizeMaskedMemIntrinPass())
 FUNCTION_PASS("scalarizer", ScalarizerPass())
diff --git a/llvm/lib/Target/AArch64/AArch64ISelLowering.cpp b/llvm/lib/Target/AArch64/AArch64ISelLowering.cpp
index 8218960406ec..b81688c35713 100644
--- a/llvm/lib/Target/AArch64/AArch64ISelLowering.cpp
+++ b/llvm/lib/Target/AArch64/AArch64ISelLowering.cpp
@@ -26025,7 +26025,7 @@ static Value *UseTlsOffset(IRBuilderBase &IRB, unsigned Offset) {
 Value *AArch64TargetLowering::getIRStackGuard(IRBuilderBase &IRB) const {
   // Android provides a fixed TLS slot for the stack cookie. See the definition
   // of TLS_SLOT_STACK_GUARD in
-  // https://android.googlesource.com/platform/bionic/+/master/libc/private/bionic_tls.h
+  // https://android.googlesource.com/platform/bionic/+/main/libc/platform/bionic/tls_defines.h
   if (Subtarget->isTargetAndroid())
     return UseTlsOffset(IRB, 0x28);
 
diff --git a/llvm/lib/Target/ARM/Thumb1InstrInfo.cpp b/llvm/lib/Target/ARM/Thumb1InstrInfo.cpp
index 85eabdb17ad1..5b0b799880a3 100644
--- a/llvm/lib/Target/ARM/Thumb1InstrInfo.cpp
+++ b/llvm/lib/Target/ARM/Thumb1InstrInfo.cpp
@@ -12,6 +12,8 @@
 
 #include "Thumb1InstrInfo.h"
 #include "ARMSubtarget.h"
+#include "llvm/ADT/BitVector.h"
+#include "llvm/CodeGen/LiveRegUnits.h"
 #include "llvm/CodeGen/MachineFrameInfo.h"
 #include "llvm/CodeGen/MachineInstrBuilder.h"
 #include "llvm/CodeGen/MachineMemOperand.h"
@@ -47,24 +49,57 @@ void Thumb1InstrInfo::copyPhysReg(MachineBasicBlock &MBB,
   assert(ARM::GPRRegClass.contains(DestReg, SrcReg) &&
          "Thumb1 can only copy GPR registers");
 
-  if (st.hasV6Ops() || ARM::hGPRRegClass.contains(SrcReg)
-      || !ARM::tGPRRegClass.contains(DestReg))
+  if (st.hasV6Ops() || ARM::hGPRRegClass.contains(SrcReg) ||
+      !ARM::tGPRRegClass.contains(DestReg))
     BuildMI(MBB, I, DL, get(ARM::tMOVr), DestReg)
         .addReg(SrcReg, getKillRegState(KillSrc))
         .add(predOps(ARMCC::AL));
   else {
-    // FIXME: Can also use 'mov hi, $src; mov $dst, hi',
-    // with hi as either r10 or r11.
-
     const TargetRegisterInfo *RegInfo = st.getRegisterInfo();
-    if (MBB.computeRegisterLiveness(RegInfo, ARM::CPSR, I)
-        == MachineBasicBlock::LQR_Dead) {
+    LiveRegUnits UsedRegs(*RegInfo);
+    UsedRegs.addLiveOuts(MBB);
+
+    auto InstUpToI = MBB.end();
+    while (InstUpToI != I)
+      // The pre-decrement is on purpose here.
+      // We want to have the liveness right before I.
+      UsedRegs.stepBackward(*--InstUpToI);
+
+    if (UsedRegs.available(ARM::CPSR)) {
       BuildMI(MBB, I, DL, get(ARM::tMOVSr), DestReg)
           .addReg(SrcReg, getKillRegState(KillSrc))
           ->addRegisterDead(ARM::CPSR, RegInfo);
       return;
     }
 
+    // Use high register to move source to destination
+    // if movs is not an option.
+    BitVector Allocatable = RegInfo->getAllocatableSet(
+        MF, RegInfo->getRegClass(ARM::hGPRRegClassID));
+
+    Register TmpReg = ARM::NoRegister;
+    // Prefer R12 as it is known to not be preserved anyway
+    if (UsedRegs.available(ARM::R12) && Allocatable.test(ARM::R12)) {
+      TmpReg = ARM::R12;
+    } else {
+      for (Register Reg : Allocatable.set_bits()) {
+        if (UsedRegs.available(Reg)) {
+          TmpReg = Reg;
+          break;
+        }
+      }
+    }
+
+    if (TmpReg) {
+      BuildMI(MBB, I, DL, get(ARM::tMOVr), TmpReg)
+          .addReg(SrcReg, getKillRegState(KillSrc))
+          .add(predOps(ARMCC::AL));
+      BuildMI(MBB, I, DL, get(ARM::tMOVr), DestReg)
+          .addReg(TmpReg, getKillRegState(true))
+          .add(predOps(ARMCC::AL));
+      return;
+    }
+
     // 'MOV lo, lo' is unpredictable on < v6, so use the stack to do it
     BuildMI(MBB, I, DL, get(ARM::tPUSH))
         .add(predOps(ARMCC::AL))
diff --git a/llvm/lib/Target/RISCV/GISel/RISCVRegisterBankInfo.cpp b/llvm/lib/Target/RISCV/GISel/RISCVRegisterBankInfo.cpp
index 8534024b6002..86e44343b508 100644
--- a/llvm/lib/Target/RISCV/GISel/RISCVRegisterBankInfo.cpp
+++ b/llvm/lib/Target/RISCV/GISel/RISCVRegisterBankInfo.cpp
@@ -290,16 +290,7 @@ RISCVRegisterBankInfo::getInstrMapping(const MachineInstr &MI) const {
 
   switch (Opc) {
   case TargetOpcode::G_ADD:
-  case TargetOpcode::G_SUB: {
-    if (MRI.getType(MI.getOperand(0).getReg()).isVector()) {
-      LLT Ty = MRI.getType(MI.getOperand(0).getReg());
-      return getInstructionMapping(
-          DefaultMappingID, /*Cost=*/1,
-          getVRBValueMapping(Ty.getSizeInBits().getKnownMinValue()),
-          NumOperands);
-    }
-  }
-    LLVM_FALLTHROUGH;
+  case TargetOpcode::G_SUB:
   case TargetOpcode::G_SHL:
   case TargetOpcode::G_ASHR:
   case TargetOpcode::G_LSHR:
@@ -320,10 +311,6 @@ RISCVRegisterBankInfo::getInstrMapping(const MachineInstr &MI) const {
   case TargetOpcode::G_PTR_ADD:
   case TargetOpcode::G_PTRTOINT:
   case TargetOpcode::G_INTTOPTR:
-  case TargetOpcode::G_SEXTLOAD:
-  case TargetOpcode::G_ZEXTLOAD:
-    return getInstructionMapping(DefaultMappingID, /*Cost=*/1, GPRValueMapping,
-                                 NumOperands);
   case TargetOpcode::G_FADD:
   case TargetOpcode::G_FSUB:
   case TargetOpcode::G_FMUL:
@@ -334,10 +321,34 @@ RISCVRegisterBankInfo::getInstrMapping(const MachineInstr &MI) const {
   case TargetOpcode::G_FMAXNUM:
   case TargetOpcode::G_FMINNUM: {
     LLT Ty = MRI.getType(MI.getOperand(0).getReg());
-    return getInstructionMapping(DefaultMappingID, /*Cost=*/1,
-                                 getFPValueMapping(Ty.getSizeInBits()),
-                                 NumOperands);
+    TypeSize Size = Ty.getSizeInBits();
+
+    const ValueMapping *Mapping;
+    if (Ty.isVector())
+      Mapping = getVRBValueMapping(Size.getKnownMinValue());
+    else if (isPreISelGenericFloatingPointOpcode(Opc))
+      Mapping = getFPValueMapping(Size.getFixedValue());
+    else
+      Mapping = GPRValueMapping;
+
+#ifndef NDEBUG
+    // Make sure all the operands are using similar size and type.
+    for (unsigned Idx = 1; Idx != NumOperands; ++Idx) {
+      LLT OpTy = MRI.getType(MI.getOperand(Idx).getReg());
+      assert(Ty.isVector() == OpTy.isVector() &&
+             "Operand has incompatible type");
+      // Don't check size for GPR.
+      if (OpTy.isVector() || isPreISelGenericFloatingPointOpcode(Opc))
+        assert(Size == OpTy.getSizeInBits() && "Operand has incompatible size");
+    }
+#endif // End NDEBUG
+
+    return getInstructionMapping(DefaultMappingID, 1, Mapping, NumOperands);
   }
+  case TargetOpcode::G_SEXTLOAD:
+  case TargetOpcode::G_ZEXTLOAD:
+    return getInstructionMapping(DefaultMappingID, /*Cost=*/1, GPRValueMapping,
+                                 NumOperands);
   case TargetOpcode::G_IMPLICIT_DEF: {
     Register Dst = MI.getOperand(0).getReg();
     LLT DstTy = MRI.getType(Dst);
diff --git a/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp b/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
index 27a4d78d6df7..aeec06313c75 100644
--- a/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
+++ b/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
@@ -245,6 +245,10 @@ RISCVTTIImpl::getIntImmCostIntrin(Intrinsic::ID IID, unsigned Idx,
   return TTI::TCC_Free;
 }
 
+bool RISCVTTIImpl::hasActiveVectorLength(unsigned, Type *DataTy, Align) const {
+  return ST->hasVInstructions();
+}
+
 TargetTransformInfo::PopcntSupportKind
 RISCVTTIImpl::getPopcntSupport(unsigned TyWidth) {
   assert(isPowerOf2_32(TyWidth) && "Ty width must be power of 2");
diff --git a/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.h b/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.h
index ac32aea4ce2b..c0169ea1ad53 100644
--- a/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.h
+++ b/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.h
@@ -78,6 +78,22 @@ public:
                                       const APInt &Imm, Type *Ty,
                                       TTI::TargetCostKind CostKind);
 
+  /// \name EVL Support for predicated vectorization.
+  /// Whether the target supports the %evl parameter of VP intrinsic efficiently
+  /// in hardware, for the given opcode and type/alignment. (see LLVM Language
+  /// Reference - "Vector Predication Intrinsics",
+  /// https://llvm.org/docs/LangRef.html#vector-predication-intrinsics and
+  /// "IR-level VP intrinsics",
+  /// https://llvm.org/docs/Proposals/VectorPredication.html#ir-level-vp-intrinsics).
+  /// \param Opcode the opcode of the instruction checked for predicated version
+  /// support.
+  /// \param DataType the type of the instruction with the \p Opcode checked for
+  /// prediction support.
+  /// \param Alignment the alignment for memory access operation checked for
+  /// predicated version support.
+  bool hasActiveVectorLength(unsigned Opcode, Type *DataType,
+                             Align Alignment) const;
+
   TargetTransformInfo::PopcntSupportKind getPopcntSupport(unsigned TyWidth);
 
   bool shouldExpandReduction(const IntrinsicInst *II) const;
diff --git a/llvm/lib/Target/X86/X86TargetTransformInfo.cpp b/llvm/lib/Target/X86/X86TargetTransformInfo.cpp
index cd6102914cc3..5d1810b5bc2c 100644
--- a/llvm/lib/Target/X86/X86TargetTransformInfo.cpp
+++ b/llvm/lib/Target/X86/X86TargetTransformInfo.cpp
@@ -5698,8 +5698,10 @@ int X86TTIImpl::getScatterOverhead() const {
 
 // Return an average cost of Gather / Scatter instruction, maybe improved later.
 // FIXME: Add TargetCostKind support.
-InstructionCost X86TTIImpl::getGSVectorCost(unsigned Opcode, Type *SrcVTy,
-                                            const Value *Ptr, Align Alignment,
+InstructionCost X86TTIImpl::getGSVectorCost(unsigned Opcode,
+                                            TTI::TargetCostKind CostKind,
+                                            Type *SrcVTy, const Value *Ptr,
+                                            Align Alignment,
                                             unsigned AddressSpace) {
 
   assert(isa<VectorType>(SrcVTy) && "Unexpected type in getGSVectorCost");
@@ -5750,8 +5752,8 @@ InstructionCost X86TTIImpl::getGSVectorCost(unsigned Opcode, Type *SrcVTy,
     // Handle splitting of vector of pointers
     auto *SplitSrcTy =
         FixedVectorType::get(SrcVTy->getScalarType(), VF / SplitFactor);
-    return SplitFactor * getGSVectorCost(Opcode, SplitSrcTy, Ptr, Alignment,
-                                         AddressSpace);
+    return SplitFactor * getGSVectorCost(Opcode, CostKind, SplitSrcTy, Ptr,
+                                         Alignment, AddressSpace);
   }
 
   // The gather / scatter cost is given by Intel architects. It is a rough
@@ -5771,15 +5773,15 @@ InstructionCost X86TTIImpl::getGSVectorCost(unsigned Opcode, Type *SrcVTy,
 /// VariableMask - The mask is non-constant at compile time.
 /// Alignment - Alignment for one element.
 /// AddressSpace - pointer[s] address space.
-///
-/// FIXME: Add TargetCostKind support.
-InstructionCost X86TTIImpl::getGSScalarCost(unsigned Opcode, Type *SrcVTy,
-                                            bool VariableMask, Align Alignment,
+/// TODO: Remove this and use getCommonMaskedMemoryOpCost directly.
+InstructionCost X86TTIImpl::getGSScalarCost(unsigned Opcode,
+                                            TTI::TargetCostKind CostKind,
+                                            Type *SrcVTy, bool VariableMask,
+                                            Align Alignment,
                                             unsigned AddressSpace) {
   Type *ScalarTy = SrcVTy->getScalarType();
   unsigned VF = cast<FixedVectorType>(SrcVTy)->getNumElements();
   APInt DemandedElts = APInt::getAllOnes(VF);
-  TTI::TargetCostKind CostKind = TTI::TCK_RecipThroughput;
 
   InstructionCost MaskUnpackCost = 0;
   if (VariableMask) {
@@ -5848,10 +5850,11 @@ InstructionCost X86TTIImpl::getGatherScatterOpCost(
        (!isLegalMaskedScatter(SrcVTy, Align(Alignment)) ||
         forceScalarizeMaskedScatter(cast<VectorType>(SrcVTy),
                                     Align(Alignment)))))
-    return getGSScalarCost(Opcode, SrcVTy, VariableMask, Alignment,
+    return getGSScalarCost(Opcode, CostKind, SrcVTy, VariableMask, Alignment,
                            AddressSpace);
 
-  return getGSVectorCost(Opcode, SrcVTy, Ptr, Alignment, AddressSpace);
+  return getGSVectorCost(Opcode, CostKind, SrcVTy, Ptr, Alignment,
+                         AddressSpace);
 }
 
 bool X86TTIImpl::isLSRCostLess(const TargetTransformInfo::LSRCost &C1,
diff --git a/llvm/lib/Target/X86/X86TargetTransformInfo.h b/llvm/lib/Target/X86/X86TargetTransformInfo.h
index 23035f655098..985b00438ce8 100644
--- a/llvm/lib/Target/X86/X86TargetTransformInfo.h
+++ b/llvm/lib/Target/X86/X86TargetTransformInfo.h
@@ -294,12 +294,12 @@ public:
 
 private:
   bool supportsGather() const;
-  InstructionCost getGSScalarCost(unsigned Opcode, Type *DataTy,
-                                  bool VariableMask, Align Alignment,
-                                  unsigned AddressSpace);
-  InstructionCost getGSVectorCost(unsigned Opcode, Type *DataTy,
-                                  const Value *Ptr, Align Alignment,
-                                  unsigned AddressSpace);
+  InstructionCost getGSScalarCost(unsigned Opcode, TTI::TargetCostKind CostKind,
+                                  Type *DataTy, bool VariableMask,
+                                  Align Alignment, unsigned AddressSpace);
+  InstructionCost getGSVectorCost(unsigned Opcode, TTI::TargetCostKind CostKind,
+                                  Type *DataTy, const Value *Ptr,
+                                  Align Alignment, unsigned AddressSpace);
 
   int getGatherOverhead() const;
   int getScatterOverhead() const;
diff --git a/llvm/lib/Transforms/Instrumentation/CMakeLists.txt b/llvm/lib/Transforms/Instrumentation/CMakeLists.txt
index b23a6ed1f084..981405329389 100644
--- a/llvm/lib/Transforms/Instrumentation/CMakeLists.txt
+++ b/llvm/lib/Transforms/Instrumentation/CMakeLists.txt
@@ -13,11 +13,11 @@ add_llvm_component_library(LLVMInstrumentation
   InstrOrderFile.cpp
   InstrProfiling.cpp
   KCFI.cpp
+  LowerAllowCheckPass.cpp
   PGOForceFunctionAttrs.cpp
   PGOInstrumentation.cpp
   PGOMemOPSizeOpt.cpp
   PoisonChecking.cpp
-  RemoveTrapsPass.cpp
   SanitizerCoverage.cpp
   SanitizerBinaryMetadata.cpp
   ValueProfileCollector.cpp
diff --git a/llvm/lib/Transforms/Instrumentation/RemoveTrapsPass.cpp b/llvm/lib/Transforms/Instrumentation/LowerAllowCheckPass.cpp
index b281468cb84e..54b63278e9dd 100644
--- a/llvm/lib/Transforms/Instrumentation/RemoveTrapsPass.cpp
+++ b/llvm/lib/Transforms/Instrumentation/LowerAllowCheckPass.cpp
@@ -1,4 +1,4 @@
-//===- RemoveTrapsPass.cpp --------------------------------------*- C++ -*-===//
+//===- LowerAllowCheckPass.cpp ----------------------------------*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
@@ -6,11 +6,12 @@
 //
 //===----------------------------------------------------------------------===//
 
-#include "llvm/Transforms/Instrumentation/RemoveTrapsPass.h"
+#include "llvm/Transforms/Instrumentation/LowerAllowCheckPass.h"
 
 #include "llvm/ADT/SmallVector.h"
 #include "llvm/ADT/Statistic.h"
 #include "llvm/Analysis/ProfileSummaryInfo.h"
+#include "llvm/IR/Constant.h"
 #include "llvm/IR/Instructions.h"
 #include "llvm/IR/IntrinsicInst.h"
 #include "llvm/IR/Intrinsics.h"
@@ -20,13 +21,14 @@
 
 using namespace llvm;
 
-#define DEBUG_TYPE "remove-traps"
+#define DEBUG_TYPE "lower-allow-check"
 
-static cl::opt<int> HotPercentileCutoff("remove-traps-percentile-cutoff-hot",
-                                        cl::desc("Hot percentile cuttoff."));
+static cl::opt<int>
+    HotPercentileCutoff("lower-allow-check-percentile-cutoff-hot",
+                        cl::desc("Hot percentile cuttoff."));
 
 static cl::opt<float>
-    RandomRate("remove-traps-random-rate",
+    RandomRate("lower-allow-check-random-rate",
                cl::desc("Probability value in the range [0.0, 1.0] of "
                         "unconditional pseudo-random checks removal."));
 
@@ -35,9 +37,11 @@ STATISTIC(NumChecksRemoved, "Number of removed checks");
 
 static bool removeUbsanTraps(Function &F, const BlockFrequencyInfo &BFI,
                              const ProfileSummaryInfo *PSI) {
-  SmallVector<IntrinsicInst *, 16> Remove;
+  SmallVector<std::pair<IntrinsicInst *, bool>, 16> ReplaceWithValue;
   std::unique_ptr<RandomNumberGenerator> Rng;
 
+  // TODO:
+  // https://github.com/llvm/llvm-project/pull/84858#discussion_r1520603139
   auto ShouldRemove = [&](bool IsHot) {
     if (!RandomRate.getNumOccurrences())
       return IsHot;
@@ -54,21 +58,23 @@ static bool removeUbsanTraps(Function &F, const BlockFrequencyInfo &BFI,
         continue;
       auto ID = II->getIntrinsicID();
       switch (ID) {
-      case Intrinsic::ubsantrap: {
+      case Intrinsic::allow_ubsan_check:
+      case Intrinsic::allow_runtime_check: {
         ++NumChecksTotal;
 
         bool IsHot = false;
         if (PSI) {
-          uint64_t Count = 0;
-          for (const auto *PR : predecessors(&BB))
-            Count += BFI.getBlockProfileCount(PR).value_or(0);
+          uint64_t Count = BFI.getBlockProfileCount(&BB).value_or(0);
           IsHot = PSI->isHotCountNthPercentile(HotPercentileCutoff, Count);
         }
 
-        if (ShouldRemove(IsHot)) {
-          Remove.push_back(II);
+        bool ToRemove = ShouldRemove(IsHot);
+        ReplaceWithValue.push_back({
+            II,
+            ToRemove,
+        });
+        if (ToRemove)
           ++NumChecksRemoved;
-        }
         break;
       }
       default:
@@ -77,14 +83,16 @@ static bool removeUbsanTraps(Function &F, const BlockFrequencyInfo &BFI,
     }
   }
 
-  for (IntrinsicInst *I : Remove)
+  for (auto [I, V] : ReplaceWithValue) {
+    I->replaceAllUsesWith(ConstantInt::getBool(I->getType(), !V));
     I->eraseFromParent();
+  }
 
-  return !Remove.empty();
+  return !ReplaceWithValue.empty();
 }
 
-PreservedAnalyses RemoveTrapsPass::run(Function &F,
-                                       FunctionAnalysisManager &AM) {
+PreservedAnalyses LowerAllowCheckPass::run(Function &F,
+                                           FunctionAnalysisManager &AM) {
   if (F.isDeclaration())
     return PreservedAnalyses::all();
   auto &MAMProxy = AM.getResult<ModuleAnalysisManagerFunctionProxy>(F);
@@ -96,7 +104,7 @@ PreservedAnalyses RemoveTrapsPass::run(Function &F,
                                        : PreservedAnalyses::all();
 }
 
-bool RemoveTrapsPass::IsRequested() {
+bool LowerAllowCheckPass::IsRequested() {
   return RandomRate.getNumOccurrences() ||
          HotPercentileCutoff.getNumOccurrences();
 }
diff --git a/llvm/lib/Transforms/Utils/LoopUtils.cpp b/llvm/lib/Transforms/Utils/LoopUtils.cpp
index 73c5d6367822..9d816c522053 100644
--- a/llvm/lib/Transforms/Utils/LoopUtils.cpp
+++ b/llvm/lib/Transforms/Utils/LoopUtils.cpp
@@ -1034,15 +1034,6 @@ CmpInst::Predicate llvm::getMinMaxReductionPredicate(RecurKind RK) {
   }
 }
 
-Value *llvm::createAnyOfOp(IRBuilderBase &Builder, Value *StartVal,
-                           RecurKind RK, Value *Left, Value *Right) {
-  if (auto VTy = dyn_cast<VectorType>(Left->getType()))
-    StartVal = Builder.CreateVectorSplat(VTy->getElementCount(), StartVal);
-  Value *Cmp =
-      Builder.CreateCmp(CmpInst::ICMP_NE, Left, StartVal, "rdx.select.cmp");
-  return Builder.CreateSelect(Cmp, Left, Right, "rdx.select");
-}
-
 Value *llvm::createMinMaxOp(IRBuilderBase &Builder, RecurKind RK, Value *Left,
                             Value *Right) {
   Type *Ty = Left->getType();
@@ -1151,16 +1142,13 @@ Value *llvm::createAnyOfTargetReduction(IRBuilderBase &Builder, Value *Src,
     NewVal = SI->getTrueValue();
   }
 
-  // Create a splat vector with the new value and compare this to the vector
-  // we want to reduce.
-  ElementCount EC = cast<VectorType>(Src->getType())->getElementCount();
-  Value *Right = Builder.CreateVectorSplat(EC, InitVal);
-  Value *Cmp =
-      Builder.CreateCmp(CmpInst::ICMP_NE, Src, Right, "rdx.select.cmp");
-
   // If any predicate is true it means that we want to select the new value.
-  Cmp = Builder.CreateOrReduce(Cmp);
-  return Builder.CreateSelect(Cmp, NewVal, InitVal, "rdx.select");
+  Value *AnyOf =
+      Src->getType()->isVectorTy() ? Builder.CreateOrReduce(Src) : Src;
+  // The compares in the loop may yield poison, which propagates through the
+  // bitwise ORs. Freeze it here before the condition is used.
+  AnyOf = Builder.CreateFreeze(AnyOf);
+  return Builder.CreateSelect(AnyOf, NewVal, InitVal, "rdx.select");
 }
 
 Value *llvm::createSimpleTargetReduction(IRBuilderBase &Builder, Value *Src,
diff --git a/llvm/lib/Transforms/Vectorize/LoopVectorizationPlanner.h b/llvm/lib/Transforms/Vectorize/LoopVectorizationPlanner.h
index ebca2d855a46..ece2a34f180c 100644
--- a/llvm/lib/Transforms/Vectorize/LoopVectorizationPlanner.h
+++ b/llvm/lib/Transforms/Vectorize/LoopVectorizationPlanner.h
@@ -68,9 +68,7 @@ class VPBuilder {
 public:
   VPBuilder() = default;
   VPBuilder(VPBasicBlock *InsertBB) { setInsertPoint(InsertBB); }
-  VPBuilder(VPRecipeBase *InsertPt) {
-    setInsertPoint(InsertPt->getParent(), InsertPt->getIterator());
-  }
+  VPBuilder(VPRecipeBase *InsertPt) { setInsertPoint(InsertPt); }
 
   /// Clear the insertion point: created instructions will not be inserted into
   /// a block.
diff --git a/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp b/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
index 0834865173b2..49bacb5ae6cc 100644
--- a/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
+++ b/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
@@ -124,6 +124,7 @@
 #include "llvm/IR/User.h"
 #include "llvm/IR/Value.h"
 #include "llvm/IR/ValueHandle.h"
+#include "llvm/IR/VectorBuilder.h"
 #include "llvm/IR/Verifier.h"
 #include "llvm/Support/Casting.h"
 #include "llvm/Support/CommandLine.h"
@@ -248,10 +249,12 @@ static cl::opt<TailFoldingStyle> ForceTailFoldingStyle(
         clEnumValN(TailFoldingStyle::DataAndControlFlow, "data-and-control",
                    "Create lane mask using active.lane.mask intrinsic, and use "
                    "it for both data and control flow"),
-        clEnumValN(
-            TailFoldingStyle::DataAndControlFlowWithoutRuntimeCheck,
-            "data-and-control-without-rt-check",
-            "Similar to data-and-control, but remove the runtime check")));
+        clEnumValN(TailFoldingStyle::DataAndControlFlowWithoutRuntimeCheck,
+                   "data-and-control-without-rt-check",
+                   "Similar to data-and-control, but remove the runtime check"),
+        clEnumValN(TailFoldingStyle::DataWithEVL, "data-with-evl",
+                   "Use predicated EVL instructions for tail folding. If EVL "
+                   "is unsupported, fallback to data-without-lane-mask.")));
 
 static cl::opt<bool> MaximizeBandwidth(
     "vectorizer-maximize-bandwidth", cl::init(false), cl::Hidden,
@@ -1505,29 +1508,62 @@ public:
 
   /// Returns the TailFoldingStyle that is best for the current loop.
   TailFoldingStyle getTailFoldingStyle(bool IVUpdateMayOverflow = true) const {
-    return IVUpdateMayOverflow ? ChosenTailFoldingStyle.first
-                               : ChosenTailFoldingStyle.second;
+    if (!ChosenTailFoldingStyle)
+      return TailFoldingStyle::None;
+    return IVUpdateMayOverflow ? ChosenTailFoldingStyle->first
+                               : ChosenTailFoldingStyle->second;
   }
 
   /// Selects and saves TailFoldingStyle for 2 options - if IV update may
   /// overflow or not.
-  void setTailFoldingStyles() {
-    assert(ChosenTailFoldingStyle.first == TailFoldingStyle::None &&
-           ChosenTailFoldingStyle.second == TailFoldingStyle::None &&
-           "Tail folding must not be selected yet.");
-    if (!Legal->prepareToFoldTailByMasking())
+  /// \param IsScalableVF true if scalable vector factors enabled.
+  /// \param UserIC User specific interleave count.
+  void setTailFoldingStyles(bool IsScalableVF, unsigned UserIC) {
+    assert(!ChosenTailFoldingStyle && "Tail folding must not be selected yet.");
+    if (!Legal->prepareToFoldTailByMasking()) {
+      ChosenTailFoldingStyle =
+          std::make_pair(TailFoldingStyle::None, TailFoldingStyle::None);
       return;
+    }
 
-    if (ForceTailFoldingStyle.getNumOccurrences()) {
-      ChosenTailFoldingStyle.first = ChosenTailFoldingStyle.second =
-          ForceTailFoldingStyle;
+    if (!ForceTailFoldingStyle.getNumOccurrences()) {
+      ChosenTailFoldingStyle = std::make_pair(
+          TTI.getPreferredTailFoldingStyle(/*IVUpdateMayOverflow=*/true),
+          TTI.getPreferredTailFoldingStyle(/*IVUpdateMayOverflow=*/false));
       return;
     }
 
-    ChosenTailFoldingStyle.first =
-        TTI.getPreferredTailFoldingStyle(/*IVUpdateMayOverflow=*/true);
-    ChosenTailFoldingStyle.second =
-        TTI.getPreferredTailFoldingStyle(/*IVUpdateMayOverflow=*/false);
+    // Set styles when forced.
+    ChosenTailFoldingStyle = std::make_pair(ForceTailFoldingStyle.getValue(),
+                                            ForceTailFoldingStyle.getValue());
+    if (ForceTailFoldingStyle != TailFoldingStyle::DataWithEVL)
+      return;
+    // Override forced styles if needed.
+    // FIXME: use actual opcode/data type for analysis here.
+    // FIXME: Investigate opportunity for fixed vector factor.
+    bool EVLIsLegal =
+        IsScalableVF && UserIC <= 1 &&
+        TTI.hasActiveVectorLength(0, nullptr, Align()) &&
+        !EnableVPlanNativePath &&
+        // FIXME: implement support for max safe dependency distance.
+        Legal->isSafeForAnyVectorWidth() &&
+        // FIXME: remove this once reductions are supported.
+        Legal->getReductionVars().empty();
+    if (!EVLIsLegal) {
+      // If for some reason EVL mode is unsupported, fallback to
+      // DataWithoutLaneMask to try to vectorize the loop with folded tail
+      // in a generic way.
+      ChosenTailFoldingStyle =
+          std::make_pair(TailFoldingStyle::DataWithoutLaneMask,
+                         TailFoldingStyle::DataWithoutLaneMask);
+      LLVM_DEBUG(
+          dbgs()
+          << "LV: Preference for VP intrinsics indicated. Will "
+             "not try to generate VP Intrinsics "
+          << (UserIC > 1
+                  ? "since interleave count specified is greater than 1.\n"
+                  : "due to non-interleaving reasons.\n"));
+    }
   }
 
   /// Returns true if all loop blocks should be masked to fold tail loop.
@@ -1544,6 +1580,18 @@ public:
     return foldTailByMasking() || Legal->blockNeedsPredication(BB);
   }
 
+  /// Returns true if VP intrinsics with explicit vector length support should
+  /// be generated in the tail folded loop.
+  bool foldTailWithEVL() const {
+    return getTailFoldingStyle() == TailFoldingStyle::DataWithEVL &&
+           // FIXME: remove this once vp_reverse is supported.
+           none_of(
+               WideningDecisions,
+               [](const std::pair<std::pair<Instruction *, ElementCount>,
+                                  std::pair<InstWidening, InstructionCost>>
+                      &Data) { return Data.second.first == CM_Widen_Reverse; });
+  }
+
   /// Returns true if the Phi is part of an inloop reduction.
   bool isInLoopReduction(PHINode *Phi) const {
     return InLoopReductions.contains(Phi);
@@ -1688,8 +1736,8 @@ private:
 
   /// Control finally chosen tail folding style. The first element is used if
   /// the IV update may overflow, the second element - if it does not.
-  std::pair<TailFoldingStyle, TailFoldingStyle> ChosenTailFoldingStyle =
-      std::make_pair(TailFoldingStyle::None, TailFoldingStyle::None);
+  std::optional<std::pair<TailFoldingStyle, TailFoldingStyle>>
+      ChosenTailFoldingStyle;
 
   /// A map holding scalar costs for different vectorization factors. The
   /// presence of a cost for an instruction in the mapping indicates that the
@@ -3007,9 +3055,8 @@ PHINode *InnerLoopVectorizer::createInductionResumeValue(
   }
 
   // Create phi nodes to merge from the  backedge-taken check block.
-  PHINode *BCResumeVal =
-      PHINode::Create(OrigPhi->getType(), 3, "bc.resume.val",
-                      LoopScalarPreHeader->getTerminator()->getIterator());
+  PHINode *BCResumeVal = PHINode::Create(OrigPhi->getType(), 3, "bc.resume.val",
+                                         LoopScalarPreHeader->getFirstNonPHI());
   // Copy original phi DL over to the new one.
   BCResumeVal->setDebugLoc(OrigPhi->getDebugLoc());
 
@@ -4647,9 +4694,24 @@ LoopVectorizationCostModel::computeMaxVF(ElementCount UserVF, unsigned UserIC) {
   // found modulo the vectorization factor is not zero, try to fold the tail
   // by masking.
   // FIXME: look for a smaller MaxVF that does divide TC rather than masking.
-  setTailFoldingStyles();
-  if (foldTailByMasking())
+  setTailFoldingStyles(MaxFactors.ScalableVF.isScalable(), UserIC);
+  if (foldTailByMasking()) {
+    if (getTailFoldingStyle() == TailFoldingStyle::DataWithEVL) {
+      LLVM_DEBUG(
+          dbgs()
+          << "LV: tail is folded with EVL, forcing unroll factor to be 1. Will "
+             "try to generate VP Intrinsics with scalable vector "
+             "factors only.\n");
+      // Tail folded loop using VP intrinsics restricts the VF to be scalable
+      // for now.
+      // TODO: extend it for fixed vectors, if required.
+      assert(MaxFactors.ScalableVF.isScalable() &&
+             "Expected scalable vector factor.");
+
+      MaxFactors.FixedVF = ElementCount::getFixed(1);
+    }
     return MaxFactors;
+  }
 
   // If there was a tail-folding hint/switch, but we can't fold the tail by
   // masking, fallback to a vectorization with a scalar epilogue.
@@ -5257,6 +5319,13 @@ LoopVectorizationCostModel::selectInterleaveCount(ElementCount VF,
   if (!isScalarEpilogueAllowed())
     return 1;
 
+  // Do not interleave if EVL is preferred and no User IC is specified.
+  if (foldTailWithEVL()) {
+    LLVM_DEBUG(dbgs() << "LV: Preference for VP intrinsics indicated. "
+                         "Unroll factor forced to be 1.\n");
+    return 1;
+  }
+
   // We used the distance for the interleave count.
   if (!Legal->isSafeForAnyVectorWidth())
     return 1;
@@ -7390,7 +7459,6 @@ static void createAndCollectMergePhiForReduction(
   auto *PhiR = cast<VPReductionPHIRecipe>(RedResult->getOperand(0));
   const RecurrenceDescriptor &RdxDesc = PhiR->getRecurrenceDescriptor();
 
-  TrackingVH<Value> ReductionStartValue = RdxDesc.getRecurrenceStartValue();
   Value *FinalValue =
       State.get(RedResult, VPIteration(State.UF - 1, VPLane::getFirstLane()));
   auto *ResumePhi =
@@ -7415,7 +7483,7 @@ static void createAndCollectMergePhiForReduction(
       BCBlockPhi->addIncoming(ResumePhi->getIncomingValueForBlock(Incoming),
                               Incoming);
     else
-      BCBlockPhi->addIncoming(ReductionStartValue, Incoming);
+      BCBlockPhi->addIncoming(RdxDesc.getRecurrenceStartValue(), Incoming);
   }
 
   auto *OrigPhi = cast<PHINode>(PhiR->getUnderlyingValue());
@@ -7708,11 +7776,10 @@ EpilogueVectorizerEpilogueLoop::createEpilogueVectorizedLoopSkeleton(
 
   // Now, compare the remaining count and if there aren't enough iterations to
   // execute the vectorized epilogue skip to the scalar part.
-  BasicBlock *VecEpilogueIterationCountCheck = LoopVectorPreHeader;
-  VecEpilogueIterationCountCheck->setName("vec.epilog.iter.check");
-  LoopVectorPreHeader =
-      SplitBlock(LoopVectorPreHeader, LoopVectorPreHeader->getTerminator(), DT,
-                 LI, nullptr, "vec.epilog.ph");
+  LoopVectorPreHeader->setName("vec.epilog.ph");
+  BasicBlock *VecEpilogueIterationCountCheck =
+      SplitBlock(LoopVectorPreHeader, LoopVectorPreHeader->begin(), DT, LI,
+                 nullptr, "vec.epilog.iter.check", true);
   emitMinimumVectorEpilogueIterCountCheck(LoopScalarPreHeader,
                                           VecEpilogueIterationCountCheck);
 
@@ -8487,6 +8554,9 @@ void LoopVectorizationPlanner::buildVPlansWithVPRecipes(ElementCount MinVF,
         VPlanTransforms::truncateToMinimalBitwidths(
             *Plan, CM.getMinimalBitwidths(), PSE.getSE()->getContext());
       VPlanTransforms::optimize(*Plan, *PSE.getSE());
+      // TODO: try to put it close to addActiveLaneMask().
+      if (CM.foldTailWithEVL())
+        VPlanTransforms::addExplicitVectorLength(*Plan);
       assert(verifyVPlanIsValid(*Plan) && "VPlan is invalid");
       VPlans.push_back(std::move(Plan));
     }
@@ -8828,6 +8898,10 @@ VPlanPtr LoopVectorizationPlanner::buildVPlan(VFRange &Range) {
 // A ComputeReductionResult recipe is added to the middle block, also for
 // in-loop reductions which compute their result in-loop, because generating
 // the subsequent bc.merge.rdx phi is driven by ComputeReductionResult recipes.
+//
+// Adjust AnyOf reductions; replace the reduction phi for the selected value
+// with a boolean reduction phi node to check if the condition is true in any
+// iteration. The final value is selected by the final ComputeReductionResult.
 void LoopVectorizationPlanner::adjustRecipesForReductions(
     VPBasicBlock *LatchVPBB, VPlanPtr &Plan, VPRecipeBuilder &RecipeBuilder,
     ElementCount MinVF) {
@@ -9001,6 +9075,41 @@ void LoopVectorizationPlanner::adjustRecipesForReductions(
       continue;
 
     const RecurrenceDescriptor &RdxDesc = PhiR->getRecurrenceDescriptor();
+    // Adjust AnyOf reductions; replace the reduction phi for the selected value
+    // with a boolean reduction phi node to check if the condition is true in
+    // any iteration. The final value is selected by the final
+    // ComputeReductionResult.
+    if (RecurrenceDescriptor::isAnyOfRecurrenceKind(
+            RdxDesc.getRecurrenceKind())) {
+      auto *Select = cast<VPRecipeBase>(*find_if(PhiR->users(), [](VPUser *U) {
+        return isa<VPWidenSelectRecipe>(U) ||
+               (isa<VPReplicateRecipe>(U) &&
+                cast<VPReplicateRecipe>(U)->getUnderlyingInstr()->getOpcode() ==
+                    Instruction::Select);
+      }));
+      VPValue *Cmp = Select->getOperand(0);
+      // If the compare is checking the reduction PHI node, adjust it to check
+      // the start value.
+      if (VPRecipeBase *CmpR = Cmp->getDefiningRecipe()) {
+        for (unsigned I = 0; I != CmpR->getNumOperands(); ++I)
+          if (CmpR->getOperand(I) == PhiR)
+            CmpR->setOperand(I, PhiR->getStartValue());
+      }
+      VPBuilder::InsertPointGuard Guard(Builder);
+      Builder.setInsertPoint(Select);
+
+      // If the true value of the select is the reduction phi, the new value is
+      // selected if the negated condition is true in any iteration.
+      if (Select->getOperand(1) == PhiR)
+        Cmp = Builder.createNot(Cmp);
+      VPValue *Or = Builder.createOr(PhiR, Cmp);
+      Select->getVPSingleValue()->replaceAllUsesWith(Or);
+
+      // Convert the reduction phi to operate on bools.
+      PhiR->setOperand(0, Plan->getOrAddLiveIn(ConstantInt::getFalse(
+                              OrigLoop->getHeader()->getContext())));
+    }
+
     // If tail is folded by masking, introduce selects between the phi
     // and the live-out instruction of each reduction, at the beginning of the
     // dedicated latch block.
@@ -9033,7 +9142,9 @@ void LoopVectorizationPlanner::adjustRecipesForReductions(
     // then extend the loop exit value to enable InstCombine to evaluate the
     // entire expression in the smaller type.
     Type *PhiTy = PhiR->getStartValue()->getLiveInIRValue()->getType();
-    if (MinVF.isVector() && PhiTy != RdxDesc.getRecurrenceType()) {
+    if (MinVF.isVector() && PhiTy != RdxDesc.getRecurrenceType() &&
+        !RecurrenceDescriptor::isAnyOfRecurrenceKind(
+            RdxDesc.getRecurrenceKind())) {
       assert(!PhiR->isInLoop() && "Unexpected truncated inloop reduction!");
       Type *RdxTy = RdxDesc.getRecurrenceType();
       auto *Trunc =
@@ -9179,7 +9290,7 @@ void VPDerivedIVRecipe::execute(VPTransformState &State) {
     State.Builder.setFastMathFlags(FPBinOp->getFastMathFlags());
 
   Value *Step = State.get(getStepValue(), VPIteration(0, 0));
-  Value *CanonicalIV = State.get(getCanonicalIV(), VPIteration(0, 0));
+  Value *CanonicalIV = State.get(getOperand(1), VPIteration(0, 0));
   Value *DerivedIV = emitTransformedIndex(
       State.Builder, CanonicalIV, getStartValue()->getLiveInIRValue(), Step,
       Kind, cast_if_present<BinaryOperator>(FPBinOp));
@@ -9307,6 +9418,52 @@ void VPReplicateRecipe::execute(VPTransformState &State) {
       State.ILV->scalarizeInstruction(UI, this, VPIteration(Part, Lane), State);
 }
 
+/// Creates either vp_store or vp_scatter intrinsics calls to represent
+/// predicated store/scatter.
+static Instruction *
+lowerStoreUsingVectorIntrinsics(IRBuilderBase &Builder, Value *Addr,
+                                Value *StoredVal, bool IsScatter, Value *Mask,
+                                Value *EVL, const Align &Alignment) {
+  CallInst *Call;
+  if (IsScatter) {
+    Call = Builder.CreateIntrinsic(Type::getVoidTy(EVL->getContext()),
+                                   Intrinsic::vp_scatter,
+                                   {StoredVal, Addr, Mask, EVL});
+  } else {
+    VectorBuilder VBuilder(Builder);
+    VBuilder.setEVL(EVL).setMask(Mask);
+    Call = cast<CallInst>(VBuilder.createVectorInstruction(
+        Instruction::Store, Type::getVoidTy(EVL->getContext()),
+        {StoredVal, Addr}));
+  }
+  Call->addParamAttr(
+      1, Attribute::getWithAlignment(Call->getContext(), Alignment));
+  return Call;
+}
+
+/// Creates either vp_load or vp_gather intrinsics calls to represent
+/// predicated load/gather.
+static Instruction *lowerLoadUsingVectorIntrinsics(IRBuilderBase &Builder,
+                                                   VectorType *DataTy,
+                                                   Value *Addr, bool IsGather,
+                                                   Value *Mask, Value *EVL,
+                                                   const Align &Alignment) {
+  CallInst *Call;
+  if (IsGather) {
+    Call =
+        Builder.CreateIntrinsic(DataTy, Intrinsic::vp_gather, {Addr, Mask, EVL},
+                                nullptr, "wide.masked.gather");
+  } else {
+    VectorBuilder VBuilder(Builder);
+    VBuilder.setEVL(EVL).setMask(Mask);
+    Call = cast<CallInst>(VBuilder.createVectorInstruction(
+        Instruction::Load, DataTy, Addr, "vp.op.load"));
+  }
+  Call->addParamAttr(
+      0, Attribute::getWithAlignment(Call->getContext(), Alignment));
+  return Call;
+}
+
 void VPWidenMemoryInstructionRecipe::execute(VPTransformState &State) {
   VPValue *StoredValue = isStore() ? getStoredValue() : nullptr;
 
@@ -9345,7 +9502,25 @@ void VPWidenMemoryInstructionRecipe::execute(VPTransformState &State) {
     for (unsigned Part = 0; Part < State.UF; ++Part) {
       Instruction *NewSI = nullptr;
       Value *StoredVal = State.get(StoredValue, Part);
-      if (CreateGatherScatter) {
+      // TODO: split this into several classes for better design.
+      if (State.EVL) {
+        assert(State.UF == 1 && "Expected only UF == 1 when vectorizing with "
+                                "explicit vector length.");
+        assert(cast<VPInstruction>(State.EVL)->getOpcode() ==
+                   VPInstruction::ExplicitVectorLength &&
+               "EVL must be VPInstruction::ExplicitVectorLength.");
+        Value *EVL = State.get(State.EVL, VPIteration(0, 0));
+        // If EVL is not nullptr, then EVL must be a valid value set during plan
+        // creation, possibly default value = whole vector register length. EVL
+        // is created only if TTI prefers predicated vectorization, thus if EVL
+        // is not nullptr it also implies preference for predicated
+        // vectorization.
+        // FIXME: Support reverse store after vp_reverse is added.
+        Value *MaskPart = isMaskRequired ? BlockInMaskParts[Part] : nullptr;
+        NewSI = lowerStoreUsingVectorIntrinsics(
+            Builder, State.get(getAddr(), Part, !CreateGatherScatter),
+            StoredVal, CreateGatherScatter, MaskPart, EVL, Alignment);
+      } else if (CreateGatherScatter) {
         Value *MaskPart = isMaskRequired ? BlockInMaskParts[Part] : nullptr;
         Value *VectorGep = State.get(getAddr(), Part);
         NewSI = Builder.CreateMaskedScatter(StoredVal, VectorGep, Alignment,
@@ -9375,7 +9550,25 @@ void VPWidenMemoryInstructionRecipe::execute(VPTransformState &State) {
   State.setDebugLocFrom(getDebugLoc());
   for (unsigned Part = 0; Part < State.UF; ++Part) {
     Value *NewLI;
-    if (CreateGatherScatter) {
+    // TODO: split this into several classes for better design.
+    if (State.EVL) {
+      assert(State.UF == 1 && "Expected only UF == 1 when vectorizing with "
+                              "explicit vector length.");
+      assert(cast<VPInstruction>(State.EVL)->getOpcode() ==
+                 VPInstruction::ExplicitVectorLength &&
+             "EVL must be VPInstruction::ExplicitVectorLength.");
+      Value *EVL = State.get(State.EVL, VPIteration(0, 0));
+      // If EVL is not nullptr, then EVL must be a valid value set during plan
+      // creation, possibly default value = whole vector register length. EVL
+      // is created only if TTI prefers predicated vectorization, thus if EVL
+      // is not nullptr it also implies preference for predicated
+      // vectorization.
+      // FIXME: Support reverse loading after vp_reverse is added.
+      Value *MaskPart = isMaskRequired ? BlockInMaskParts[Part] : nullptr;
+      NewLI = lowerLoadUsingVectorIntrinsics(
+          Builder, DataTy, State.get(getAddr(), Part, !CreateGatherScatter),
+          CreateGatherScatter, MaskPart, EVL, Alignment);
+    } else if (CreateGatherScatter) {
       Value *MaskPart = isMaskRequired ? BlockInMaskParts[Part] : nullptr;
       Value *VectorGep = State.get(getAddr(), Part);
       NewLI = Builder.CreateMaskedGather(DataTy, VectorGep, Alignment, MaskPart,
@@ -10043,9 +10236,19 @@ bool LoopVectorizePass::processLoop(Loop *L) {
           Value *ResumeV = nullptr;
           // TODO: Move setting of resume values to prepareToExecute.
           if (auto *ReductionPhi = dyn_cast<VPReductionPHIRecipe>(&R)) {
-            ResumeV = ReductionResumeValues
-                          .find(&ReductionPhi->getRecurrenceDescriptor())
-                          ->second;
+            const RecurrenceDescriptor &RdxDesc =
+                ReductionPhi->getRecurrenceDescriptor();
+            RecurKind RK = RdxDesc.getRecurrenceKind();
+            ResumeV = ReductionResumeValues.find(&RdxDesc)->second;
+            if (RecurrenceDescriptor::isAnyOfRecurrenceKind(RK)) {
+              // VPReductionPHIRecipes for AnyOf reductions expect a boolean as
+              // start value; compare the final value from the main vector loop
+              // to the start value.
+              IRBuilder<> Builder(
+                  cast<Instruction>(ResumeV)->getParent()->getFirstNonPHI());
+              ResumeV = Builder.CreateICmpNE(ResumeV,
+                                             RdxDesc.getRecurrenceStartValue());
+            }
           } else {
             // Create induction resume values for both widened pointer and
             // integer/fp inductions and update the start value of the induction
diff --git a/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp b/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp
index bdd26acfd2f8..332877f35081 100644
--- a/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp
+++ b/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp
@@ -14146,6 +14146,7 @@ bool BoUpSLP::collectValuesToDemote(
                 return getTreeEntry(U) ||
                        (UserIgnoreList && UserIgnoreList->contains(U)) ||
                        (U->getType()->isSized() &&
+                        !U->getType()->isScalableTy() &&
                         DL->getTypeSizeInBits(U->getType()) <= BitWidth);
               }) &&
       !IsPotentiallyTruncated(I, BitWidth))
diff --git a/llvm/lib/Transforms/Vectorize/VPlan.cpp b/llvm/lib/Transforms/Vectorize/VPlan.cpp
index f0b7008992d7..8ebd75da3465 100644
--- a/llvm/lib/Transforms/Vectorize/VPlan.cpp
+++ b/llvm/lib/Transforms/Vectorize/VPlan.cpp
@@ -871,13 +871,15 @@ void VPlan::execute(VPTransformState *State) {
     // only a single part is generated, which provides the last part from the
     // previous iteration. For non-ordered reductions all UF parts are
     // generated.
-    bool SinglePartNeeded = isa<VPCanonicalIVPHIRecipe>(PhiR) ||
-                            isa<VPFirstOrderRecurrencePHIRecipe>(PhiR) ||
-                            (isa<VPReductionPHIRecipe>(PhiR) &&
-                             cast<VPReductionPHIRecipe>(PhiR)->isOrdered());
-    bool NeedsScalar = isa<VPCanonicalIVPHIRecipe>(PhiR) ||
-                       (isa<VPReductionPHIRecipe>(PhiR) &&
-                        cast<VPReductionPHIRecipe>(PhiR)->isInLoop());
+    bool SinglePartNeeded =
+        isa<VPCanonicalIVPHIRecipe>(PhiR) ||
+        isa<VPFirstOrderRecurrencePHIRecipe, VPEVLBasedIVPHIRecipe>(PhiR) ||
+        (isa<VPReductionPHIRecipe>(PhiR) &&
+         cast<VPReductionPHIRecipe>(PhiR)->isOrdered());
+    bool NeedsScalar =
+        isa<VPCanonicalIVPHIRecipe, VPEVLBasedIVPHIRecipe>(PhiR) ||
+        (isa<VPReductionPHIRecipe>(PhiR) &&
+         cast<VPReductionPHIRecipe>(PhiR)->isInLoop());
     unsigned LastPartForNewPhi = SinglePartNeeded ? 1 : State->UF;
 
     for (unsigned Part = 0; Part < LastPartForNewPhi; ++Part) {
diff --git a/llvm/lib/Transforms/Vectorize/VPlan.h b/llvm/lib/Transforms/Vectorize/VPlan.h
index 813ebda29ffd..77577b516ae2 100644
--- a/llvm/lib/Transforms/Vectorize/VPlan.h
+++ b/llvm/lib/Transforms/Vectorize/VPlan.h
@@ -242,6 +242,15 @@ struct VPTransformState {
   ElementCount VF;
   unsigned UF;
 
+  /// If EVL (Explicit Vector Length) is not nullptr, then EVL must be a valid
+  /// value set during plan transformation, possibly a default value = whole
+  /// vector register length. EVL is created only if TTI prefers predicated
+  /// vectorization, thus if EVL is not nullptr it also implies preference for
+  /// predicated vectorization.
+  /// TODO: this is a temporarily solution, the EVL must be explicitly used by
+  /// the recipes and must be removed here.
+  VPValue *EVL = nullptr;
+
   /// Hold the indices to generate specific scalar instructions. Null indicates
   /// that all instances are to be generated, using either scalar or vector
   /// instructions.
@@ -1159,6 +1168,7 @@ public:
     SLPLoad,
     SLPStore,
     ActiveLaneMask,
+    ExplicitVectorLength,
     CalculateTripCountMinusVF,
     // Increment the canonical IV separately for each unrolled part.
     CanonicalIVIncrementForPart,
@@ -2489,6 +2499,45 @@ public:
 #endif
 };
 
+/// A recipe for generating the phi node for the current index of elements,
+/// adjusted in accordance with EVL value. It starts at the start value of the
+/// canonical induction and gets incremented by EVL in each iteration of the
+/// vector loop.
+class VPEVLBasedIVPHIRecipe : public VPHeaderPHIRecipe {
+public:
+  VPEVLBasedIVPHIRecipe(VPValue *StartIV, DebugLoc DL)
+      : VPHeaderPHIRecipe(VPDef::VPEVLBasedIVPHISC, nullptr, StartIV, DL) {}
+
+  ~VPEVLBasedIVPHIRecipe() override = default;
+
+  VPEVLBasedIVPHIRecipe *clone() override {
+    llvm_unreachable("cloning not implemented yet");
+  }
+
+  VP_CLASSOF_IMPL(VPDef::VPEVLBasedIVPHISC)
+
+  static inline bool classof(const VPHeaderPHIRecipe *D) {
+    return D->getVPDefID() == VPDef::VPEVLBasedIVPHISC;
+  }
+
+  /// Generate phi for handling IV based on EVL over iterations correctly.
+  /// TODO: investigate if it can share the code with VPCanonicalIVPHIRecipe.
+  void execute(VPTransformState &State) override;
+
+  /// Returns true if the recipe only uses the first lane of operand \p Op.
+  bool onlyFirstLaneUsed(const VPValue *Op) const override {
+    assert(is_contained(operands(), Op) &&
+           "Op must be an operand of the recipe");
+    return true;
+  }
+
+#if !defined(NDEBUG) || defined(LLVM_ENABLE_DUMP)
+  /// Print the recipe.
+  void print(raw_ostream &O, const Twine &Indent,
+             VPSlotTracker &SlotTracker) const override;
+#endif
+};
+
 /// A Recipe for widening the canonical induction variable of the vector loop.
 class VPWidenCanonicalIVRecipe : public VPSingleDefRecipe {
 public:
@@ -2522,8 +2571,8 @@ public:
   }
 };
 
-/// A recipe for converting the canonical IV value to the corresponding value of
-/// an IV with different start and step values, using Start + CanonicalIV *
+/// A recipe for converting the input value \p IV value to the corresponding
+/// value of an IV with different start and step values, using Start + IV *
 /// Step.
 class VPDerivedIVRecipe : public VPSingleDefRecipe {
   /// Kind of the induction.
@@ -2541,16 +2590,16 @@ public:
             Start, CanonicalIV, Step) {}
 
   VPDerivedIVRecipe(InductionDescriptor::InductionKind Kind,
-                    const FPMathOperator *FPBinOp, VPValue *Start,
-                    VPCanonicalIVPHIRecipe *CanonicalIV, VPValue *Step)
-      : VPSingleDefRecipe(VPDef::VPDerivedIVSC, {Start, CanonicalIV, Step}),
-        Kind(Kind), FPBinOp(FPBinOp) {}
+                    const FPMathOperator *FPBinOp, VPValue *Start, VPValue *IV,
+                    VPValue *Step)
+      : VPSingleDefRecipe(VPDef::VPDerivedIVSC, {Start, IV, Step}), Kind(Kind),
+        FPBinOp(FPBinOp) {}
 
   ~VPDerivedIVRecipe() override = default;
 
   VPRecipeBase *clone() override {
-    return new VPDerivedIVRecipe(Kind, FPBinOp, getStartValue(),
-                                 getCanonicalIV(), getStepValue());
+    return new VPDerivedIVRecipe(Kind, FPBinOp, getStartValue(), getOperand(1),
+                                 getStepValue());
   }
 
   VP_CLASSOF_IMPL(VPDef::VPDerivedIVSC)
@@ -2570,9 +2619,6 @@ public:
   }
 
   VPValue *getStartValue() const { return getOperand(0); }
-  VPCanonicalIVPHIRecipe *getCanonicalIV() const {
-    return cast<VPCanonicalIVPHIRecipe>(getOperand(1));
-  }
   VPValue *getStepValue() const { return getOperand(2); }
 
   /// Returns true if the recipe only uses the first lane of operand \p Op.
diff --git a/llvm/lib/Transforms/Vectorize/VPlanAnalysis.cpp b/llvm/lib/Transforms/Vectorize/VPlanAnalysis.cpp
index 04e30312dc23..c8ae2ee5a30f 100644
--- a/llvm/lib/Transforms/Vectorize/VPlanAnalysis.cpp
+++ b/llvm/lib/Transforms/Vectorize/VPlanAnalysis.cpp
@@ -216,14 +216,14 @@ Type *VPTypeAnalysis::inferScalarType(const VPValue *V) {
   Type *ResultTy =
       TypeSwitch<const VPRecipeBase *, Type *>(V->getDefiningRecipe())
           .Case<VPCanonicalIVPHIRecipe, VPFirstOrderRecurrencePHIRecipe,
-                VPReductionPHIRecipe, VPWidenPointerInductionRecipe>(
-              [this](const auto *R) {
-                // Handle header phi recipes, except VPWienIntOrFpInduction
-                // which needs special handling due it being possibly truncated.
-                // TODO: consider inferring/caching type of siblings, e.g.,
-                // backedge value, here and in cases below.
-                return inferScalarType(R->getStartValue());
-              })
+                VPReductionPHIRecipe, VPWidenPointerInductionRecipe,
+                VPEVLBasedIVPHIRecipe>([this](const auto *R) {
+            // Handle header phi recipes, except VPWidenIntOrFpInduction
+            // which needs special handling due it being possibly truncated.
+            // TODO: consider inferring/caching type of siblings, e.g.,
+            // backedge value, here and in cases below.
+            return inferScalarType(R->getStartValue());
+          })
           .Case<VPWidenIntOrFpInductionRecipe, VPDerivedIVRecipe>(
               [](const auto *R) { return R->getScalarType(); })
           .Case<VPPredInstPHIRecipe, VPWidenPHIRecipe, VPScalarIVStepsRecipe,
diff --git a/llvm/lib/Transforms/Vectorize/VPlanRecipes.cpp b/llvm/lib/Transforms/Vectorize/VPlanRecipes.cpp
index 124ae3108d8a..fc8323480d9a 100644
--- a/llvm/lib/Transforms/Vectorize/VPlanRecipes.cpp
+++ b/llvm/lib/Transforms/Vectorize/VPlanRecipes.cpp
@@ -286,6 +286,7 @@ bool VPInstruction::canGenerateScalarForFirstLane() const {
   case VPInstruction::CanonicalIVIncrementForPart:
   case VPInstruction::ComputeReductionResult:
   case VPInstruction::PtrAdd:
+  case VPInstruction::ExplicitVectorLength:
     return true;
   default:
     return false;
@@ -386,6 +387,33 @@ Value *VPInstruction::generatePerPart(VPTransformState &State, unsigned Part) {
     Value *Zero = ConstantInt::get(ScalarTC->getType(), 0);
     return Builder.CreateSelect(Cmp, Sub, Zero);
   }
+  case VPInstruction::ExplicitVectorLength: {
+    // Compute EVL
+    auto GetEVL = [=](VPTransformState &State, Value *AVL) {
+      assert(AVL->getType()->isIntegerTy() &&
+             "Requested vector length should be an integer.");
+
+      // TODO: Add support for MaxSafeDist for correct loop emission.
+      assert(State.VF.isScalable() && "Expected scalable vector factor.");
+      Value *VFArg = State.Builder.getInt32(State.VF.getKnownMinValue());
+
+      Value *EVL = State.Builder.CreateIntrinsic(
+          State.Builder.getInt32Ty(), Intrinsic::experimental_get_vector_length,
+          {AVL, VFArg, State.Builder.getTrue()});
+      return EVL;
+    };
+    // TODO: Restructure this code with an explicit remainder loop, vsetvli can
+    // be outside of the main loop.
+    assert(Part == 0 && "No unrolling expected for predicated vectorization.");
+    // Compute VTC - IV as the AVL (requested vector length).
+    Value *Index = State.get(getOperand(0), VPIteration(0, 0));
+    Value *TripCount = State.get(getOperand(1), VPIteration(0, 0));
+    Value *AVL = State.Builder.CreateSub(TripCount, Index);
+    Value *EVL = GetEVL(State, AVL);
+    assert(!State.EVL && "multiple EVL recipes");
+    State.EVL = this;
+    return EVL;
+  }
   case VPInstruction::CanonicalIVIncrementForPart: {
     auto *IV = State.get(getOperand(0), VPIteration(0, 0));
     if (Part == 0)
@@ -473,6 +501,8 @@ Value *VPInstruction::generatePerPart(VPTransformState &State, unsigned Part) {
     // Reduce all of the unrolled parts into a single vector.
     Value *ReducedPartRdx = RdxParts[0];
     unsigned Op = RecurrenceDescriptor::getOpcode(RK);
+    if (RecurrenceDescriptor::isAnyOfRecurrenceKind(RK))
+      Op = Instruction::Or;
 
     if (PhiR->isOrdered()) {
       ReducedPartRdx = RdxParts[State.UF - 1];
@@ -485,19 +515,16 @@ Value *VPInstruction::generatePerPart(VPTransformState &State, unsigned Part) {
         if (Op != Instruction::ICmp && Op != Instruction::FCmp)
           ReducedPartRdx = Builder.CreateBinOp(
               (Instruction::BinaryOps)Op, RdxPart, ReducedPartRdx, "bin.rdx");
-        else if (RecurrenceDescriptor::isAnyOfRecurrenceKind(RK)) {
-          TrackingVH<Value> ReductionStartValue =
-              RdxDesc.getRecurrenceStartValue();
-          ReducedPartRdx = createAnyOfOp(Builder, ReductionStartValue, RK,
-                                         ReducedPartRdx, RdxPart);
-        } else
+        else
           ReducedPartRdx = createMinMaxOp(Builder, RK, ReducedPartRdx, RdxPart);
       }
     }
 
     // Create the reduction after the loop. Note that inloop reductions create
     // the target reduction in the loop using a Reduction recipe.
-    if (State.VF.isVector() && !PhiR->isInLoop()) {
+    if ((State.VF.isVector() ||
+         RecurrenceDescriptor::isAnyOfRecurrenceKind(RK)) &&
+        !PhiR->isInLoop()) {
       ReducedPartRdx =
           createTargetReduction(Builder, RdxDesc, ReducedPartRdx, OrigPhi);
       // If the reduction can be performed in a smaller type, we need to extend
@@ -592,6 +619,7 @@ bool VPInstruction::onlyFirstLaneUsed(const VPValue *Op) const {
     // TODO: Cover additional opcodes.
     return vputils::onlyFirstLaneUsed(this);
   case VPInstruction::ActiveLaneMask:
+  case VPInstruction::ExplicitVectorLength:
   case VPInstruction::CalculateTripCountMinusVF:
   case VPInstruction::CanonicalIVIncrementForPart:
   case VPInstruction::BranchOnCount:
@@ -628,6 +656,9 @@ void VPInstruction::print(raw_ostream &O, const Twine &Indent,
   case VPInstruction::ActiveLaneMask:
     O << "active lane mask";
     break;
+  case VPInstruction::ExplicitVectorLength:
+    O << "EXPLICIT-VECTOR-LENGTH";
+    break;
   case VPInstruction::FirstOrderRecurrenceSplice:
     O << "first-order splice";
     break;
@@ -1184,7 +1215,7 @@ void VPDerivedIVRecipe::print(raw_ostream &O, const Twine &Indent,
   O << Indent << "= DERIVED-IV ";
   getStartValue()->printAsOperand(O, SlotTracker);
   O << " + ";
-  getCanonicalIV()->printAsOperand(O, SlotTracker);
+  getOperand(1)->printAsOperand(O, SlotTracker);
   O << " * ";
   getStepValue()->printAsOperand(O, SlotTracker);
 }
@@ -1974,3 +2005,25 @@ void VPActiveLaneMaskPHIRecipe::print(raw_ostream &O, const Twine &Indent,
   printOperands(O, SlotTracker);
 }
 #endif
+
+void VPEVLBasedIVPHIRecipe::execute(VPTransformState &State) {
+  BasicBlock *VectorPH = State.CFG.getPreheaderBBFor(this);
+  assert(State.UF == 1 && "Expected unroll factor 1 for VP vectorization.");
+  Value *Start = State.get(getOperand(0), VPIteration(0, 0));
+  PHINode *EntryPart =
+      State.Builder.CreatePHI(Start->getType(), 2, "evl.based.iv");
+  EntryPart->addIncoming(Start, VectorPH);
+  EntryPart->setDebugLoc(getDebugLoc());
+  State.set(this, EntryPart, 0, /*IsScalar=*/true);
+}
+
+#if !defined(NDEBUG) || defined(LLVM_ENABLE_DUMP)
+void VPEVLBasedIVPHIRecipe::print(raw_ostream &O, const Twine &Indent,
+                                  VPSlotTracker &SlotTracker) const {
+  O << Indent << "EXPLICIT-VECTOR-LENGTH-BASED-IV-PHI ";
+
+  printAsOperand(O, SlotTracker);
+  O << " = phi ";
+  printOperands(O, SlotTracker);
+}
+#endif
diff --git a/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp b/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp
index 7d4e54d09945..1256e4d8fda5 100644
--- a/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp
+++ b/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp
@@ -1209,6 +1209,45 @@ static VPActiveLaneMaskPHIRecipe *addVPLaneMaskPhiAndUpdateExitBranch(
   return LaneMaskPhi;
 }
 
+/// Replaces (ICMP_ULE, WideCanonicalIV, backedge-taken-count) pattern using
+/// the given \p Idiom.
+static void
+replaceHeaderPredicateWith(VPlan &Plan, VPValue &Idiom,
+                           function_ref<bool(VPUser &, unsigned)> Cond = {}) {
+  auto *FoundWidenCanonicalIVUser =
+      find_if(Plan.getCanonicalIV()->users(),
+              [](VPUser *U) { return isa<VPWidenCanonicalIVRecipe>(U); });
+  if (FoundWidenCanonicalIVUser == Plan.getCanonicalIV()->users().end())
+    return;
+  auto *WideCanonicalIV =
+      cast<VPWidenCanonicalIVRecipe>(*FoundWidenCanonicalIVUser);
+  // Walk users of WideCanonicalIV and replace all compares of the form
+  // (ICMP_ULE, WideCanonicalIV, backedge-taken-count) with
+  // the given idiom VPValue.
+  VPValue *BTC = Plan.getOrCreateBackedgeTakenCount();
+  for (VPUser *U : SmallVector<VPUser *>(WideCanonicalIV->users())) {
+    auto *CompareToReplace = dyn_cast<VPInstruction>(U);
+    if (!CompareToReplace ||
+        CompareToReplace->getOpcode() != Instruction::ICmp ||
+        CompareToReplace->getPredicate() != CmpInst::ICMP_ULE ||
+        CompareToReplace->getOperand(1) != BTC)
+      continue;
+
+    assert(CompareToReplace->getOperand(0) == WideCanonicalIV &&
+           "WidenCanonicalIV must be the first operand of the compare");
+    if (Cond) {
+      CompareToReplace->replaceUsesWithIf(&Idiom, Cond);
+      if (!CompareToReplace->getNumUsers())
+        CompareToReplace->eraseFromParent();
+    } else {
+      CompareToReplace->replaceAllUsesWith(&Idiom);
+      CompareToReplace->eraseFromParent();
+    }
+  }
+  if (!WideCanonicalIV->getNumUsers())
+    WideCanonicalIV->eraseFromParent();
+}
+
 void VPlanTransforms::addActiveLaneMask(
     VPlan &Plan, bool UseActiveLaneMaskForControlFlow,
     bool DataAndControlFlowWithoutRuntimeCheck) {
@@ -1238,20 +1277,77 @@ void VPlanTransforms::addActiveLaneMask(
   // Walk users of WideCanonicalIV and replace all compares of the form
   // (ICMP_ULE, WideCanonicalIV, backedge-taken-count) with an
   // active-lane-mask.
-  VPValue *BTC = Plan.getOrCreateBackedgeTakenCount();
-  for (VPUser *U : SmallVector<VPUser *>(WideCanonicalIV->users())) {
-    auto *CompareToReplace = dyn_cast<VPInstruction>(U);
-    if (!CompareToReplace ||
-        CompareToReplace->getOpcode() != Instruction::ICmp ||
-        CompareToReplace->getPredicate() != CmpInst::ICMP_ULE ||
-        CompareToReplace->getOperand(1) != BTC)
-      continue;
+  replaceHeaderPredicateWith(Plan, *LaneMask);
+}
 
-    assert(CompareToReplace->getOperand(0) == WideCanonicalIV &&
-           "WidenCanonicalIV must be the first operand of the compare");
-    CompareToReplace->replaceAllUsesWith(LaneMask);
-    CompareToReplace->eraseFromParent();
+/// Add a VPEVLBasedIVPHIRecipe and related recipes to \p Plan and
+/// replaces all uses except the canonical IV increment of
+/// VPCanonicalIVPHIRecipe with a VPEVLBasedIVPHIRecipe. VPCanonicalIVPHIRecipe
+/// is used only for loop iterations counting after this transformation.
+///
+/// The function uses the following definitions:
+///  %StartV is the canonical induction start value.
+///
+/// The function adds the following recipes:
+///
+/// vector.ph:
+/// ...
+///
+/// vector.body:
+/// ...
+/// %EVLPhi = EXPLICIT-VECTOR-LENGTH-BASED-IV-PHI [ %StartV, %vector.ph ],
+///                                               [ %NextEVLIV, %vector.body ]
+/// %VPEVL = EXPLICIT-VECTOR-LENGTH %EVLPhi, original TC
+/// ...
+/// %NextEVLIV = add IVSize (cast i32 %VPEVVL to IVSize), %EVLPhi
+/// ...
+///
+void VPlanTransforms::addExplicitVectorLength(VPlan &Plan) {
+  VPBasicBlock *Header = Plan.getVectorLoopRegion()->getEntryBasicBlock();
+  auto *CanonicalIVPHI = Plan.getCanonicalIV();
+  VPValue *StartV = CanonicalIVPHI->getStartValue();
+
+  // TODO: revisit this and try to remove the mask operand.
+  // Walk VPWidenMemoryInstructionRecipe users of WideCanonicalIV and replace
+  // all compares of the form (ICMP_ULE, WideCanonicalIV, backedge-taken-count),
+  // used as mask in VPWidenMemoryInstructionRecipe, with an all-true-mask.
+  Value *TrueMask =
+      ConstantInt::getTrue(CanonicalIVPHI->getScalarType()->getContext());
+  VPValue *VPTrueMask = Plan.getOrAddLiveIn(TrueMask);
+  replaceHeaderPredicateWith(Plan, *VPTrueMask, [](VPUser &U, unsigned) {
+    return isa<VPWidenMemoryInstructionRecipe>(U);
+  });
+  // Now create the ExplicitVectorLengthPhi recipe in the main loop.
+  auto *EVLPhi = new VPEVLBasedIVPHIRecipe(StartV, DebugLoc());
+  EVLPhi->insertAfter(CanonicalIVPHI);
+  auto *VPEVL = new VPInstruction(VPInstruction::ExplicitVectorLength,
+                                  {EVLPhi, Plan.getTripCount()});
+  VPEVL->insertBefore(*Header, Header->getFirstNonPhi());
+
+  auto *CanonicalIVIncrement =
+      cast<VPInstruction>(CanonicalIVPHI->getBackedgeValue());
+  VPSingleDefRecipe *OpVPEVL = VPEVL;
+  if (unsigned IVSize = CanonicalIVPHI->getScalarType()->getScalarSizeInBits();
+      IVSize != 32) {
+    OpVPEVL = new VPScalarCastRecipe(IVSize < 32 ? Instruction::Trunc
+                                                 : Instruction::ZExt,
+                                     OpVPEVL, CanonicalIVPHI->getScalarType());
+    OpVPEVL->insertBefore(CanonicalIVIncrement);
   }
+  auto *NextEVLIV =
+      new VPInstruction(Instruction::Add, {OpVPEVL, EVLPhi},
+                        {CanonicalIVIncrement->hasNoUnsignedWrap(),
+                         CanonicalIVIncrement->hasNoSignedWrap()},
+                        CanonicalIVIncrement->getDebugLoc(), "index.evl.next");
+  NextEVLIV->insertBefore(CanonicalIVIncrement);
+  EVLPhi->addOperand(NextEVLIV);
+
+  // Replace all uses of VPCanonicalIVPHIRecipe by
+  // VPEVLBasedIVPHIRecipe except for the canonical IV increment.
+  CanonicalIVPHI->replaceAllUsesWith(EVLPhi);
+  CanonicalIVIncrement->setOperand(0, CanonicalIVPHI);
+  // TODO: support unroll factor > 1.
+  Plan.setUF(1);
 }
 
 void VPlanTransforms::dropPoisonGeneratingRecipes(
@@ -1277,9 +1373,7 @@ void VPlanTransforms::dropPoisonGeneratingRecipes(
       // handled.
       if (isa<VPWidenMemoryInstructionRecipe>(CurRec) ||
           isa<VPInterleaveRecipe>(CurRec) ||
-          isa<VPScalarIVStepsRecipe>(CurRec) ||
-          isa<VPCanonicalIVPHIRecipe>(CurRec) ||
-          isa<VPActiveLaneMaskPHIRecipe>(CurRec))
+          isa<VPScalarIVStepsRecipe>(CurRec) || isa<VPHeaderPHIRecipe>(CurRec))
         continue;
 
       // This recipe contributes to the address computation of a widen
diff --git a/llvm/lib/Transforms/Vectorize/VPlanTransforms.h b/llvm/lib/Transforms/Vectorize/VPlanTransforms.h
index ff83c3f083b0..0cbc70713d9c 100644
--- a/llvm/lib/Transforms/Vectorize/VPlanTransforms.h
+++ b/llvm/lib/Transforms/Vectorize/VPlanTransforms.h
@@ -98,6 +98,13 @@ struct VPlanTransforms {
   ///       VPlan directly.
   static void dropPoisonGeneratingRecipes(
       VPlan &Plan, function_ref<bool(BasicBlock *)> BlockNeedsPredication);
+
+  /// Add a VPEVLBasedIVPHIRecipe and related recipes to \p Plan and
+  /// replaces all uses except the canonical IV increment of
+  /// VPCanonicalIVPHIRecipe with a VPEVLBasedIVPHIRecipe.
+  /// VPCanonicalIVPHIRecipe is only used to control the loop after
+  /// this transformation.
+  static void addExplicitVectorLength(VPlan &Plan);
 };
 
 } // namespace llvm
diff --git a/llvm/lib/Transforms/Vectorize/VPlanValue.h b/llvm/lib/Transforms/Vectorize/VPlanValue.h
index 1d2c17e91b7a..8b221d30e525 100644
--- a/llvm/lib/Transforms/Vectorize/VPlanValue.h
+++ b/llvm/lib/Transforms/Vectorize/VPlanValue.h
@@ -368,6 +368,7 @@ public:
     // VPHeaderPHIRecipe need to be kept together.
     VPCanonicalIVPHISC,
     VPActiveLaneMaskPHISC,
+    VPEVLBasedIVPHISC,
     VPFirstOrderRecurrencePHISC,
     VPWidenIntOrFpInductionSC,
     VPWidenPointerInductionSC,
diff --git a/llvm/lib/Transforms/Vectorize/VPlanVerifier.cpp b/llvm/lib/Transforms/Vectorize/VPlanVerifier.cpp
index 7ebdb914fb85..12d37fa711db 100644
--- a/llvm/lib/Transforms/Vectorize/VPlanVerifier.cpp
+++ b/llvm/lib/Transforms/Vectorize/VPlanVerifier.cpp
@@ -92,7 +92,50 @@ static bool verifyVPBasicBlock(const VPBasicBlock *VPBB,
   for (const VPRecipeBase &R : *VPBB)
     RecipeNumbering[&R] = Cnt++;
 
+  // Set of recipe types along with VPInstruction Opcodes of all EVL-related
+  // recipes that must appear at most once in the header block.
+  DenseSet<unsigned> EVLFound;
+  const VPRecipeBase *VPWidenMemRecipe = nullptr;
+  const VPlan *Plan = VPBB->getPlan();
+  bool IsHeader = Plan->getEntry()->getNumSuccessors() == 1 &&
+                  Plan->getVectorLoopRegion()->getEntry() == VPBB;
+  auto CheckEVLRecipiesInsts = [&](const VPRecipeBase *R) {
+    if (isa<VPEVLBasedIVPHIRecipe>(R)) {
+      if (!IsHeader) {
+        errs() << "EVL PHI recipe not in entry block!\n";
+        return false;
+      }
+      if (!EVLFound.insert(VPDef::VPEVLBasedIVPHISC).second) {
+        errs() << "EVL PHI recipe inserted more than once!\n";
+        return false;
+      }
+      return true;
+    }
+    if (const auto *RInst = dyn_cast<VPInstruction>(R);
+        RInst && RInst->getOpcode() == VPInstruction::ExplicitVectorLength) {
+      if (!IsHeader) {
+        errs() << "EVL instruction not in the header block!\n";
+        return false;
+      }
+      if (!EVLFound.insert(RInst->getOpcode() + VPDef::VPLastPHISC).second) {
+        errs() << "EVL instruction inserted more than once!\n";
+        return false;
+      }
+      if (VPWidenMemRecipe) {
+        errs() << "Use of EVL instruction by widen memory recipe before "
+                  "definition!\n";
+        return false;
+      }
+      return true;
+    }
+    if (isa<VPWidenMemoryInstructionRecipe>(R))
+      VPWidenMemRecipe = R;
+    return true;
+  };
+
   for (const VPRecipeBase &R : *VPBB) {
+    if (!CheckEVLRecipiesInsts(&R))
+      return false;
     for (const VPValue *V : R.definedValues()) {
       for (const VPUser *U : V->users()) {
         auto *UI = dyn_cast<VPRecipeBase>(U);
diff --git a/llvm/test/Analysis/CostModel/X86/masked-intrinsic-codesize.ll b/llvm/test/Analysis/CostModel/X86/masked-intrinsic-codesize.ll
index 55fdaafc24ed..d97c3d550efd 100644
--- a/llvm/test/Analysis/CostModel/X86/masked-intrinsic-codesize.ll
+++ b/llvm/test/Analysis/CostModel/X86/masked-intrinsic-codesize.ll
@@ -8,1454 +8,1508 @@
 ; RUN: opt < %s -S -mtriple=x86_64-apple-darwin -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=code-size -mcpu=knl      | FileCheck %s --check-prefixes=AVX512,KNL
 ; RUN: opt < %s -S -mtriple=x86_64-apple-darwin -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=code-size -mcpu=skx      | FileCheck %s --check-prefixes=AVX512,SKX
 
-define i32 @masked_load() {
+define i32 @masked_load(<1 x i1> %m1, <2 x i1> %m2, <3 x i1> %m3, <4 x i1> %m4, <5 x i1> %m5, <6 x i1> %m6, <7 x i1> %m7, <8 x i1> %m8, <9 x i1> %m9, <10 x i1> %m10, <11 x i1> %m11, <12 x i1> %m12, <13 x i1> %m13, <14 x i1> %m14, <15 x i1> %m15, <16 x i1> %m16, <32 x i1> %m32, <64 x i1> %m64) {
 ; SSE2-LABEL: 'masked_load'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 91 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 107 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 101 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 95 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 89 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 83 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 77 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 53 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 190 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 95 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 440 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 220 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 110 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 91 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 107 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 101 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 95 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 89 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 83 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 77 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 53 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 190 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 95 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 440 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 220 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 110 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; SSE42-LABEL: 'masked_load'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 75 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 70 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 55 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 320 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 75 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 70 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 55 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 320 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; AVX-LABEL: 'masked_load'
-; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 163 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 324 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i64> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i64> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i64> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i64> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 163 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 324 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; KNL-LABEL: 'masked_load'
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 164 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 326 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 164 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 326 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; SKX-LABEL: 'masked_load'
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
-  %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x double> undef)
-  %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x double> undef)
-  %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x double> undef)
-  %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x double> undef)
-  %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x double> undef)
-  %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x double> undef)
-  %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x double> undef)
-  %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x double> undef)
-
-  %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x float> undef)
-  %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x float> undef)
-  %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x float> undef)
-  %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x float> undef)
-  %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x float> undef)
-  %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x float> undef)
-  %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x float> undef)
-  %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x float> undef)
-  %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x float> undef)
-  %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x float> undef)
-  %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x float> undef)
-  %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x float> undef)
-  %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x float> undef)
-  %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x float> undef)
-  %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x float> undef)
-  %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x float> undef)
-
-  %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-  %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i64> undef)
-  %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i64> undef)
-  %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i64> undef)
-  %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-  %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i64> undef)
-  %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-  %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-
-  %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-  %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x i32> undef)
-  %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x i32> undef)
-  %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x i32> undef)
-  %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x i32> undef)
-  %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x i32> undef)
-  %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x i32> undef)
-  %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x i32> undef)
-  %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-  %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i32> undef)
-  %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i32> undef)
-  %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i32> undef)
-  %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-  %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i32> undef)
-  %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-  %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i32> undef)
-
-  %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-  %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-  %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-  %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-
-  %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-  %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-  %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-  %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+  %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+  %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x double> undef)
+  %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x double> undef)
+  %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x double> undef)
+  %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+  %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x double> undef)
+  %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+  %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+
+  %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+  %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x float> undef)
+  %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x float> undef)
+  %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x float> undef)
+  %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x float> undef)
+  %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x float> undef)
+  %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x float> undef)
+  %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x float> undef)
+  %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+  %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x float> undef)
+  %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x float> undef)
+  %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x float> undef)
+  %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+  %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x float> undef)
+  %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+  %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x float> undef)
+
+  %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+  %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i64> undef)
+  %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i64> undef)
+  %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i64> undef)
+  %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+  %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i64> undef)
+  %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+  %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+
+  %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+  %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x i32> undef)
+  %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x i32> undef)
+  %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x i32> undef)
+  %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x i32> undef)
+  %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x i32> undef)
+  %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x i32> undef)
+  %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x i32> undef)
+  %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+  %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i32> undef)
+  %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i32> undef)
+  %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i32> undef)
+  %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+  %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i32> undef)
+  %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+  %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i32> undef)
+
+  %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+  %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+  %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+  %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+
+  %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+  %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+  %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+  %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 
   ret i32 0
 }
 
-define i32 @masked_store() {
+define i32 @masked_store(<1 x i1> %m1, <2 x i1> %m2, <3 x i1> %m3, <4 x i1> %m4, <5 x i1> %m5, <6 x i1> %m6, <7 x i1> %m7, <8 x i1> %m8, <9 x i1> %m9, <10 x i1> %m10, <11 x i1> %m11, <12 x i1> %m12, <13 x i1> %m13, <14 x i1> %m14, <15 x i1> %m15, <16 x i1> %m16, <32 x i1> %m32, <64 x i1> %m64) {
 ; SSE2-LABEL: 'masked_store'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 91 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 85 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 79 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 73 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 62 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 107 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 93 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 73 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 59 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 53 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 190 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 95 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 440 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 220 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 110 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 91 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 85 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> %m15)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 79 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> %m14)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 73 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> %m13)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> %m12)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 62 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> %m11)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> %m10)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> %m9)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 107 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> %m15)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 93 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> %m14)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> %m13)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> %m12)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 73 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> %m11)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> %m10)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 59 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> %m9)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 53 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 190 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> %m32)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 95 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 440 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> %m64)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 220 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> %m32)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 110 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> %m8)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; SSE42-LABEL: 'masked_store'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 75 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 70 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 55 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 320 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> %m15)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> %m14)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> %m13)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> %m12)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> %m11)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> %m10)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> %m9)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 75 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> %m15)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 70 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> %m14)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> %m13)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> %m12)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 55 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> %m11)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> %m10)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> %m9)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> %m32)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 320 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> %m64)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> %m32)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> %m8)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; AVX-LABEL: 'masked_store'
-; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 163 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 324 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> %m8)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> %m7)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> %m6)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> %m5)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> %m4)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> %m3)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> %m2)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> %m1)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> %m16)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> %m15)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> %m14)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> %m13)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> %m12)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> %m11)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> %m10)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> %m9)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> %m8)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> %m7)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> %m6)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> %m5)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> %m4)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> %m3)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> %m2)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> %m1)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> %m8)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> %m7)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> %m6)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> %m5)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> %m4)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> %m3)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> %m2)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> %m1)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> %m16)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> %m15)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> %m14)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> %m13)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> %m12)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> %m11)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> %m10)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> %m9)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> %m8)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> %m7)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> %m6)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> %m5)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> %m4)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> %m3)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> %m2)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> %m1)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 163 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> %m32)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> %m16)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> %m8)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> %m4)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 324 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> %m64)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> %m32)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> %m16)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> %m8)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; KNL-LABEL: 'masked_store'
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 164 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 326 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> %m7)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> %m6)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> %m5)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> %m3)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> %m2)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> %m1)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> %m16)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> %m15)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> %m14)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> %m13)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> %m12)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> %m11)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> %m10)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> %m9)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> %m7)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> %m6)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> %m5)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> %m3)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> %m2)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> %m1)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> %m7)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> %m6)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> %m5)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> %m3)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> %m2)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> %m1)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> %m16)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> %m15)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> %m14)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> %m13)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> %m12)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> %m11)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> %m10)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> %m9)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> %m7)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> %m6)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> %m5)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> %m3)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> %m2)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> %m1)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 164 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> %m32)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> %m16)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 326 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> %m64)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> %m32)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> %m16)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> %m8)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; SKX-LABEL: 'masked_store'
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> %m15)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> %m14)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> %m13)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> %m12)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> %m11)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> %m10)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> %m9)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> %m15)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> %m14)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> %m13)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> %m12)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> %m11)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> %m10)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> %m9)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> %m32)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> %m64)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> %m32)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> %m8)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
-  call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> undef)
-  call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> undef)
-  call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> undef)
-  call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> undef)
-  call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> undef)
-  call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> undef)
-  call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> undef)
-
-  call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> undef)
-  call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> undef)
-  call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> undef)
-  call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> undef)
-  call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> undef)
-  call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> undef)
-  call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> undef)
-  call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> undef)
-  call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> undef)
-  call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> undef)
-  call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> undef)
-  call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> undef)
-  call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> undef)
-  call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> undef)
-  call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> undef)
-
-  call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> undef)
-  call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> undef)
-  call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> undef)
-  call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> undef)
-  call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> undef)
-  call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> undef)
-  call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> undef)
-
-  call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> undef)
-  call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> undef)
-  call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> undef)
-  call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> undef)
-  call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> undef)
-  call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> undef)
-  call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> undef)
-  call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> undef)
-  call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> undef)
-  call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> undef)
-  call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> undef)
-  call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> undef)
-  call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> undef)
-  call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> undef)
-  call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> undef)
-
-  call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> undef)
-  call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> undef)
-  call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> undef)
-
-  call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> undef)
-  call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> undef)
-  call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> undef)
-  call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> undef)
+  call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> %m7)
+  call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> %m6)
+  call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> %m5)
+  call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> %m4)
+  call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> %m3)
+  call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> %m2)
+  call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> %m1)
+
+  call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> %m16)
+  call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> %m15)
+  call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> %m14)
+  call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> %m13)
+  call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> %m12)
+  call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> %m11)
+  call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> %m10)
+  call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> %m9)
+  call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> %m7)
+  call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> %m6)
+  call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> %m5)
+  call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> %m4)
+  call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> %m3)
+  call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> %m2)
+  call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> %m1)
+
+  call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> %m7)
+  call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> %m6)
+  call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> %m5)
+  call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> %m4)
+  call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> %m3)
+  call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> %m2)
+  call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> %m1)
+
+  call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> %m16)
+  call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> %m15)
+  call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> %m14)
+  call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> %m13)
+  call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> %m12)
+  call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> %m11)
+  call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> %m10)
+  call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> %m9)
+  call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> %m7)
+  call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> %m6)
+  call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> %m5)
+  call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> %m4)
+  call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> %m3)
+  call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> %m2)
+  call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> %m1)
+
+  call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> %m32)
+  call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> %m16)
+  call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> %m4)
+
+  call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> %m64)
+  call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> %m32)
+  call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> %m16)
+  call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> %m8)
 
   ret i32 0
 }
 
-define i32 @masked_gather() {
+define i32 @masked_gather(<1 x i1> %m1, <2 x i1> %m2, <4 x i1> %m4, <8 x i1> %m8, <16 x i1> %m16, <32 x i1> %m32, <64 x i1> %m64) {
 ; SSE2-LABEL: 'masked_gather'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 252 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 126 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 63 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 92 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 380 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 190 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 95 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; SSE42-LABEL: 'masked_gather'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 192 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 320 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; AVX1-LABEL: 'masked_gather'
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 98 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 194 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 97 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 78 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 82 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 322 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 161 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; AVX2-LABEL: 'masked_gather'
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 98 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 194 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 97 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 78 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 82 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 322 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 161 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; SKL-LABEL: 'masked_gather'
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 98 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 194 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 97 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 322 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 161 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; KNL-LABEL: 'masked_gather'
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 99 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 195 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 97 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 163 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 323 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 161 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; SKX-LABEL: 'masked_gather'
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 99 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 195 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 97 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 163 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 323 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 161 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
-  %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x double> undef)
-  %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x double> undef)
-  %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x double> undef)
-  %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x double> undef)
+  %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+  %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+  %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+  %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x double> undef)
 
-  %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x float> undef)
-  %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x float> undef)
-  %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x float> undef)
-  %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x float> undef)
+  %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+  %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+  %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+  %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x float> undef)
 
-  %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-  %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-  %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-  %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
+  %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+  %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+  %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+  %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
 
-  %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-  %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-  %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-  %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
+  %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+  %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+  %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+  %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
 
-  %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-  %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-  %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-  %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
+  %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+  %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+  %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+  %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
 
-  %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-  %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-  %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-  %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+  %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+  %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+  %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+  %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 
   ret i32 0
 }
 
-define i32 @masked_scatter() {
+define i32 @masked_scatter(<1 x i1> %m1, <2 x i1> %m2, <4 x i1> %m4, <8 x i1> %m8, <16 x i1> %m16, <32 x i1> %m32, <64 x i1> %m64) {
 ; SSE2-LABEL: 'masked_scatter'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 252 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 126 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 63 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 92 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 380 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> %m64)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 190 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 95 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; SSE42-LABEL: 'masked_scatter'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 192 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 320 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> %m64)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; AVX-LABEL: 'masked_scatter'
-; AVX-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 98 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 194 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 97 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 78 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 82 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 322 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> %m64)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 161 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; KNL-LABEL: 'masked_scatter'
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 99 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 195 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 97 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 163 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 323 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> %m64)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 161 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; SKX-LABEL: 'masked_scatter'
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 99 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 195 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 97 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 163 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 323 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> %m64)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 161 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
-  call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-  call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-  call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
+  call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+  call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+  call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
 
-  call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-  call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-  call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
+  call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+  call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+  call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
 
-  call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-  call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-  call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
+  call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+  call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+  call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
 
-  call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-  call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-  call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
+  call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+  call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+  call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
 
-  call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-  call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-  call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
+  call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+  call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+  call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
 
-  call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> undef)
-  call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-  call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-  call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
+  call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> %m64)
+  call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+  call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+  call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
 
   ret i32 0
 }
 
-define i32 @masked_expandload() {
+define i32 @masked_expandload(<1 x i1> %m1, <2 x i1> %m2, <4 x i1> %m4, <8 x i1> %m8, <16 x i1> %m16, <32 x i1> %m32, <64 x i1> %m64) {
 ; SSE2-LABEL: 'masked_expandload'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> undef, <8 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> undef, <4 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> undef, <2 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> undef, <1 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> undef, <16 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> undef, <8 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> undef, <4 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> undef, <2 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> undef, <8 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> undef, <4 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> undef, <2 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> undef, <1 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> undef, <16 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> undef, <8 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> undef, <4 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> undef, <2 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> undef, <32 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> undef, <16 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> undef, <8 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> undef, <4 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 188 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> undef, <64 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 94 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> undef, <32 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> undef, <16 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> undef, <8 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> %m8, <8 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> %m4, <4 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> %m2, <2 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> %m1, <1 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> %m16, <16 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> %m8, <8 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> %m4, <4 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> %m2, <2 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> %m8, <8 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> %m4, <4 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> %m2, <2 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> %m1, <1 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> %m16, <16 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> %m8, <8 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> %m4, <4 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> %m2, <2 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> %m32, <32 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> %m16, <16 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> %m8, <8 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> %m4, <4 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 192 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> %m64, <64 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> %m32, <32 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> %m16, <16 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> %m8, <8 x i8> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; SSE42-LABEL: 'masked_expandload'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> undef, <8 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> undef, <4 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> undef, <2 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> undef, <1 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> undef, <16 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> undef, <8 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> undef, <4 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> undef, <2 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> undef, <8 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> undef, <4 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> undef, <2 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> undef, <1 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> undef, <16 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> undef, <8 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> undef, <4 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> undef, <2 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> undef, <32 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> undef, <16 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> undef, <8 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> undef, <4 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> undef, <64 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> undef, <32 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> undef, <16 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> undef, <8 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> %m8, <8 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> %m4, <4 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> %m2, <2 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> %m1, <1 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> %m16, <16 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> %m8, <8 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> %m4, <4 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> %m2, <2 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> %m8, <8 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> %m4, <4 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> %m2, <2 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> %m1, <1 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> %m16, <16 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> %m8, <8 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> %m4, <4 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> %m2, <2 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> %m32, <32 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> %m16, <16 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> %m8, <8 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> %m4, <4 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> %m64, <64 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> %m32, <32 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> %m16, <16 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> %m8, <8 x i8> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
-; AVX-LABEL: 'masked_expandload'
-; AVX-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> undef, <8 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> undef, <4 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> undef, <2 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> undef, <1 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> undef, <16 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> undef, <8 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> undef, <4 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> undef, <2 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> undef, <8 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> undef, <4 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> undef, <2 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> undef, <1 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> undef, <16 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> undef, <8 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> undef, <4 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> undef, <2 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> undef, <32 x i16> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> undef, <16 x i16> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> undef, <8 x i16> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> undef, <4 x i16> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 130 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> undef, <64 x i8> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> undef, <32 x i8> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> undef, <16 x i8> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> undef, <8 x i8> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
+; AVX1-LABEL: 'masked_expandload'
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> %m8, <8 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> %m4, <4 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> %m2, <2 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> %m1, <1 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> %m16, <16 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> %m8, <8 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> %m4, <4 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> %m2, <2 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> %m8, <8 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> %m4, <4 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> %m2, <2 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> %m1, <1 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> %m16, <16 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> %m8, <8 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> %m4, <4 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> %m2, <2 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> %m32, <32 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> %m16, <16 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> %m8, <8 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> %m4, <4 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 134 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> %m64, <64 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> %m32, <32 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> %m16, <16 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> %m8, <8 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
+;
+; AVX2-LABEL: 'masked_expandload'
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> %m8, <8 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> %m4, <4 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> %m2, <2 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> %m1, <1 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> %m16, <16 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> %m8, <8 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> %m4, <4 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> %m2, <2 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> %m8, <8 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> %m4, <4 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> %m2, <2 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> %m1, <1 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> %m16, <16 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> %m8, <8 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> %m4, <4 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> %m2, <2 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> %m32, <32 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> %m16, <16 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> %m8, <8 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> %m4, <4 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> %m64, <64 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> %m32, <32 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> %m16, <16 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> %m8, <8 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
+;
+; SKL-LABEL: 'masked_expandload'
+; SKL-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> %m8, <8 x double> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> %m4, <4 x double> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> %m2, <2 x double> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> %m1, <1 x double> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> %m16, <16 x float> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> %m8, <8 x float> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> %m4, <4 x float> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> %m2, <2 x float> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> %m8, <8 x i64> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> %m4, <4 x i64> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> %m2, <2 x i64> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> %m1, <1 x i64> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> %m16, <16 x i32> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> %m8, <8 x i32> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> %m4, <4 x i32> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> %m2, <2 x i32> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> %m32, <32 x i16> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> %m16, <16 x i16> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> %m8, <8 x i16> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> %m4, <4 x i16> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> %m64, <64 x i8> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> %m32, <32 x i8> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> %m16, <16 x i8> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> %m8, <8 x i8> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; AVX512-LABEL: 'masked_expandload'
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> undef, <8 x double> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> undef, <4 x double> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> undef, <2 x double> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> undef, <1 x double> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> undef, <16 x float> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> undef, <8 x float> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> undef, <4 x float> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> undef, <2 x float> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> undef, <8 x i64> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> undef, <4 x i64> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> undef, <2 x i64> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> undef, <1 x i64> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> undef, <16 x i32> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> undef, <8 x i32> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> undef, <4 x i32> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> undef, <2 x i32> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> undef, <32 x i16> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> undef, <16 x i16> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> undef, <8 x i16> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> undef, <4 x i16> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 131 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> undef, <64 x i8> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> undef, <32 x i8> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> undef, <16 x i8> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> undef, <8 x i8> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> %m8, <8 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> %m4, <4 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> %m2, <2 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> %m1, <1 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> %m16, <16 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> %m8, <8 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> %m4, <4 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> %m2, <2 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> %m8, <8 x i64> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> %m4, <4 x i64> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> %m2, <2 x i64> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> %m1, <1 x i64> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> %m16, <16 x i32> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> %m8, <8 x i32> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> %m4, <4 x i32> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> %m2, <2 x i32> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 99 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> %m32, <32 x i16> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> %m16, <16 x i16> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> %m8, <8 x i16> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> %m4, <4 x i16> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 195 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> %m64, <64 x i8> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 97 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> %m32, <32 x i8> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> %m16, <16 x i8> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> %m8, <8 x i8> undef)
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
-  %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> undef, <8 x double> undef)
-  %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> undef, <4 x double> undef)
-  %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> undef, <2 x double> undef)
-  %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> undef, <1 x double> undef)
+  %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> %m8, <8 x double> undef)
+  %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> %m4, <4 x double> undef)
+  %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> %m2, <2 x double> undef)
+  %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> %m1, <1 x double> undef)
 
-  %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> undef, <16 x float> undef)
-  %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> undef, <8 x float> undef)
-  %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> undef, <4 x float> undef)
-  %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> undef, <2 x float> undef)
+  %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> %m16, <16 x float> undef)
+  %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> %m8, <8 x float> undef)
+  %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> %m4, <4 x float> undef)
+  %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> %m2, <2 x float> undef)
 
-  %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> undef, <8 x i64> undef)
-  %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> undef, <4 x i64> undef)
-  %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> undef, <2 x i64> undef)
-  %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> undef, <1 x i64> undef)
+  %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> %m8, <8 x i64> undef)
+  %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> %m4, <4 x i64> undef)
+  %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> %m2, <2 x i64> undef)
+  %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> %m1, <1 x i64> undef)
 
-  %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> undef, <16 x i32> undef)
-  %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> undef, <8 x i32> undef)
-  %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> undef, <4 x i32> undef)
-  %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> undef, <2 x i32> undef)
+  %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> %m16, <16 x i32> undef)
+  %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> %m8, <8 x i32> undef)
+  %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> %m4, <4 x i32> undef)
+  %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> %m2, <2 x i32> undef)
 
-  %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> undef, <32 x i16> undef)
-  %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> undef, <16 x i16> undef)
-  %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> undef, <8 x i16> undef)
-  %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> undef, <4 x i16> undef)
+  %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> %m32, <32 x i16> undef)
+  %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> %m16, <16 x i16> undef)
+  %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> %m8, <8 x i16> undef)
+  %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> %m4, <4 x i16> undef)
 
-  %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> undef, <64 x i8> undef)
-  %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> undef, <32 x i8> undef)
-  %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> undef, <16 x i8> undef)
-  %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> undef, <8 x i8> undef)
+  %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> %m64, <64 x i8> undef)
+  %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> %m32, <32 x i8> undef)
+  %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> %m16, <16 x i8> undef)
+  %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> %m8, <8 x i8> undef)
 
   ret i32 0
 }
 
-define i32 @masked_compressstore() {
+define i32 @masked_compressstore(<1 x i1> %m1, <2 x i1> %m2, <4 x i1> %m4, <8 x i1> %m8, <16 x i1> %m16, <32 x i1> %m32, <64 x i1> %m64) {
 ; SSE2-LABEL: 'masked_compressstore'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 192 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> %m1)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> %m1)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> %m32)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 192 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> %m64)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> %m32)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> %m8)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; SSE42-LABEL: 'masked_compressstore'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> %m1)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> %m1)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> %m32)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> %m64)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> %m32)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> %m8)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; AVX1-LABEL: 'masked_compressstore'
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 134 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> %m8)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> %m4)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> %m2)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> %m1)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> %m16)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> %m8)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> %m4)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> %m2)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> %m8)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> %m4)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> %m2)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> %m1)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> %m16)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> %m8)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> %m4)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> %m2)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> %m32)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> %m16)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> %m8)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> %m4)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 134 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> %m64)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> %m32)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> %m16)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> %m8)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; AVX2-LABEL: 'masked_compressstore'
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> %m8)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> %m4)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> %m2)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> %m1)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> %m16)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> %m8)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> %m4)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> %m2)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> %m8)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> %m4)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> %m2)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> %m1)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> %m16)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> %m8)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> %m4)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> %m2)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> %m32)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> %m16)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> %m8)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> %m4)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> %m64)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> %m32)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> %m16)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> %m8)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; SKL-LABEL: 'masked_compressstore'
-; SKL-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> %m8)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> %m4)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> %m2)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> %m1)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> %m16)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> %m8)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> %m4)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> %m2)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> %m8)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> %m4)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> %m2)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> %m1)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> %m16)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> %m8)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> %m4)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> %m2)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> %m32)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> %m16)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> %m8)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> %m4)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> %m64)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> %m32)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> %m16)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> %m8)
 ; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; AVX512-LABEL: 'masked_compressstore'
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 99 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 195 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 97 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> %m8)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> %m4)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> %m2)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> %m1)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> %m16)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> %m8)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> %m4)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> %m2)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> %m8)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> %m4)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> %m2)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> %m1)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> %m16)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> %m8)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> %m4)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> %m2)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 99 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> %m32)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> %m16)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> %m8)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> %m4)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 195 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> %m64)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 97 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> %m32)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> %m16)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> %m8)
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
-  call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> undef)
-  call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> undef)
-  call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> undef)
-  call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> undef)
-
-  call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> undef)
-  call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> undef)
-  call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> undef)
-  call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> undef)
-
-  call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> undef)
-  call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> undef)
-  call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> undef)
-  call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> undef)
-
-  call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> undef)
-  call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> undef)
-  call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> undef)
-  call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> undef)
-
-  call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> undef)
-  call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> undef)
-  call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> undef)
-  call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> undef)
-
-  call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> undef)
-  call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> undef)
-  call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> undef)
-  call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> undef)
+  call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> %m8)
+  call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> %m4)
+  call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> %m2)
+  call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> %m1)
+
+  call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> %m16)
+  call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> %m8)
+  call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> %m4)
+  call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> %m2)
+
+  call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> %m8)
+  call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> %m4)
+  call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> %m2)
+  call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> %m1)
+
+  call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> %m16)
+  call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> %m8)
+  call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> %m4)
+  call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> %m2)
+
+  call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> %m32)
+  call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> %m16)
+  call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> %m8)
+  call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> %m4)
+
+  call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> %m64)
+  call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> %m32)
+  call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> %m16)
+  call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> %m8)
 
   ret i32 0
 }
diff --git a/llvm/test/Analysis/CostModel/X86/masked-intrinsic-cost-inseltpoison.ll b/llvm/test/Analysis/CostModel/X86/masked-intrinsic-cost-inseltpoison.ll
index ad56c28510f5..b0f9f6d32a56 100644
--- a/llvm/test/Analysis/CostModel/X86/masked-intrinsic-cost-inseltpoison.ll
+++ b/llvm/test/Analysis/CostModel/X86/masked-intrinsic-cost-inseltpoison.ll
@@ -8,1454 +8,1562 @@
 ; RUN: opt < %s -S -mtriple=x86_64-apple-darwin -passes="print<cost-model>" 2>&1 -disable-output -mcpu=knl      | FileCheck %s --check-prefixes=AVX512,KNL
 ; RUN: opt < %s -S -mtriple=x86_64-apple-darwin -passes="print<cost-model>" 2>&1 -disable-output -mcpu=skx      | FileCheck %s --check-prefixes=AVX512,SKX
 
-define i32 @masked_load() {
+define i32 @masked_load(<1 x i1> %m1, <2 x i1> %m2, <3 x i1> %m3, <4 x i1> %m4, <5 x i1> %m5, <6 x i1> %m6, <7 x i1> %m7, <8 x i1> %m8, <9 x i1> %m9, <10 x i1> %m10, <11 x i1> %m11, <12 x i1> %m12, <13 x i1> %m13, <14 x i1> %m14, <15 x i1> %m15, <16 x i1> %m16, <32 x i1> %m32, <64 x i1> %m64) {
 ; SSE2-LABEL: 'masked_load'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 75 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 63 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 59 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 55 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 91 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 158 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 79 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 376 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 188 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 94 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 75 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 63 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 59 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 55 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 91 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 158 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 79 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 376 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 188 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 94 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; SSE42-LABEL: 'masked_load'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 256 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 256 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; AVX-LABEL: 'masked_load'
-; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 131 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 260 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 130 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i64> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i64> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i64> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i64> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 131 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 260 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 130 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; KNL-LABEL: 'masked_load'
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 262 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 130 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 262 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 130 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; SKX-LABEL: 'masked_load'
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
-  %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x double> undef)
-  %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x double> undef)
-  %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x double> undef)
-  %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x double> undef)
-  %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x double> undef)
-  %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x double> undef)
-  %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x double> undef)
-  %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x double> undef)
-
-  %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x float> undef)
-  %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x float> undef)
-  %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x float> undef)
-  %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x float> undef)
-  %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x float> undef)
-  %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x float> undef)
-  %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x float> undef)
-  %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x float> undef)
-  %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x float> undef)
-  %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x float> undef)
-  %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x float> undef)
-  %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x float> undef)
-  %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x float> undef)
-  %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x float> undef)
-  %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x float> undef)
-  %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x float> undef)
-
-  %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-  %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i64> undef)
-  %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i64> undef)
-  %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i64> undef)
-  %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-  %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i64> undef)
-  %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-  %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-
-  %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-  %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x i32> undef)
-  %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x i32> undef)
-  %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x i32> undef)
-  %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x i32> undef)
-  %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x i32> undef)
-  %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x i32> undef)
-  %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x i32> undef)
-  %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-  %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i32> undef)
-  %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i32> undef)
-  %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i32> undef)
-  %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-  %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i32> undef)
-  %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-  %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i32> undef)
-
-  %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-  %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-  %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-  %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-
-  %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-  %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-  %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-  %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+  %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+  %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x double> undef)
+  %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x double> undef)
+  %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x double> undef)
+  %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+  %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x double> undef)
+  %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+  %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+
+  %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+  %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x float> undef)
+  %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x float> undef)
+  %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x float> undef)
+  %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x float> undef)
+  %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x float> undef)
+  %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x float> undef)
+  %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x float> undef)
+  %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+  %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x float> undef)
+  %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x float> undef)
+  %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x float> undef)
+  %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+  %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x float> undef)
+  %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+  %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x float> undef)
+
+  %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+  %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i64> undef)
+  %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i64> undef)
+  %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i64> undef)
+  %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+  %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i64> undef)
+  %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+  %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+
+  %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+  %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x i32> undef)
+  %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x i32> undef)
+  %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x i32> undef)
+  %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x i32> undef)
+  %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x i32> undef)
+  %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x i32> undef)
+  %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x i32> undef)
+  %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+  %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i32> undef)
+  %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i32> undef)
+  %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i32> undef)
+  %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+  %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i32> undef)
+  %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+  %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i32> undef)
+
+  %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+  %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+  %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+  %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+
+  %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+  %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+  %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+  %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 
   ret i32 0
 }
 
-define i32 @masked_store() {
+define i32 @masked_store(<1 x i1> %m1, <2 x i1> %m2, <3 x i1> %m3, <4 x i1> %m4, <5 x i1> %m5, <6 x i1> %m6, <7 x i1> %m7, <8 x i1> %m8, <9 x i1> %m9, <10 x i1> %m10, <11 x i1> %m11, <12 x i1> %m12, <13 x i1> %m13, <14 x i1> %m14, <15 x i1> %m15, <16 x i1> %m16, <32 x i1> %m32, <64 x i1> %m64) {
 ; SSE2-LABEL: 'masked_store'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 75 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 70 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 91 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 85 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 79 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 73 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 62 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 158 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 79 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 376 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 188 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 94 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 75 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 70 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> %m15)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> %m14)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> %m13)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> %m12)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> %m11)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> %m10)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> %m9)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 91 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 85 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> %m15)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 79 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> %m14)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 73 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> %m13)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> %m12)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 62 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> %m11)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> %m10)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> %m9)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 158 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> %m32)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 79 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 376 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> %m64)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 188 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> %m32)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 94 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> %m8)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; SSE42-LABEL: 'masked_store'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 256 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> %m15)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> %m14)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> %m13)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> %m12)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> %m11)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> %m10)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> %m9)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> %m15)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> %m14)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> %m13)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> %m12)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> %m11)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> %m10)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> %m9)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> %m32)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 256 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> %m64)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> %m32)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> %m8)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; AVX-LABEL: 'masked_store'
-; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 131 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 260 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 130 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> %m8)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> %m7)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> %m6)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> %m5)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> %m4)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> %m3)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> %m2)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> %m1)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> %m16)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> %m15)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> %m14)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> %m13)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> %m12)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> %m11)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> %m10)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> %m9)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> %m8)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> %m7)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> %m6)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> %m5)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> %m4)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> %m3)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> %m2)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> %m1)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> %m8)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> %m7)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> %m6)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> %m5)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> %m4)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> %m3)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> %m2)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> %m1)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> %m16)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> %m15)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> %m14)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> %m13)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> %m12)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> %m11)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> %m10)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> %m9)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> %m8)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> %m7)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> %m6)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> %m5)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> %m4)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> %m3)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> %m2)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> %m1)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 131 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> %m32)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> %m16)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> %m8)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> %m4)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 260 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> %m64)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 130 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> %m32)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> %m16)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> %m8)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; KNL-LABEL: 'masked_store'
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 262 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 130 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> %m7)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> %m6)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> %m5)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> %m3)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> %m2)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> %m1)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> %m16)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> %m15)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> %m14)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> %m13)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> %m12)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> %m11)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> %m10)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> %m9)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> %m7)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> %m6)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> %m5)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> %m3)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> %m2)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> %m1)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> %m7)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> %m6)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> %m5)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> %m3)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> %m2)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> %m1)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> %m16)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> %m15)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> %m14)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> %m13)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> %m12)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> %m11)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> %m10)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> %m9)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> %m7)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> %m6)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> %m5)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> %m3)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> %m2)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> %m1)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> %m32)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> %m16)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 262 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> %m64)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 130 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> %m32)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> %m16)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> %m8)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; SKX-LABEL: 'masked_store'
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> %m15)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> %m14)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> %m13)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> %m12)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> %m11)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> %m10)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> %m9)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> %m15)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> %m14)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> %m13)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> %m12)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> %m11)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> %m10)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> %m9)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> %m32)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> %m64)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> %m32)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> %m8)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
-  call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> undef)
-  call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> undef)
-  call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> undef)
-  call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> undef)
-  call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> undef)
-  call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> undef)
-  call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> undef)
-
-  call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> undef)
-  call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> undef)
-  call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> undef)
-  call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> undef)
-  call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> undef)
-  call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> undef)
-  call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> undef)
-  call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> undef)
-  call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> undef)
-  call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> undef)
-  call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> undef)
-  call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> undef)
-  call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> undef)
-  call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> undef)
-  call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> undef)
-
-  call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> undef)
-  call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> undef)
-  call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> undef)
-  call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> undef)
-  call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> undef)
-  call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> undef)
-  call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> undef)
-
-  call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> undef)
-  call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> undef)
-  call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> undef)
-  call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> undef)
-  call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> undef)
-  call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> undef)
-  call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> undef)
-  call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> undef)
-  call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> undef)
-  call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> undef)
-  call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> undef)
-  call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> undef)
-  call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> undef)
-  call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> undef)
-  call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> undef)
-
-  call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> undef)
-  call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> undef)
-  call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> undef)
-
-  call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> undef)
-  call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> undef)
-  call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> undef)
-  call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> undef)
+  call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> %m7)
+  call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> %m6)
+  call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> %m5)
+  call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> %m4)
+  call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> %m3)
+  call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> %m2)
+  call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> %m1)
+
+  call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> %m16)
+  call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> %m15)
+  call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> %m14)
+  call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> %m13)
+  call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> %m12)
+  call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> %m11)
+  call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> %m10)
+  call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> %m9)
+  call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> %m7)
+  call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> %m6)
+  call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> %m5)
+  call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> %m4)
+  call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> %m3)
+  call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> %m2)
+  call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> %m1)
+
+  call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> %m7)
+  call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> %m6)
+  call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> %m5)
+  call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> %m4)
+  call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> %m3)
+  call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> %m2)
+  call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> %m1)
+
+  call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> %m16)
+  call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> %m15)
+  call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> %m14)
+  call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> %m13)
+  call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> %m12)
+  call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> %m11)
+  call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> %m10)
+  call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> %m9)
+  call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> %m7)
+  call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> %m6)
+  call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> %m5)
+  call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> %m4)
+  call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> %m3)
+  call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> %m2)
+  call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> %m1)
+
+  call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> %m32)
+  call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> %m16)
+  call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> %m4)
+
+  call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> %m64)
+  call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> %m32)
+  call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> %m16)
+  call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> %m8)
 
   ret i32 0
 }
 
-define i32 @masked_gather() {
+define i32 @masked_gather(<1 x i1> %m1, <2 x i1> %m2, <4 x i1> %m4, <8 x i1> %m8, <16 x i1> %m16, <32 x i1> %m32, <64 x i1> %m64) {
 ; SSE2-LABEL: 'masked_gather'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 316 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 158 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 79 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 77 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 93 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 384 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 192 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; SSE42-LABEL: 'masked_gather'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 192 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 130 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 260 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 130 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; AVX1-LABEL: 'masked_gather'
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 106 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 53 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 210 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 105 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 140 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 70 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 278 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 139 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 69 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; AVX2-LABEL: 'masked_gather'
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 106 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 53 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 210 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 105 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 139 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 70 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 276 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 138 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 69 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; SKL-LABEL: 'masked_gather'
-; SKL-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 106 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 53 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 210 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 105 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 139 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 70 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 276 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 138 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 69 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; SKL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; KNL-LABEL: 'masked_gather'
-; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 111 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 55 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 219 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 109 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 175 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 87 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 347 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 173 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; SKX-LABEL: 'masked_gather'
-; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 111 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 55 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 219 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 109 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 175 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 87 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 347 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 173 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
-  %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x double> undef)
-  %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x double> undef)
-  %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x double> undef)
-  %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x double> undef)
+  %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+  %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+  %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+  %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x double> undef)
 
-  %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x float> undef)
-  %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x float> undef)
-  %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x float> undef)
-  %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x float> undef)
+  %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+  %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+  %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+  %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x float> undef)
 
-  %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-  %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-  %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-  %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
+  %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+  %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+  %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+  %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
 
-  %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-  %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-  %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-  %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
+  %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+  %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+  %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+  %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
 
-  %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-  %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-  %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-  %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
+  %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+  %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+  %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+  %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
 
-  %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-  %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-  %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-  %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+  %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+  %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+  %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+  %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 
   ret i32 0
 }
 
-define i32 @masked_scatter() {
+define i32 @masked_scatter(<1 x i1> %m1, <2 x i1> %m2, <4 x i1> %m4, <8 x i1> %m8, <16 x i1> %m16, <32 x i1> %m32, <64 x i1> %m64) {
 ; SSE2-LABEL: 'masked_scatter'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 316 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 158 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 79 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 77 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 93 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 384 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> %m64)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 192 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; SSE42-LABEL: 'masked_scatter'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 192 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 130 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 260 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> %m64)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 130 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
-; AVX-LABEL: 'masked_scatter'
-; AVX-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 106 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 53 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 210 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 105 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
+; AVX1-LABEL: 'masked_scatter'
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 140 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 70 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 278 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> %m64)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 139 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 69 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
+;
+; AVX2-LABEL: 'masked_scatter'
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 139 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 70 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 276 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> %m64)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 138 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 69 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
+;
+; SKL-LABEL: 'masked_scatter'
+; SKL-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 139 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 70 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 276 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> %m64)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 138 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 69 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; KNL-LABEL: 'masked_scatter'
-; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 111 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 55 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 219 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 109 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 175 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 87 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 347 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> %m64)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 173 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; SKX-LABEL: 'masked_scatter'
-; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 111 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 55 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 219 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 109 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 175 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 87 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 347 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> %m64)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 173 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
-  call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-  call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-  call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
+  call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+  call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+  call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
 
-  call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-  call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-  call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
+  call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+  call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+  call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
 
-  call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-  call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-  call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
+  call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+  call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+  call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
 
-  call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-  call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-  call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
+  call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+  call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+  call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
 
-  call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-  call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-  call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
+  call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+  call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+  call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
 
-  call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> undef)
-  call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-  call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-  call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
+  call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> %m64)
+  call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+  call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+  call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
 
   ret i32 0
 }
 
-define i32 @masked_expandload() {
+define i32 @masked_expandload(<1 x i1> %m1, <2 x i1> %m2, <4 x i1> %m4, <8 x i1> %m8, <16 x i1> %m16, <32 x i1> %m32, <64 x i1> %m64) {
 ; SSE2-LABEL: 'masked_expandload'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> undef, <8 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> undef, <4 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> undef, <2 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> undef, <1 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> undef, <16 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> undef, <8 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> undef, <4 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> undef, <2 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> undef, <8 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> undef, <4 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> undef, <2 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> undef, <1 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> undef, <16 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> undef, <8 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> undef, <4 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> undef, <2 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> undef, <32 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> undef, <16 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> undef, <8 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> undef, <4 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 188 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> undef, <64 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 94 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> undef, <32 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> undef, <16 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> undef, <8 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> %m8, <8 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> %m4, <4 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> %m2, <2 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> %m1, <1 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> %m16, <16 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> %m8, <8 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> %m4, <4 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> %m2, <2 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> %m8, <8 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> %m4, <4 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> %m2, <2 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> %m1, <1 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> %m16, <16 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> %m8, <8 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> %m4, <4 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> %m2, <2 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> %m32, <32 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> %m16, <16 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> %m8, <8 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> %m4, <4 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 192 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> %m64, <64 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> %m32, <32 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> %m16, <16 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> %m8, <8 x i8> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; SSE42-LABEL: 'masked_expandload'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> undef, <8 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> undef, <4 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> undef, <2 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> undef, <1 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> undef, <16 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> undef, <8 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> undef, <4 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> undef, <2 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> undef, <8 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> undef, <4 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> undef, <2 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> undef, <1 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> undef, <16 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> undef, <8 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> undef, <4 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> undef, <2 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> undef, <32 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> undef, <16 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> undef, <8 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> undef, <4 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> undef, <64 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> undef, <32 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> undef, <16 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> undef, <8 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> %m8, <8 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> %m4, <4 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> %m2, <2 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> %m1, <1 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> %m16, <16 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> %m8, <8 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> %m4, <4 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> %m2, <2 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> %m8, <8 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> %m4, <4 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> %m2, <2 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> %m1, <1 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> %m16, <16 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> %m8, <8 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> %m4, <4 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> %m2, <2 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> %m32, <32 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> %m16, <16 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> %m8, <8 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> %m4, <4 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> %m64, <64 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> %m32, <32 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> %m16, <16 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> %m8, <8 x i8> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
-; AVX-LABEL: 'masked_expandload'
-; AVX-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> undef, <8 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> undef, <4 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> undef, <2 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> undef, <1 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> undef, <16 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> undef, <8 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> undef, <4 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> undef, <2 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> undef, <8 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> undef, <4 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> undef, <2 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> undef, <1 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> undef, <16 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> undef, <8 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> undef, <4 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> undef, <2 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> undef, <32 x i16> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> undef, <16 x i16> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> undef, <8 x i16> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> undef, <4 x i16> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 130 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> undef, <64 x i8> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> undef, <32 x i8> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> undef, <16 x i8> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> undef, <8 x i8> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
+; AVX1-LABEL: 'masked_expandload'
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> %m8, <8 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> %m4, <4 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> %m2, <2 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> %m1, <1 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> %m16, <16 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> %m8, <8 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> %m4, <4 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> %m2, <2 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> %m8, <8 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> %m4, <4 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> %m2, <2 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> %m1, <1 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> %m16, <16 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> %m8, <8 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> %m4, <4 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> %m2, <2 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> %m32, <32 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> %m16, <16 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> %m8, <8 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> %m4, <4 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 134 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> %m64, <64 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> %m32, <32 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> %m16, <16 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> %m8, <8 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
+;
+; AVX2-LABEL: 'masked_expandload'
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> %m8, <8 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> %m4, <4 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> %m2, <2 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> %m1, <1 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> %m16, <16 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> %m8, <8 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> %m4, <4 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> %m2, <2 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> %m8, <8 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> %m4, <4 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> %m2, <2 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> %m1, <1 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> %m16, <16 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> %m8, <8 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> %m4, <4 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> %m2, <2 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> %m32, <32 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> %m16, <16 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> %m8, <8 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> %m4, <4 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> %m64, <64 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> %m32, <32 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> %m16, <16 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> %m8, <8 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
+;
+; SKL-LABEL: 'masked_expandload'
+; SKL-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> %m8, <8 x double> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> %m4, <4 x double> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> %m2, <2 x double> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> %m1, <1 x double> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> %m16, <16 x float> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> %m8, <8 x float> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> %m4, <4 x float> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> %m2, <2 x float> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> %m8, <8 x i64> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> %m4, <4 x i64> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> %m2, <2 x i64> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> %m1, <1 x i64> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> %m16, <16 x i32> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> %m8, <8 x i32> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> %m4, <4 x i32> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> %m2, <2 x i32> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> %m32, <32 x i16> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> %m16, <16 x i16> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> %m8, <8 x i16> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> %m4, <4 x i16> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> %m64, <64 x i8> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> %m32, <32 x i8> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> %m16, <16 x i8> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> %m8, <8 x i8> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; AVX512-LABEL: 'masked_expandload'
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> undef, <8 x double> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> undef, <4 x double> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> undef, <2 x double> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> undef, <1 x double> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> undef, <16 x float> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> undef, <8 x float> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> undef, <4 x float> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> undef, <2 x float> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> undef, <8 x i64> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> undef, <4 x i64> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> undef, <2 x i64> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> undef, <1 x i64> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> undef, <16 x i32> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> undef, <8 x i32> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> undef, <4 x i32> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> undef, <2 x i32> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> undef, <32 x i16> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> undef, <16 x i16> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> undef, <8 x i16> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> undef, <4 x i16> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 131 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> undef, <64 x i8> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> undef, <32 x i8> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> undef, <16 x i8> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> undef, <8 x i8> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> %m8, <8 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> %m4, <4 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> %m2, <2 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> %m1, <1 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> %m16, <16 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> %m8, <8 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> %m4, <4 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> %m2, <2 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> %m8, <8 x i64> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> %m4, <4 x i64> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> %m2, <2 x i64> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> %m1, <1 x i64> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> %m16, <16 x i32> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> %m8, <8 x i32> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> %m4, <4 x i32> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> %m2, <2 x i32> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 99 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> %m32, <32 x i16> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> %m16, <16 x i16> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> %m8, <8 x i16> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> %m4, <4 x i16> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 195 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> %m64, <64 x i8> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 97 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> %m32, <32 x i8> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> %m16, <16 x i8> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> %m8, <8 x i8> undef)
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
-  %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> undef, <8 x double> undef)
-  %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> undef, <4 x double> undef)
-  %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> undef, <2 x double> undef)
-  %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> undef, <1 x double> undef)
+  %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> %m8, <8 x double> undef)
+  %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> %m4, <4 x double> undef)
+  %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> %m2, <2 x double> undef)
+  %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> %m1, <1 x double> undef)
 
-  %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> undef, <16 x float> undef)
-  %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> undef, <8 x float> undef)
-  %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> undef, <4 x float> undef)
-  %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> undef, <2 x float> undef)
+  %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> %m16, <16 x float> undef)
+  %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> %m8, <8 x float> undef)
+  %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> %m4, <4 x float> undef)
+  %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> %m2, <2 x float> undef)
 
-  %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> undef, <8 x i64> undef)
-  %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> undef, <4 x i64> undef)
-  %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> undef, <2 x i64> undef)
-  %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> undef, <1 x i64> undef)
+  %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> %m8, <8 x i64> undef)
+  %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> %m4, <4 x i64> undef)
+  %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> %m2, <2 x i64> undef)
+  %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> %m1, <1 x i64> undef)
 
-  %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> undef, <16 x i32> undef)
-  %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> undef, <8 x i32> undef)
-  %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> undef, <4 x i32> undef)
-  %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> undef, <2 x i32> undef)
+  %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> %m16, <16 x i32> undef)
+  %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> %m8, <8 x i32> undef)
+  %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> %m4, <4 x i32> undef)
+  %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> %m2, <2 x i32> undef)
 
-  %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> undef, <32 x i16> undef)
-  %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> undef, <16 x i16> undef)
-  %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> undef, <8 x i16> undef)
-  %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> undef, <4 x i16> undef)
+  %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> %m32, <32 x i16> undef)
+  %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> %m16, <16 x i16> undef)
+  %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> %m8, <8 x i16> undef)
+  %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> %m4, <4 x i16> undef)
 
-  %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> undef, <64 x i8> undef)
-  %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> undef, <32 x i8> undef)
-  %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> undef, <16 x i8> undef)
-  %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> undef, <8 x i8> undef)
+  %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> %m64, <64 x i8> undef)
+  %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> %m32, <32 x i8> undef)
+  %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> %m16, <16 x i8> undef)
+  %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> %m8, <8 x i8> undef)
 
   ret i32 0
 }
 
-define i32 @masked_compressstore() {
+define i32 @masked_compressstore(<1 x i1> %m1, <2 x i1> %m2, <4 x i1> %m4, <8 x i1> %m8, <16 x i1> %m16, <32 x i1> %m32, <64 x i1> %m64) {
 ; SSE2-LABEL: 'masked_compressstore'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 192 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> %m1)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> %m1)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> %m32)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 192 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> %m64)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> %m32)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> %m8)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; SSE42-LABEL: 'masked_compressstore'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> %m1)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> %m1)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> %m32)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> %m64)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> %m32)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> %m8)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; AVX1-LABEL: 'masked_compressstore'
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 134 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> %m8)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> %m4)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> %m2)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> %m1)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> %m16)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> %m8)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> %m4)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> %m2)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> %m8)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> %m4)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> %m2)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> %m1)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> %m16)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> %m8)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> %m4)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> %m2)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> %m32)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> %m16)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> %m8)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> %m4)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 134 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> %m64)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> %m32)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> %m16)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> %m8)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; AVX2-LABEL: 'masked_compressstore'
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> %m8)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> %m4)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> %m2)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> %m1)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> %m16)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> %m8)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> %m4)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> %m2)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> %m8)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> %m4)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> %m2)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> %m1)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> %m16)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> %m8)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> %m4)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> %m2)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> %m32)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> %m16)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> %m8)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> %m4)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> %m64)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> %m32)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> %m16)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> %m8)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; SKL-LABEL: 'masked_compressstore'
-; SKL-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> %m8)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> %m4)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> %m2)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> %m1)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> %m16)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> %m8)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> %m4)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> %m2)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> %m8)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> %m4)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> %m2)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> %m1)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> %m16)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> %m8)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> %m4)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> %m2)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> %m32)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> %m16)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> %m8)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> %m4)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> %m64)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> %m32)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> %m16)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> %m8)
 ; SKL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; AVX512-LABEL: 'masked_compressstore'
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 99 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 195 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 97 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> %m8)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> %m4)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> %m2)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> %m1)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> %m16)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> %m8)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> %m4)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> %m2)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> %m8)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> %m4)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> %m2)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> %m1)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> %m16)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> %m8)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> %m4)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> %m2)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 99 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> %m32)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> %m16)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> %m8)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> %m4)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 195 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> %m64)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 97 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> %m32)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> %m16)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> %m8)
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
-  call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> undef)
-  call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> undef)
-  call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> undef)
-  call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> undef)
-
-  call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> undef)
-  call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> undef)
-  call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> undef)
-  call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> undef)
-
-  call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> undef)
-  call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> undef)
-  call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> undef)
-  call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> undef)
-
-  call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> undef)
-  call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> undef)
-  call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> undef)
-  call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> undef)
-
-  call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> undef)
-  call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> undef)
-  call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> undef)
-  call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> undef)
-
-  call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> undef)
-  call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> undef)
-  call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> undef)
-  call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> undef)
+  call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> %m8)
+  call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> %m4)
+  call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> %m2)
+  call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> %m1)
+
+  call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> %m16)
+  call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> %m8)
+  call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> %m4)
+  call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> %m2)
+
+  call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> %m8)
+  call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> %m4)
+  call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> %m2)
+  call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> %m1)
+
+  call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> %m16)
+  call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> %m8)
+  call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> %m4)
+  call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> %m2)
+
+  call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> %m32)
+  call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> %m16)
+  call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> %m8)
+  call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> %m4)
+
+  call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> %m64)
+  call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> %m32)
+  call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> %m16)
+  call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> %m8)
 
   ret i32 0
 }
diff --git a/llvm/test/Analysis/CostModel/X86/masked-intrinsic-cost.ll b/llvm/test/Analysis/CostModel/X86/masked-intrinsic-cost.ll
index c7e7c46482c4..46123e9f6057 100644
--- a/llvm/test/Analysis/CostModel/X86/masked-intrinsic-cost.ll
+++ b/llvm/test/Analysis/CostModel/X86/masked-intrinsic-cost.ll
@@ -8,1454 +8,1562 @@
 ; RUN: opt < %s -S -mtriple=x86_64-apple-darwin -passes="print<cost-model>" 2>&1 -disable-output -mcpu=knl      | FileCheck %s --check-prefixes=AVX512,KNL
 ; RUN: opt < %s -S -mtriple=x86_64-apple-darwin -passes="print<cost-model>" 2>&1 -disable-output -mcpu=skx      | FileCheck %s --check-prefixes=AVX512,SKX
 
-define i32 @masked_load() {
+define i32 @masked_load(<1 x i1> %m1, <2 x i1> %m2, <3 x i1> %m3, <4 x i1> %m4, <5 x i1> %m5, <6 x i1> %m6, <7 x i1> %m7, <8 x i1> %m8, <9 x i1> %m9, <10 x i1> %m10, <11 x i1> %m11, <12 x i1> %m12, <13 x i1> %m13, <14 x i1> %m14, <15 x i1> %m15, <16 x i1> %m16, <32 x i1> %m32, <64 x i1> %m64) {
 ; SSE2-LABEL: 'masked_load'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 75 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 63 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 59 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 55 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 91 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 158 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 79 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 376 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 188 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 94 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 75 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 63 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 59 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 55 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 91 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 158 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 79 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 376 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 188 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 94 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; SSE42-LABEL: 'masked_load'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 256 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 256 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; AVX-LABEL: 'masked_load'
-; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 131 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 260 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 130 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i64> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i64> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i64> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i64> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 131 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 260 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 130 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; KNL-LABEL: 'masked_load'
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 262 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 130 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 262 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 130 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; SKX-LABEL: 'masked_load'
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
-  %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x double> undef)
-  %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x double> undef)
-  %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x double> undef)
-  %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x double> undef)
-  %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x double> undef)
-  %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x double> undef)
-  %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x double> undef)
-  %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x double> undef)
-
-  %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x float> undef)
-  %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x float> undef)
-  %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x float> undef)
-  %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x float> undef)
-  %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x float> undef)
-  %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x float> undef)
-  %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x float> undef)
-  %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x float> undef)
-  %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x float> undef)
-  %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x float> undef)
-  %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x float> undef)
-  %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x float> undef)
-  %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x float> undef)
-  %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x float> undef)
-  %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x float> undef)
-  %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x float> undef)
-
-  %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-  %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i64> undef)
-  %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i64> undef)
-  %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i64> undef)
-  %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-  %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i64> undef)
-  %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-  %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-
-  %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-  %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x i32> undef)
-  %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x i32> undef)
-  %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x i32> undef)
-  %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x i32> undef)
-  %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x i32> undef)
-  %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x i32> undef)
-  %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x i32> undef)
-  %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-  %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i32> undef)
-  %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i32> undef)
-  %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i32> undef)
-  %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-  %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i32> undef)
-  %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-  %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i32> undef)
-
-  %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-  %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-  %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-  %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-
-  %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-  %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-  %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-  %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+  %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+  %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x double> undef)
+  %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x double> undef)
+  %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x double> undef)
+  %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+  %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x double> undef)
+  %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+  %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+
+  %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+  %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x float> undef)
+  %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x float> undef)
+  %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x float> undef)
+  %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x float> undef)
+  %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x float> undef)
+  %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x float> undef)
+  %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x float> undef)
+  %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+  %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x float> undef)
+  %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x float> undef)
+  %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x float> undef)
+  %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+  %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x float> undef)
+  %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+  %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x float> undef)
+
+  %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+  %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i64> undef)
+  %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i64> undef)
+  %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i64> undef)
+  %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+  %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i64> undef)
+  %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+  %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+
+  %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+  %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x i32> undef)
+  %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x i32> undef)
+  %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x i32> undef)
+  %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x i32> undef)
+  %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x i32> undef)
+  %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x i32> undef)
+  %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x i32> undef)
+  %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+  %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i32> undef)
+  %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i32> undef)
+  %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i32> undef)
+  %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+  %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i32> undef)
+  %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+  %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i32> undef)
+
+  %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+  %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+  %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+  %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+
+  %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+  %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+  %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+  %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 
   ret i32 0
 }
 
-define i32 @masked_store() {
+define i32 @masked_store(<1 x i1> %m1, <2 x i1> %m2, <3 x i1> %m3, <4 x i1> %m4, <5 x i1> %m5, <6 x i1> %m6, <7 x i1> %m7, <8 x i1> %m8, <9 x i1> %m9, <10 x i1> %m10, <11 x i1> %m11, <12 x i1> %m12, <13 x i1> %m13, <14 x i1> %m14, <15 x i1> %m15, <16 x i1> %m16, <32 x i1> %m32, <64 x i1> %m64) {
 ; SSE2-LABEL: 'masked_store'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 75 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 70 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 91 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 85 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 79 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 73 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 62 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 158 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 79 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 376 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 188 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 94 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 75 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 70 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> %m15)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> %m14)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> %m13)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> %m12)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> %m11)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> %m10)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> %m9)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 91 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 85 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> %m15)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 79 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> %m14)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 73 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> %m13)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> %m12)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 62 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> %m11)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> %m10)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> %m9)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 158 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> %m32)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 79 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 376 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> %m64)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 188 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> %m32)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 94 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> %m8)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; SSE42-LABEL: 'masked_store'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 256 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> %m15)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> %m14)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> %m13)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> %m12)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> %m11)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> %m10)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> %m9)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> %m15)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> %m14)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> %m13)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> %m12)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> %m11)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> %m10)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> %m9)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> %m32)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 256 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> %m64)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> %m32)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> %m8)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; AVX-LABEL: 'masked_store'
-; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 131 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 260 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 130 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> %m8)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> %m7)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> %m6)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> %m5)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> %m4)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> %m3)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> %m2)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> %m1)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> %m16)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> %m15)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> %m14)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> %m13)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> %m12)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> %m11)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> %m10)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> %m9)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> %m8)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> %m7)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> %m6)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> %m5)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> %m4)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> %m3)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> %m2)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> %m1)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> %m8)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> %m7)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> %m6)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> %m5)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> %m4)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> %m3)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> %m2)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> %m1)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> %m16)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> %m15)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> %m14)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> %m13)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> %m12)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> %m11)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> %m10)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> %m9)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> %m8)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> %m7)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> %m6)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> %m5)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> %m4)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> %m3)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> %m2)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> %m1)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 131 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> %m32)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> %m16)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> %m8)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> %m4)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 260 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> %m64)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 130 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> %m32)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> %m16)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> %m8)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; KNL-LABEL: 'masked_store'
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 262 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 130 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> %m7)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> %m6)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> %m5)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> %m3)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> %m2)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> %m1)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> %m16)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> %m15)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> %m14)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> %m13)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> %m12)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> %m11)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> %m10)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> %m9)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> %m7)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> %m6)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> %m5)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> %m3)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> %m2)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> %m1)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> %m7)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> %m6)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> %m5)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> %m3)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> %m2)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> %m1)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> %m16)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> %m15)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> %m14)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> %m13)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> %m12)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> %m11)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> %m10)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> %m9)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> %m7)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> %m6)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> %m5)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> %m3)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> %m2)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> %m1)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> %m32)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> %m16)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 262 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> %m64)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 130 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> %m32)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> %m16)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> %m8)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; SKX-LABEL: 'masked_store'
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> %m15)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> %m14)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> %m13)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> %m12)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> %m11)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> %m10)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> %m9)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> %m15)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> %m14)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> %m13)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> %m12)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> %m11)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> %m10)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> %m9)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> %m32)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> %m64)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> %m32)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> %m8)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
-  call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> undef)
-  call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> undef)
-  call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> undef)
-  call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> undef)
-  call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> undef)
-  call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> undef)
-  call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> undef)
-
-  call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> undef)
-  call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> undef)
-  call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> undef)
-  call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> undef)
-  call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> undef)
-  call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> undef)
-  call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> undef)
-  call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> undef)
-  call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> undef)
-  call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> undef)
-  call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> undef)
-  call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> undef)
-  call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> undef)
-  call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> undef)
-  call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> undef)
-
-  call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> undef)
-  call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> undef)
-  call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> undef)
-  call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> undef)
-  call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> undef)
-  call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> undef)
-  call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> undef)
-
-  call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> undef)
-  call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> undef)
-  call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> undef)
-  call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> undef)
-  call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> undef)
-  call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> undef)
-  call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> undef)
-  call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> undef)
-  call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> undef)
-  call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> undef)
-  call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> undef)
-  call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> undef)
-  call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> undef)
-  call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> undef)
-  call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> undef)
-
-  call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> undef)
-  call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> undef)
-  call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> undef)
-
-  call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> undef)
-  call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> undef)
-  call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> undef)
-  call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> undef)
+  call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> %m7)
+  call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> %m6)
+  call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> %m5)
+  call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> %m4)
+  call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> %m3)
+  call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> %m2)
+  call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> %m1)
+
+  call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> %m16)
+  call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> %m15)
+  call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> %m14)
+  call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> %m13)
+  call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> %m12)
+  call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> %m11)
+  call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> %m10)
+  call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> %m9)
+  call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> %m7)
+  call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> %m6)
+  call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> %m5)
+  call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> %m4)
+  call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> %m3)
+  call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> %m2)
+  call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> %m1)
+
+  call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> %m7)
+  call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> %m6)
+  call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> %m5)
+  call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> %m4)
+  call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> %m3)
+  call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> %m2)
+  call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> %m1)
+
+  call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> %m16)
+  call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> %m15)
+  call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> %m14)
+  call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> %m13)
+  call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> %m12)
+  call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> %m11)
+  call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> %m10)
+  call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> %m9)
+  call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> %m7)
+  call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> %m6)
+  call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> %m5)
+  call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> %m4)
+  call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> %m3)
+  call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> %m2)
+  call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> %m1)
+
+  call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> %m32)
+  call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> %m16)
+  call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> %m4)
+
+  call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> %m64)
+  call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> %m32)
+  call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> %m16)
+  call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> %m8)
 
   ret i32 0
 }
 
-define i32 @masked_gather() {
+define i32 @masked_gather(<1 x i1> %m1, <2 x i1> %m2, <4 x i1> %m4, <8 x i1> %m8, <16 x i1> %m16, <32 x i1> %m32, <64 x i1> %m64) {
 ; SSE2-LABEL: 'masked_gather'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 316 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 158 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 79 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 77 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 93 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 384 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 192 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; SSE42-LABEL: 'masked_gather'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 192 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 130 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 260 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 130 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; AVX1-LABEL: 'masked_gather'
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 106 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 53 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 210 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 105 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 140 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 70 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 278 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 139 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 69 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; AVX2-LABEL: 'masked_gather'
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 106 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 53 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 210 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 105 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 139 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 70 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 276 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 138 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 69 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; SKL-LABEL: 'masked_gather'
-; SKL-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 106 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 53 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 210 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 105 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 139 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 70 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 276 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 138 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 69 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; SKL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; KNL-LABEL: 'masked_gather'
-; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 111 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 55 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 219 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 109 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 175 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 87 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 347 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 173 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; SKX-LABEL: 'masked_gather'
-; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 111 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 55 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 219 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 109 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 175 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 87 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 347 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 173 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
-  %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x double> undef)
-  %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x double> undef)
-  %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x double> undef)
-  %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x double> undef)
+  %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+  %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+  %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+  %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x double> undef)
 
-  %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x float> undef)
-  %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x float> undef)
-  %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x float> undef)
-  %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x float> undef)
+  %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+  %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+  %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+  %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x float> undef)
 
-  %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-  %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-  %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-  %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
+  %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+  %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+  %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+  %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
 
-  %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-  %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-  %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-  %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
+  %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+  %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+  %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+  %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
 
-  %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-  %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-  %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-  %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
+  %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+  %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+  %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+  %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
 
-  %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-  %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-  %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-  %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+  %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+  %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+  %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+  %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 
   ret i32 0
 }
 
-define i32 @masked_scatter() {
+define i32 @masked_scatter(<1 x i1> %m1, <2 x i1> %m2, <4 x i1> %m4, <8 x i1> %m8, <16 x i1> %m16, <32 x i1> %m32, <64 x i1> %m64) {
 ; SSE2-LABEL: 'masked_scatter'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 316 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 158 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 79 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 77 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 93 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 384 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> %m64)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 192 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; SSE42-LABEL: 'masked_scatter'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 192 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 130 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 260 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> %m64)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 130 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
-; AVX-LABEL: 'masked_scatter'
-; AVX-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 106 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 53 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 210 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 105 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
+; AVX1-LABEL: 'masked_scatter'
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 140 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 70 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 278 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> %m64)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 139 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 69 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
+;
+; AVX2-LABEL: 'masked_scatter'
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 139 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 70 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 276 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> %m64)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 138 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 69 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
+;
+; SKL-LABEL: 'masked_scatter'
+; SKL-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 139 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 70 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 276 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> %m64)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 138 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 69 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; KNL-LABEL: 'masked_scatter'
-; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 111 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 55 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 219 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 109 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 175 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 87 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 347 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> %m64)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 173 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; SKX-LABEL: 'masked_scatter'
-; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 111 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 55 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 219 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 109 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 175 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 87 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 347 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> %m64)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 173 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
-  call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-  call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-  call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
+  call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+  call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+  call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
 
-  call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-  call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-  call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
+  call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+  call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+  call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
 
-  call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-  call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-  call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
+  call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+  call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+  call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
 
-  call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-  call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-  call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
+  call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+  call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+  call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
 
-  call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-  call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-  call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
+  call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+  call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+  call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
 
-  call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> undef)
-  call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-  call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-  call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
+  call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> %m64)
+  call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+  call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+  call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
 
   ret i32 0
 }
 
-define i32 @masked_expandload() {
+define i32 @masked_expandload(<1 x i1> %m1, <2 x i1> %m2, <4 x i1> %m4, <8 x i1> %m8, <16 x i1> %m16, <32 x i1> %m32, <64 x i1> %m64) {
 ; SSE2-LABEL: 'masked_expandload'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> undef, <8 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> undef, <4 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> undef, <2 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> undef, <1 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> undef, <16 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> undef, <8 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> undef, <4 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> undef, <2 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> undef, <8 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> undef, <4 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> undef, <2 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> undef, <1 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> undef, <16 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> undef, <8 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> undef, <4 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> undef, <2 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> undef, <32 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> undef, <16 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> undef, <8 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> undef, <4 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 188 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> undef, <64 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 94 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> undef, <32 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> undef, <16 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> undef, <8 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> %m8, <8 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> %m4, <4 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> %m2, <2 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> %m1, <1 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> %m16, <16 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> %m8, <8 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> %m4, <4 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> %m2, <2 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> %m8, <8 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> %m4, <4 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> %m2, <2 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> %m1, <1 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> %m16, <16 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> %m8, <8 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> %m4, <4 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> %m2, <2 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> %m32, <32 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> %m16, <16 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> %m8, <8 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> %m4, <4 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 192 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> %m64, <64 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> %m32, <32 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> %m16, <16 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> %m8, <8 x i8> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; SSE42-LABEL: 'masked_expandload'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> undef, <8 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> undef, <4 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> undef, <2 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> undef, <1 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> undef, <16 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> undef, <8 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> undef, <4 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> undef, <2 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> undef, <8 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> undef, <4 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> undef, <2 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> undef, <1 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> undef, <16 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> undef, <8 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> undef, <4 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> undef, <2 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> undef, <32 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> undef, <16 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> undef, <8 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> undef, <4 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> undef, <64 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> undef, <32 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> undef, <16 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> undef, <8 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> %m8, <8 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> %m4, <4 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> %m2, <2 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> %m1, <1 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> %m16, <16 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> %m8, <8 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> %m4, <4 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> %m2, <2 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> %m8, <8 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> %m4, <4 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> %m2, <2 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> %m1, <1 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> %m16, <16 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> %m8, <8 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> %m4, <4 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> %m2, <2 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> %m32, <32 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> %m16, <16 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> %m8, <8 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> %m4, <4 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> %m64, <64 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> %m32, <32 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> %m16, <16 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> %m8, <8 x i8> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
-; AVX-LABEL: 'masked_expandload'
-; AVX-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> undef, <8 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> undef, <4 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> undef, <2 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> undef, <1 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> undef, <16 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> undef, <8 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> undef, <4 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> undef, <2 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> undef, <8 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> undef, <4 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> undef, <2 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> undef, <1 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> undef, <16 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> undef, <8 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> undef, <4 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> undef, <2 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> undef, <32 x i16> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> undef, <16 x i16> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> undef, <8 x i16> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> undef, <4 x i16> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 130 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> undef, <64 x i8> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> undef, <32 x i8> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> undef, <16 x i8> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> undef, <8 x i8> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
+; AVX1-LABEL: 'masked_expandload'
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> %m8, <8 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> %m4, <4 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> %m2, <2 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> %m1, <1 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> %m16, <16 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> %m8, <8 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> %m4, <4 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> %m2, <2 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> %m8, <8 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> %m4, <4 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> %m2, <2 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> %m1, <1 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> %m16, <16 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> %m8, <8 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> %m4, <4 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> %m2, <2 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> %m32, <32 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> %m16, <16 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> %m8, <8 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> %m4, <4 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 134 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> %m64, <64 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> %m32, <32 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> %m16, <16 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> %m8, <8 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
+;
+; AVX2-LABEL: 'masked_expandload'
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> %m8, <8 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> %m4, <4 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> %m2, <2 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> %m1, <1 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> %m16, <16 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> %m8, <8 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> %m4, <4 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> %m2, <2 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> %m8, <8 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> %m4, <4 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> %m2, <2 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> %m1, <1 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> %m16, <16 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> %m8, <8 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> %m4, <4 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> %m2, <2 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> %m32, <32 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> %m16, <16 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> %m8, <8 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> %m4, <4 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> %m64, <64 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> %m32, <32 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> %m16, <16 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> %m8, <8 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
+;
+; SKL-LABEL: 'masked_expandload'
+; SKL-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> %m8, <8 x double> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> %m4, <4 x double> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> %m2, <2 x double> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> %m1, <1 x double> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> %m16, <16 x float> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> %m8, <8 x float> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> %m4, <4 x float> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> %m2, <2 x float> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> %m8, <8 x i64> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> %m4, <4 x i64> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> %m2, <2 x i64> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> %m1, <1 x i64> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> %m16, <16 x i32> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> %m8, <8 x i32> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> %m4, <4 x i32> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> %m2, <2 x i32> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> %m32, <32 x i16> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> %m16, <16 x i16> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> %m8, <8 x i16> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> %m4, <4 x i16> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> %m64, <64 x i8> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> %m32, <32 x i8> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> %m16, <16 x i8> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> %m8, <8 x i8> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; AVX512-LABEL: 'masked_expandload'
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> undef, <8 x double> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> undef, <4 x double> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> undef, <2 x double> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> undef, <1 x double> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> undef, <16 x float> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> undef, <8 x float> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> undef, <4 x float> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> undef, <2 x float> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> undef, <8 x i64> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> undef, <4 x i64> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> undef, <2 x i64> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> undef, <1 x i64> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> undef, <16 x i32> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> undef, <8 x i32> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> undef, <4 x i32> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> undef, <2 x i32> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> undef, <32 x i16> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> undef, <16 x i16> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> undef, <8 x i16> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> undef, <4 x i16> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 131 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> undef, <64 x i8> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> undef, <32 x i8> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> undef, <16 x i8> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> undef, <8 x i8> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> %m8, <8 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> %m4, <4 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> %m2, <2 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> %m1, <1 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> %m16, <16 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> %m8, <8 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> %m4, <4 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> %m2, <2 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> %m8, <8 x i64> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> %m4, <4 x i64> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> %m2, <2 x i64> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> %m1, <1 x i64> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> %m16, <16 x i32> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> %m8, <8 x i32> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> %m4, <4 x i32> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> %m2, <2 x i32> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 99 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> %m32, <32 x i16> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> %m16, <16 x i16> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> %m8, <8 x i16> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> %m4, <4 x i16> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 195 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> %m64, <64 x i8> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 97 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> %m32, <32 x i8> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> %m16, <16 x i8> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> %m8, <8 x i8> undef)
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
-  %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> undef, <8 x double> undef)
-  %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> undef, <4 x double> undef)
-  %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> undef, <2 x double> undef)
-  %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> undef, <1 x double> undef)
+  %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> %m8, <8 x double> undef)
+  %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> %m4, <4 x double> undef)
+  %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> %m2, <2 x double> undef)
+  %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> %m1, <1 x double> undef)
 
-  %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> undef, <16 x float> undef)
-  %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> undef, <8 x float> undef)
-  %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> undef, <4 x float> undef)
-  %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> undef, <2 x float> undef)
+  %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> %m16, <16 x float> undef)
+  %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> %m8, <8 x float> undef)
+  %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> %m4, <4 x float> undef)
+  %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> %m2, <2 x float> undef)
 
-  %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> undef, <8 x i64> undef)
-  %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> undef, <4 x i64> undef)
-  %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> undef, <2 x i64> undef)
-  %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> undef, <1 x i64> undef)
+  %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> %m8, <8 x i64> undef)
+  %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> %m4, <4 x i64> undef)
+  %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> %m2, <2 x i64> undef)
+  %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> %m1, <1 x i64> undef)
 
-  %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> undef, <16 x i32> undef)
-  %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> undef, <8 x i32> undef)
-  %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> undef, <4 x i32> undef)
-  %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> undef, <2 x i32> undef)
+  %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> %m16, <16 x i32> undef)
+  %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> %m8, <8 x i32> undef)
+  %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> %m4, <4 x i32> undef)
+  %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> %m2, <2 x i32> undef)
 
-  %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> undef, <32 x i16> undef)
-  %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> undef, <16 x i16> undef)
-  %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> undef, <8 x i16> undef)
-  %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> undef, <4 x i16> undef)
+  %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> %m32, <32 x i16> undef)
+  %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> %m16, <16 x i16> undef)
+  %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> %m8, <8 x i16> undef)
+  %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> %m4, <4 x i16> undef)
 
-  %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> undef, <64 x i8> undef)
-  %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> undef, <32 x i8> undef)
-  %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> undef, <16 x i8> undef)
-  %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> undef, <8 x i8> undef)
+  %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> %m64, <64 x i8> undef)
+  %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> %m32, <32 x i8> undef)
+  %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> %m16, <16 x i8> undef)
+  %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> %m8, <8 x i8> undef)
 
   ret i32 0
 }
 
-define i32 @masked_compressstore() {
+define i32 @masked_compressstore(<1 x i1> %m1, <2 x i1> %m2, <4 x i1> %m4, <8 x i1> %m8, <16 x i1> %m16, <32 x i1> %m32, <64 x i1> %m64) {
 ; SSE2-LABEL: 'masked_compressstore'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 192 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> %m1)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> %m1)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> %m32)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 192 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> %m64)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> %m32)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> %m8)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; SSE42-LABEL: 'masked_compressstore'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> %m1)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> %m1)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> %m32)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> %m64)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> %m32)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> %m8)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; AVX1-LABEL: 'masked_compressstore'
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 134 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> %m8)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> %m4)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> %m2)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> %m1)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> %m16)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> %m8)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> %m4)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> %m2)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> %m8)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> %m4)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> %m2)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> %m1)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> %m16)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> %m8)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> %m4)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> %m2)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> %m32)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> %m16)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> %m8)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> %m4)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 134 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> %m64)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> %m32)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> %m16)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> %m8)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; AVX2-LABEL: 'masked_compressstore'
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> %m8)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> %m4)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> %m2)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> %m1)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> %m16)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> %m8)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> %m4)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> %m2)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> %m8)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> %m4)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> %m2)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> %m1)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> %m16)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> %m8)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> %m4)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> %m2)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> %m32)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> %m16)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> %m8)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> %m4)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> %m64)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> %m32)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> %m16)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> %m8)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; SKL-LABEL: 'masked_compressstore'
-; SKL-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> %m8)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> %m4)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> %m2)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> %m1)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> %m16)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> %m8)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> %m4)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> %m2)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> %m8)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> %m4)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> %m2)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> %m1)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> %m16)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> %m8)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> %m4)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> %m2)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> %m32)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> %m16)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> %m8)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> %m4)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> %m64)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> %m32)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> %m16)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> %m8)
 ; SKL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
 ; AVX512-LABEL: 'masked_compressstore'
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 99 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 195 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 97 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> %m8)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> %m4)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> %m2)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> %m1)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> %m16)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> %m8)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> %m4)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> %m2)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> %m8)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> %m4)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> %m2)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> %m1)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> %m16)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> %m8)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> %m4)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> %m2)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 99 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> %m32)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> %m16)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> %m8)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> %m4)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 195 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> %m64)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 97 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> %m32)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> %m16)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> %m8)
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
-  call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> undef)
-  call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> undef)
-  call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> undef)
-  call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> undef)
-
-  call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> undef)
-  call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> undef)
-  call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> undef)
-  call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> undef)
-
-  call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> undef)
-  call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> undef)
-  call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> undef)
-  call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> undef)
-
-  call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> undef)
-  call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> undef)
-  call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> undef)
-  call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> undef)
-
-  call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> undef)
-  call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> undef)
-  call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> undef)
-  call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> undef)
-
-  call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> undef)
-  call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> undef)
-  call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> undef)
-  call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> undef)
+  call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> %m8)
+  call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> %m4)
+  call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> %m2)
+  call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> %m1)
+
+  call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> %m16)
+  call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> %m8)
+  call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> %m4)
+  call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> %m2)
+
+  call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> %m8)
+  call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> %m4)
+  call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> %m2)
+  call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> %m1)
+
+  call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> %m16)
+  call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> %m8)
+  call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> %m4)
+  call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> %m2)
+
+  call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> %m32)
+  call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> %m16)
+  call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> %m8)
+  call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> %m4)
+
+  call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> %m64)
+  call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> %m32)
+  call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> %m16)
+  call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> %m8)
 
   ret i32 0
 }
diff --git a/llvm/test/Analysis/CostModel/X86/masked-intrinsic-latency.ll b/llvm/test/Analysis/CostModel/X86/masked-intrinsic-latency.ll
index edb05ad71a40..416005b4f5f4 100644
--- a/llvm/test/Analysis/CostModel/X86/masked-intrinsic-latency.ll
+++ b/llvm/test/Analysis/CostModel/X86/masked-intrinsic-latency.ll
@@ -8,1454 +8,1508 @@
 ; RUN: opt < %s -S -mtriple=x86_64-apple-darwin -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=latency -mcpu=knl      | FileCheck %s --check-prefixes=AVX512,KNL
 ; RUN: opt < %s -S -mtriple=x86_64-apple-darwin -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=latency -mcpu=skx      | FileCheck %s --check-prefixes=AVX512,SKX
 
-define i32 @masked_load() {
+define i32 @masked_load(<1 x i1> %m1, <2 x i1> %m2, <3 x i1> %m3, <4 x i1> %m4, <5 x i1> %m5, <6 x i1> %m6, <7 x i1> %m7, <8 x i1> %m8, <9 x i1> %m9, <10 x i1> %m10, <11 x i1> %m11, <12 x i1> %m12, <13 x i1> %m13, <14 x i1> %m14, <15 x i1> %m15, <16 x i1> %m16, <32 x i1> %m32, <64 x i1> %m64) {
 ; SSE2-LABEL: 'masked_load'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 91 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 107 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 101 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 95 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 89 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 83 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 77 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 53 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 190 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 95 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 440 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 220 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 110 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 91 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 107 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 101 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 95 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 89 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 83 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 77 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 53 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 190 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 95 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 440 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 220 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 110 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; SSE42-LABEL: 'masked_load'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 75 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 70 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 55 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 320 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 75 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 70 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 55 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 320 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; AVX-LABEL: 'masked_load'
-; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 163 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 324 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i64> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i64> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i64> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i64> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 163 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 324 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; KNL-LABEL: 'masked_load'
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 164 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 326 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 164 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 326 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; SKX-LABEL: 'masked_load'
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
-  %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x double> undef)
-  %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x double> undef)
-  %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x double> undef)
-  %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x double> undef)
-  %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x double> undef)
-  %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x double> undef)
-  %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x double> undef)
-  %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x double> undef)
-
-  %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x float> undef)
-  %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x float> undef)
-  %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x float> undef)
-  %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x float> undef)
-  %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x float> undef)
-  %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x float> undef)
-  %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x float> undef)
-  %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x float> undef)
-  %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x float> undef)
-  %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x float> undef)
-  %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x float> undef)
-  %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x float> undef)
-  %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x float> undef)
-  %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x float> undef)
-  %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x float> undef)
-  %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x float> undef)
-
-  %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-  %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i64> undef)
-  %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i64> undef)
-  %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i64> undef)
-  %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-  %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i64> undef)
-  %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-  %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-
-  %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-  %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x i32> undef)
-  %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x i32> undef)
-  %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x i32> undef)
-  %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x i32> undef)
-  %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x i32> undef)
-  %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x i32> undef)
-  %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x i32> undef)
-  %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-  %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i32> undef)
-  %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i32> undef)
-  %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i32> undef)
-  %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-  %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i32> undef)
-  %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-  %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i32> undef)
-
-  %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-  %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-  %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-  %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-
-  %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-  %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-  %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-  %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+  %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+  %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x double> undef)
+  %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x double> undef)
+  %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x double> undef)
+  %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+  %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x double> undef)
+  %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+  %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+
+  %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+  %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x float> undef)
+  %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x float> undef)
+  %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x float> undef)
+  %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x float> undef)
+  %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x float> undef)
+  %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x float> undef)
+  %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x float> undef)
+  %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+  %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x float> undef)
+  %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x float> undef)
+  %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x float> undef)
+  %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+  %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x float> undef)
+  %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+  %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x float> undef)
+
+  %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+  %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i64> undef)
+  %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i64> undef)
+  %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i64> undef)
+  %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+  %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i64> undef)
+  %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+  %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+
+  %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+  %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x i32> undef)
+  %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x i32> undef)
+  %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x i32> undef)
+  %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x i32> undef)
+  %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x i32> undef)
+  %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x i32> undef)
+  %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x i32> undef)
+  %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+  %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i32> undef)
+  %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i32> undef)
+  %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i32> undef)
+  %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+  %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i32> undef)
+  %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+  %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i32> undef)
+
+  %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+  %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+  %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+  %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+
+  %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+  %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+  %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+  %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 
   ret i32 0
 }
 
-define i32 @masked_store() {
+define i32 @masked_store(<1 x i1> %m1, <2 x i1> %m2, <3 x i1> %m3, <4 x i1> %m4, <5 x i1> %m5, <6 x i1> %m6, <7 x i1> %m7, <8 x i1> %m8, <9 x i1> %m9, <10 x i1> %m10, <11 x i1> %m11, <12 x i1> %m12, <13 x i1> %m13, <14 x i1> %m14, <15 x i1> %m15, <16 x i1> %m16, <32 x i1> %m32, <64 x i1> %m64) {
 ; SSE2-LABEL: 'masked_store'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 91 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 85 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 79 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 73 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 62 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 107 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 93 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 73 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 59 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 53 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 190 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 95 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 440 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 220 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 110 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 91 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 85 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> %m15)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 79 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> %m14)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 73 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> %m13)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> %m12)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 62 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> %m11)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> %m10)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> %m9)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 107 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> %m15)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 93 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> %m14)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> %m13)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> %m12)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 73 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> %m11)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> %m10)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 59 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> %m9)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 53 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 190 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> %m32)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 95 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 440 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> %m64)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 220 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> %m32)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 110 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> %m8)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; SSE42-LABEL: 'masked_store'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 75 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 70 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 55 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 320 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> %m15)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> %m14)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> %m13)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> %m12)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> %m11)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> %m10)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> %m9)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 75 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> %m15)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 70 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> %m14)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> %m13)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> %m12)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 55 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> %m11)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> %m10)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> %m9)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> %m32)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 320 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> %m64)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> %m32)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> %m8)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; AVX-LABEL: 'masked_store'
-; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 163 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 324 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> %m8)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> %m7)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> %m6)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> %m5)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> %m4)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> %m3)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> %m2)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> %m1)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> %m16)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> %m15)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> %m14)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> %m13)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> %m12)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> %m11)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> %m10)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> %m9)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> %m8)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> %m7)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> %m6)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> %m5)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> %m4)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> %m3)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> %m2)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> %m1)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> %m8)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> %m7)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> %m6)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> %m5)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> %m4)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> %m3)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> %m2)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> %m1)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> %m16)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> %m15)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> %m14)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> %m13)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> %m12)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> %m11)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> %m10)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> %m9)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> %m8)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> %m7)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> %m6)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> %m5)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> %m4)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> %m3)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> %m2)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> %m1)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 163 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> %m32)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> %m16)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> %m8)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> %m4)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 324 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> %m64)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> %m32)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> %m16)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> %m8)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; KNL-LABEL: 'masked_store'
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 164 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 326 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> %m7)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> %m6)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> %m5)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> %m3)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> %m2)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> %m1)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> %m16)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> %m15)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> %m14)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> %m13)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> %m12)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> %m11)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> %m10)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> %m9)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> %m7)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> %m6)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> %m5)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> %m3)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> %m2)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> %m1)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> %m7)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> %m6)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> %m5)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> %m3)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> %m2)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> %m1)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> %m16)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> %m15)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> %m14)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> %m13)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> %m12)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> %m11)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> %m10)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> %m9)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> %m7)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> %m6)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> %m5)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> %m3)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> %m2)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> %m1)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 164 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> %m32)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> %m16)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 326 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> %m64)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> %m32)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> %m16)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> %m8)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; SKX-LABEL: 'masked_store'
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> %m15)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> %m14)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> %m13)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> %m12)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> %m11)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> %m10)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> %m9)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> %m15)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> %m14)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> %m13)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> %m12)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> %m11)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> %m10)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> %m9)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> %m32)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> %m64)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> %m32)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> %m8)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
-  call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> undef)
-  call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> undef)
-  call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> undef)
-  call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> undef)
-  call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> undef)
-  call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> undef)
-  call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> undef)
-
-  call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> undef)
-  call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> undef)
-  call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> undef)
-  call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> undef)
-  call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> undef)
-  call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> undef)
-  call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> undef)
-  call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> undef)
-  call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> undef)
-  call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> undef)
-  call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> undef)
-  call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> undef)
-  call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> undef)
-  call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> undef)
-  call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> undef)
-
-  call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> undef)
-  call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> undef)
-  call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> undef)
-  call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> undef)
-  call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> undef)
-  call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> undef)
-  call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> undef)
-
-  call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> undef)
-  call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> undef)
-  call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> undef)
-  call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> undef)
-  call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> undef)
-  call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> undef)
-  call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> undef)
-  call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> undef)
-  call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> undef)
-  call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> undef)
-  call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> undef)
-  call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> undef)
-  call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> undef)
-  call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> undef)
-  call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> undef)
-
-  call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> undef)
-  call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> undef)
-  call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> undef)
-
-  call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> undef)
-  call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> undef)
-  call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> undef)
-  call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> undef)
+  call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> %m7)
+  call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> %m6)
+  call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> %m5)
+  call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> %m4)
+  call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> %m3)
+  call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> %m2)
+  call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> %m1)
+
+  call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> %m16)
+  call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> %m15)
+  call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> %m14)
+  call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> %m13)
+  call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> %m12)
+  call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> %m11)
+  call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> %m10)
+  call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> %m9)
+  call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> %m7)
+  call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> %m6)
+  call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> %m5)
+  call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> %m4)
+  call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> %m3)
+  call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> %m2)
+  call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> %m1)
+
+  call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> %m7)
+  call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> %m6)
+  call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> %m5)
+  call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> %m4)
+  call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> %m3)
+  call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> %m2)
+  call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> %m1)
+
+  call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> %m16)
+  call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> %m15)
+  call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> %m14)
+  call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> %m13)
+  call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> %m12)
+  call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> %m11)
+  call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> %m10)
+  call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> %m9)
+  call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> %m7)
+  call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> %m6)
+  call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> %m5)
+  call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> %m4)
+  call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> %m3)
+  call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> %m2)
+  call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> %m1)
+
+  call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> %m32)
+  call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> %m16)
+  call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> %m4)
+
+  call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> %m64)
+  call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> %m32)
+  call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> %m16)
+  call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> %m8)
 
   ret i32 0
 }
 
-define i32 @masked_gather() {
+define i32 @masked_gather(<1 x i1> %m1, <2 x i1> %m2, <4 x i1> %m4, <8 x i1> %m8, <16 x i1> %m16, <32 x i1> %m32, <64 x i1> %m64) {
 ; SSE2-LABEL: 'masked_gather'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 252 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 126 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 63 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 92 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 380 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 190 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 95 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; SSE42-LABEL: 'masked_gather'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 192 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 320 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; AVX1-LABEL: 'masked_gather'
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 98 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 194 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 97 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 78 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 82 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 322 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 161 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; AVX2-LABEL: 'masked_gather'
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 98 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 194 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 97 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 78 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 82 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 322 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 161 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; SKL-LABEL: 'masked_gather'
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 98 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 194 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 97 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 322 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 161 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; KNL-LABEL: 'masked_gather'
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 99 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 195 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 97 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 163 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 323 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 161 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; SKX-LABEL: 'masked_gather'
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 99 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 195 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 97 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 163 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 323 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 161 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
-  %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x double> undef)
-  %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x double> undef)
-  %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x double> undef)
-  %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x double> undef)
+  %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+  %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+  %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+  %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x double> undef)
 
-  %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x float> undef)
-  %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x float> undef)
-  %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x float> undef)
-  %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x float> undef)
+  %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+  %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+  %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+  %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x float> undef)
 
-  %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-  %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-  %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-  %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
+  %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+  %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+  %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+  %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
 
-  %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-  %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-  %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-  %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
+  %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+  %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+  %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+  %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
 
-  %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-  %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-  %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-  %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
+  %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+  %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+  %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+  %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
 
-  %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-  %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-  %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-  %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+  %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+  %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+  %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+  %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 
   ret i32 0
 }
 
-define i32 @masked_scatter() {
+define i32 @masked_scatter(<1 x i1> %m1, <2 x i1> %m2, <4 x i1> %m4, <8 x i1> %m8, <16 x i1> %m16, <32 x i1> %m32, <64 x i1> %m64) {
 ; SSE2-LABEL: 'masked_scatter'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 252 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 126 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 63 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 92 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 380 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> %m64)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 190 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 95 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; SSE42-LABEL: 'masked_scatter'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 192 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 320 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> %m64)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; AVX-LABEL: 'masked_scatter'
-; AVX-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 98 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 194 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 97 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 78 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 82 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 322 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> %m64)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 161 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; KNL-LABEL: 'masked_scatter'
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 99 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 195 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 97 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 163 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 323 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> %m64)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 161 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; SKX-LABEL: 'masked_scatter'
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 99 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 195 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 97 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 163 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 323 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> %m64)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 161 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
-  call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-  call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-  call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
+  call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+  call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+  call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
 
-  call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-  call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-  call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
+  call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+  call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+  call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
 
-  call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-  call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-  call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
+  call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+  call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+  call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
 
-  call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-  call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-  call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
+  call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+  call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+  call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
 
-  call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-  call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-  call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
+  call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+  call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+  call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
 
-  call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> undef)
-  call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-  call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-  call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
+  call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> %m64)
+  call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+  call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+  call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
 
   ret i32 0
 }
 
-define i32 @masked_expandload() {
+define i32 @masked_expandload(<1 x i1> %m1, <2 x i1> %m2, <4 x i1> %m4, <8 x i1> %m8, <16 x i1> %m16, <32 x i1> %m32, <64 x i1> %m64) {
 ; SSE2-LABEL: 'masked_expandload'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> undef, <8 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> undef, <4 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> undef, <2 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> undef, <1 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> undef, <16 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> undef, <8 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> undef, <4 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> undef, <2 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> undef, <8 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> undef, <4 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> undef, <2 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> undef, <1 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> undef, <16 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> undef, <8 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> undef, <4 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> undef, <2 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> undef, <32 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> undef, <16 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> undef, <8 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> undef, <4 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 188 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> undef, <64 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 94 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> undef, <32 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> undef, <16 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> undef, <8 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> %m8, <8 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> %m4, <4 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> %m2, <2 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> %m1, <1 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> %m16, <16 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> %m8, <8 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> %m4, <4 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> %m2, <2 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> %m8, <8 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> %m4, <4 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> %m2, <2 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> %m1, <1 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> %m16, <16 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> %m8, <8 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> %m4, <4 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> %m2, <2 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> %m32, <32 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> %m16, <16 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> %m8, <8 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> %m4, <4 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 192 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> %m64, <64 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> %m32, <32 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> %m16, <16 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> %m8, <8 x i8> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; SSE42-LABEL: 'masked_expandload'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> undef, <8 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> undef, <4 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> undef, <2 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> undef, <1 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> undef, <16 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> undef, <8 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> undef, <4 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> undef, <2 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> undef, <8 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> undef, <4 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> undef, <2 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> undef, <1 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> undef, <16 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> undef, <8 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> undef, <4 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> undef, <2 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> undef, <32 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> undef, <16 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> undef, <8 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> undef, <4 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> undef, <64 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> undef, <32 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> undef, <16 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> undef, <8 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> %m8, <8 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> %m4, <4 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> %m2, <2 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> %m1, <1 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> %m16, <16 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> %m8, <8 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> %m4, <4 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> %m2, <2 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> %m8, <8 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> %m4, <4 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> %m2, <2 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> %m1, <1 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> %m16, <16 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> %m8, <8 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> %m4, <4 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> %m2, <2 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> %m32, <32 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> %m16, <16 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> %m8, <8 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> %m4, <4 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> %m64, <64 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> %m32, <32 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> %m16, <16 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> %m8, <8 x i8> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
-; AVX-LABEL: 'masked_expandload'
-; AVX-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> undef, <8 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> undef, <4 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> undef, <2 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> undef, <1 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> undef, <16 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> undef, <8 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> undef, <4 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> undef, <2 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> undef, <8 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> undef, <4 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> undef, <2 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> undef, <1 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> undef, <16 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> undef, <8 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> undef, <4 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> undef, <2 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> undef, <32 x i16> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> undef, <16 x i16> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> undef, <8 x i16> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> undef, <4 x i16> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 130 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> undef, <64 x i8> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> undef, <32 x i8> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> undef, <16 x i8> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> undef, <8 x i8> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
+; AVX1-LABEL: 'masked_expandload'
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> %m8, <8 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> %m4, <4 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> %m2, <2 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> %m1, <1 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> %m16, <16 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> %m8, <8 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> %m4, <4 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> %m2, <2 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> %m8, <8 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> %m4, <4 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> %m2, <2 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> %m1, <1 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> %m16, <16 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> %m8, <8 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> %m4, <4 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> %m2, <2 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> %m32, <32 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> %m16, <16 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> %m8, <8 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> %m4, <4 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 134 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> %m64, <64 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> %m32, <32 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> %m16, <16 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> %m8, <8 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
+;
+; AVX2-LABEL: 'masked_expandload'
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> %m8, <8 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> %m4, <4 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> %m2, <2 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> %m1, <1 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> %m16, <16 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> %m8, <8 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> %m4, <4 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> %m2, <2 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> %m8, <8 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> %m4, <4 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> %m2, <2 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> %m1, <1 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> %m16, <16 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> %m8, <8 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> %m4, <4 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> %m2, <2 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> %m32, <32 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> %m16, <16 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> %m8, <8 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> %m4, <4 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> %m64, <64 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> %m32, <32 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> %m16, <16 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> %m8, <8 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
+;
+; SKL-LABEL: 'masked_expandload'
+; SKL-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> %m8, <8 x double> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> %m4, <4 x double> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> %m2, <2 x double> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> %m1, <1 x double> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> %m16, <16 x float> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> %m8, <8 x float> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> %m4, <4 x float> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> %m2, <2 x float> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> %m8, <8 x i64> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> %m4, <4 x i64> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> %m2, <2 x i64> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> %m1, <1 x i64> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> %m16, <16 x i32> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> %m8, <8 x i32> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> %m4, <4 x i32> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> %m2, <2 x i32> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> %m32, <32 x i16> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> %m16, <16 x i16> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> %m8, <8 x i16> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> %m4, <4 x i16> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> %m64, <64 x i8> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> %m32, <32 x i8> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> %m16, <16 x i8> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> %m8, <8 x i8> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; AVX512-LABEL: 'masked_expandload'
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> undef, <8 x double> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> undef, <4 x double> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> undef, <2 x double> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> undef, <1 x double> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> undef, <16 x float> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> undef, <8 x float> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> undef, <4 x float> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> undef, <2 x float> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> undef, <8 x i64> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> undef, <4 x i64> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> undef, <2 x i64> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> undef, <1 x i64> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> undef, <16 x i32> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> undef, <8 x i32> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> undef, <4 x i32> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> undef, <2 x i32> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> undef, <32 x i16> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> undef, <16 x i16> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> undef, <8 x i16> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> undef, <4 x i16> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 131 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> undef, <64 x i8> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> undef, <32 x i8> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> undef, <16 x i8> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> undef, <8 x i8> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> %m8, <8 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> %m4, <4 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> %m2, <2 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> %m1, <1 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> %m16, <16 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> %m8, <8 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> %m4, <4 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> %m2, <2 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> %m8, <8 x i64> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> %m4, <4 x i64> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> %m2, <2 x i64> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> %m1, <1 x i64> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> %m16, <16 x i32> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> %m8, <8 x i32> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> %m4, <4 x i32> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> %m2, <2 x i32> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 99 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> %m32, <32 x i16> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> %m16, <16 x i16> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> %m8, <8 x i16> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> %m4, <4 x i16> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 195 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> %m64, <64 x i8> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 97 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> %m32, <32 x i8> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> %m16, <16 x i8> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> %m8, <8 x i8> undef)
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
-  %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> undef, <8 x double> undef)
-  %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> undef, <4 x double> undef)
-  %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> undef, <2 x double> undef)
-  %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> undef, <1 x double> undef)
+  %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> %m8, <8 x double> undef)
+  %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> %m4, <4 x double> undef)
+  %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> %m2, <2 x double> undef)
+  %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> %m1, <1 x double> undef)
 
-  %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> undef, <16 x float> undef)
-  %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> undef, <8 x float> undef)
-  %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> undef, <4 x float> undef)
-  %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> undef, <2 x float> undef)
+  %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> %m16, <16 x float> undef)
+  %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> %m8, <8 x float> undef)
+  %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> %m4, <4 x float> undef)
+  %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> %m2, <2 x float> undef)
 
-  %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> undef, <8 x i64> undef)
-  %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> undef, <4 x i64> undef)
-  %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> undef, <2 x i64> undef)
-  %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> undef, <1 x i64> undef)
+  %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> %m8, <8 x i64> undef)
+  %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> %m4, <4 x i64> undef)
+  %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> %m2, <2 x i64> undef)
+  %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> %m1, <1 x i64> undef)
 
-  %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> undef, <16 x i32> undef)
-  %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> undef, <8 x i32> undef)
-  %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> undef, <4 x i32> undef)
-  %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> undef, <2 x i32> undef)
+  %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> %m16, <16 x i32> undef)
+  %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> %m8, <8 x i32> undef)
+  %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> %m4, <4 x i32> undef)
+  %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> %m2, <2 x i32> undef)
 
-  %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> undef, <32 x i16> undef)
-  %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> undef, <16 x i16> undef)
-  %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> undef, <8 x i16> undef)
-  %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> undef, <4 x i16> undef)
+  %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> %m32, <32 x i16> undef)
+  %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> %m16, <16 x i16> undef)
+  %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> %m8, <8 x i16> undef)
+  %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> %m4, <4 x i16> undef)
 
-  %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> undef, <64 x i8> undef)
-  %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> undef, <32 x i8> undef)
-  %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> undef, <16 x i8> undef)
-  %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> undef, <8 x i8> undef)
+  %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> %m64, <64 x i8> undef)
+  %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> %m32, <32 x i8> undef)
+  %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> %m16, <16 x i8> undef)
+  %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> %m8, <8 x i8> undef)
 
   ret i32 0
 }
 
-define i32 @masked_compressstore() {
+define i32 @masked_compressstore(<1 x i1> %m1, <2 x i1> %m2, <4 x i1> %m4, <8 x i1> %m8, <16 x i1> %m16, <32 x i1> %m32, <64 x i1> %m64) {
 ; SSE2-LABEL: 'masked_compressstore'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 192 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> %m1)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> %m1)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> %m32)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 192 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> %m64)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> %m32)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> %m8)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; SSE42-LABEL: 'masked_compressstore'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> %m1)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> %m1)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> %m32)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> %m64)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> %m32)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> %m8)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; AVX1-LABEL: 'masked_compressstore'
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 134 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> %m8)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> %m4)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> %m2)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> %m1)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> %m16)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> %m8)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> %m4)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> %m2)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> %m8)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> %m4)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> %m2)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> %m1)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> %m16)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> %m8)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> %m4)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> %m2)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> %m32)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> %m16)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> %m8)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> %m4)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 134 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> %m64)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> %m32)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> %m16)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> %m8)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; AVX2-LABEL: 'masked_compressstore'
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> %m8)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> %m4)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> %m2)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> %m1)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> %m16)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> %m8)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> %m4)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> %m2)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> %m8)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> %m4)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> %m2)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> %m1)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> %m16)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> %m8)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> %m4)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> %m2)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> %m32)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> %m16)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> %m8)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> %m4)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> %m64)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> %m32)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> %m16)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> %m8)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; SKL-LABEL: 'masked_compressstore'
-; SKL-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> %m8)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> %m4)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> %m2)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> %m1)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> %m16)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> %m8)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> %m4)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> %m2)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> %m8)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> %m4)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> %m2)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> %m1)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> %m16)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> %m8)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> %m4)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> %m2)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> %m32)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> %m16)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> %m8)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> %m4)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> %m64)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> %m32)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> %m16)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> %m8)
 ; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; AVX512-LABEL: 'masked_compressstore'
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 99 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 195 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 97 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> %m8)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> %m4)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> %m2)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> %m1)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> %m16)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> %m8)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> %m4)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> %m2)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> %m8)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> %m4)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> %m2)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> %m1)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> %m16)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> %m8)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> %m4)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> %m2)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 99 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> %m32)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> %m16)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> %m8)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> %m4)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 195 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> %m64)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 97 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> %m32)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> %m16)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> %m8)
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
-  call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> undef)
-  call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> undef)
-  call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> undef)
-  call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> undef)
-
-  call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> undef)
-  call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> undef)
-  call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> undef)
-  call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> undef)
-
-  call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> undef)
-  call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> undef)
-  call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> undef)
-  call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> undef)
-
-  call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> undef)
-  call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> undef)
-  call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> undef)
-  call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> undef)
-
-  call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> undef)
-  call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> undef)
-  call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> undef)
-  call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> undef)
-
-  call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> undef)
-  call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> undef)
-  call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> undef)
-  call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> undef)
+  call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> %m8)
+  call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> %m4)
+  call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> %m2)
+  call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> %m1)
+
+  call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> %m16)
+  call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> %m8)
+  call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> %m4)
+  call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> %m2)
+
+  call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> %m8)
+  call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> %m4)
+  call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> %m2)
+  call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> %m1)
+
+  call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> %m16)
+  call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> %m8)
+  call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> %m4)
+  call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> %m2)
+
+  call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> %m32)
+  call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> %m16)
+  call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> %m8)
+  call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> %m4)
+
+  call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> %m64)
+  call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> %m32)
+  call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> %m16)
+  call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> %m8)
 
   ret i32 0
 }
diff --git a/llvm/test/Analysis/CostModel/X86/masked-intrinsic-sizelatency.ll b/llvm/test/Analysis/CostModel/X86/masked-intrinsic-sizelatency.ll
index 3ebd9cc41c09..b460df341b44 100644
--- a/llvm/test/Analysis/CostModel/X86/masked-intrinsic-sizelatency.ll
+++ b/llvm/test/Analysis/CostModel/X86/masked-intrinsic-sizelatency.ll
@@ -8,1454 +8,1508 @@
 ; RUN: opt < %s -S -mtriple=x86_64-apple-darwin -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=size-latency -mcpu=knl      | FileCheck %s --check-prefixes=AVX512,KNL
 ; RUN: opt < %s -S -mtriple=x86_64-apple-darwin -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=size-latency -mcpu=skx      | FileCheck %s --check-prefixes=AVX512,SKX
 
-define i32 @masked_load() {
+define i32 @masked_load(<1 x i1> %m1, <2 x i1> %m2, <3 x i1> %m3, <4 x i1> %m4, <5 x i1> %m5, <6 x i1> %m6, <7 x i1> %m7, <8 x i1> %m8, <9 x i1> %m9, <10 x i1> %m10, <11 x i1> %m11, <12 x i1> %m12, <13 x i1> %m13, <14 x i1> %m14, <15 x i1> %m15, <16 x i1> %m16, <32 x i1> %m32, <64 x i1> %m64) {
 ; SSE2-LABEL: 'masked_load'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 91 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 107 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 101 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 95 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 89 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 83 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 77 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 53 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 190 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 95 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 440 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 220 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 110 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 91 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 107 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 101 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 95 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 89 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 83 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 77 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 53 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 190 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 95 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 440 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 220 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 110 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; SSE42-LABEL: 'masked_load'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 75 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 70 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 55 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 320 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 75 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 70 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 55 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 320 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; AVX-LABEL: 'masked_load'
-; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 163 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 324 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i64> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i64> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i64> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i64> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 163 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 324 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; KNL-LABEL: 'masked_load'
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 164 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 326 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 164 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 326 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; SKX-LABEL: 'masked_load'
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
-  %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x double> undef)
-  %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x double> undef)
-  %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x double> undef)
-  %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x double> undef)
-  %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x double> undef)
-  %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x double> undef)
-  %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x double> undef)
-  %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x double> undef)
-
-  %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x float> undef)
-  %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x float> undef)
-  %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x float> undef)
-  %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x float> undef)
-  %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x float> undef)
-  %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x float> undef)
-  %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x float> undef)
-  %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x float> undef)
-  %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x float> undef)
-  %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x float> undef)
-  %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x float> undef)
-  %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x float> undef)
-  %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x float> undef)
-  %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x float> undef)
-  %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x float> undef)
-  %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x float> undef)
-
-  %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-  %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i64> undef)
-  %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i64> undef)
-  %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i64> undef)
-  %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-  %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i64> undef)
-  %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-  %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-
-  %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-  %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> undef, <15 x i32> undef)
-  %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> undef, <14 x i32> undef)
-  %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> undef, <13 x i32> undef)
-  %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> undef, <12 x i32> undef)
-  %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> undef, <11 x i32> undef)
-  %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> undef, <10 x i32> undef)
-  %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> undef, <9 x i32> undef)
-  %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-  %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> undef, <7 x i32> undef)
-  %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> undef, <6 x i32> undef)
-  %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> undef, <5 x i32> undef)
-  %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-  %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> undef, <3 x i32> undef)
-  %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-  %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> undef, <1 x i32> undef)
-
-  %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-  %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-  %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-  %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-
-  %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-  %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-  %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-  %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+  %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+  %V7F64 = call <7 x double> @llvm.masked.load.v7f64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x double> undef)
+  %V6F64 = call <6 x double> @llvm.masked.load.v6f64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x double> undef)
+  %V5F64 = call <5 x double> @llvm.masked.load.v5f64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x double> undef)
+  %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+  %V3F64 = call <3 x double> @llvm.masked.load.v3f64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x double> undef)
+  %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+  %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+
+  %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+  %V15F32 = call <15 x float> @llvm.masked.load.v15f32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x float> undef)
+  %V14F32 = call <14 x float> @llvm.masked.load.v14f32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x float> undef)
+  %V13F32 = call <13 x float> @llvm.masked.load.v13f32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x float> undef)
+  %V12F32 = call <12 x float> @llvm.masked.load.v12f32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x float> undef)
+  %V11F32 = call <11 x float> @llvm.masked.load.v11f32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x float> undef)
+  %V10F32 = call <10 x float> @llvm.masked.load.v10f32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x float> undef)
+  %V9F32 = call <9 x float> @llvm.masked.load.v9f32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x float> undef)
+  %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+  %V7F32 = call <7 x float> @llvm.masked.load.v7f32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x float> undef)
+  %V6F32 = call <6 x float> @llvm.masked.load.v6f32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x float> undef)
+  %V5F32 = call <5 x float> @llvm.masked.load.v5f32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x float> undef)
+  %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+  %V3F32 = call <3 x float> @llvm.masked.load.v3f32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x float> undef)
+  %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+  %V1F32 = call <1 x float> @llvm.masked.load.v1f32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x float> undef)
+
+  %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+  %V7I64 = call <7 x i64> @llvm.masked.load.v7i64.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i64> undef)
+  %V6I64 = call <6 x i64> @llvm.masked.load.v6i64.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i64> undef)
+  %V5I64 = call <5 x i64> @llvm.masked.load.v5i64.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i64> undef)
+  %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+  %V3I64 = call <3 x i64> @llvm.masked.load.v3i64.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i64> undef)
+  %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+  %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+
+  %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+  %V15I32 = call <15 x i32> @llvm.masked.load.v15i32.p0(ptr undef, i32 1, <15 x i1> %m15, <15 x i32> undef)
+  %V14I32 = call <14 x i32> @llvm.masked.load.v14i32.p0(ptr undef, i32 1, <14 x i1> %m14, <14 x i32> undef)
+  %V13I32 = call <13 x i32> @llvm.masked.load.v13i32.p0(ptr undef, i32 1, <13 x i1> %m13, <13 x i32> undef)
+  %V12I32 = call <12 x i32> @llvm.masked.load.v12i32.p0(ptr undef, i32 1, <12 x i1> %m12, <12 x i32> undef)
+  %V11I32 = call <11 x i32> @llvm.masked.load.v11i32.p0(ptr undef, i32 1, <11 x i1> %m11, <11 x i32> undef)
+  %V10I32 = call <10 x i32> @llvm.masked.load.v10i32.p0(ptr undef, i32 1, <10 x i1> %m10, <10 x i32> undef)
+  %V9I32 = call <9 x i32> @llvm.masked.load.v9i32.p0(ptr undef, i32 1, <9 x i1> %m9, <9 x i32> undef)
+  %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+  %V7I32 = call <7 x i32> @llvm.masked.load.v7i32.p0(ptr undef, i32 1, <7 x i1> %m7, <7 x i32> undef)
+  %V6I32 = call <6 x i32> @llvm.masked.load.v6i32.p0(ptr undef, i32 1, <6 x i1> %m6, <6 x i32> undef)
+  %V5I32 = call <5 x i32> @llvm.masked.load.v5i32.p0(ptr undef, i32 1, <5 x i1> %m5, <5 x i32> undef)
+  %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+  %V3I32 = call <3 x i32> @llvm.masked.load.v3i32.p0(ptr undef, i32 1, <3 x i1> %m3, <3 x i32> undef)
+  %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0(ptr undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+  %V1I32 = call <1 x i32> @llvm.masked.load.v1i32.p0(ptr undef, i32 1, <1 x i1> %m1, <1 x i32> undef)
+
+  %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+  %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+  %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+  %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+
+  %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0(ptr undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+  %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0(ptr undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+  %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0(ptr undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+  %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 
   ret i32 0
 }
 
-define i32 @masked_store() {
+define i32 @masked_store(<1 x i1> %m1, <2 x i1> %m2, <3 x i1> %m3, <4 x i1> %m4, <5 x i1> %m5, <6 x i1> %m6, <7 x i1> %m7, <8 x i1> %m8, <9 x i1> %m9, <10 x i1> %m10, <11 x i1> %m11, <12 x i1> %m12, <13 x i1> %m13, <14 x i1> %m14, <15 x i1> %m15, <16 x i1> %m16, <32 x i1> %m32, <64 x i1> %m64) {
 ; SSE2-LABEL: 'masked_store'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 91 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 85 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 79 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 73 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 62 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 107 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 93 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 73 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 59 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 53 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 190 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 95 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 440 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 220 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 110 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 91 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 85 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> %m15)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 79 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> %m14)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 73 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> %m13)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> %m12)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 62 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> %m11)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> %m10)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> %m9)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 107 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> %m15)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 93 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> %m14)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> %m13)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> %m12)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 73 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> %m11)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> %m10)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 59 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> %m9)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 53 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 190 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> %m32)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 95 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 440 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> %m64)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 220 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> %m32)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 110 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> %m8)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; SSE42-LABEL: 'masked_store'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 75 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 70 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 55 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 320 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> %m15)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> %m14)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> %m13)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> %m12)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> %m11)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> %m10)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> %m9)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 75 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> %m15)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 70 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> %m14)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> %m13)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> %m12)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 55 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> %m11)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> %m10)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> %m9)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> %m32)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 320 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> %m64)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> %m32)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> %m8)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; AVX-LABEL: 'masked_store'
-; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 163 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 324 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> %m8)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> %m7)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> %m6)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> %m5)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> %m4)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> %m3)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> %m2)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> %m1)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> %m16)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> %m15)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> %m14)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> %m13)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> %m12)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> %m11)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> %m10)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> %m9)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> %m8)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> %m7)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> %m6)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> %m5)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> %m4)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> %m3)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> %m2)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> %m1)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> %m8)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> %m7)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> %m6)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> %m5)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> %m4)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> %m3)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> %m2)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> %m1)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> %m16)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> %m15)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> %m14)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> %m13)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> %m12)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> %m11)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> %m10)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> %m9)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> %m8)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> %m7)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> %m6)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> %m5)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> %m4)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> %m3)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> %m2)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> %m1)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 163 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> %m32)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> %m16)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> %m8)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> %m4)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 324 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> %m64)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> %m32)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> %m16)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> %m8)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; KNL-LABEL: 'masked_store'
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 164 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 326 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> %m7)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> %m6)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> %m5)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> %m3)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> %m2)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> %m1)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> %m16)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> %m15)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> %m14)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> %m13)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> %m12)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> %m11)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> %m10)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> %m9)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> %m7)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> %m6)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> %m5)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> %m3)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> %m2)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> %m1)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> %m7)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> %m6)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> %m5)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> %m3)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> %m2)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> %m1)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> %m16)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> %m15)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> %m14)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> %m13)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> %m12)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> %m11)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> %m10)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> %m9)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> %m7)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> %m6)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> %m5)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> %m3)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> %m2)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> %m1)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 164 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> %m32)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> %m16)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 326 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> %m64)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> %m32)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> %m16)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> %m8)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; SKX-LABEL: 'masked_store'
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> %m15)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> %m14)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> %m13)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> %m12)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> %m11)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> %m10)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> %m9)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> %m15)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> %m14)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> %m13)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> %m12)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> %m11)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> %m10)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> %m9)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> %m7)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> %m6)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> %m5)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> %m3)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> %m2)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> %m1)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> %m32)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> %m64)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> %m32)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> %m16)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> %m8)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
-  call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> undef)
-  call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> undef)
-  call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> undef)
-  call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> undef)
-  call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> undef)
-  call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> undef)
-  call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> undef)
-
-  call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> undef)
-  call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> undef)
-  call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> undef)
-  call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> undef)
-  call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> undef)
-  call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> undef)
-  call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> undef)
-  call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> undef)
-  call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> undef)
-  call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> undef)
-  call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> undef)
-  call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> undef)
-  call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> undef)
-  call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> undef)
-  call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> undef)
-
-  call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> undef)
-  call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> undef)
-  call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> undef)
-  call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> undef)
-  call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> undef)
-  call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> undef)
-  call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> undef)
-
-  call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> undef)
-  call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> undef)
-  call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> undef)
-  call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> undef)
-  call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> undef)
-  call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> undef)
-  call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> undef)
-  call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> undef)
-  call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> undef)
-  call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> undef)
-  call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> undef)
-  call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> undef)
-  call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> undef)
-  call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> undef)
-  call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> undef)
-
-  call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> undef)
-  call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> undef)
-  call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> undef)
-
-  call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> undef)
-  call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> undef)
-  call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> undef)
-  call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> undef)
+  call void @llvm.masked.store.v8f64.p0(<8 x double> undef, ptr undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.store.v7f64.p0(<7 x double> undef, ptr undef, i32 1, <7 x i1> %m7)
+  call void @llvm.masked.store.v6f64.p0(<6 x double> undef, ptr undef, i32 1, <6 x i1> %m6)
+  call void @llvm.masked.store.v5f64.p0(<5 x double> undef, ptr undef, i32 1, <5 x i1> %m5)
+  call void @llvm.masked.store.v4f64.p0(<4 x double> undef, ptr undef, i32 1, <4 x i1> %m4)
+  call void @llvm.masked.store.v3f64.p0(<3 x double> undef, ptr undef, i32 1, <3 x i1> %m3)
+  call void @llvm.masked.store.v2f64.p0(<2 x double> undef, ptr undef, i32 1, <2 x i1> %m2)
+  call void @llvm.masked.store.v1f64.p0(<1 x double> undef, ptr undef, i32 1, <1 x i1> %m1)
+
+  call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr undef, i32 1, <16 x i1> %m16)
+  call void @llvm.masked.store.v15f32.p0(<15 x float> undef, ptr undef, i32 1, <15 x i1> %m15)
+  call void @llvm.masked.store.v14f32.p0(<14 x float> undef, ptr undef, i32 1, <14 x i1> %m14)
+  call void @llvm.masked.store.v13f32.p0(<13 x float> undef, ptr undef, i32 1, <13 x i1> %m13)
+  call void @llvm.masked.store.v12f32.p0(<12 x float> undef, ptr undef, i32 1, <12 x i1> %m12)
+  call void @llvm.masked.store.v11f32.p0(<11 x float> undef, ptr undef, i32 1, <11 x i1> %m11)
+  call void @llvm.masked.store.v10f32.p0(<10 x float> undef, ptr undef, i32 1, <10 x i1> %m10)
+  call void @llvm.masked.store.v9f32.p0(<9 x float> undef, ptr undef, i32 1, <9 x i1> %m9)
+  call void @llvm.masked.store.v8f32.p0(<8 x float> undef, ptr undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.store.v7f32.p0(<7 x float> undef, ptr undef, i32 1, <7 x i1> %m7)
+  call void @llvm.masked.store.v6f32.p0(<6 x float> undef, ptr undef, i32 1, <6 x i1> %m6)
+  call void @llvm.masked.store.v5f32.p0(<5 x float> undef, ptr undef, i32 1, <5 x i1> %m5)
+  call void @llvm.masked.store.v4f32.p0(<4 x float> undef, ptr undef, i32 1, <4 x i1> %m4)
+  call void @llvm.masked.store.v3f32.p0(<3 x float> undef, ptr undef, i32 1, <3 x i1> %m3)
+  call void @llvm.masked.store.v2f32.p0(<2 x float> undef, ptr undef, i32 1, <2 x i1> %m2)
+  call void @llvm.masked.store.v1f32.p0(<1 x float> undef, ptr undef, i32 1, <1 x i1> %m1)
+
+  call void @llvm.masked.store.v8i64.p0(<8 x i64> undef, ptr undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.store.v7i64.p0(<7 x i64> undef, ptr undef, i32 1, <7 x i1> %m7)
+  call void @llvm.masked.store.v6i64.p0(<6 x i64> undef, ptr undef, i32 1, <6 x i1> %m6)
+  call void @llvm.masked.store.v5i64.p0(<5 x i64> undef, ptr undef, i32 1, <5 x i1> %m5)
+  call void @llvm.masked.store.v4i64.p0(<4 x i64> undef, ptr undef, i32 1, <4 x i1> %m4)
+  call void @llvm.masked.store.v3i64.p0(<3 x i64> undef, ptr undef, i32 1, <3 x i1> %m3)
+  call void @llvm.masked.store.v2i64.p0(<2 x i64> undef, ptr undef, i32 1, <2 x i1> %m2)
+  call void @llvm.masked.store.v1i64.p0(<1 x i64> undef, ptr undef, i32 1, <1 x i1> %m1)
+
+  call void @llvm.masked.store.v16i32.p0(<16 x i32> undef, ptr undef, i32 1, <16 x i1> %m16)
+  call void @llvm.masked.store.v15i32.p0(<15 x i32> undef, ptr undef, i32 1, <15 x i1> %m15)
+  call void @llvm.masked.store.v14i32.p0(<14 x i32> undef, ptr undef, i32 1, <14 x i1> %m14)
+  call void @llvm.masked.store.v13i32.p0(<13 x i32> undef, ptr undef, i32 1, <13 x i1> %m13)
+  call void @llvm.masked.store.v12i32.p0(<12 x i32> undef, ptr undef, i32 1, <12 x i1> %m12)
+  call void @llvm.masked.store.v11i32.p0(<11 x i32> undef, ptr undef, i32 1, <11 x i1> %m11)
+  call void @llvm.masked.store.v10i32.p0(<10 x i32> undef, ptr undef, i32 1, <10 x i1> %m10)
+  call void @llvm.masked.store.v9i32.p0(<9 x i32> undef, ptr undef, i32 1, <9 x i1> %m9)
+  call void @llvm.masked.store.v8i32.p0(<8 x i32> undef, ptr undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.store.v7i32.p0(<7 x i32> undef, ptr undef, i32 1, <7 x i1> %m7)
+  call void @llvm.masked.store.v6i32.p0(<6 x i32> undef, ptr undef, i32 1, <6 x i1> %m6)
+  call void @llvm.masked.store.v5i32.p0(<5 x i32> undef, ptr undef, i32 1, <5 x i1> %m5)
+  call void @llvm.masked.store.v4i32.p0(<4 x i32> undef, ptr undef, i32 1, <4 x i1> %m4)
+  call void @llvm.masked.store.v3i32.p0(<3 x i32> undef, ptr undef, i32 1, <3 x i1> %m3)
+  call void @llvm.masked.store.v2i32.p0(<2 x i32> undef, ptr undef, i32 1, <2 x i1> %m2)
+  call void @llvm.masked.store.v1i32.p0(<1 x i32> undef, ptr undef, i32 1, <1 x i1> %m1)
+
+  call void @llvm.masked.store.v32i16.p0(<32 x i16> undef, ptr undef, i32 1, <32 x i1> %m32)
+  call void @llvm.masked.store.v16i16.p0(<16 x i16> undef, ptr undef, i32 1, <16 x i1> %m16)
+  call void @llvm.masked.store.v8i16.p0(<8 x i16> undef, ptr undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.store.v4i16.p0(<4 x i16> undef, ptr undef, i32 1, <4 x i1> %m4)
+
+  call void @llvm.masked.store.v64i8.p0(<64 x i8> undef, ptr undef, i32 1, <64 x i1> %m64)
+  call void @llvm.masked.store.v32i8.p0(<32 x i8> undef, ptr undef, i32 1, <32 x i1> %m32)
+  call void @llvm.masked.store.v16i8.p0(<16 x i8> undef, ptr undef, i32 1, <16 x i1> %m16)
+  call void @llvm.masked.store.v8i8.p0(<8 x i8> undef, ptr undef, i32 1, <8 x i1> %m8)
 
   ret i32 0
 }
 
-define i32 @masked_gather() {
+define i32 @masked_gather(<1 x i1> %m1, <2 x i1> %m2, <4 x i1> %m4, <8 x i1> %m8, <16 x i1> %m16, <32 x i1> %m32, <64 x i1> %m64) {
 ; SSE2-LABEL: 'masked_gather'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 252 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 126 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 63 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 92 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 380 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 190 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 95 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; SSE42-LABEL: 'masked_gather'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 192 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 320 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; AVX1-LABEL: 'masked_gather'
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 98 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 194 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 97 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 78 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 82 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 322 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 161 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; AVX2-LABEL: 'masked_gather'
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 98 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 194 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 97 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 78 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 82 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 322 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 161 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; SKL-LABEL: 'masked_gather'
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 98 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 194 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 97 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 322 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 161 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; KNL-LABEL: 'masked_gather'
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 99 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 195 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 97 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 163 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 323 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 161 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; SKX-LABEL: 'masked_gather'
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x double> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 99 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 195 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 97 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x double> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 163 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 323 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 161 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
-  %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x double> undef)
-  %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x double> undef)
-  %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x double> undef)
-  %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x double> undef)
+  %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x double> undef)
+  %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x double> undef)
+  %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x double> undef)
+  %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x double> undef)
 
-  %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x float> undef)
-  %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x float> undef)
-  %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x float> undef)
-  %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x float> undef)
+  %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x float> undef)
+  %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x float> undef)
+  %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x float> undef)
+  %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x float> undef)
 
-  %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
-  %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
-  %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
-  %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
+  %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i64> undef)
+  %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i64> undef)
+  %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i64> undef)
+  %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0(<1 x ptr> undef, i32 1, <1 x i1> %m1, <1 x i64> undef)
 
-  %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
-  %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
-  %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-  %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
+  %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i32> undef)
+  %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i32> undef)
+  %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i32> undef)
+  %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0(<2 x ptr> undef, i32 1, <2 x i1> %m2, <2 x i32> undef)
 
-  %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
-  %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
-  %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-  %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
+  %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i16> undef)
+  %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i16> undef)
+  %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i16> undef)
+  %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0(<4 x ptr> undef, i32 1, <4 x i1> %m4, <4 x i16> undef)
 
-  %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
-  %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
-  %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-  %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+  %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0(<64 x ptr> undef, i32 1, <64 x i1> %m64, <64 x i8> undef)
+  %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> undef, i32 1, <32 x i1> %m32, <32 x i8> undef)
+  %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> undef, i32 1, <16 x i1> %m16, <16 x i8> undef)
+  %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0(<8 x ptr> undef, i32 1, <8 x i1> %m8, <8 x i8> undef)
 
   ret i32 0
 }
 
-define i32 @masked_scatter() {
+define i32 @masked_scatter(<1 x i1> %m1, <2 x i1> %m2, <4 x i1> %m4, <8 x i1> %m8, <16 x i1> %m16, <32 x i1> %m32, <64 x i1> %m64) {
 ; SSE2-LABEL: 'masked_scatter'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 252 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 126 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 63 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 92 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 380 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> %m64)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 190 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 95 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; SSE42-LABEL: 'masked_scatter'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 192 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 320 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> %m64)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; AVX-LABEL: 'masked_scatter'
-; AVX-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 98 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 194 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 97 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 78 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 82 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 322 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> %m64)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 161 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; KNL-LABEL: 'masked_scatter'
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 99 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 195 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 97 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 163 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 323 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> %m64)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 161 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; SKX-LABEL: 'masked_scatter'
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 99 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 195 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 97 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 163 for instruction: call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 81 for instruction: call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 323 for instruction: call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> %m64)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 161 for instruction: call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
-  call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-  call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-  call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
+  call void @llvm.masked.scatter.v8f64.v8p0(<8 x double> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+  call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+  call void @llvm.masked.scatter.v1f64.v1p0(<1 x double> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
 
-  call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-  call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-  call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
+  call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+  call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+  call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
 
-  call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-  call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
-  call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> undef)
+  call void @llvm.masked.scatter.v8i64.v8p0(<8 x i64> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.scatter.v4i64.v4p0(<4 x i64> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+  call void @llvm.masked.scatter.v2i64.v2p0(<2 x i64> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
+  call void @llvm.masked.scatter.v1i64.v1p0(<1 x i64> undef, <1 x ptr> undef, i32 1, <1 x i1> %m1)
 
-  call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-  call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
-  call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> undef)
+  call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+  call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
+  call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> undef, <2 x ptr> undef, i32 1, <2 x i1> %m2)
 
-  call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-  call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-  call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
-  call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> undef)
+  call void @llvm.masked.scatter.v32i16.v32p0(<32 x i16> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+  call void @llvm.masked.scatter.v16i16.v16p0(<16 x i16> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+  call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
+  call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> undef, <4 x ptr> undef, i32 1, <4 x i1> %m4)
 
-  call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> undef)
-  call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> undef)
-  call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> undef)
-  call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> undef)
+  call void @llvm.masked.scatter.v64i8.v64p0(<64 x i8> undef, <64 x ptr> undef, i32 1, <64 x i1> %m64)
+  call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> undef, <32 x ptr> undef, i32 1, <32 x i1> %m32)
+  call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> undef, <16 x ptr> undef, i32 1, <16 x i1> %m16)
+  call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> undef, <8 x ptr> undef, i32 1, <8 x i1> %m8)
 
   ret i32 0
 }
 
-define i32 @masked_expandload() {
+define i32 @masked_expandload(<1 x i1> %m1, <2 x i1> %m2, <4 x i1> %m4, <8 x i1> %m8, <16 x i1> %m16, <32 x i1> %m32, <64 x i1> %m64) {
 ; SSE2-LABEL: 'masked_expandload'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> undef, <8 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> undef, <4 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> undef, <2 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> undef, <1 x double> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> undef, <16 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> undef, <8 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> undef, <4 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> undef, <2 x float> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> undef, <8 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> undef, <4 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> undef, <2 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> undef, <1 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> undef, <16 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> undef, <8 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> undef, <4 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> undef, <2 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> undef, <32 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> undef, <16 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> undef, <8 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> undef, <4 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 188 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> undef, <64 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 94 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> undef, <32 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> undef, <16 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> undef, <8 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> %m8, <8 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> %m4, <4 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> %m2, <2 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> %m1, <1 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> %m16, <16 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> %m8, <8 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> %m4, <4 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> %m2, <2 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> %m8, <8 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> %m4, <4 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> %m2, <2 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> %m1, <1 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> %m16, <16 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> %m8, <8 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> %m4, <4 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> %m2, <2 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> %m32, <32 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> %m16, <16 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> %m8, <8 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> %m4, <4 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 192 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> %m64, <64 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> %m32, <32 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> %m16, <16 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> %m8, <8 x i8> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; SSE42-LABEL: 'masked_expandload'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> undef, <8 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> undef, <4 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> undef, <2 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> undef, <1 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> undef, <16 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> undef, <8 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> undef, <4 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> undef, <2 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> undef, <8 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> undef, <4 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> undef, <2 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> undef, <1 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> undef, <16 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> undef, <8 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> undef, <4 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> undef, <2 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> undef, <32 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> undef, <16 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> undef, <8 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> undef, <4 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> undef, <64 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> undef, <32 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> undef, <16 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> undef, <8 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> %m8, <8 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> %m4, <4 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> %m2, <2 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> %m1, <1 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> %m16, <16 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> %m8, <8 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> %m4, <4 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> %m2, <2 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> %m8, <8 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> %m4, <4 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> %m2, <2 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> %m1, <1 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> %m16, <16 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> %m8, <8 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> %m4, <4 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> %m2, <2 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> %m32, <32 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> %m16, <16 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> %m8, <8 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> %m4, <4 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> %m64, <64 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> %m32, <32 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> %m16, <16 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> %m8, <8 x i8> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
-; AVX-LABEL: 'masked_expandload'
-; AVX-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> undef, <8 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> undef, <4 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> undef, <2 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> undef, <1 x double> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> undef, <16 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> undef, <8 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> undef, <4 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> undef, <2 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> undef, <8 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> undef, <4 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> undef, <2 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> undef, <1 x i64> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> undef, <16 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> undef, <8 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> undef, <4 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> undef, <2 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> undef, <32 x i16> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> undef, <16 x i16> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> undef, <8 x i16> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> undef, <4 x i16> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 130 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> undef, <64 x i8> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> undef, <32 x i8> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> undef, <16 x i8> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> undef, <8 x i8> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
+; AVX1-LABEL: 'masked_expandload'
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> %m8, <8 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> %m4, <4 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> %m2, <2 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> %m1, <1 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> %m16, <16 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> %m8, <8 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> %m4, <4 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> %m2, <2 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> %m8, <8 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> %m4, <4 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> %m2, <2 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> %m1, <1 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> %m16, <16 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> %m8, <8 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> %m4, <4 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> %m2, <2 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> %m32, <32 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> %m16, <16 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> %m8, <8 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> %m4, <4 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 134 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> %m64, <64 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> %m32, <32 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> %m16, <16 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> %m8, <8 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
+;
+; AVX2-LABEL: 'masked_expandload'
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> %m8, <8 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> %m4, <4 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> %m2, <2 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> %m1, <1 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> %m16, <16 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> %m8, <8 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> %m4, <4 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> %m2, <2 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> %m8, <8 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> %m4, <4 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> %m2, <2 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> %m1, <1 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> %m16, <16 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> %m8, <8 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> %m4, <4 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> %m2, <2 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> %m32, <32 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> %m16, <16 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> %m8, <8 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> %m4, <4 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> %m64, <64 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> %m32, <32 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> %m16, <16 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> %m8, <8 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
+;
+; SKL-LABEL: 'masked_expandload'
+; SKL-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> %m8, <8 x double> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> %m4, <4 x double> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> %m2, <2 x double> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> %m1, <1 x double> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> %m16, <16 x float> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> %m8, <8 x float> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> %m4, <4 x float> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> %m2, <2 x float> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> %m8, <8 x i64> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> %m4, <4 x i64> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> %m2, <2 x i64> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> %m1, <1 x i64> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> %m16, <16 x i32> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> %m8, <8 x i32> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> %m4, <4 x i32> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> %m2, <2 x i32> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> %m32, <32 x i16> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> %m16, <16 x i16> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> %m8, <8 x i16> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> %m4, <4 x i16> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> %m64, <64 x i8> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> %m32, <32 x i8> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> %m16, <16 x i8> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> %m8, <8 x i8> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; AVX512-LABEL: 'masked_expandload'
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> undef, <8 x double> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> undef, <4 x double> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> undef, <2 x double> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> undef, <1 x double> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> undef, <16 x float> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> undef, <8 x float> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> undef, <4 x float> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> undef, <2 x float> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> undef, <8 x i64> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> undef, <4 x i64> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> undef, <2 x i64> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> undef, <1 x i64> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> undef, <16 x i32> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> undef, <8 x i32> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> undef, <4 x i32> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> undef, <2 x i32> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> undef, <32 x i16> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> undef, <16 x i16> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> undef, <8 x i16> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> undef, <4 x i16> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 131 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> undef, <64 x i8> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> undef, <32 x i8> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> undef, <16 x i8> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> undef, <8 x i8> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> %m8, <8 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> %m4, <4 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> %m2, <2 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> %m1, <1 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> %m16, <16 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> %m8, <8 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> %m4, <4 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> %m2, <2 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> %m8, <8 x i64> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> %m4, <4 x i64> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> %m2, <2 x i64> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> %m1, <1 x i64> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> %m16, <16 x i32> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> %m8, <8 x i32> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> %m4, <4 x i32> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> %m2, <2 x i32> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 99 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> %m32, <32 x i16> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> %m16, <16 x i16> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> %m8, <8 x i16> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> %m4, <4 x i16> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 195 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> %m64, <64 x i8> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 97 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> %m32, <32 x i8> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> %m16, <16 x i8> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> %m8, <8 x i8> undef)
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
-  %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> undef, <8 x double> undef)
-  %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> undef, <4 x double> undef)
-  %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> undef, <2 x double> undef)
-  %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> undef, <1 x double> undef)
+  %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(ptr undef, <8 x i1> %m8, <8 x double> undef)
+  %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(ptr undef, <4 x i1> %m4, <4 x double> undef)
+  %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(ptr undef, <2 x i1> %m2, <2 x double> undef)
+  %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(ptr undef, <1 x i1> %m1, <1 x double> undef)
 
-  %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> undef, <16 x float> undef)
-  %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> undef, <8 x float> undef)
-  %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> undef, <4 x float> undef)
-  %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> undef, <2 x float> undef)
+  %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(ptr undef, <16 x i1> %m16, <16 x float> undef)
+  %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(ptr undef, <8 x i1> %m8, <8 x float> undef)
+  %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(ptr undef, <4 x i1> %m4, <4 x float> undef)
+  %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(ptr undef, <2 x i1> %m2, <2 x float> undef)
 
-  %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> undef, <8 x i64> undef)
-  %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> undef, <4 x i64> undef)
-  %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> undef, <2 x i64> undef)
-  %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> undef, <1 x i64> undef)
+  %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(ptr undef, <8 x i1> %m8, <8 x i64> undef)
+  %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(ptr undef, <4 x i1> %m4, <4 x i64> undef)
+  %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(ptr undef, <2 x i1> %m2, <2 x i64> undef)
+  %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(ptr undef, <1 x i1> %m1, <1 x i64> undef)
 
-  %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> undef, <16 x i32> undef)
-  %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> undef, <8 x i32> undef)
-  %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> undef, <4 x i32> undef)
-  %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> undef, <2 x i32> undef)
+  %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(ptr undef, <16 x i1> %m16, <16 x i32> undef)
+  %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(ptr undef, <8 x i1> %m8, <8 x i32> undef)
+  %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(ptr undef, <4 x i1> %m4, <4 x i32> undef)
+  %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(ptr undef, <2 x i1> %m2, <2 x i32> undef)
 
-  %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> undef, <32 x i16> undef)
-  %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> undef, <16 x i16> undef)
-  %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> undef, <8 x i16> undef)
-  %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> undef, <4 x i16> undef)
+  %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(ptr undef, <32 x i1> %m32, <32 x i16> undef)
+  %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(ptr undef, <16 x i1> %m16, <16 x i16> undef)
+  %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(ptr undef, <8 x i1> %m8, <8 x i16> undef)
+  %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(ptr undef, <4 x i1> %m4, <4 x i16> undef)
 
-  %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> undef, <64 x i8> undef)
-  %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> undef, <32 x i8> undef)
-  %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> undef, <16 x i8> undef)
-  %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> undef, <8 x i8> undef)
+  %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(ptr undef, <64 x i1> %m64, <64 x i8> undef)
+  %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(ptr undef, <32 x i1> %m32, <32 x i8> undef)
+  %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(ptr undef, <16 x i1> %m16, <16 x i8> undef)
+  %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(ptr undef, <8 x i1> %m8, <8 x i8> undef)
 
   ret i32 0
 }
 
-define i32 @masked_compressstore() {
+define i32 @masked_compressstore(<1 x i1> %m1, <2 x i1> %m2, <4 x i1> %m4, <8 x i1> %m8, <16 x i1> %m16, <32 x i1> %m32, <64 x i1> %m64) {
 ; SSE2-LABEL: 'masked_compressstore'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 192 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> %m1)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> %m1)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> %m2)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> %m32)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> %m8)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> %m4)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 192 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> %m64)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> %m32)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> %m16)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> %m8)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; SSE42-LABEL: 'masked_compressstore'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> %m1)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> %m1)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> %m2)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> %m32)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> %m8)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> %m4)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> %m64)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> %m32)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> %m16)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> %m8)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; AVX1-LABEL: 'masked_compressstore'
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 134 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> %m8)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> %m4)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> %m2)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> %m1)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> %m16)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> %m8)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> %m4)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> %m2)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> %m8)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> %m4)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> %m2)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> %m1)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> %m16)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> %m8)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> %m4)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> %m2)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> %m32)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> %m16)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> %m8)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> %m4)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 134 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> %m64)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> %m32)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> %m16)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> %m8)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; AVX2-LABEL: 'masked_compressstore'
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> %m8)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> %m4)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> %m2)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> %m1)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> %m16)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> %m8)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> %m4)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> %m2)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> %m8)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> %m4)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> %m2)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> %m1)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> %m16)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> %m8)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> %m4)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> %m2)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> %m32)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> %m16)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> %m8)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> %m4)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> %m64)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> %m32)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> %m16)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> %m8)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; SKL-LABEL: 'masked_compressstore'
-; SKL-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> undef)
-; SKL-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> undef)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> %m8)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> %m4)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> %m2)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> %m1)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> %m16)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> %m8)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> %m4)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> %m2)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> %m8)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> %m4)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> %m2)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> %m1)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> %m16)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> %m8)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> %m4)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> %m2)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> %m32)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> %m16)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> %m8)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> %m4)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> %m64)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> %m32)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> %m16)
+; SKL-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> %m8)
 ; SKL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
 ; AVX512-LABEL: 'masked_compressstore'
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 99 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 195 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 97 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> %m8)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> %m4)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> %m2)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> %m1)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> %m16)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> %m8)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> %m4)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> %m2)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> %m8)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> %m4)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> %m2)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> %m1)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> %m16)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> %m8)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> %m4)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> %m2)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 99 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> %m32)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> %m16)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> %m8)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> %m4)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 195 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> %m64)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 97 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> %m32)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> %m16)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> %m8)
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 0
 ;
-  call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> undef)
-  call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> undef)
-  call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> undef)
-  call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> undef)
-
-  call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> undef)
-  call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> undef)
-  call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> undef)
-  call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> undef)
-
-  call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> undef)
-  call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> undef)
-  call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> undef)
-  call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> undef)
-
-  call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> undef)
-  call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> undef)
-  call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> undef)
-  call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> undef)
-
-  call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> undef)
-  call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> undef)
-  call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> undef)
-  call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> undef)
-
-  call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> undef)
-  call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> undef)
-  call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> undef)
-  call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> undef)
+  call void @llvm.masked.compressstore.v8f64(<8 x double> undef, ptr undef, <8 x i1> %m8)
+  call void @llvm.masked.compressstore.v4f64(<4 x double> undef, ptr undef, <4 x i1> %m4)
+  call void @llvm.masked.compressstore.v2f64(<2 x double> undef, ptr undef, <2 x i1> %m2)
+  call void @llvm.masked.compressstore.v1f64(<1 x double> undef, ptr undef, <1 x i1> %m1)
+
+  call void @llvm.masked.compressstore.v16f32(<16 x float> undef, ptr undef, <16 x i1> %m16)
+  call void @llvm.masked.compressstore.v8f32(<8 x float> undef, ptr undef, <8 x i1> %m8)
+  call void @llvm.masked.compressstore.v4f32(<4 x float> undef, ptr undef, <4 x i1> %m4)
+  call void @llvm.masked.compressstore.v2f32(<2 x float> undef, ptr undef, <2 x i1> %m2)
+
+  call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, ptr undef, <8 x i1> %m8)
+  call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, ptr undef, <4 x i1> %m4)
+  call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, ptr undef, <2 x i1> %m2)
+  call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, ptr undef, <1 x i1> %m1)
+
+  call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, ptr undef, <16 x i1> %m16)
+  call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, ptr undef, <8 x i1> %m8)
+  call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, ptr undef, <4 x i1> %m4)
+  call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, ptr undef, <2 x i1> %m2)
+
+  call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, ptr undef, <32 x i1> %m32)
+  call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, ptr undef, <16 x i1> %m16)
+  call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, ptr undef, <8 x i1> %m8)
+  call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, ptr undef, <4 x i1> %m4)
+
+  call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, ptr undef, <64 x i1> %m64)
+  call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, ptr undef, <32 x i1> %m32)
+  call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, ptr undef, <16 x i1> %m16)
+  call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, ptr undef, <8 x i1> %m8)
 
   ret i32 0
 }
diff --git a/llvm/test/Bitcode/dbg-record-roundtrip.ll b/llvm/test/Bitcode/dbg-record-roundtrip.ll
index bd347cac7206..cc83fdd4fa53 100644
--- a/llvm/test/Bitcode/dbg-record-roundtrip.ll
+++ b/llvm/test/Bitcode/dbg-record-roundtrip.ll
@@ -15,6 +15,16 @@
 ; RUN: | llvm-dis --load-bitcode-into-experimental-debuginfo-iterators=true --write-experimental-debuginfo=true \
 ; RUN: | FileCheck %s --check-prefixes=RECORDS
 
+;; When preserving, we should output the format the bitcode was written in
+;; regardless of the value of the write flag.
+; RUN: llvm-as --write-experimental-debuginfo-iterators-to-bitcode=true %s -o - \
+; RUN: | llvm-dis --preserve-input-debuginfo-format=true --write-experimental-debuginfo=false \
+; RUN: | FileCheck %s --check-prefixes=RECORDS
+
+; RUN: llvm-as --write-experimental-debuginfo-iterators-to-bitcode=false %s -o - \
+; RUN: | llvm-dis --preserve-input-debuginfo-format=true --write-experimental-debuginfo=true \
+; RUN: | FileCheck %s
+
 ;; Check that verify-uselistorder passes regardless of input format.
 ; RUN: llvm-as %s --write-experimental-debuginfo-iterators-to-bitcode=true -o - | verify-uselistorder
 ; RUN: verify-uselistorder %s
diff --git a/llvm/test/CodeGen/AArch64/GlobalISel/combine-commute-const-infinite-loop.mir b/llvm/test/CodeGen/AArch64/GlobalISel/combine-commute-const-infinite-loop.mir
new file mode 100644
index 000000000000..6efdc5e5e8c3
--- /dev/null
+++ b/llvm/test/CodeGen/AArch64/GlobalISel/combine-commute-const-infinite-loop.mir
@@ -0,0 +1,29 @@
+# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py UTC_ARGS: --version 4
+# RUN: llc -mtriple aarch64 -run-pass=aarch64-prelegalizer-combiner %s -o - \
+# RUN:     --aarch64prelegalizercombiner-disable-rule=constant_fold_binop | FileCheck %s
+# REQUIRES: asserts
+
+# `constant_fold_binop` is disabled to trigger the infinite loop in `commute_int_constant_to_rhs`.
+
+---
+name:            add
+tracksRegLiveness: true
+body:             |
+  bb.0:
+    liveins: $s0
+
+    ; CHECK-LABEL: name: add
+    ; CHECK: liveins: $s0
+    ; CHECK-NEXT: {{  $}}
+    ; CHECK-NEXT: %c0:_(s32) = G_CONSTANT i32 1
+    ; CHECK-NEXT: %c1:_(s32) = G_CONSTANT i32 2
+    ; CHECK-NEXT: %add:_(s32) = G_ADD %c0, %c1
+    ; CHECK-NEXT: $s0 = COPY %add(s32)
+    ; CHECK-NEXT: RET_ReallyLR
+    %c0:_(s32) = G_CONSTANT i32 1
+    %c1:_(s32) = G_CONSTANT i32 2
+    %add:_(s32) = G_ADD %c0, %c1
+    $s0 = COPY %add
+    RET_ReallyLR
+
+...
diff --git a/llvm/test/CodeGen/ARM/sadd_sat.ll b/llvm/test/CodeGen/ARM/sadd_sat.ll
index 1632c4e86c76..0060b4458081 100644
--- a/llvm/test/CodeGen/ARM/sadd_sat.ll
+++ b/llvm/test/CodeGen/ARM/sadd_sat.ll
@@ -130,8 +130,8 @@ define i64 @func2(i64 %x, i64 %y) nounwind {
 ; CHECK-T15TE-NEXT:    bics r4, r1
 ; CHECK-T15TE-NEXT:    asrs r1, r3, #31
 ; CHECK-T15TE-NEXT:    cmp r4, #0
-; CHECK-T15TE-NEXT:    push {r1}
-; CHECK-T15TE-NEXT:    pop {r0}
+; CHECK-T15TE-NEXT:    mov r12, r1
+; CHECK-T15TE-NEXT:    mov r0, r12
 ; CHECK-T15TE-NEXT:    bmi .LBB1_2
 ; CHECK-T15TE-NEXT:  @ %bb.1:
 ; CHECK-T15TE-NEXT:    movs r0, r2
diff --git a/llvm/test/CodeGen/ARM/select_const.ll b/llvm/test/CodeGen/ARM/select_const.ll
index e12dd02f16c2..df95af313eac 100644
--- a/llvm/test/CodeGen/ARM/select_const.ll
+++ b/llvm/test/CodeGen/ARM/select_const.ll
@@ -665,8 +665,8 @@ define i64 @opaque_constant1(i1 %cond, i64 %x) {
 ; THUMB-NEXT:    movs r7, #1
 ; THUMB-NEXT:    ands r0, r7
 ; THUMB-NEXT:    subs r1, r0, #1
-; THUMB-NEXT:    push {r0}
-; THUMB-NEXT:    pop {r4}
+; THUMB-NEXT:    mov r12, r0
+; THUMB-NEXT:    mov r4, r12
 ; THUMB-NEXT:    sbcs r4, r1
 ; THUMB-NEXT:    cmp r0, #0
 ; THUMB-NEXT:    bne .LBB24_2
@@ -681,8 +681,8 @@ define i64 @opaque_constant1(i1 %cond, i64 %x) {
 ; THUMB-NEXT:    ands r5, r0
 ; THUMB-NEXT:    movs r6, #0
 ; THUMB-NEXT:    subs r0, r5, #1
-; THUMB-NEXT:    push {r4}
-; THUMB-NEXT:    pop {r1}
+; THUMB-NEXT:    mov r12, r4
+; THUMB-NEXT:    mov r1, r12
 ; THUMB-NEXT:    sbcs r1, r6
 ; THUMB-NEXT:    eors r3, r7
 ; THUMB-NEXT:    ldr r6, .LCPI24_0
@@ -786,11 +786,11 @@ define i64 @func(i64 %arg) {
 ; THUMB-NEXT:    push {r4, lr}
 ; THUMB-NEXT:    movs r2, #0
 ; THUMB-NEXT:    adds r3, r0, #1
-; THUMB-NEXT:    push {r1}
-; THUMB-NEXT:    pop {r3}
+; THUMB-NEXT:    mov r12, r1
+; THUMB-NEXT:    mov r3, r12
 ; THUMB-NEXT:    adcs r3, r2
-; THUMB-NEXT:    push {r2}
-; THUMB-NEXT:    pop {r3}
+; THUMB-NEXT:    mov r12, r2
+; THUMB-NEXT:    mov r3, r12
 ; THUMB-NEXT:    adcs r3, r2
 ; THUMB-NEXT:    subs r4, r3, #1
 ; THUMB-NEXT:    adds r0, r0, #1
diff --git a/llvm/test/CodeGen/ARM/wide-compares.ll b/llvm/test/CodeGen/ARM/wide-compares.ll
index 6584f0c7616c..09e3592b6d42 100644
--- a/llvm/test/CodeGen/ARM/wide-compares.ll
+++ b/llvm/test/CodeGen/ARM/wide-compares.ll
@@ -257,12 +257,12 @@ define {i32, i32} @test_slt_not(i32 %c, i32 %d, i64 %a, i64 %b) {
 ; CHECK-THUMB1-NOMOV-NEXT:    ldr r5, [sp, #16]
 ; CHECK-THUMB1-NOMOV-NEXT:    subs r2, r2, r5
 ; CHECK-THUMB1-NOMOV-NEXT:    sbcs r3, r0
-; CHECK-THUMB1-NOMOV-NEXT:    push {r1}
-; CHECK-THUMB1-NOMOV-NEXT:    pop {r0}
+; CHECK-THUMB1-NOMOV-NEXT:    mov r12, r1
+; CHECK-THUMB1-NOMOV-NEXT:    mov r0, r12
 ; CHECK-THUMB1-NOMOV-NEXT:    blt .LBB3_2
 ; CHECK-THUMB1-NOMOV-NEXT:  @ %bb.1: @ %entry
-; CHECK-THUMB1-NOMOV-NEXT:    push {r4}
-; CHECK-THUMB1-NOMOV-NEXT:    pop {r0}
+; CHECK-THUMB1-NOMOV-NEXT:    mov r12, r4
+; CHECK-THUMB1-NOMOV-NEXT:    mov r0, r12
 ; CHECK-THUMB1-NOMOV-NEXT:  .LBB3_2: @ %entry
 ; CHECK-THUMB1-NOMOV-NEXT:    bge .LBB3_4
 ; CHECK-THUMB1-NOMOV-NEXT:  @ %bb.3: @ %entry
diff --git a/llvm/test/CodeGen/Thumb/pr35836.ll b/llvm/test/CodeGen/Thumb/pr35836.ll
index 96a6fe5d1420..ba33a8184bcc 100644
--- a/llvm/test/CodeGen/Thumb/pr35836.ll
+++ b/llvm/test/CodeGen/Thumb/pr35836.ll
@@ -35,18 +35,18 @@ while.body:
   br label %while.body
 }
 ; CHECK: adds	r3, r0, r1
-; CHECK: push	{r5}
-; CHECK: pop	{r1}
+; CHECK: mov	r12, r5
+; CHECK: mov	r1, r12
 ; CHECK: adcs	r1, r5
 ; CHECK: ldr	r0, [sp, #12]           @ 4-byte Reload
 ; CHECK: ldr	r2, [sp, #8]            @ 4-byte Reload
 ; CHECK: adds	r2, r0, r2
-; CHECK: push	{r5}
-; CHECK: pop	{r4}
+; CHECK: mov	r12, r5
+; CHECK: mov	r4, r12
 ; CHECK: adcs	r4, r5
 ; CHECK: adds	r0, r2, r5
-; CHECK: push	{r3}
-; CHECK: pop	{r0}
+; CHECK: mov	r12, r3
+; CHECK: mov	r0, r12
 ; CHECK: adcs	r0, r4
 ; CHECK: ldr	r6, [sp, #4]            @ 4-byte Reload
 ; CHECK: str	r0, [r6]
diff --git a/llvm/test/CodeGen/Thumb/urem-seteq-illegal-types.ll b/llvm/test/CodeGen/Thumb/urem-seteq-illegal-types.ll
index aa5deb6542b2..61a741445b81 100644
--- a/llvm/test/CodeGen/Thumb/urem-seteq-illegal-types.ll
+++ b/llvm/test/CodeGen/Thumb/urem-seteq-illegal-types.ll
@@ -122,8 +122,8 @@ define <3 x i1> @test_urem_vec(<3 x i11> %X) nounwind {
 ; CHECK-NEXT:    movs r3, #1
 ; CHECK-NEXT:    movs r4, #0
 ; CHECK-NEXT:    cmp r0, #170
-; CHECK-NEXT:    push {r3}
-; CHECK-NEXT:    pop {r0}
+; CHECK-NEXT:    mov r12, r3
+; CHECK-NEXT:    mov r0, r12
 ; CHECK-NEXT:    bhi .LBB4_2
 ; CHECK-NEXT:  @ %bb.1:
 ; CHECK-NEXT:    movs r0, r4
@@ -134,8 +134,8 @@ define <3 x i1> @test_urem_vec(<3 x i11> %X) nounwind {
 ; CHECK-NEXT:    movs r1, #73
 ; CHECK-NEXT:    lsls r1, r1, #23
 ; CHECK-NEXT:    cmp r5, r1
-; CHECK-NEXT:    push {r3}
-; CHECK-NEXT:    pop {r1}
+; CHECK-NEXT:    mov r12, r3
+; CHECK-NEXT:    mov r1, r12
 ; CHECK-NEXT:    bhi .LBB4_4
 ; CHECK-NEXT:  @ %bb.3:
 ; CHECK-NEXT:    movs r1, r4
diff --git a/llvm/test/CodeGen/X86/bool-vector.ll b/llvm/test/CodeGen/X86/bool-vector.ll
index e4deb878aa46..d52b455eb2e6 100644
--- a/llvm/test/CodeGen/X86/bool-vector.ll
+++ b/llvm/test/CodeGen/X86/bool-vector.ll
@@ -6,7 +6,7 @@
 ; RUN: llc < %s -mtriple=i686-unknown-unknown -mattr=+avx2 | FileCheck %s --check-prefix=AVX2
 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx2 | FileCheck %s --check-prefix=AVX2
 
-define i32 @PR15215_bad(<4 x i32> %input) {
+define i32 @PR15215_bad(<4 x i32> %input) nounwind {
 ; X86-LABEL: PR15215_bad:
 ; X86:       # %bb.0: # %entry
 ; X86-NEXT:    movzbl {{[0-9]+}}(%esp), %eax
@@ -59,12 +59,10 @@ entry:
   ret i32 %2
 }
 
-define i32 @PR15215_good(<4 x i32> %input) {
+define i32 @PR15215_good(<4 x i32> %input) nounwind {
 ; X86-LABEL: PR15215_good:
 ; X86:       # %bb.0: # %entry
 ; X86-NEXT:    pushl %esi
-; X86-NEXT:    .cfi_def_cfa_offset 8
-; X86-NEXT:    .cfi_offset %esi, -8
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %ecx
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %edx
@@ -77,7 +75,6 @@ define i32 @PR15215_good(<4 x i32> %input) {
 ; X86-NEXT:    leal (%edx,%ecx,4), %ecx
 ; X86-NEXT:    leal (%ecx,%eax,8), %eax
 ; X86-NEXT:    popl %esi
-; X86-NEXT:    .cfi_def_cfa_offset 4
 ; X86-NEXT:    retl
 ;
 ; X64-LABEL: PR15215_good:
diff --git a/llvm/test/CodeGen/X86/pr45995.ll b/llvm/test/CodeGen/X86/pr45995.ll
index f9a9ab768f1a..997ad6be84b9 100644
--- a/llvm/test/CodeGen/X86/pr45995.ll
+++ b/llvm/test/CodeGen/X86/pr45995.ll
@@ -1,18 +1,12 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc -O3 --x86-asm-syntax=intel -mtriple=x86_64 -mattr=avx < %s | FileCheck %s
 
-define void @extracter0([4 x <4 x i1>] %matrix) {
+define void @extracter0([4 x <4 x i1>] %matrix) nounwind {
 ; CHECK-LABEL: extracter0:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    push rbp
-; CHECK-NEXT:    .cfi_def_cfa_offset 16
 ; CHECK-NEXT:    push r14
-; CHECK-NEXT:    .cfi_def_cfa_offset 24
 ; CHECK-NEXT:    push rbx
-; CHECK-NEXT:    .cfi_def_cfa_offset 32
-; CHECK-NEXT:    .cfi_offset rbx, -32
-; CHECK-NEXT:    .cfi_offset r14, -24
-; CHECK-NEXT:    .cfi_offset rbp, -16
 ; CHECK-NEXT:    vpslld xmm0, xmm0, 31
 ; CHECK-NEXT:    vmovmskps edi, xmm0
 ; CHECK-NEXT:    mov ebx, edi
@@ -31,11 +25,8 @@ define void @extracter0([4 x <4 x i1>] %matrix) {
 ; CHECK-NEXT:    movzx edi, bl
 ; CHECK-NEXT:    call print_i1@PLT
 ; CHECK-NEXT:    pop rbx
-; CHECK-NEXT:    .cfi_def_cfa_offset 24
 ; CHECK-NEXT:    pop r14
-; CHECK-NEXT:    .cfi_def_cfa_offset 16
 ; CHECK-NEXT:    pop rbp
-; CHECK-NEXT:    .cfi_def_cfa_offset 8
 ; CHECK-NEXT:    ret
   %1 = extractvalue [4 x <4 x i1>] %matrix, 0
   %2 = extractelement <4 x i1> %1, i64 0
@@ -49,29 +40,16 @@ define void @extracter0([4 x <4 x i1>] %matrix) {
   ret void
 }
 
-define void @extracter1([4 x <4 x i1>] %matrix) {
+define void @extracter1([4 x <4 x i1>] %matrix) nounwind {
 ; CHECK-LABEL: extracter1:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    push rbp
-; CHECK-NEXT:    .cfi_def_cfa_offset 16
 ; CHECK-NEXT:    push r15
-; CHECK-NEXT:    .cfi_def_cfa_offset 24
 ; CHECK-NEXT:    push r14
-; CHECK-NEXT:    .cfi_def_cfa_offset 32
 ; CHECK-NEXT:    push r13
-; CHECK-NEXT:    .cfi_def_cfa_offset 40
 ; CHECK-NEXT:    push r12
-; CHECK-NEXT:    .cfi_def_cfa_offset 48
 ; CHECK-NEXT:    push rbx
-; CHECK-NEXT:    .cfi_def_cfa_offset 56
 ; CHECK-NEXT:    push rax
-; CHECK-NEXT:    .cfi_def_cfa_offset 64
-; CHECK-NEXT:    .cfi_offset rbx, -56
-; CHECK-NEXT:    .cfi_offset r12, -48
-; CHECK-NEXT:    .cfi_offset r13, -40
-; CHECK-NEXT:    .cfi_offset r14, -32
-; CHECK-NEXT:    .cfi_offset r15, -24
-; CHECK-NEXT:    .cfi_offset rbp, -16
 ; CHECK-NEXT:    vpslld xmm1, xmm1, 31
 ; CHECK-NEXT:    vmovmskps ebx, xmm1
 ; CHECK-NEXT:    mov eax, ebx
@@ -109,19 +87,12 @@ define void @extracter1([4 x <4 x i1>] %matrix) {
 ; CHECK-NEXT:    movzx edi, byte ptr [rsp + 7] # 1-byte Folded Reload
 ; CHECK-NEXT:    call print_i1@PLT
 ; CHECK-NEXT:    add rsp, 8
-; CHECK-NEXT:    .cfi_def_cfa_offset 56
 ; CHECK-NEXT:    pop rbx
-; CHECK-NEXT:    .cfi_def_cfa_offset 48
 ; CHECK-NEXT:    pop r12
-; CHECK-NEXT:    .cfi_def_cfa_offset 40
 ; CHECK-NEXT:    pop r13
-; CHECK-NEXT:    .cfi_def_cfa_offset 32
 ; CHECK-NEXT:    pop r14
-; CHECK-NEXT:    .cfi_def_cfa_offset 24
 ; CHECK-NEXT:    pop r15
-; CHECK-NEXT:    .cfi_def_cfa_offset 16
 ; CHECK-NEXT:    pop rbp
-; CHECK-NEXT:    .cfi_def_cfa_offset 8
 ; CHECK-NEXT:    ret
   %1 = extractvalue [4 x <4 x i1>] %matrix, 0
   %2 = extractelement <4 x i1> %1, i64 0
diff --git a/llvm/test/DebugInfo/roundtrip-non-instruction-debug-info.ll b/llvm/test/DebugInfo/roundtrip-non-instruction-debug-info.ll
index b15b76d1690c..b154a4ff162d 100644
--- a/llvm/test/DebugInfo/roundtrip-non-instruction-debug-info.ll
+++ b/llvm/test/DebugInfo/roundtrip-non-instruction-debug-info.ll
@@ -18,6 +18,14 @@
 ; RUN: opt --passes=verify -S --try-experimental-debuginfo-iterators --write-experimental-debuginfo=true < %s \
 ; RUN:   | FileCheck %s --check-prefixes=CHECK,NEWDBG --implicit-check-not=llvm.dbg --implicit-check-not=#dbg
 
+;; Test that the preserving flag overrides the write flag.
+; RUN: opt --passes=verify -S --preserve-input-debuginfo-format=true --write-experimental-debuginfo=true < %s \
+; RUN:   | FileCheck %s --check-prefixes=CHECK,OLDDBG  --implicit-check-not=llvm.dbg --implicit-check-not=#dbg
+
+; RUN: opt --passes=verify -S --write-experimental-debuginfo=true < %s \
+; RUN:   | opt --passes=verify -S --preserve-input-debuginfo-format=true --write-experimental-debuginfo=false \
+; RUN:   | FileCheck %s --check-prefixes=CHECK,NEWDBG  --implicit-check-not=llvm.dbg --implicit-check-not=#dbg
+
 ; CHECK: @f(i32 %[[VAL_A:[0-9a-zA-Z]+]])
 ; CHECK-NEXT: entry:
 ; OLDDBG-NEXT: call void @llvm.dbg.value(metadata i32 %[[VAL_A]], metadata ![[VAR_A:[0-9]+]], metadata !DIExpression()), !dbg ![[LOC_1:[0-9]+]]
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/sve-select-cmp.ll b/llvm/test/Transforms/LoopVectorize/AArch64/sve-select-cmp.ll
index 1c26ee8479e5..2470bca1e17b 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/sve-select-cmp.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/sve-select-cmp.ll
@@ -8,39 +8,41 @@ target triple = "aarch64-linux-gnu"
 define i32 @select_const_i32_from_icmp(ptr nocapture readonly %v, i64 %n) #0 {
 ; CHECK-VF4IC1-LABEL: @select_const_i32_from_icmp
 ; CHECK-VF4IC1:      vector.body:
-; CHECK-VF4IC1:        [[VEC_PHI:%.*]] = phi <vscale x 4 x i32> [ shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 3, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer), %vector.ph ], [ [[VEC_SEL:%.*]], %vector.body ]
+; CHECK-VF4IC1:        [[VEC_PHI:%.*]] = phi <vscale x 4 x i1> [ zeroinitializer, %vector.ph ], [ [[VEC_SEL:%.*]], %vector.body ]
 ; CHECK-VF4IC1:        [[VEC_LOAD:%.*]] = load <vscale x 4 x i32>
 ; CHECK-VF4IC1-NEXT:   [[VEC_ICMP:%.*]] = icmp eq <vscale x 4 x i32> [[VEC_LOAD]], shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 3, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)
-; CHECK-VF4IC1-NEXT:   [[VEC_SEL]] = select <vscale x 4 x i1> [[VEC_ICMP]], <vscale x 4 x i32> [[VEC_PHI]], <vscale x 4 x i32> shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 7, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)
+; CHECK-VF4IC1-NEXT:   [[NOT:%*]] = xor <vscale x 4 x i1> [[VEC_ICMP]], shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer)
+; CHECK-VF4IC1-NEXT:   [[VEC_SEL]] = or <vscale x 4 x i1> [[VEC_PHI]], [[NOT]]
 ; CHECK-VF4IC1:      middle.block:
-; CHECK-VF4IC1-NEXT:   [[FIN_ICMP:%.*]] = icmp ne <vscale x 4 x i32> [[VEC_SEL]], shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 3, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)
-; CHECK-VF4IC1-NEXT:   [[OR_RDX:%.*]] = call i1 @llvm.vector.reduce.or.nxv4i1(<vscale x 4 x i1> [[FIN_ICMP]])
-; CHECK-VF4IC1-NEXT:   {{.*}} = select i1 [[OR_RDX]], i32 7, i32 3
+; CHECK-VF4IC1-NEXT:   [[OR_RDX:%.*]] = call i1 @llvm.vector.reduce.or.nxv4i1(<vscale x 4 x i1> [[VEC_SEL]])
+; CHECK-VF4IC1-NEXT:   [[FR:%.*]] = freeze i1 [[OR_RDX]]
+; CHECK-VF4IC1-NEXT:   {{.*}} = select i1 [[FR]], i32 7, i32 3
 
 ; CHECK-VF4IC4-LABEL: @select_const_i32_from_icmp
 ; CHECK-VF4IC4:      vector.body:
-; CHECK-VF4IC4:        [[VEC_PHI1:%.*]] = phi <vscale x 4 x i32> [ shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 3, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer), %vector.ph ], [ [[VEC_SEL1:%.*]], %vector.body ]
-; CHECK-VF4IC4:        [[VEC_PHI2:%.*]] = phi <vscale x 4 x i32> [ shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 3, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer), %vector.ph ], [ [[VEC_SEL2:%.*]], %vector.body ]
-; CHECK-VF4IC4:        [[VEC_PHI3:%.*]] = phi <vscale x 4 x i32> [ shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 3, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer), %vector.ph ], [ [[VEC_SEL3:%.*]], %vector.body ]
-; CHECK-VF4IC4:        [[VEC_PHI4:%.*]] = phi <vscale x 4 x i32> [ shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 3, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer), %vector.ph ], [ [[VEC_SEL4:%.*]], %vector.body ]
+; CHECK-VF4IC4:        [[VEC_PHI1:%.*]] = phi <vscale x 4 x i1> [ zeroinitializer, %vector.ph ], [ [[VEC_SEL1:%.*]], %vector.body ]
+; CHECK-VF4IC4:        [[VEC_PHI2:%.*]] = phi <vscale x 4 x i1> [ zeroinitializer, %vector.ph ], [ [[VEC_SEL2:%.*]], %vector.body ]
+; CHECK-VF4IC4:        [[VEC_PHI3:%.*]] = phi <vscale x 4 x i1> [ zeroinitializer, %vector.ph ], [ [[VEC_SEL3:%.*]], %vector.body ]
+; CHECK-VF4IC4:        [[VEC_PHI4:%.*]] = phi <vscale x 4 x i1> [ zeroinitializer, %vector.ph ], [ [[VEC_SEL4:%.*]], %vector.body ]
 ; CHECK-VF4IC4:        [[VEC_ICMP1:%.*]] = icmp eq <vscale x 4 x i32> {{.*}}, shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 3, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)
 ; CHECK-VF4IC4-NEXT:   [[VEC_ICMP2:%.*]] = icmp eq <vscale x 4 x i32> {{.*}}, shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 3, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)
 ; CHECK-VF4IC4-NEXT:   [[VEC_ICMP3:%.*]] = icmp eq <vscale x 4 x i32> {{.*}}, shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 3, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)
 ; CHECK-VF4IC4-NEXT:   [[VEC_ICMP4:%.*]] = icmp eq <vscale x 4 x i32> {{.*}}, shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 3, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)
-; CHECK-VF4IC4-NEXT:   [[VEC_SEL1]] = select <vscale x 4 x i1> [[VEC_ICMP1]], <vscale x 4 x i32> [[VEC_PHI1]], <vscale x 4 x i32> shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 7, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)
-; CHECK-VF4IC4-NEXT:   [[VEC_SEL2]] = select <vscale x 4 x i1> [[VEC_ICMP2]], <vscale x 4 x i32> [[VEC_PHI2]], <vscale x 4 x i32> shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 7, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)
-; CHECK-VF4IC4-NEXT:   [[VEC_SEL3]] = select <vscale x 4 x i1> [[VEC_ICMP3]], <vscale x 4 x i32> [[VEC_PHI3]], <vscale x 4 x i32> shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 7, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)
-; CHECK-VF4IC4-NEXT:   [[VEC_SEL4]] = select <vscale x 4 x i1> [[VEC_ICMP4]], <vscale x 4 x i32> [[VEC_PHI4]], <vscale x 4 x i32> shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 7, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)
+; CHECK-VF4IC4-NEXT:    [[NOT1:%.*]] = xor <vscale x 4 x i1> [[VEC_ICMP1]], shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer)
+; CHECK-VF4IC4-NEXT:    [[NOT2:%.*]] = xor <vscale x 4 x i1> [[VEC_ICMP2]], shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer)
+; CHECK-VF4IC4-NEXT:    [[NOT3:%.*]] = xor <vscale x 4 x i1> [[VEC_ICMP3]], shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer)
+; CHECK-VF4IC4-NEXT:    [[NOT4:%.*]] = xor <vscale x 4 x i1> [[VEC_ICMP4]], shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer)
+; CHECK-VF4IC4-NEXT:   [[VEC_SEL1:%.*]] = or <vscale x 4 x i1> [[VEC_PHI1]], [[NOT1]]
+; CHECK-VF4IC4-NEXT:   [[VEC_SEL2:%.*]] = or <vscale x 4 x i1> [[VEC_PHI2]], [[NOT2]]
+; CHECK-VF4IC4-NEXT:   [[VEC_SEL3:%.*]] = or <vscale x 4 x i1> [[VEC_PHI3]], [[NOT3]]
+; CHECK-VF4IC4-NEXT:   [[VEC_SEL4:%.*]] = or <vscale x 4 x i1> [[VEC_PHI4]], [[NOT4]]
 ; CHECK-VF4IC4:      middle.block:
-; CHECK-VF4IC4-NEXT:   [[VEC_ICMP5:%.*]] = icmp ne <vscale x 4 x i32> [[VEC_SEL1]], shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 3, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)
-; CHECK-VF4IC4-NEXT:   [[VEC_SEL5:%.*]] = select <vscale x 4 x i1> [[VEC_ICMP5]], <vscale x 4 x i32> [[VEC_SEL1]], <vscale x 4 x i32> [[VEC_SEL2]]
-; CHECK-VF4IC4-NEXT:   [[VEC_ICMP6:%.*]] = icmp ne <vscale x 4 x i32> [[VEC_SEL5]], shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 3, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)
-; CHECK-VF4IC4-NEXT:   [[VEC_SEL6:%.*]] = select <vscale x 4 x i1> [[VEC_ICMP6]], <vscale x 4 x i32> [[VEC_SEL5]], <vscale x 4 x i32> [[VEC_SEL3]]
-; CHECK-VF4IC4-NEXT:   [[VEC_ICMP7:%.*]] = icmp ne <vscale x 4 x i32> [[VEC_SEL6]], shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 3, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)
-; CHECK-VF4IC4-NEXT:   [[VEC_SEL7:%.*]] = select <vscale x 4 x i1> [[VEC_ICMP7]], <vscale x 4 x i32> [[VEC_SEL6]], <vscale x 4 x i32> [[VEC_SEL4]]
-; CHECK-VF4IC4-NEXT:   [[FIN_ICMP:%.*]] = icmp ne <vscale x 4 x i32> [[VEC_SEL7]], shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 3, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)
-; CHECK-VF4IC4-NEXT:   [[OR_RDX:%.*]] = call i1 @llvm.vector.reduce.or.nxv4i1(<vscale x 4 x i1> [[FIN_ICMP]])
-; CHECK-VF4IC4-NEXT:   {{.*}} = select i1 [[OR_RDX]], i32 7, i32 3
+; CHECK-VF4IC4-NEXT:   [[OR1:%.*]] = or <vscale x 4 x i1> [[VEC_SEL2]], [[VEC_SEL1]]
+; CHECK-VF4IC4-NEXT:   [[OR2:%.*]] = or <vscale x 4 x i1> [[VEC_SEL3]], [[OR1]]
+; CHECK-VF4IC4-NEXT:   [[OR3:%.*]] = or <vscale x 4 x i1> [[VEC_SEL4]], [[OR2]]
+; CHECK-VF4IC4-NEXT:   [[OR_RDX:%.*]] = call i1 @llvm.vector.reduce.or.nxv4i1(<vscale x 4 x i1> [[OR3]])
+; CHECK-VF4IC4-NEXT:   [[FR:%.*]] = freeze i1 [[OR_RDX]]
+; CHECK-VF4IC4-NEXT:   {{.*}} = select i1 [[FR]], i32 7, i32 3
 entry:
   br label %for.body
 
@@ -62,21 +64,18 @@ exit:                                     ; preds = %for.body
 define i32 @select_i32_from_icmp(ptr nocapture readonly %v, i32 %a, i32 %b, i64 %n) #0 {
 ; CHECK-VF4IC1-LABEL: @select_i32_from_icmp
 ; CHECK-VF4IC1:      vector.ph:
-; CHECK-VF4IC1:        [[TMP1:%.*]] = insertelement <vscale x 4 x i32> poison, i32 %a, i64 0
-; CHECK-VF4IC1-NEXT:   [[SPLAT_OF_A:%.*]] = shufflevector <vscale x 4 x i32> [[TMP1]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
-; CHECK-VF4IC1-NEXT:   [[TMP2:%.*]] = insertelement <vscale x 4 x i32> poison, i32 %b, i64 0
-; CHECK-VF4IC1-NEXT:   [[SPLAT_OF_B:%.*]] = shufflevector <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
+; CHECK-VF4IC1-NOT:    shufflevector <vscale x 4 x i32>
+; CHECK-VF4IC1-NOT:    shufflevector <vscale x 4 x i32>
 ; CHECK-VF4IC1:      vector.body:
-; CHECK-VF4IC1:        [[VEC_PHI:%.*]] = phi <vscale x 4 x i32> [ [[SPLAT_OF_A]], %vector.ph ], [ [[VEC_SEL:%.*]], %vector.body ]
+; CHECK-VF4IC1:        [[VEC_PHI:%.*]] = phi <vscale x 4 x i1> [ zeroinitializer, %vector.ph ], [ [[VEC_SEL:%.*]], %vector.body ]
 ; CHECK-VF4IC1:        [[VEC_LOAD:%.*]] = load <vscale x 4 x i32>
 ; CHECK-VF4IC1-NEXT:   [[VEC_ICMP:%.*]] = icmp eq <vscale x 4 x i32> [[VEC_LOAD]], shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 3, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)
-; CHECK-VF4IC1-NEXT:   [[VEC_SEL]] = select <vscale x 4 x i1> [[VEC_ICMP]], <vscale x 4 x i32> [[VEC_PHI]], <vscale x 4 x i32> [[SPLAT_OF_B]]
+; CHECK-VF4IC1-NEXT:   [[NOT:%*]] = xor <vscale x 4 x i1> [[VEC_ICMP]], shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer)
+; CHECK-VF4IC1-NEXT:   [[VEC_SEL]] = or <vscale x 4 x i1> [[VEC_PHI]], [[NOT]]
 ; CHECK-VF4IC1:      middle.block:
-; CHECK-VF4IC1-NEXT:   [[FIN_INS:%.*]] = insertelement <vscale x 4 x i32> poison, i32 %a, i64 0
-; CHECK-VF4IC1-NEXT:   [[FIN_SPLAT:%.*]] = shufflevector <vscale x 4 x i32> [[FIN_INS]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
-; CHECK-VF4IC1-NEXT:   [[FIN_CMP:%.*]] = icmp ne <vscale x 4 x i32> [[VEC_SEL]], [[FIN_SPLAT]]
-; CHECK-VF4IC1-NEXT:   [[OR_RDX:%.*]] = call i1 @llvm.vector.reduce.or.nxv4i1(<vscale x 4 x i1> [[FIN_CMP]])
-; CHECK-VF4IC1-NEXT:   {{.*}} = select i1 [[OR_RDX]], i32 %b, i32 %a
+; CHECK-VF4IC1-NEXT:   [[OR_RDX:%.*]] = call i1 @llvm.vector.reduce.or.nxv4i1(<vscale x 4 x i1> [[VEC_SEL]])
+; CHECK-VF4IC1-NEXT:   [[FR:%.*]] = freeze i1 [[OR_RDX]]
+; CHECK-VF4IC1-NEXT:   {{.*}} = select i1 [[FR]], i32 %b, i32 %a
 
 ; CHECK-VF4IC4-LABEL: @select_i32_from_icmp
 ; CHECK-VF4IC4:      vector.body:
@@ -101,14 +100,15 @@ exit:                                     ; preds = %for.body
 define i32 @select_const_i32_from_fcmp(ptr nocapture readonly %v, i64 %n) #0 {
 ; CHECK-VF4IC1-LABEL: @select_const_i32_from_fcmp
 ; CHECK-VF4IC1:      vector.body:
-; CHECK-VF4IC1:        [[VEC_PHI:%.*]] = phi <vscale x 4 x i32> [ shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 2, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer), %vector.ph ], [ [[VEC_SEL:%.*]], %vector.body ]
+; CHECK-VF4IC1:        [[VEC_PHI:%.*]] = phi <vscale x 4 x i1> [ zeroinitializer, %vector.ph ], [ [[VEC_SEL:%.*]], %vector.body ]
 ; CHECK-VF4IC1:        [[VEC_LOAD:%.*]] = load <vscale x 4 x float>
 ; CHECK-VF4IC1-NEXT:   [[VEC_ICMP:%.*]] = fcmp fast ueq <vscale x 4 x float> [[VEC_LOAD]], shufflevector (<vscale x 4 x float> insertelement (<vscale x 4 x float> poison, float 3.000000e+00, i64 0), <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer)
-; CHECK-VF4IC1-NEXT:   [[VEC_SEL]] = select <vscale x 4 x i1> [[VEC_ICMP]], <vscale x 4 x i32> [[VEC_PHI]], <vscale x 4 x i32> shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 1, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)
+; CHECK-VF4IC1-NEXT:   [[NOT:%*]] = xor <vscale x 4 x i1> [[VEC_ICMP]], shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer)
+; CHECK-VF4IC1-NEXT:   [[VEC_SEL]] = or <vscale x 4 x i1> [[VEC_PHI]], [[NOT]]
 ; CHECK-VF4IC1:      middle.block:
-; CHECK-VF4IC1-NEXT:   [[FIN_ICMP:%.*]] = icmp ne <vscale x 4 x i32> [[VEC_SEL]], shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 2, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)
-; CHECK-VF4IC1-NEXT:   [[OR_RDX:%.*]] = call i1 @llvm.vector.reduce.or.nxv4i1(<vscale x 4 x i1> [[FIN_ICMP]])
-; CHECK-VF4IC1-NEXT:   {{.*}} = select i1 [[OR_RDX]], i32 1, i32 2
+; CHECK-VF4IC1-NEXT:   [[OR_RDX:%.*]] = call i1 @llvm.vector.reduce.or.nxv4i1(<vscale x 4 x i1> [[VEC_SEL]])
+; CHECK-VF4IC1-NEXT:   [[FR:%.*]] = freeze i1 [[OR_RDX]]
+; CHECK-VF4IC1-NEXT:   {{.*}} = select i1 [[FR]], i32 1, i32 2
 
 ; CHECK-VF4IC4-LABEL: @select_const_i32_from_fcmp
 ; CHECK-VF4IC4:      vector.body:
@@ -156,17 +156,17 @@ exit:                                     ; preds = %for.body
 define i32 @pred_select_const_i32_from_icmp(ptr noalias nocapture readonly %src1, ptr noalias nocapture readonly %src2, i64 %n) #0 {
 ; CHECK-VF4IC1-LABEL: @pred_select_const_i32_from_icmp
 ; CHECK-VF4IC1:      vector.body:
-; CHECK-VF4IC1:        [[VEC_PHI:%.*]] = phi <vscale x 4 x i32> [ zeroinitializer, %vector.ph ], [ [[VEC_SEL:%.*]], %vector.body ]
+; CHECK-VF4IC1:        [[VEC_PHI:%.*]] = phi <vscale x 4 x i1> [ zeroinitializer, %vector.ph ], [ [[VEC_SEL:%.*]], %vector.body ]
 ; CHECK-VF4IC1:        [[VEC_LOAD:%.*]] = load <vscale x 4 x i32>
 ; CHECK-VF4IC1:        [[MASK:%.*]] = icmp sgt <vscale x 4 x i32> [[VEC_LOAD]], shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 35, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)
 ; CHECK-VF4IC1:        [[MASKED_LOAD:%.*]] = call <vscale x 4 x i32> @llvm.masked.load.nxv4i32.p0(ptr {{%.*}}, i32 4, <vscale x 4 x i1> [[MASK]], <vscale x 4 x i32> poison)
 ; CHECK-VF4IC1-NEXT:   [[VEC_ICMP:%.*]] = icmp eq <vscale x 4 x i32> [[MASKED_LOAD]], shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 2, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)
-; CHECK-VF4IC1-NEXT:   [[VEC_SEL_TMP:%.*]] = select <vscale x 4 x i1> [[VEC_ICMP]], <vscale x 4 x i32> shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 1, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer), <vscale x 4 x i32> [[VEC_PHI]]
-; CHECK-VF4IC1:        [[VEC_SEL:%.*]] = select <vscale x 4 x i1> [[MASK]], <vscale x 4 x i32> [[VEC_SEL_TMP]], <vscale x 4 x i32> [[VEC_PHI]]
+; CHECK-VF4IC1-NEXT:   [[VEC_SEL_TMP:%.*]] = or <vscale x 4 x i1> [[VEC_PHI]], [[VEC_ICMP]]
+; CHECK-VF4IC1:        [[VEC_SEL:%.*]] = select <vscale x 4 x i1> [[MASK]], <vscale x 4 x i1> [[VEC_SEL_TMP]], <vscale x 4 x i1> [[VEC_PHI]]
 ; CHECK-VF4IC1:      middle.block:
-; CHECK-VF4IC1-NEXT:   [[FIN_ICMP:%.*]] = icmp ne <vscale x 4 x i32> [[VEC_SEL]], zeroinitializer
-; CHECK-VF4IC1-NEXT:   [[OR_RDX:%.*]] = call i1 @llvm.vector.reduce.or.nxv4i1(<vscale x 4 x i1> [[FIN_ICMP]])
-; CHECK-VF4IC1-NEXT:   {{.*}} = select i1 [[OR_RDX]], i32 1, i32 0
+; CHECK-VF4IC1-NEXT:   [[OR_RDX:%.*]] = call i1 @llvm.vector.reduce.or.nxv4i1(<vscale x 4 x i1> [[VEC_SEL]])
+; CHECK-VF4IC1-NEXT:   [[FR:%.*]] = freeze i1 [[OR_RDX]]
+; CHECK-VF4IC1-NEXT:   {{.*}} = select i1 [[FR]], i32 1, i32 0
 
 ; CHECK-VF4IC4-LABEL: @pred_select_const_i32_from_icmp
 ; CHECK-VF4IC4:      vector.body:
diff --git a/llvm/test/Transforms/LoopVectorize/PowerPC/vectorize-force-tail-with-evl.ll b/llvm/test/Transforms/LoopVectorize/PowerPC/vectorize-force-tail-with-evl.ll
new file mode 100644
index 000000000000..2ce2a45a811a
--- /dev/null
+++ b/llvm/test/Transforms/LoopVectorize/PowerPC/vectorize-force-tail-with-evl.ll
@@ -0,0 +1,51 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
+; RUN: opt -passes=loop-vectorize \
+; RUN: -force-tail-folding-style=data-with-evl \
+; RUN: -prefer-predicate-over-epilogue=predicate-dont-vectorize \
+; RUN: -mtriple=powerpc64le-unknown-linux-gnu \
+; RUN: -mcpu=pwr10 -S < %s | FileCheck %s
+
+; RUN: opt -passes=loop-vectorize \
+; RUN: -force-tail-folding-style=none \
+; RUN: -prefer-predicate-over-epilogue=predicate-dont-vectorize \
+; RUN: -mtriple=powerpc64le-unknown-linux-gnu \
+; RUN: -mcpu=pwr10 -S < %s | FileCheck %s
+
+define void @foo(ptr noalias %a, ptr noalias %b, ptr noalias %c, i64 %N) {
+; CHECK-LABEL: @foo(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    br label [[FOR_BODY:%.*]]
+; CHECK:       for.body:
+; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
+; CHECK-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 [[IV]]
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
+; CHECK-NEXT:    [[ARRAYIDX2:%.*]] = getelementptr inbounds i32, ptr [[C:%.*]], i64 [[IV]]
+; CHECK-NEXT:    [[TMP1:%.*]] = load i32, ptr [[ARRAYIDX2]], align 4
+; CHECK-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP1]], [[TMP0]]
+; CHECK-NEXT:    [[ARRAYIDX4:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[IV]]
+; CHECK-NEXT:    store i32 [[ADD]], ptr [[ARRAYIDX4]], align 4
+; CHECK-NEXT:    [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
+; CHECK-NEXT:    [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], [[N:%.*]]
+; CHECK-NEXT:    br i1 [[EXITCOND_NOT]], label [[FOR_COND_CLEANUP:%.*]], label [[FOR_BODY]]
+; CHECK:       for.cond.cleanup:
+; CHECK-NEXT:    ret void
+;
+entry:
+  br label %for.body
+
+for.body:
+  %iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]
+  %arrayidx = getelementptr inbounds i32, ptr %b, i64 %iv
+  %0 = load i32, ptr %arrayidx, align 4
+  %arrayidx2 = getelementptr inbounds i32, ptr %c, i64 %iv
+  %1 = load i32, ptr %arrayidx2, align 4
+  %add = add nsw i32 %1, %0
+  %arrayidx4 = getelementptr inbounds i32, ptr %a, i64 %iv
+  store i32 %add, ptr %arrayidx4, align 4
+  %iv.next = add nuw nsw i64 %iv, 1
+  %exitcond.not = icmp eq i64 %iv.next, %N
+  br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup:
+  ret void
+}
diff --git a/llvm/test/Transforms/LoopVectorize/PowerPC/vplan-force-tail-with-evl.ll b/llvm/test/Transforms/LoopVectorize/PowerPC/vplan-force-tail-with-evl.ll
new file mode 100644
index 000000000000..5d1a471c5d16
--- /dev/null
+++ b/llvm/test/Transforms/LoopVectorize/PowerPC/vplan-force-tail-with-evl.ll
@@ -0,0 +1,117 @@
+; REQUIRES: asserts
+
+; RUN: opt -passes=loop-vectorize -debug-only=loop-vectorize \
+; RUN: -force-tail-folding-style=data-with-evl \
+; RUN: -prefer-predicate-over-epilogue=predicate-dont-vectorize \
+; RUN: -mtriple=powerpc64le-unknown-linux-gnu \
+; RUN: -mcpu=pwr10 -disable-output < %s 2>&1 | FileCheck %s
+
+define void @foo(ptr noalias %a, ptr noalias %b, ptr noalias %c, i64 %N) {
+; CHECK-LABEL: VPlan 'Initial VPlan for VF={2,4},UF>=1' {
+; CHECK-NEXT: Live-in vp<%0> = VF * UF
+; CHECK-NEXT: Live-in vp<%1> = vector-trip-count
+; CHECK-NEXT: Live-in vp<%2> = backedge-taken count
+; CHECK-NEXT: Live-in ir<%N> = original trip-count
+; CHECK-EMPTY:
+; CHECK-NEXT: vector.ph:
+; CHECK-NEXT: Successor(s): vector loop
+; CHECK-EMPTY:
+; CHECK-NEXT: <x1> vector loop: {
+; CHECK-NEXT:   vector.body:
+; CHECK-NEXT:     EMIT vp<%3> = CANONICAL-INDUCTION ir<0>, vp<%16>
+; CHECK-NEXT:     WIDEN-INDUCTION %iv = phi 0, %iv.next, ir<1>
+; CHECK-NEXT:     EMIT vp<%5> = icmp ule ir<%iv>, vp<%2>
+; CHECK-NEXT:   Successor(s): pred.store
+; CHECK-EMPTY:
+; CHECK-NEXT:  <xVFxUF> pred.store: {
+; CHECK-NEXT:    pred.store.entry:
+; CHECK-NEXT:      BRANCH-ON-MASK vp<%5>
+; CHECK-NEXT:    Successor(s): pred.store.if, pred.store.continue
+; CHECK-EMPTY:
+; CHECK-NEXT:    pred.store.if:
+; CHECK-NEXT:      vp<%6> = SCALAR-STEPS vp<%3>, ir<1>
+; CHECK-NEXT:      REPLICATE ir<%arrayidx> = getelementptr inbounds ir<%b>, vp<%6>
+; CHECK-NEXT:      REPLICATE ir<%0> = load ir<%arrayidx>
+; CHECK-NEXT:      REPLICATE ir<%arrayidx2> = getelementptr inbounds ir<%c>, vp<%6>
+; CHECK-NEXT:      REPLICATE ir<%1> = load ir<%arrayidx2>
+; CHECK-NEXT:      REPLICATE ir<%arrayidx4> = getelementptr inbounds ir<%a>, vp<%6>
+; CHECK-NEXT:      REPLICATE ir<%add> = add nsw ir<%1>, ir<%0>
+; CHECK-NEXT:      REPLICATE store ir<%add>, ir<%arrayidx4>
+; CHECK-NEXT:    Successor(s): pred.store.continue
+; CHECK-EMPTY:
+; CHECK-NEXT:    pred.store.continue:
+; CHECK-NEXT:      PHI-PREDICATED-INSTRUCTION vp<%14> = ir<%0>
+; CHECK-NEXT:      PHI-PREDICATED-INSTRUCTION vp<%15> = ir<%1>
+; CHECK-NEXT:    No successors
+; CHECK-NEXT:  }
+; CHECK-NEXT:  Successor(s): for.body.2
+; CHECK-EMPTY:
+; CHECK-NEXT:  for.body.2:
+; CHECK-NEXT:     EMIT vp<%16> = add vp<%3>, vp<%0>
+; CHECK-NEXT:     EMIT branch-on-count vp<%16>, vp<%1>
+; CHECK-NEXT:   No successors
+; CHECK-NEXT: }
+;
+entry:
+  br label %for.body
+
+for.body:
+  %iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]
+  %arrayidx = getelementptr inbounds i32, ptr %b, i64 %iv
+  %0 = load i32, ptr %arrayidx, align 4
+  %arrayidx2 = getelementptr inbounds i32, ptr %c, i64 %iv
+  %1 = load i32, ptr %arrayidx2, align 4
+  %add = add nsw i32 %1, %0
+  %arrayidx4 = getelementptr inbounds i32, ptr %a, i64 %iv
+  store i32 %add, ptr %arrayidx4, align 4
+  %iv.next = add nuw nsw i64 %iv, 1
+  %exitcond.not = icmp eq i64 %iv.next, %N
+  br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup:
+  ret void
+}
+
+define void @safe_dep(ptr %p) {
+; CHECK-LABEL: VPlan 'Initial VPlan for VF={2},UF>=1' {
+; CHECK-NEXT: Live-in vp<%0> = VF * UF
+; CHECK-NEXT: Live-in vp<%1> = vector-trip-count
+; CHECK-NEXT: Live-in ir<512> = original trip-count
+; CHECK-EMPTY:
+; CHECK-NEXT: vector.ph:
+; CHECK-NEXT: Successor(s): vector loop
+; CHECK-EMPTY:
+; CHECK-NEXT: <x1> vector loop: {
+; CHECK-NEXT:   vector.body:
+; CHECK-NEXT:     EMIT vp<%2> = CANONICAL-INDUCTION ir<0>, vp<%10>
+; CHECK-NEXT:     vp<%3> = SCALAR-STEPS vp<%2>, ir<1>
+; CHECK-NEXT:     CLONE ir<%a1> = getelementptr ir<%p>, vp<%3>
+; CHECK-NEXT:     vp<%5> = vector-pointer ir<%a1>
+; CHECK-NEXT:     WIDEN ir<%v> = load vp<%5>
+; CHECK-NEXT:     CLONE ir<%offset> = add vp<%3>, ir<100>
+; CHECK-NEXT:     CLONE ir<%a2> = getelementptr ir<%p>, ir<%offset>
+; CHECK-NEXT:     vp<%9> = vector-pointer ir<%a2>
+; CHECK-NEXT:     WIDEN store vp<%9>, ir<%v>
+; CHECK-NEXT:     EMIT vp<%10> = add nuw vp<%2>, vp<%0>
+; CHECK-NEXT:     EMIT branch-on-count vp<%10>, vp<%1>
+; CHECK-NEXT:   No successors
+; CHECK-NEXT: }
+;
+entry:
+  br label %loop
+
+loop:
+  %iv = phi i64 [0, %entry], [%iv.next, %loop]
+  %a1 = getelementptr i64, ptr %p, i64 %iv
+  %v = load i64, ptr %a1, align 32
+  %offset = add i64 %iv, 100
+  %a2 = getelementptr i64, ptr %p, i64 %offset
+  store i64 %v, ptr %a2, align 32
+  %iv.next = add i64 %iv, 1
+  %cmp = icmp ne i64 %iv, 511
+  br i1 %cmp, label %loop, label %exit
+
+exit:
+  ret void
+}
+
diff --git a/llvm/test/Transforms/LoopVectorize/RISCV/inloop-reduction.ll b/llvm/test/Transforms/LoopVectorize/RISCV/inloop-reduction.ll
index 57e1dc9051f4..b876e9d2c1a5 100644
--- a/llvm/test/Transforms/LoopVectorize/RISCV/inloop-reduction.ll
+++ b/llvm/test/Transforms/LoopVectorize/RISCV/inloop-reduction.ll
@@ -1,11 +1,13 @@
 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
 ; RUN: opt -mtriple riscv64-linux-gnu -mattr=+v,+d -passes=loop-vectorize < %s -S -o - | FileCheck %s -check-prefix=OUTLOOP
 ; RUN: opt -mtriple riscv64-linux-gnu -mattr=+v,+d -passes=loop-vectorize -prefer-inloop-reductions < %s -S -o - | FileCheck %s -check-prefix=INLOOP
-
+; RUN: opt -passes=loop-vectorize -force-tail-folding-style=data-with-evl -prefer-predicate-over-epilogue=predicate-dont-vectorize -mtriple=riscv64 -mattr=+v -S < %s 2>&1 | FileCheck --check-prefix=IF-EVL %s
 
 target datalayout = "e-m:e-p:64:64-i64:64-i128:128-n64-S128"
 target triple = "riscv64"
 
+; FIXME: inloop reductions are not supported yet with predicated vectorization.
+
 define i32 @add_i16_i32(ptr nocapture readonly %x, i32 %n) {
 ; OUTLOOP-LABEL: @add_i16_i32(
 ; OUTLOOP-NEXT:  entry:
@@ -115,6 +117,70 @@ define i32 @add_i16_i32(ptr nocapture readonly %x, i32 %n) {
 ; INLOOP-NEXT:    [[R_0_LCSSA:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[ADD_LCSSA]], [[FOR_COND_CLEANUP_LOOPEXIT]] ]
 ; INLOOP-NEXT:    ret i32 [[R_0_LCSSA]]
 ;
+; IF-EVL-LABEL: @add_i16_i32(
+; IF-EVL-NEXT:  entry:
+; IF-EVL-NEXT:    [[CMP6:%.*]] = icmp sgt i32 [[N:%.*]], 0
+; IF-EVL-NEXT:    br i1 [[CMP6]], label [[FOR_BODY_PREHEADER:%.*]], label [[FOR_COND_CLEANUP:%.*]]
+; IF-EVL:       for.body.preheader:
+; IF-EVL-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; IF-EVL:       vector.ph:
+; IF-EVL-NEXT:    [[TMP0:%.*]] = call i32 @llvm.vscale.i32()
+; IF-EVL-NEXT:    [[TMP1:%.*]] = mul i32 [[TMP0]], 4
+; IF-EVL-NEXT:    [[TMP2:%.*]] = call i32 @llvm.vscale.i32()
+; IF-EVL-NEXT:    [[TMP3:%.*]] = mul i32 [[TMP2]], 4
+; IF-EVL-NEXT:    [[TMP4:%.*]] = sub i32 [[TMP3]], 1
+; IF-EVL-NEXT:    [[N_RND_UP:%.*]] = add i32 [[N]], [[TMP4]]
+; IF-EVL-NEXT:    [[N_MOD_VF:%.*]] = urem i32 [[N_RND_UP]], [[TMP1]]
+; IF-EVL-NEXT:    [[N_VEC:%.*]] = sub i32 [[N_RND_UP]], [[N_MOD_VF]]
+; IF-EVL-NEXT:    [[TRIP_COUNT_MINUS_1:%.*]] = sub i32 [[N]], 1
+; IF-EVL-NEXT:    [[TMP5:%.*]] = call i32 @llvm.vscale.i32()
+; IF-EVL-NEXT:    [[TMP6:%.*]] = mul i32 [[TMP5]], 4
+; IF-EVL-NEXT:    [[BROADCAST_SPLATINSERT1:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[TRIP_COUNT_MINUS_1]], i64 0
+; IF-EVL-NEXT:    [[BROADCAST_SPLAT2:%.*]] = shufflevector <vscale x 4 x i32> [[BROADCAST_SPLATINSERT1]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
+; IF-EVL-NEXT:    br label [[VECTOR_BODY:%.*]]
+; IF-EVL:       vector.body:
+; IF-EVL-NEXT:    [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; IF-EVL-NEXT:    [[VEC_PHI:%.*]] = phi <vscale x 4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP14:%.*]], [[VECTOR_BODY]] ]
+; IF-EVL-NEXT:    [[TMP7:%.*]] = add i32 [[INDEX]], 0
+; IF-EVL-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[INDEX]], i64 0
+; IF-EVL-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 4 x i32> [[BROADCAST_SPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
+; IF-EVL-NEXT:    [[TMP8:%.*]] = call <vscale x 4 x i32> @llvm.experimental.stepvector.nxv4i32()
+; IF-EVL-NEXT:    [[TMP9:%.*]] = add <vscale x 4 x i32> zeroinitializer, [[TMP8]]
+; IF-EVL-NEXT:    [[VEC_IV:%.*]] = add <vscale x 4 x i32> [[BROADCAST_SPLAT]], [[TMP9]]
+; IF-EVL-NEXT:    [[TMP10:%.*]] = icmp ule <vscale x 4 x i32> [[VEC_IV]], [[BROADCAST_SPLAT2]]
+; IF-EVL-NEXT:    [[TMP11:%.*]] = getelementptr inbounds i16, ptr [[X:%.*]], i32 [[TMP7]]
+; IF-EVL-NEXT:    [[TMP12:%.*]] = getelementptr inbounds i16, ptr [[TMP11]], i32 0
+; IF-EVL-NEXT:    [[WIDE_MASKED_LOAD:%.*]] = call <vscale x 4 x i16> @llvm.masked.load.nxv4i16.p0(ptr [[TMP12]], i32 2, <vscale x 4 x i1> [[TMP10]], <vscale x 4 x i16> poison)
+; IF-EVL-NEXT:    [[TMP13:%.*]] = sext <vscale x 4 x i16> [[WIDE_MASKED_LOAD]] to <vscale x 4 x i32>
+; IF-EVL-NEXT:    [[TMP14]] = add <vscale x 4 x i32> [[VEC_PHI]], [[TMP13]]
+; IF-EVL-NEXT:    [[TMP15:%.*]] = select <vscale x 4 x i1> [[TMP10]], <vscale x 4 x i32> [[TMP14]], <vscale x 4 x i32> [[VEC_PHI]]
+; IF-EVL-NEXT:    [[INDEX_NEXT]] = add i32 [[INDEX]], [[TMP6]]
+; IF-EVL-NEXT:    [[TMP16:%.*]] = icmp eq i32 [[INDEX_NEXT]], [[N_VEC]]
+; IF-EVL-NEXT:    br i1 [[TMP16]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
+; IF-EVL:       middle.block:
+; IF-EVL-NEXT:    [[TMP17:%.*]] = call i32 @llvm.vector.reduce.add.nxv4i32(<vscale x 4 x i32> [[TMP15]])
+; IF-EVL-NEXT:    br i1 true, label [[FOR_COND_CLEANUP_LOOPEXIT:%.*]], label [[SCALAR_PH]]
+; IF-EVL:       scalar.ph:
+; IF-EVL-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i32 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[FOR_BODY_PREHEADER]] ]
+; IF-EVL-NEXT:    [[BC_MERGE_RDX:%.*]] = phi i32 [ 0, [[FOR_BODY_PREHEADER]] ], [ [[TMP17]], [[MIDDLE_BLOCK]] ]
+; IF-EVL-NEXT:    br label [[FOR_BODY:%.*]]
+; IF-EVL:       for.body:
+; IF-EVL-NEXT:    [[I_08:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
+; IF-EVL-NEXT:    [[R_07:%.*]] = phi i32 [ [[ADD:%.*]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
+; IF-EVL-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i16, ptr [[X]], i32 [[I_08]]
+; IF-EVL-NEXT:    [[TMP18:%.*]] = load i16, ptr [[ARRAYIDX]], align 2
+; IF-EVL-NEXT:    [[CONV:%.*]] = sext i16 [[TMP18]] to i32
+; IF-EVL-NEXT:    [[ADD]] = add nsw i32 [[R_07]], [[CONV]]
+; IF-EVL-NEXT:    [[INC]] = add nuw nsw i32 [[I_08]], 1
+; IF-EVL-NEXT:    [[EXITCOND:%.*]] = icmp eq i32 [[INC]], [[N]]
+; IF-EVL-NEXT:    br i1 [[EXITCOND]], label [[FOR_COND_CLEANUP_LOOPEXIT]], label [[FOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
+; IF-EVL:       for.cond.cleanup.loopexit:
+; IF-EVL-NEXT:    [[ADD_LCSSA:%.*]] = phi i32 [ [[ADD]], [[FOR_BODY]] ], [ [[TMP17]], [[MIDDLE_BLOCK]] ]
+; IF-EVL-NEXT:    br label [[FOR_COND_CLEANUP]]
+; IF-EVL:       for.cond.cleanup:
+; IF-EVL-NEXT:    [[R_0_LCSSA:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[ADD_LCSSA]], [[FOR_COND_CLEANUP_LOOPEXIT]] ]
+; IF-EVL-NEXT:    ret i32 [[R_0_LCSSA]]
+;
 entry:
   %cmp6 = icmp sgt i32 %n, 0
   br i1 %cmp6, label %for.body, label %for.cond.cleanup
diff --git a/llvm/test/Transforms/LoopVectorize/RISCV/select-cmp-reduction.ll b/llvm/test/Transforms/LoopVectorize/RISCV/select-cmp-reduction.ll
index 34a7987bb40a..a235da069b32 100644
--- a/llvm/test/Transforms/LoopVectorize/RISCV/select-cmp-reduction.ll
+++ b/llvm/test/Transforms/LoopVectorize/RISCV/select-cmp-reduction.ll
@@ -12,25 +12,24 @@ define i32 @select_icmp(i32 %x, i32 %y, ptr nocapture readonly %c, i64 %n) #0 {
 ; CHECK-NEXT:    [[N_VEC:%.*]] = sub i64 %n, [[N_MOD_VF]]
 ; CHECK-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <4 x i32> poison, i32 [[X:%.*]], i64 0
 ; CHECK-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <4 x i32> [[BROADCAST_SPLATINSERT]], <4 x i32> poison, <4 x i32> zeroinitializer
-; CHECK-NEXT:    [[BROADCAST_SPLATINSERT1:%.*]] = insertelement <4 x i32> poison, i32 [[Y:%.*]], i64 0
-; CHECK-NEXT:    [[BROADCAST_SPLAT2:%.*]] = shufflevector <4 x i32> [[BROADCAST_SPLATINSERT1]], <4 x i32> poison, <4 x i32> zeroinitializer
 ; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
 ; CHECK:       vector.body:
 ; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP5:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <4 x i1> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP5:%.*]], [[VECTOR_BODY]] ]
 ; CHECK-NEXT:    [[TMP0:%.*]] = add i64 [[INDEX]], 0
 ; CHECK-NEXT:    [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[C:%.*]], i64 [[TMP0]]
 ; CHECK-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i32 0
 ; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP2]], align 4
 ; CHECK-NEXT:    [[TMP4:%.*]] = icmp slt <4 x i32> [[WIDE_LOAD]], [[BROADCAST_SPLAT]]
-; CHECK-NEXT:    [[TMP5]] = select <4 x i1> [[TMP4]], <4 x i32> [[VEC_PHI]], <4 x i32> [[BROADCAST_SPLAT2]]
+; CHECK-NEXT:    [[NOT:%.*]] = xor <4 x i1> [[TMP4]], <i1 true, i1 true, i1 true, i1 true>
+; CHECK-NEXT:    [[TMP5]] = or <4 x i1> [[VEC_PHI]], [[NOT]]
 ; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
 ; CHECK-NEXT:    [[TMP6:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
 ; CHECK-NEXT:    br i1 [[TMP6]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
 ; CHECK:       middle.block:
-; CHECK-NEXT:    [[RDX_SELECT_CMP:%.*]] = icmp ne <4 x i32> [[TMP5]], zeroinitializer
-; CHECK-NEXT:    [[TMP7:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[RDX_SELECT_CMP]])
-; CHECK-NEXT:    [[RDX_SELECT:%.*]] = select i1 [[TMP7]], i32 [[Y]], i32 0
+; CHECK-NEXT:    [[TMP7:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[TMP5]])
+; CHECK-NEXT:    [[FR:%.*]] = freeze i1 [[TMP7]]
+; CHECK-NEXT:    [[RDX_SELECT:%.*]] = select i1 [[FR]], i32 %y, i32 0
 ;
 ; SCALABLE-LABEL: @select_icmp
 ; SCALABLE:       vector.ph:
@@ -42,25 +41,24 @@ define i32 @select_icmp(i32 %x, i32 %y, ptr nocapture readonly %c, i64 %n) #0 {
 ; SCALABLE-NEXT:    [[TMP11:%.*]] = mul i64 [[TMP10]], 4
 ; SCALABLE-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[X:%.*]], i64 0
 ; SCALABLE-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 4 x i32> [[BROADCAST_SPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
-; SCALABLE-NEXT:    [[BROADCAST_SPLATINSERT1:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[Y:%.*]], i64 0
-; SCALABLE-NEXT:    [[BROADCAST_SPLAT2:%.*]] = shufflevector <vscale x 4 x i32> [[BROADCAST_SPLATINSERT1]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 ; SCALABLE-NEXT:    br label [[VECTOR_BODY:%.*]]
 ; SCALABLE:       vector.body:
 ; SCALABLE-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; SCALABLE-NEXT:    [[VEC_PHI:%.*]] = phi <vscale x 4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP9:%.*]], [[VECTOR_BODY]] ]
+; SCALABLE-NEXT:    [[VEC_PHI:%.*]] = phi <vscale x 4 x i1> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP9:%.*]], [[VECTOR_BODY]] ]
 ; SCALABLE-NEXT:    [[TMP4:%.*]] = add i64 [[INDEX]], 0
 ; SCALABLE-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[C:%.*]], i64 [[TMP4]]
 ; SCALABLE-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[TMP5]], i32 0
 ; SCALABLE-NEXT:    [[WIDE_LOAD:%.*]] = load <vscale x 4 x i32>, ptr [[TMP6]], align 4
 ; SCALABLE-NEXT:    [[TMP8:%.*]] = icmp slt <vscale x 4 x i32> [[WIDE_LOAD]], [[BROADCAST_SPLAT]]
-; SCALABLE-NEXT:    [[TMP9]] = select <vscale x 4 x i1> [[TMP8]], <vscale x 4 x i32> [[VEC_PHI]], <vscale x 4 x i32> [[BROADCAST_SPLAT2]]
+; SCALABLE-NEXT:    [[NOT:%.*]] = xor <vscale x 4 x i1> [[TMP8]], shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer)
+; SCALABLE-NEXT:    [[TMP9]] = or <vscale x 4 x i1> [[VEC_PHI]], [[NOT]]
 ; SCALABLE-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP11]]
 ; SCALABLE-NEXT:    [[TMP12:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
 ; SCALABLE-NEXT:    br i1 [[TMP12]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
 ; SCALABLE:       middle.block:
-; SCALABLE-NEXT:    [[RDX_SELECT_CMP:%.*]] = icmp ne <vscale x 4 x i32> [[TMP9]], zeroinitializer
-; SCALABLE-NEXT:    [[TMP13:%.*]] = call i1 @llvm.vector.reduce.or.nxv4i1(<vscale x 4 x i1> [[RDX_SELECT_CMP]])
-; SCALABLE-NEXT:    [[RDX_SELECT:%.*]] = select i1 [[TMP13]], i32 [[Y]], i32 0
+; SCALABLE-NEXT:    [[TMP13:%.*]] = call i1 @llvm.vector.reduce.or.nxv4i1(<vscale x 4 x i1> [[TMP9]])
+; SCALABLE-NEXT:    [[FR:%.*]] = freeze i1 [[TMP13]]
+; SCALABLE-NEXT:    [[RDX_SELECT:%.*]] = select i1 [[FR]], i32 %y, i32 0
 ;
 entry:
   br label %for.body
@@ -87,25 +85,24 @@ define i32 @select_fcmp(float %x, i32 %y, ptr nocapture readonly %c, i64 %n) #0
 ; CHECK-NEXT:    [[N_VEC:%.*]] = sub i64 %n, [[N_MOD_VF]]
 ; CHECK-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <4 x float> poison, float [[X:%.*]], i64 0
 ; CHECK-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <4 x float> [[BROADCAST_SPLATINSERT]], <4 x float> poison, <4 x i32> zeroinitializer
-; CHECK-NEXT:    [[BROADCAST_SPLATINSERT1:%.*]] = insertelement <4 x i32> poison, i32 [[Y:%.*]], i64 0
-; CHECK-NEXT:    [[BROADCAST_SPLAT2:%.*]] = shufflevector <4 x i32> [[BROADCAST_SPLATINSERT1]], <4 x i32> poison, <4 x i32> zeroinitializer
 ; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
 ; CHECK:       vector.body:
 ; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP5:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <4 x i1> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP5:%.*]], [[VECTOR_BODY]] ]
 ; CHECK-NEXT:    [[TMP0:%.*]] = add i64 [[INDEX]], 0
 ; CHECK-NEXT:    [[TMP1:%.*]] = getelementptr inbounds float, ptr [[C:%.*]], i64 [[TMP0]]
 ; CHECK-NEXT:    [[TMP2:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i32 0
 ; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <4 x float>, ptr [[TMP2]], align 4
 ; CHECK-NEXT:    [[TMP4:%.*]] = fcmp fast olt <4 x float> [[WIDE_LOAD]], [[BROADCAST_SPLAT]]
-; CHECK-NEXT:    [[TMP5]] = select <4 x i1> [[TMP4]], <4 x i32> [[VEC_PHI]], <4 x i32> [[BROADCAST_SPLAT2]]
+; CHECK-NEXT:    [[NOT:%.*]] = xor <4 x i1> [[TMP4]], <i1 true, i1 true, i1 true, i1 true>
+; CHECK-NEXT:    [[TMP5]] = or <4 x i1> [[VEC_PHI]], [[NOT]]
 ; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
 ; CHECK-NEXT:    [[TMP6:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
 ; CHECK-NEXT:    br i1 [[TMP6]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
 ; CHECK:       middle.block:
-; CHECK-NEXT:    [[RDX_SELECT_CMP:%.*]] = icmp ne <4 x i32> [[TMP5]], zeroinitializer
-; CHECK-NEXT:    [[TMP7:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[RDX_SELECT_CMP]])
-; CHECK-NEXT:    [[RDX_SELECT:%.*]] = select i1 [[TMP7]], i32 [[Y]], i32 0
+; CHECK-NEXT:    [[TMP7:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[TMP5]])
+; CHECK-NEXT:    [[FR:%.*]] = freeze i1 [[TMP7]]
+; CHECK-NEXT:    [[RDX_SELECT:%.*]] = select i1 [[FR]], i32 %y, i32 0
 ;
 ; SCALABLE-LABEL: @select_fcmp
 ; SCALABLE:       vector.ph:
@@ -117,25 +114,24 @@ define i32 @select_fcmp(float %x, i32 %y, ptr nocapture readonly %c, i64 %n) #0
 ; SCALABLE-NEXT:    [[TMP11:%.*]] = mul i64 [[TMP10]], 4
 ; SCALABLE-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[X:%.*]], i64 0
 ; SCALABLE-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 4 x float> [[BROADCAST_SPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
-; SCALABLE-NEXT:    [[BROADCAST_SPLATINSERT1:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[Y:%.*]], i64 0
-; SCALABLE-NEXT:    [[BROADCAST_SPLAT2:%.*]] = shufflevector <vscale x 4 x i32> [[BROADCAST_SPLATINSERT1]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 ; SCALABLE-NEXT:    br label [[VECTOR_BODY:%.*]]
 ; SCALABLE:       vector.body:
 ; SCALABLE-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; SCALABLE-NEXT:    [[VEC_PHI:%.*]] = phi <vscale x 4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP9:%.*]], [[VECTOR_BODY]] ]
+; SCALABLE-NEXT:    [[VEC_PHI:%.*]] = phi <vscale x 4 x i1> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP9:%.*]], [[VECTOR_BODY]] ]
 ; SCALABLE-NEXT:    [[TMP4:%.*]] = add i64 [[INDEX]], 0
 ; SCALABLE-NEXT:    [[TMP5:%.*]] = getelementptr inbounds float, ptr [[C:%.*]], i64 [[TMP4]]
 ; SCALABLE-NEXT:    [[TMP6:%.*]] = getelementptr inbounds float, ptr [[TMP5]], i32 0
 ; SCALABLE-NEXT:    [[WIDE_LOAD:%.*]] = load <vscale x 4 x float>, ptr [[TMP6]], align 4
 ; SCALABLE-NEXT:    [[TMP8:%.*]] = fcmp fast olt <vscale x 4 x float> [[WIDE_LOAD]], [[BROADCAST_SPLAT]]
-; SCALABLE-NEXT:    [[TMP9]] = select <vscale x 4 x i1> [[TMP8]], <vscale x 4 x i32> [[VEC_PHI]], <vscale x 4 x i32> [[BROADCAST_SPLAT2]]
+; SCALABLE-NEXT:    [[NOT:%.*]] = xor <vscale x 4 x i1> [[TMP8]], shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer)
+; SCALABLE-NEXT:    [[TMP9]] = or <vscale x 4 x i1> [[VEC_PHI]], [[NOT]]
 ; SCALABLE-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP11]]
 ; SCALABLE-NEXT:    [[TMP12:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
 ; SCALABLE-NEXT:    br i1 [[TMP12]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
 ; SCALABLE:       middle.block:
-; SCALABLE-NEXT:    [[RDX_SELECT_CMP:%.*]] = icmp ne <vscale x 4 x i32> [[TMP9]], zeroinitializer
-; SCALABLE-NEXT:    [[TMP13:%.*]] = call i1 @llvm.vector.reduce.or.nxv4i1(<vscale x 4 x i1> [[RDX_SELECT_CMP]])
-; SCALABLE-NEXT:    [[RDX_SELECT:%.*]] = select i1 [[TMP13]], i32 [[Y]], i32 0
+; SCALABLE-NEXT:    [[TMP13:%.*]] = call i1 @llvm.vector.reduce.or.nxv4i1(<vscale x 4 x i1> [[TMP9]])
+; SCALABLE-NEXT:    [[FR:%.*]] = freeze i1 [[TMP13]]
+; SCALABLE-NEXT:    [[RDX_SELECT:%.*]] = select i1 [[FR]], i32 %y, i32 0
 ;
 entry:
   br label %for.body
@@ -163,20 +159,21 @@ define i32 @select_const_i32_from_icmp(ptr nocapture readonly %v, i64 %n) #0 {
 ; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
 ; CHECK:       vector.body:
 ; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <4 x i32> [ <i32 3, i32 3, i32 3, i32 3>, [[VECTOR_PH]] ], [ [[TMP5:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <4 x i1> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP5:%.*]], [[VECTOR_BODY]] ]
 ; CHECK-NEXT:    [[TMP0:%.*]] = add i64 [[INDEX]], 0
 ; CHECK-NEXT:    [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[V:%.*]], i64 [[TMP0]]
 ; CHECK-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i32 0
 ; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP2]], align 4
 ; CHECK-NEXT:    [[TMP4:%.*]] = icmp eq <4 x i32> [[WIDE_LOAD]], <i32 3, i32 3, i32 3, i32 3>
-; CHECK-NEXT:    [[TMP5]] = select <4 x i1> [[TMP4]], <4 x i32> [[VEC_PHI]], <4 x i32> <i32 7, i32 7, i32 7, i32 7>
+; CHECK-NEXT:    [[NOT:%.*]] = xor <4 x i1> [[TMP4]], <i1 true, i1 true, i1 true, i1 true>
+; CHECK-NEXT:    [[TMP5]] = or <4 x i1> [[VEC_PHI]], [[NOT]]
 ; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
 ; CHECK-NEXT:    [[TMP6:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
 ; CHECK-NEXT:    br i1 [[TMP6]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
 ; CHECK:       middle.block:
-; CHECK-NEXT:    [[RDX_SELECT_CMP:%.*]] = icmp ne <4 x i32> [[TMP5]], <i32 3, i32 3, i32 3, i32 3>
-; CHECK-NEXT:    [[TMP7:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[RDX_SELECT_CMP]])
-; CHECK-NEXT:    [[RDX_SELECT:%.*]] = select i1 [[TMP7]], i32 7, i32 3
+; CHECK-NEXT:    [[TMP7:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[TMP5]])
+; CHECK-NEXT:    [[FR:%.*]] = freeze i1 [[TMP7]]
+; CHECK-NEXT:    [[RDX_SELECT:%.*]] = select i1 [[FR]], i32 7, i32 3
 ;
 ; SCALABLE-LABEL: @select_const_i32_from_icmp
 ; SCALABLE:       vector.ph:
@@ -189,20 +186,21 @@ define i32 @select_const_i32_from_icmp(ptr nocapture readonly %v, i64 %n) #0 {
 ; SCALABLE-NEXT:    br label [[VECTOR_BODY:%.*]]
 ; SCALABLE:       vector.body:
 ; SCALABLE-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; SCALABLE-NEXT:    [[VEC_PHI:%.*]] = phi <vscale x 4 x i32> [ shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 3, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer), [[VECTOR_PH]] ], [ [[TMP9:%.*]], [[VECTOR_BODY]] ]
+; SCALABLE-NEXT:    [[VEC_PHI:%.*]] = phi <vscale x 4 x i1> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP9:%.*]], [[VECTOR_BODY]] ]
 ; SCALABLE-NEXT:    [[TMP4:%.*]] = add i64 [[INDEX]], 0
 ; SCALABLE-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[V:%.*]], i64 [[TMP4]]
 ; SCALABLE-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[TMP5]], i32 0
 ; SCALABLE-NEXT:    [[WIDE_LOAD:%.*]] = load <vscale x 4 x i32>, ptr [[TMP6]], align 4
 ; SCALABLE-NEXT:    [[TMP8:%.*]] = icmp eq <vscale x 4 x i32> [[WIDE_LOAD]], shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 3, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)
-; SCALABLE-NEXT:    [[TMP9]] = select <vscale x 4 x i1> [[TMP8]], <vscale x 4 x i32> [[VEC_PHI]], <vscale x 4 x i32> shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 7, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)
+; SCALABLE-NEXT:    [[NOT:%.*]] = xor <vscale x 4 x i1> [[TMP8]], shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer)
+; SCALABLE-NEXT:    [[TMP9]] = or <vscale x 4 x i1> [[VEC_PHI]], [[NOT]]
 ; SCALABLE-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP11]]
 ; SCALABLE-NEXT:    [[TMP12:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
 ; SCALABLE-NEXT:    br i1 [[TMP12]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
 ; SCALABLE:       middle.block:
-; SCALABLE-NEXT:    [[RDX_SELECT_CMP:%.*]] = icmp ne <vscale x 4 x i32> [[TMP9]], shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 3, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)
-; SCALABLE-NEXT:    [[TMP13:%.*]] = call i1 @llvm.vector.reduce.or.nxv4i1(<vscale x 4 x i1> [[RDX_SELECT_CMP]])
-; SCALABLE-NEXT:    [[RDX_SELECT:%.*]] = select i1 [[TMP13]], i32 7, i32 3
+; SCALABLE-NEXT:    [[TMP13:%.*]] = call i1 @llvm.vector.reduce.or.nxv4i1(<vscale x 4 x i1> [[TMP9]])
+; SCALABLE-NEXT:    [[FR:%.*]] = freeze i1 [[TMP13]]
+; SCALABLE-NEXT:    [[RDX_SELECT:%.*]] = select i1 [[FR]], i32 7, i32 3
 ;
 entry:
   br label %for.body
@@ -227,29 +225,24 @@ define i32 @select_i32_from_icmp(ptr nocapture readonly %v, i32 %a, i32 %b, i64
 ; CHECK:       vector.ph:
 ; CHECK-NEXT:    [[N_MOD_VF:%.*]] = urem i64 %n, 4
 ; CHECK-NEXT:    [[N_VEC:%.*]] = sub i64 %n, [[N_MOD_VF]]
-; CHECK-NEXT:    [[MINMAX_IDENT_SPLATINSERT:%.*]] = insertelement <4 x i32> poison, i32 [[A:%.*]], i64 0
-; CHECK-NEXT:    [[MINMAX_IDENT_SPLAT:%.*]] = shufflevector <4 x i32> [[MINMAX_IDENT_SPLATINSERT]], <4 x i32> poison, <4 x i32> zeroinitializer
-; CHECK-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <4 x i32> poison, i32 [[B:%.*]], i64 0
-; CHECK-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <4 x i32> [[BROADCAST_SPLATINSERT]], <4 x i32> poison, <4 x i32> zeroinitializer
 ; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
 ; CHECK:       vector.body:
 ; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <4 x i32> [ [[MINMAX_IDENT_SPLAT]], [[VECTOR_PH]] ], [ [[TMP5:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <4 x i1> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP5:%.*]], [[VECTOR_BODY]] ]
 ; CHECK-NEXT:    [[TMP0:%.*]] = add i64 [[INDEX]], 0
 ; CHECK-NEXT:    [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[V:%.*]], i64 [[TMP0]]
 ; CHECK-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i32 0
 ; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP2]], align 4
 ; CHECK-NEXT:    [[TMP4:%.*]] = icmp eq <4 x i32> [[WIDE_LOAD]], <i32 3, i32 3, i32 3, i32 3>
-; CHECK-NEXT:    [[TMP5]] = select <4 x i1> [[TMP4]], <4 x i32> [[VEC_PHI]], <4 x i32> [[BROADCAST_SPLAT]]
+; CHECK-NEXT:    [[NOT:%.*]] = xor <4 x i1> [[TMP4]], <i1 true, i1 true, i1 true, i1 true>
+; CHECK-NEXT:    [[TMP5]] = or <4 x i1> [[VEC_PHI]], [[NOT]]
 ; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
 ; CHECK-NEXT:    [[TMP6:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
 ; CHECK-NEXT:    br i1 [[TMP6]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP8:![0-9]+]]
 ; CHECK:       middle.block:
-; CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <4 x i32> poison, i32 [[A]], i64 0
-; CHECK-NEXT:    [[DOTSPLAT:%.*]] = shufflevector <4 x i32> [[DOTSPLATINSERT]], <4 x i32> poison, <4 x i32> zeroinitializer
-; CHECK-NEXT:    [[RDX_SELECT_CMP:%.*]] = icmp ne <4 x i32> [[TMP5]], [[DOTSPLAT]]
-; CHECK-NEXT:    [[TMP7:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[RDX_SELECT_CMP]])
-; CHECK-NEXT:    [[RDX_SELECT:%.*]] = select i1 [[TMP7]], i32 [[B]], i32 [[A]]
+; CHECK-NEXT:    [[TMP7:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[TMP5]])
+; CHECK-NEXT:    [[FR:%.*]] = freeze i1 [[TMP7]]
+; CHECK-NEXT:    [[RDX_SELECT:%.*]] = select i1 [[FR]], i32 %b, i32 %a
 ;
 ; SCALABLE-LABEL: @select_i32_from_icmp
 ; SCALABLE:       vector.ph:
@@ -259,29 +252,24 @@ define i32 @select_i32_from_icmp(ptr nocapture readonly %v, i32 %a, i32 %b, i64
 ; SCALABLE-NEXT:    [[N_VEC:%.*]] = sub i64 %n, [[N_MOD_VF]]
 ; SCALABLE-NEXT:    [[TMP10:%.*]] = call i64 @llvm.vscale.i64()
 ; SCALABLE-NEXT:    [[TMP11:%.*]] = mul i64 [[TMP10]], 4
-; SCALABLE-NEXT:    [[MINMAX_IDENT_SPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[A:%.*]], i64 0
-; SCALABLE-NEXT:    [[MINMAX_IDENT_SPLAT:%.*]] = shufflevector <vscale x 4 x i32> [[MINMAX_IDENT_SPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
-; SCALABLE-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[B:%.*]], i64 0
-; SCALABLE-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 4 x i32> [[BROADCAST_SPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 ; SCALABLE-NEXT:    br label [[VECTOR_BODY:%.*]]
 ; SCALABLE:       vector.body:
 ; SCALABLE-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; SCALABLE-NEXT:    [[VEC_PHI:%.*]] = phi <vscale x 4 x i32> [ [[MINMAX_IDENT_SPLAT]], [[VECTOR_PH]] ], [ [[TMP9:%.*]], [[VECTOR_BODY]] ]
+; SCALABLE-NEXT:    [[VEC_PHI:%.*]] = phi <vscale x 4 x i1> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP9:%.*]], [[VECTOR_BODY]] ]
 ; SCALABLE-NEXT:    [[TMP4:%.*]] = add i64 [[INDEX]], 0
 ; SCALABLE-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[V:%.*]], i64 [[TMP4]]
 ; SCALABLE-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[TMP5]], i32 0
 ; SCALABLE-NEXT:    [[WIDE_LOAD:%.*]] = load <vscale x 4 x i32>, ptr [[TMP6]], align 4
 ; SCALABLE-NEXT:    [[TMP8:%.*]] = icmp eq <vscale x 4 x i32> [[WIDE_LOAD]], shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 3, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)
-; SCALABLE-NEXT:    [[TMP9]] = select <vscale x 4 x i1> [[TMP8]], <vscale x 4 x i32> [[VEC_PHI]], <vscale x 4 x i32> [[BROADCAST_SPLAT]]
+; SCALABLE-NEXT:    [[NOT:%.*]] = xor <vscale x 4 x i1> [[TMP8]], shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer)
+; SCALABLE-NEXT:    [[TMP9]] = or <vscale x 4 x i1> [[VEC_PHI]], [[NOT]]
 ; SCALABLE-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP11]]
 ; SCALABLE-NEXT:    [[TMP12:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
 ; SCALABLE-NEXT:    br i1 [[TMP12]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP8:![0-9]+]]
 ; SCALABLE:       middle.block:
-; SCALABLE-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[A]], i64 0
-; SCALABLE-NEXT:    [[DOTSPLAT:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
-; SCALABLE-NEXT:    [[RDX_SELECT_CMP:%.*]] = icmp ne <vscale x 4 x i32> [[TMP9]], [[DOTSPLAT]]
-; SCALABLE-NEXT:    [[TMP13:%.*]] = call i1 @llvm.vector.reduce.or.nxv4i1(<vscale x 4 x i1> [[RDX_SELECT_CMP]])
-; SCALABLE-NEXT:    [[RDX_SELECT:%.*]] = select i1 [[TMP13]], i32 [[B]], i32 [[A]]
+; SCALABLE-NEXT:    [[TMP13:%.*]] = call i1 @llvm.vector.reduce.or.nxv4i1(<vscale x 4 x i1> [[TMP9]])
+; SCALABLE-NEXT:    [[FR:%.*]] = freeze i1 [[TMP13]]
+; SCALABLE-NEXT:    [[RDX_SELECT:%.*]] = select i1 [[FR]], i32 %b, i32 %a
 ;
 entry:
   br label %for.body
@@ -309,20 +297,21 @@ define i32 @select_const_i32_from_fcmp(ptr nocapture readonly %v, i64 %n) #0 {
 ; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
 ; CHECK:       vector.body:
 ; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <4 x i32> [ <i32 2, i32 2, i32 2, i32 2>, [[VECTOR_PH]] ], [ [[TMP5:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <4 x i1> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP5:%.*]], [[VECTOR_BODY]] ]
 ; CHECK-NEXT:    [[TMP0:%.*]] = add i64 [[INDEX]], 0
 ; CHECK-NEXT:    [[TMP1:%.*]] = getelementptr inbounds float, ptr [[V:%.*]], i64 [[TMP0]]
 ; CHECK-NEXT:    [[TMP2:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i32 0
 ; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <4 x float>, ptr [[TMP2]], align 4
 ; CHECK-NEXT:    [[TMP4:%.*]] = fcmp fast ueq <4 x float> [[WIDE_LOAD]], <float 3.000000e+00, float 3.000000e+00, float 3.000000e+00, float 3.000000e+00>
-; CHECK-NEXT:    [[TMP5]] = select <4 x i1> [[TMP4]], <4 x i32> [[VEC_PHI]], <4 x i32> <i32 1, i32 1, i32 1, i32 1>
+; CHECK-NEXT:    [[NOT:%.*]] = xor <4 x i1> [[TMP4]], <i1 true, i1 true, i1 true, i1 true>
+; CHECK-NEXT:    [[TMP5]] = or <4 x i1> [[VEC_PHI]], [[NOT]]
 ; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
 ; CHECK-NEXT:    [[TMP6:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
 ; CHECK-NEXT:    br i1 [[TMP6]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP10:![0-9]+]]
 ; CHECK:       middle.block:
-; CHECK-NEXT:    [[RDX_SELECT_CMP:%.*]] = icmp ne <4 x i32> [[TMP5]], <i32 2, i32 2, i32 2, i32 2>
-; CHECK-NEXT:    [[TMP7:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[RDX_SELECT_CMP]])
-; CHECK-NEXT:    [[RDX_SELECT:%.*]] = select i1 [[TMP7]], i32 1, i32 2
+; CHECK-NEXT:    [[TMP7:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[TMP5]])
+; CHECK-NEXT:    [[FR:%.*]] = freeze i1 [[TMP7]]
+; CHECK-NEXT:    [[RDX_SELECT:%.*]] = select i1 [[FR]], i32 1, i32 2
 ;
 ; SCALABLE-LABEL: @select_const_i32_from_fcmp
 ; SCALABLE:       vector.ph:
@@ -335,20 +324,21 @@ define i32 @select_const_i32_from_fcmp(ptr nocapture readonly %v, i64 %n) #0 {
 ; SCALABLE-NEXT:    br label [[VECTOR_BODY:%.*]]
 ; SCALABLE:       vector.body:
 ; SCALABLE-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; SCALABLE-NEXT:    [[VEC_PHI:%.*]] = phi <vscale x 4 x i32> [ shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 2, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer), [[VECTOR_PH]] ], [ [[TMP9:%.*]], [[VECTOR_BODY]] ]
+; SCALABLE-NEXT:    [[VEC_PHI:%.*]] = phi <vscale x 4 x i1> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP9:%.*]], [[VECTOR_BODY]] ]
 ; SCALABLE-NEXT:    [[TMP4:%.*]] = add i64 [[INDEX]], 0
 ; SCALABLE-NEXT:    [[TMP5:%.*]] = getelementptr inbounds float, ptr [[V:%.*]], i64 [[TMP4]]
 ; SCALABLE-NEXT:    [[TMP6:%.*]] = getelementptr inbounds float, ptr [[TMP5]], i32 0
 ; SCALABLE-NEXT:    [[WIDE_LOAD:%.*]] = load <vscale x 4 x float>, ptr [[TMP6]], align 4
 ; SCALABLE-NEXT:    [[TMP8:%.*]] = fcmp fast ueq <vscale x 4 x float> [[WIDE_LOAD]], shufflevector (<vscale x 4 x float> insertelement (<vscale x 4 x float> poison, float 3.000000e+00, i64 0), <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer)
-; SCALABLE-NEXT:    [[TMP9]] = select <vscale x 4 x i1> [[TMP8]], <vscale x 4 x i32> [[VEC_PHI]], <vscale x 4 x i32> shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 1, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)
+; SCALABLE-NEXT:    [[NOT:%.*]] = xor <vscale x 4 x i1> [[TMP8]], shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer)
+; SCALABLE-NEXT:    [[TMP9]] = or <vscale x 4 x i1> [[VEC_PHI]], [[NOT]]
 ; SCALABLE-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP11]]
 ; SCALABLE-NEXT:    [[TMP12:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
 ; SCALABLE-NEXT:    br i1 [[TMP12]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP10:![0-9]+]]
 ; SCALABLE:       middle.block:
-; SCALABLE-NEXT:    [[RDX_SELECT_CMP:%.*]] = icmp ne <vscale x 4 x i32> [[TMP9]], shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 2, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)
-; SCALABLE-NEXT:    [[TMP13:%.*]] = call i1 @llvm.vector.reduce.or.nxv4i1(<vscale x 4 x i1> [[RDX_SELECT_CMP]])
-; SCALABLE-NEXT:    [[RDX_SELECT:%.*]] = select i1 [[TMP13]], i32 1, i32 2
+; SCALABLE-NEXT:    [[TMP13:%.*]] = call i1 @llvm.vector.reduce.or.nxv4i1(<vscale x 4 x i1> [[TMP9]])
+; SCALABLE-NEXT:    [[FR:%.*]] = freeze i1 [[TMP13]]
+; SCALABLE-NEXT:    [[RDX_SELECT:%.*]] = select i1 [[FR]], i32 1, i32 2
 ;
 entry:
   br label %for.body
@@ -401,7 +391,7 @@ define i32 @pred_select_const_i32_from_icmp(ptr noalias nocapture readonly %src1
 ; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
 ; CHECK:       vector.body:
 ; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[PREDPHI:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <4 x i1> [ zeroinitializer, [[VECTOR_PH]] ], [ [[PREDPHI:%.*]], [[VECTOR_BODY]] ]
 ; CHECK-NEXT:    [[TMP0:%.*]] = add i64 [[INDEX]], 0
 ; CHECK-NEXT:    [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[SRC1:%.*]], i64 [[TMP0]]
 ; CHECK-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i32 0
@@ -411,16 +401,16 @@ define i32 @pred_select_const_i32_from_icmp(ptr noalias nocapture readonly %src1
 ; CHECK-NEXT:    [[TMP6:%.*]] = getelementptr i32, ptr [[TMP5]], i32 0
 ; CHECK-NEXT:    [[WIDE_MASKED_LOAD:%.*]] = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr [[TMP6]], i32 4, <4 x i1> [[TMP4]], <4 x i32> poison)
 ; CHECK-NEXT:    [[TMP8:%.*]] = icmp eq <4 x i32> [[WIDE_MASKED_LOAD]], <i32 2, i32 2, i32 2, i32 2>
-; CHECK-NEXT:    [[TMP9:%.*]] = select <4 x i1> [[TMP8]], <4 x i32> <i32 1, i32 1, i32 1, i32 1>, <4 x i32> [[VEC_PHI]]
+; CHECK-NEXT:    [[TMP9:%.*]] = or <4 x i1> [[VEC_PHI]], [[TMP8]]
 ; CHECK-NEXT:    [[TMP10:%.*]] = xor <4 x i1> [[TMP4]], <i1 true, i1 true, i1 true, i1 true>
-; CHECK-NEXT:    [[PREDPHI]] = select <4 x i1> [[TMP4]], <4 x i32> [[TMP9]], <4 x i32> [[VEC_PHI]]
+; CHECK-NEXT:    [[PREDPHI]] = select <4 x i1> [[TMP4]], <4 x i1> [[TMP9]], <4 x i1> [[VEC_PHI]]
 ; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
 ; CHECK-NEXT:    [[TMP11:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
 ; CHECK-NEXT:    br i1 [[TMP11]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP12:![0-9]+]]
 ; CHECK:       middle.block:
-; CHECK-NEXT:    [[RDX_SELECT_CMP:%.*]] = icmp ne <4 x i32> [[PREDPHI]], zeroinitializer
-; CHECK-NEXT:    [[TMP12:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[RDX_SELECT_CMP]])
-; CHECK-NEXT:    [[RDX_SELECT:%.*]] = select i1 [[TMP12]], i32 1, i32 0
+; CHECK-NEXT:    [[TMP12:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[PREDPHI]])
+; CHECK-NEXT:    [[FR:%.*]] = freeze i1 [[TMP12]]
+; CHECK-NEXT:    [[RDX_SELECT:%.*]] = select i1 [[FR]], i32 1, i32 0
 ;
 ; SCALABLE-LABEL: @pred_select_const_i32_from_icmp
 ; SCALABLE:       vector.ph:
@@ -433,7 +423,7 @@ define i32 @pred_select_const_i32_from_icmp(ptr noalias nocapture readonly %src1
 ; SCALABLE-NEXT:    br label [[VECTOR_BODY:%.*]]
 ; SCALABLE:       vector.body:
 ; SCALABLE-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; SCALABLE-NEXT:    [[VEC_PHI:%.*]] = phi <vscale x 4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[PREDPHI:%.*]], [[VECTOR_BODY]] ]
+; SCALABLE-NEXT:    [[VEC_PHI:%.*]] = phi <vscale x 4 x i1> [ zeroinitializer, [[VECTOR_PH]] ], [ [[PREDPHI:%.*]], [[VECTOR_BODY]] ]
 ; SCALABLE-NEXT:    [[TMP4:%.*]] = add i64 [[INDEX]], 0
 ; SCALABLE-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[SRC1:%.*]], i64 [[TMP4]]
 ; SCALABLE-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[TMP5]], i32 0
@@ -443,16 +433,16 @@ define i32 @pred_select_const_i32_from_icmp(ptr noalias nocapture readonly %src1
 ; SCALABLE-NEXT:    [[TMP10:%.*]] = getelementptr i32, ptr [[TMP9]], i32 0
 ; SCALABLE-NEXT:    [[WIDE_MASKED_LOAD:%.*]] = call <vscale x 4 x i32> @llvm.masked.load.nxv4i32.p0(ptr [[TMP10]], i32 4, <vscale x 4 x i1> [[TMP8]], <vscale x 4 x i32> poison)
 ; SCALABLE-NEXT:    [[TMP12:%.*]] = icmp eq <vscale x 4 x i32> [[WIDE_MASKED_LOAD]], shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 2, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)
-; SCALABLE-NEXT:    [[TMP13:%.*]] = select <vscale x 4 x i1> [[TMP12]], <vscale x 4 x i32> shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 1, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer), <vscale x 4 x i32> [[VEC_PHI]]
+; SCALABLE-NEXT:    [[TMP13:%.*]] = or <vscale x 4 x i1> [[VEC_PHI]], [[TMP12]]
 ; SCALABLE-NEXT:    [[TMP14:%.*]] = xor <vscale x 4 x i1> [[TMP8]], shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer)
-; SCALABLE-NEXT:    [[PREDPHI]] = select <vscale x 4 x i1> [[TMP8]], <vscale x 4 x i32> [[TMP13]], <vscale x 4 x i32> [[VEC_PHI]]
+; SCALABLE-NEXT:    [[PREDPHI]] = select <vscale x 4 x i1> [[TMP8]], <vscale x 4 x i1> [[TMP13]], <vscale x 4 x i1> [[VEC_PHI]]
 ; SCALABLE-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP16]]
 ; SCALABLE-NEXT:    [[TMP17:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
 ; SCALABLE-NEXT:    br i1 [[TMP17]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP12:![0-9]+]]
 ; SCALABLE:       middle.block:
-; SCALABLE-NEXT:    [[RDX_SELECT_CMP:%.*]] = icmp ne <vscale x 4 x i32> [[PREDPHI]], zeroinitializer
-; SCALABLE-NEXT:    [[TMP18:%.*]] = call i1 @llvm.vector.reduce.or.nxv4i1(<vscale x 4 x i1> [[RDX_SELECT_CMP]])
-; SCALABLE-NEXT:    [[RDX_SELECT:%.*]] = select i1 [[TMP18]], i32 1, i32 0
+; SCALABLE-NEXT:    [[TMP18:%.*]] = call i1 @llvm.vector.reduce.or.nxv4i1(<vscale x 4 x i1> [[PREDPHI]])
+; SCALABLE-NEXT:    [[FR:%.*]] = freeze i1 [[TMP18]]
+; SCALABLE-NEXT:    [[RDX_SELECT:%.*]] = select i1 [[FR]], i32 1, i32 0
 ;
 entry:
   br label %for.body
diff --git a/llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-gather-scatter.ll b/llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-gather-scatter.ll
new file mode 100644
index 000000000000..835ff3756881
--- /dev/null
+++ b/llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-gather-scatter.ll
@@ -0,0 +1,116 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
+; RUN: opt -passes=loop-vectorize \
+; RUN: -force-tail-folding-style=data-with-evl \
+; RUN: -prefer-predicate-over-epilogue=predicate-dont-vectorize \
+; RUN: -mtriple=riscv64 -mattr=+v -S < %s | FileCheck %s --check-prefix=IF-EVL
+
+; RUN: opt -passes=loop-vectorize \
+; RUN: -force-tail-folding-style=none \
+; RUN: -prefer-predicate-over-epilogue=predicate-dont-vectorize \
+; RUN: -mtriple=riscv64 -mattr=+v -S < %s | FileCheck %s --check-prefix=NO-VP
+
+define void @gather_scatter(ptr noalias %in, ptr noalias %out, ptr noalias %index, i64 %n) {
+; IF-EVL-LABEL: @gather_scatter(
+; IF-EVL-NEXT:  entry:
+; IF-EVL-NEXT:    [[TMP0:%.*]] = sub i64 -1, [[N:%.*]]
+; IF-EVL-NEXT:    [[TMP1:%.*]] = call i64 @llvm.vscale.i64()
+; IF-EVL-NEXT:    [[TMP2:%.*]] = mul i64 [[TMP1]], 2
+; IF-EVL-NEXT:    [[TMP3:%.*]] = icmp ult i64 [[TMP0]], [[TMP2]]
+; IF-EVL-NEXT:    br i1 [[TMP3]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; IF-EVL:       vector.ph:
+; IF-EVL-NEXT:    [[TMP4:%.*]] = call i64 @llvm.vscale.i64()
+; IF-EVL-NEXT:    [[TMP5:%.*]] = mul i64 [[TMP4]], 2
+; IF-EVL-NEXT:    [[TMP6:%.*]] = call i64 @llvm.vscale.i64()
+; IF-EVL-NEXT:    [[TMP7:%.*]] = mul i64 [[TMP6]], 2
+; IF-EVL-NEXT:    [[TMP8:%.*]] = sub i64 [[TMP7]], 1
+; IF-EVL-NEXT:    [[N_RND_UP:%.*]] = add i64 [[N]], [[TMP8]]
+; IF-EVL-NEXT:    [[N_MOD_VF:%.*]] = urem i64 [[N_RND_UP]], [[TMP5]]
+; IF-EVL-NEXT:    [[N_VEC:%.*]] = sub i64 [[N_RND_UP]], [[N_MOD_VF]]
+; IF-EVL-NEXT:    [[TRIP_COUNT_MINUS_1:%.*]] = sub i64 [[N]], 1
+; IF-EVL-NEXT:    [[TMP9:%.*]] = call i64 @llvm.vscale.i64()
+; IF-EVL-NEXT:    [[TMP10:%.*]] = mul i64 [[TMP9]], 2
+; IF-EVL-NEXT:    [[TMP11:%.*]] = call <vscale x 2 x i64> @llvm.experimental.stepvector.nxv2i64()
+; IF-EVL-NEXT:    [[TMP12:%.*]] = add <vscale x 2 x i64> [[TMP11]], zeroinitializer
+; IF-EVL-NEXT:    [[TMP13:%.*]] = mul <vscale x 2 x i64> [[TMP12]], shufflevector (<vscale x 2 x i64> insertelement (<vscale x 2 x i64> poison, i64 1, i64 0), <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer)
+; IF-EVL-NEXT:    [[INDUCTION:%.*]] = add <vscale x 2 x i64> zeroinitializer, [[TMP13]]
+; IF-EVL-NEXT:    [[TMP14:%.*]] = call i64 @llvm.vscale.i64()
+; IF-EVL-NEXT:    [[TMP15:%.*]] = mul i64 [[TMP14]], 2
+; IF-EVL-NEXT:    [[TMP16:%.*]] = mul i64 1, [[TMP15]]
+; IF-EVL-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[TMP16]], i64 0
+; IF-EVL-NEXT:    [[DOTSPLAT:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
+; IF-EVL-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[TRIP_COUNT_MINUS_1]], i64 0
+; IF-EVL-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 2 x i64> [[BROADCAST_SPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
+; IF-EVL-NEXT:    br label [[VECTOR_BODY:%.*]]
+; IF-EVL:       vector.body:
+; IF-EVL-NEXT:    [[INDEX1:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; IF-EVL-NEXT:    [[EVL_BASED_IV:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_EVL_NEXT:%.*]], [[VECTOR_BODY]] ]
+; IF-EVL-NEXT:    [[VEC_IND:%.*]] = phi <vscale x 2 x i64> [ [[INDUCTION]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; IF-EVL-NEXT:    [[TMP17:%.*]] = sub i64 [[N]], [[EVL_BASED_IV]]
+; IF-EVL-NEXT:    [[TMP18:%.*]] = call i32 @llvm.experimental.get.vector.length.i64(i64 [[TMP17]], i32 2, i1 true)
+; IF-EVL-NEXT:    [[TMP19:%.*]] = icmp ule <vscale x 2 x i64> [[VEC_IND]], [[BROADCAST_SPLAT]]
+; IF-EVL-NEXT:    [[TMP20:%.*]] = getelementptr inbounds i32, ptr [[INDEX:%.*]], <vscale x 2 x i64> [[VEC_IND]]
+; IF-EVL-NEXT:    [[WIDE_MASKED_GATHER:%.*]] = call <vscale x 2 x i64> @llvm.vp.gather.nxv2i64.nxv2p0(<vscale x 2 x ptr> align 8 [[TMP20]], <vscale x 2 x i1> [[TMP19]], i32 [[TMP18]])
+; IF-EVL-NEXT:    [[TMP21:%.*]] = getelementptr inbounds float, ptr [[IN:%.*]], <vscale x 2 x i64> [[WIDE_MASKED_GATHER]]
+; IF-EVL-NEXT:    [[WIDE_MASKED_GATHER2:%.*]] = call <vscale x 2 x float> @llvm.vp.gather.nxv2f32.nxv2p0(<vscale x 2 x ptr> align 4 [[TMP21]], <vscale x 2 x i1> [[TMP19]], i32 [[TMP18]])
+; IF-EVL-NEXT:    [[TMP22:%.*]] = getelementptr inbounds float, ptr [[OUT:%.*]], <vscale x 2 x i64> [[WIDE_MASKED_GATHER]]
+; IF-EVL-NEXT:    call void @llvm.vp.scatter.nxv2f32.nxv2p0(<vscale x 2 x float> [[WIDE_MASKED_GATHER2]], <vscale x 2 x ptr> align 4 [[TMP22]], <vscale x 2 x i1> [[TMP19]], i32 [[TMP18]])
+; IF-EVL-NEXT:    [[TMP23:%.*]] = zext i32 [[TMP18]] to i64
+; IF-EVL-NEXT:    [[INDEX_EVL_NEXT]] = add i64 [[TMP23]], [[EVL_BASED_IV]]
+; IF-EVL-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX1]], [[TMP10]]
+; IF-EVL-NEXT:    [[VEC_IND_NEXT]] = add <vscale x 2 x i64> [[VEC_IND]], [[DOTSPLAT]]
+; IF-EVL-NEXT:    [[TMP24:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
+; IF-EVL-NEXT:    br i1 [[TMP24]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
+; IF-EVL:       middle.block:
+; IF-EVL-NEXT:    br i1 true, label [[FOR_END:%.*]], label [[SCALAR_PH]]
+; IF-EVL:       scalar.ph:
+; IF-EVL-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; IF-EVL-NEXT:    br label [[FOR_BODY:%.*]]
+; IF-EVL:       for.body:
+; IF-EVL-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
+; IF-EVL-NEXT:    [[ARRAYIDX3:%.*]] = getelementptr inbounds i32, ptr [[INDEX]], i64 [[INDVARS_IV]]
+; IF-EVL-NEXT:    [[TMP25:%.*]] = load i64, ptr [[ARRAYIDX3]], align 8
+; IF-EVL-NEXT:    [[ARRAYIDX5:%.*]] = getelementptr inbounds float, ptr [[IN]], i64 [[TMP25]]
+; IF-EVL-NEXT:    [[TMP26:%.*]] = load float, ptr [[ARRAYIDX5]], align 4
+; IF-EVL-NEXT:    [[ARRAYIDX7:%.*]] = getelementptr inbounds float, ptr [[OUT]], i64 [[TMP25]]
+; IF-EVL-NEXT:    store float [[TMP26]], ptr [[ARRAYIDX7]], align 4
+; IF-EVL-NEXT:    [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
+; IF-EVL-NEXT:    [[EXITCOND_NOT:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], [[N]]
+; IF-EVL-NEXT:    br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
+; IF-EVL:       for.end:
+; IF-EVL-NEXT:    ret void
+;
+; NO-VP-LABEL: @gather_scatter(
+; NO-VP-NEXT:  entry:
+; NO-VP-NEXT:    br label [[FOR_BODY:%.*]]
+; NO-VP:       for.body:
+; NO-VP-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
+; NO-VP-NEXT:    [[ARRAYIDX3:%.*]] = getelementptr inbounds i32, ptr [[INDEX:%.*]], i64 [[INDVARS_IV]]
+; NO-VP-NEXT:    [[TMP0:%.*]] = load i64, ptr [[ARRAYIDX3]], align 8
+; NO-VP-NEXT:    [[ARRAYIDX5:%.*]] = getelementptr inbounds float, ptr [[IN:%.*]], i64 [[TMP0]]
+; NO-VP-NEXT:    [[TMP1:%.*]] = load float, ptr [[ARRAYIDX5]], align 4
+; NO-VP-NEXT:    [[ARRAYIDX7:%.*]] = getelementptr inbounds float, ptr [[OUT:%.*]], i64 [[TMP0]]
+; NO-VP-NEXT:    store float [[TMP1]], ptr [[ARRAYIDX7]], align 4
+; NO-VP-NEXT:    [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
+; NO-VP-NEXT:    [[EXITCOND_NOT:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], [[N:%.*]]
+; NO-VP-NEXT:    br i1 [[EXITCOND_NOT]], label [[FOR_END:%.*]], label [[FOR_BODY]]
+; NO-VP:       for.end:
+; NO-VP-NEXT:    ret void
+;
+entry:
+  br label %for.body
+
+for.body:
+  %indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ]
+  %arrayidx3 = getelementptr inbounds i32, ptr %index, i64 %indvars.iv
+  %0 = load i64, ptr %arrayidx3, align 8
+  %arrayidx5 = getelementptr inbounds float, ptr %in, i64 %0
+  %1 = load float, ptr %arrayidx5, align 4
+  %arrayidx7 = getelementptr inbounds float, ptr %out, i64 %0
+  store float %1, ptr %arrayidx7, align 4
+  %indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
+  %exitcond.not = icmp eq i64 %indvars.iv.next, %n
+  br i1 %exitcond.not, label %for.end, label %for.body
+
+for.end:
+  ret void
+}
diff --git a/llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-interleave.ll b/llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-interleave.ll
new file mode 100644
index 000000000000..0b495bc680f0
--- /dev/null
+++ b/llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-interleave.ll
@@ -0,0 +1,175 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
+; RUN: opt -passes=loop-vectorize \
+; RUN: -force-tail-folding-style=data-with-evl \
+; RUN: -prefer-predicate-over-epilogue=predicate-dont-vectorize \
+; RUN: -mtriple=riscv64 -mattr=+v -S < %s | FileCheck --check-prefix=IF-EVL %s
+
+; RUN: opt -passes=loop-vectorize \
+; RUN: -force-tail-folding-style=none \
+; RUN: -prefer-predicate-over-epilogue=predicate-else-scalar-epilogue \
+; RUN: -mtriple=riscv64 -mattr=+v -S < %s | FileCheck --check-prefix=NO-VP %s
+
+; FIXME: interleaved accesses are not supported yet with predicated vectorization.
+define void @interleave(ptr noalias %a, ptr noalias %b, i64 %N) {
+; IF-EVL-LABEL: @interleave(
+; IF-EVL-NEXT:  entry:
+; IF-EVL-NEXT:    [[TMP17:%.*]] = sub i64 -1, [[N:%.*]]
+; IF-EVL-NEXT:    [[TMP31:%.*]] = call i64 @llvm.vscale.i64()
+; IF-EVL-NEXT:    [[TMP2:%.*]] = mul i64 [[TMP31]], 8
+; IF-EVL-NEXT:    [[TMP3:%.*]] = icmp ult i64 [[TMP17]], [[TMP2]]
+; IF-EVL-NEXT:    br i1 [[TMP3]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; IF-EVL:       vector.ph:
+; IF-EVL-NEXT:    [[TMP4:%.*]] = call i64 @llvm.vscale.i64()
+; IF-EVL-NEXT:    [[TMP5:%.*]] = mul i64 [[TMP4]], 8
+; IF-EVL-NEXT:    [[TMP6:%.*]] = call i64 @llvm.vscale.i64()
+; IF-EVL-NEXT:    [[TMP7:%.*]] = mul i64 [[TMP6]], 8
+; IF-EVL-NEXT:    [[TMP32:%.*]] = sub i64 [[TMP7]], 1
+; IF-EVL-NEXT:    [[N_RND_UP:%.*]] = add i64 [[N]], [[TMP32]]
+; IF-EVL-NEXT:    [[N_MOD_VF:%.*]] = urem i64 [[N_RND_UP]], [[TMP5]]
+; IF-EVL-NEXT:    [[N_VEC:%.*]] = sub i64 [[N_RND_UP]], [[N_MOD_VF]]
+; IF-EVL-NEXT:    [[TRIP_COUNT_MINUS_1:%.*]] = sub i64 [[N]], 1
+; IF-EVL-NEXT:    [[TMP9:%.*]] = call i64 @llvm.vscale.i64()
+; IF-EVL-NEXT:    [[TMP10:%.*]] = mul i64 [[TMP9]], 8
+; IF-EVL-NEXT:    [[TMP11:%.*]] = call <vscale x 4 x i64> @llvm.experimental.stepvector.nxv4i64()
+; IF-EVL-NEXT:    [[TMP12:%.*]] = add <vscale x 4 x i64> [[TMP11]], zeroinitializer
+; IF-EVL-NEXT:    [[TMP13:%.*]] = mul <vscale x 4 x i64> [[TMP12]], shufflevector (<vscale x 4 x i64> insertelement (<vscale x 4 x i64> poison, i64 1, i64 0), <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer)
+; IF-EVL-NEXT:    [[INDUCTION:%.*]] = add <vscale x 4 x i64> zeroinitializer, [[TMP13]]
+; IF-EVL-NEXT:    [[TMP14:%.*]] = call i64 @llvm.vscale.i64()
+; IF-EVL-NEXT:    [[TMP15:%.*]] = mul i64 [[TMP14]], 4
+; IF-EVL-NEXT:    [[TMP37:%.*]] = mul i64 1, [[TMP15]]
+; IF-EVL-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i64> poison, i64 [[TMP37]], i64 0
+; IF-EVL-NEXT:    [[DOTSPLAT:%.*]] = shufflevector <vscale x 4 x i64> [[DOTSPLATINSERT]], <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer
+; IF-EVL-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 4 x i64> poison, i64 [[TRIP_COUNT_MINUS_1]], i64 0
+; IF-EVL-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 4 x i64> [[BROADCAST_SPLATINSERT]], <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer
+; IF-EVL-NEXT:    br label [[VECTOR_BODY:%.*]]
+; IF-EVL:       vector.body:
+; IF-EVL-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; IF-EVL-NEXT:    [[VEC_IND:%.*]] = phi <vscale x 4 x i64> [ [[INDUCTION]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; IF-EVL-NEXT:    [[STEP_ADD:%.*]] = add <vscale x 4 x i64> [[VEC_IND]], [[DOTSPLAT]]
+; IF-EVL-NEXT:    [[TMP0:%.*]] = add i64 [[INDEX]], 0
+; IF-EVL-NEXT:    [[TMP18:%.*]] = call i64 @llvm.vscale.i64()
+; IF-EVL-NEXT:    [[TMP19:%.*]] = mul i64 [[TMP18]], 4
+; IF-EVL-NEXT:    [[TMP38:%.*]] = add i64 [[TMP19]], 0
+; IF-EVL-NEXT:    [[TMP39:%.*]] = mul i64 [[TMP38]], 1
+; IF-EVL-NEXT:    [[TMP1:%.*]] = add i64 [[INDEX]], [[TMP39]]
+; IF-EVL-NEXT:    [[TMP23:%.*]] = icmp ule <vscale x 4 x i64> [[VEC_IND]], [[BROADCAST_SPLAT]]
+; IF-EVL-NEXT:    [[TMP24:%.*]] = icmp ule <vscale x 4 x i64> [[STEP_ADD]], [[BROADCAST_SPLAT]]
+; IF-EVL-NEXT:    [[TMP25:%.*]] = getelementptr inbounds [2 x i32], ptr [[B:%.*]], <vscale x 4 x i64> [[VEC_IND]], i32 0
+; IF-EVL-NEXT:    [[TMP26:%.*]] = getelementptr inbounds [2 x i32], ptr [[B]], <vscale x 4 x i64> [[STEP_ADD]], i32 0
+; IF-EVL-NEXT:    [[WIDE_MASKED_GATHER:%.*]] = call <vscale x 4 x i32> @llvm.masked.gather.nxv4i32.nxv4p0(<vscale x 4 x ptr> [[TMP25]], i32 4, <vscale x 4 x i1> [[TMP23]], <vscale x 4 x i32> poison)
+; IF-EVL-NEXT:    [[WIDE_MASKED_GATHER2:%.*]] = call <vscale x 4 x i32> @llvm.masked.gather.nxv4i32.nxv4p0(<vscale x 4 x ptr> [[TMP26]], i32 4, <vscale x 4 x i1> [[TMP24]], <vscale x 4 x i32> poison)
+; IF-EVL-NEXT:    [[TMP27:%.*]] = getelementptr inbounds [2 x i32], ptr [[B]], <vscale x 4 x i64> [[VEC_IND]], i32 1
+; IF-EVL-NEXT:    [[TMP28:%.*]] = getelementptr inbounds [2 x i32], ptr [[B]], <vscale x 4 x i64> [[STEP_ADD]], i32 1
+; IF-EVL-NEXT:    [[WIDE_MASKED_GATHER3:%.*]] = call <vscale x 4 x i32> @llvm.masked.gather.nxv4i32.nxv4p0(<vscale x 4 x ptr> [[TMP27]], i32 4, <vscale x 4 x i1> [[TMP23]], <vscale x 4 x i32> poison)
+; IF-EVL-NEXT:    [[WIDE_MASKED_GATHER4:%.*]] = call <vscale x 4 x i32> @llvm.masked.gather.nxv4i32.nxv4p0(<vscale x 4 x ptr> [[TMP28]], i32 4, <vscale x 4 x i1> [[TMP24]], <vscale x 4 x i32> poison)
+; IF-EVL-NEXT:    [[TMP29:%.*]] = add nsw <vscale x 4 x i32> [[WIDE_MASKED_GATHER3]], [[WIDE_MASKED_GATHER]]
+; IF-EVL-NEXT:    [[TMP30:%.*]] = add nsw <vscale x 4 x i32> [[WIDE_MASKED_GATHER4]], [[WIDE_MASKED_GATHER2]]
+; IF-EVL-NEXT:    [[TMP8:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP0]]
+; IF-EVL-NEXT:    [[TMP16:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP1]]
+; IF-EVL-NEXT:    [[TMP33:%.*]] = getelementptr inbounds i32, ptr [[TMP8]], i32 0
+; IF-EVL-NEXT:    [[TMP34:%.*]] = call i64 @llvm.vscale.i64()
+; IF-EVL-NEXT:    [[TMP35:%.*]] = mul i64 [[TMP34]], 4
+; IF-EVL-NEXT:    [[TMP36:%.*]] = getelementptr inbounds i32, ptr [[TMP8]], i64 [[TMP35]]
+; IF-EVL-NEXT:    call void @llvm.masked.store.nxv4i32.p0(<vscale x 4 x i32> [[TMP29]], ptr [[TMP33]], i32 4, <vscale x 4 x i1> [[TMP23]])
+; IF-EVL-NEXT:    call void @llvm.masked.store.nxv4i32.p0(<vscale x 4 x i32> [[TMP30]], ptr [[TMP36]], i32 4, <vscale x 4 x i1> [[TMP24]])
+; IF-EVL-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP10]]
+; IF-EVL-NEXT:    [[VEC_IND_NEXT]] = add <vscale x 4 x i64> [[STEP_ADD]], [[DOTSPLAT]]
+; IF-EVL-NEXT:    [[TMP20:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
+; IF-EVL-NEXT:    br i1 [[TMP20]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
+; IF-EVL:       middle.block:
+; IF-EVL-NEXT:    br i1 true, label [[FOR_COND_CLEANUP:%.*]], label [[SCALAR_PH]]
+; IF-EVL:       scalar.ph:
+; IF-EVL-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; IF-EVL-NEXT:    br label [[FOR_BODY:%.*]]
+; IF-EVL:       for.body:
+; IF-EVL-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
+; IF-EVL-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], ptr [[B]], i64 [[IV]], i32 0
+; IF-EVL-NEXT:    [[TMP21:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
+; IF-EVL-NEXT:    [[ARRAYIDX2:%.*]] = getelementptr inbounds [2 x i32], ptr [[B]], i64 [[IV]], i32 1
+; IF-EVL-NEXT:    [[TMP22:%.*]] = load i32, ptr [[ARRAYIDX2]], align 4
+; IF-EVL-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP22]], [[TMP21]]
+; IF-EVL-NEXT:    [[ARRAYIDX4:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[IV]]
+; IF-EVL-NEXT:    store i32 [[ADD]], ptr [[ARRAYIDX4]], align 4
+; IF-EVL-NEXT:    [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
+; IF-EVL-NEXT:    [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
+; IF-EVL-NEXT:    br i1 [[EXITCOND_NOT]], label [[FOR_COND_CLEANUP]], label [[FOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
+; IF-EVL:       for.cond.cleanup:
+; IF-EVL-NEXT:    ret void
+;
+; NO-VP-LABEL: @interleave(
+; NO-VP-NEXT:  entry:
+; NO-VP-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[N:%.*]], 16
+; NO-VP-NEXT:    br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; NO-VP:       vector.ph:
+; NO-VP-NEXT:    [[N_MOD_VF:%.*]] = urem i64 [[N]], 16
+; NO-VP-NEXT:    [[N_VEC:%.*]] = sub i64 [[N]], [[N_MOD_VF]]
+; NO-VP-NEXT:    br label [[VECTOR_BODY:%.*]]
+; NO-VP:       vector.body:
+; NO-VP-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; NO-VP-NEXT:    [[TMP10:%.*]] = add i64 [[INDEX]], 0
+; NO-VP-NEXT:    [[TMP1:%.*]] = add i64 [[INDEX]], 8
+; NO-VP-NEXT:    [[TMP2:%.*]] = getelementptr inbounds [2 x i32], ptr [[B:%.*]], i64 [[TMP10]], i32 0
+; NO-VP-NEXT:    [[TMP3:%.*]] = getelementptr inbounds [2 x i32], ptr [[B]], i64 [[TMP1]], i32 0
+; NO-VP-NEXT:    [[TMP4:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i32 0
+; NO-VP-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[TMP3]], i32 0
+; NO-VP-NEXT:    [[WIDE_VEC:%.*]] = load <16 x i32>, ptr [[TMP4]], align 4
+; NO-VP-NEXT:    [[WIDE_VEC1:%.*]] = load <16 x i32>, ptr [[TMP5]], align 4
+; NO-VP-NEXT:    [[STRIDED_VEC:%.*]] = shufflevector <16 x i32> [[WIDE_VEC]], <16 x i32> poison, <8 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14>
+; NO-VP-NEXT:    [[STRIDED_VEC2:%.*]] = shufflevector <16 x i32> [[WIDE_VEC1]], <16 x i32> poison, <8 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14>
+; NO-VP-NEXT:    [[STRIDED_VEC3:%.*]] = shufflevector <16 x i32> [[WIDE_VEC]], <16 x i32> poison, <8 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15>
+; NO-VP-NEXT:    [[STRIDED_VEC4:%.*]] = shufflevector <16 x i32> [[WIDE_VEC1]], <16 x i32> poison, <8 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15>
+; NO-VP-NEXT:    [[TMP6:%.*]] = add nsw <8 x i32> [[STRIDED_VEC3]], [[STRIDED_VEC]]
+; NO-VP-NEXT:    [[TMP7:%.*]] = add nsw <8 x i32> [[STRIDED_VEC4]], [[STRIDED_VEC2]]
+; NO-VP-NEXT:    [[TMP24:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP10]]
+; NO-VP-NEXT:    [[TMP13:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP1]]
+; NO-VP-NEXT:    [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[TMP24]], i32 0
+; NO-VP-NEXT:    [[TMP11:%.*]] = getelementptr inbounds i32, ptr [[TMP24]], i32 8
+; NO-VP-NEXT:    store <8 x i32> [[TMP6]], ptr [[TMP12]], align 4
+; NO-VP-NEXT:    store <8 x i32> [[TMP7]], ptr [[TMP11]], align 4
+; NO-VP-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 16
+; NO-VP-NEXT:    [[TMP28:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
+; NO-VP-NEXT:    br i1 [[TMP28]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
+; NO-VP:       middle.block:
+; NO-VP-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[N]], [[N_VEC]]
+; NO-VP-NEXT:    br i1 [[CMP_N]], label [[FOR_COND_CLEANUP:%.*]], label [[SCALAR_PH]]
+; NO-VP:       scalar.ph:
+; NO-VP-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; NO-VP-NEXT:    br label [[FOR_BODY:%.*]]
+; NO-VP:       for.body:
+; NO-VP-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
+; NO-VP-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], ptr [[B]], i64 [[IV]], i32 0
+; NO-VP-NEXT:    [[TMP29:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
+; NO-VP-NEXT:    [[ARRAYIDX2:%.*]] = getelementptr inbounds [2 x i32], ptr [[B]], i64 [[IV]], i32 1
+; NO-VP-NEXT:    [[TMP30:%.*]] = load i32, ptr [[ARRAYIDX2]], align 4
+; NO-VP-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP30]], [[TMP29]]
+; NO-VP-NEXT:    [[ARRAYIDX4:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[IV]]
+; NO-VP-NEXT:    store i32 [[ADD]], ptr [[ARRAYIDX4]], align 4
+; NO-VP-NEXT:    [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
+; NO-VP-NEXT:    [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
+; NO-VP-NEXT:    br i1 [[EXITCOND_NOT]], label [[FOR_COND_CLEANUP]], label [[FOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
+; NO-VP:       for.cond.cleanup:
+; NO-VP-NEXT:    ret void
+;
+entry:
+  br label %for.body
+
+for.body:
+  %iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]
+  %arrayidx = getelementptr inbounds [2 x i32], ptr %b, i64 %iv, i32 0
+  %0 = load i32, ptr %arrayidx, align 4
+  %arrayidx2 = getelementptr inbounds [2 x i32], ptr %b, i64 %iv, i32 1
+  %1 = load i32, ptr %arrayidx2, align 4
+  %add = add nsw i32 %1, %0
+  %arrayidx4 = getelementptr inbounds i32, ptr %a, i64 %iv
+  store i32 %add, ptr %arrayidx4, align 4
+  %iv.next = add nuw nsw i64 %iv, 1
+  %exitcond.not = icmp eq i64 %iv.next, %N
+  br i1 %exitcond.not, label %for.cond.cleanup, label %for.body, !llvm.loop !0
+
+for.cond.cleanup:
+  ret void
+}
+
+!0 = distinct !{!0, !1, !2}
+!1 = !{!"llvm.loop.interleave.count", i32 2}
+!2 = !{!"llvm.loop.vectorize.enable", i1 true}
diff --git a/llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-iv32.ll b/llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-iv32.ll
new file mode 100644
index 000000000000..d5ad99f5cff8
--- /dev/null
+++ b/llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-iv32.ll
@@ -0,0 +1,124 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
+; RUN: opt -passes=loop-vectorize \
+; RUN: -force-tail-folding-style=data-with-evl \
+; RUN: -prefer-predicate-over-epilogue=predicate-dont-vectorize \
+; RUN: -mtriple=riscv64 -mattr=+v -S < %s | FileCheck --check-prefix=IF-EVL %s
+
+; RUN: opt -passes=loop-vectorize \
+; RUN: -force-tail-folding-style=none \
+; RUN: -prefer-predicate-over-epilogue=predicate-else-scalar-epilogue \
+; RUN: -mtriple=riscv64 -mattr=+v -S < %s | FileCheck --check-prefix=NO-VP %s
+
+define void @iv32(ptr noalias %a, ptr noalias %b, i32 %N) {
+; IF-EVL-LABEL: @iv32(
+; IF-EVL-NEXT:  entry:
+; IF-EVL-NEXT:    [[TMP19:%.*]] = sub i32 -1, [[N:%.*]]
+; IF-EVL-NEXT:    [[TMP1:%.*]] = call i32 @llvm.vscale.i32()
+; IF-EVL-NEXT:    [[TMP2:%.*]] = mul i32 [[TMP1]], 4
+; IF-EVL-NEXT:    [[TMP3:%.*]] = icmp ult i32 [[TMP19]], [[TMP2]]
+; IF-EVL-NEXT:    br i1 [[TMP3]], label [[SCALAR_PH:%.*]], label [[ENTRY:%.*]]
+; IF-EVL:       vector.ph:
+; IF-EVL-NEXT:    [[TMP4:%.*]] = call i32 @llvm.vscale.i32()
+; IF-EVL-NEXT:    [[TMP5:%.*]] = mul i32 [[TMP4]], 4
+; IF-EVL-NEXT:    [[TMP6:%.*]] = call i32 @llvm.vscale.i32()
+; IF-EVL-NEXT:    [[TMP7:%.*]] = mul i32 [[TMP6]], 4
+; IF-EVL-NEXT:    [[TMP8:%.*]] = sub i32 [[TMP7]], 1
+; IF-EVL-NEXT:    [[N_RND_UP:%.*]] = add i32 [[N]], [[TMP8]]
+; IF-EVL-NEXT:    [[N_MOD_VF:%.*]] = urem i32 [[N_RND_UP]], [[TMP5]]
+; IF-EVL-NEXT:    [[N_VEC:%.*]] = sub i32 [[N_RND_UP]], [[N_MOD_VF]]
+; IF-EVL-NEXT:    [[TMP9:%.*]] = call i32 @llvm.vscale.i32()
+; IF-EVL-NEXT:    [[TMP10:%.*]] = mul i32 [[TMP9]], 4
+; IF-EVL-NEXT:    br label [[FOR_BODY:%.*]]
+; IF-EVL:       vector.body:
+; IF-EVL-NEXT:    [[IV:%.*]] = phi i32 [ 0, [[ENTRY]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
+; IF-EVL-NEXT:    [[EVL_BASED_IV:%.*]] = phi i32 [ 0, [[ENTRY]] ], [ [[INDEX_EVL_NEXT:%.*]], [[FOR_BODY]] ]
+; IF-EVL-NEXT:    [[TMP11:%.*]] = sub i32 [[N]], [[EVL_BASED_IV]]
+; IF-EVL-NEXT:    [[TMP12:%.*]] = call i32 @llvm.experimental.get.vector.length.i32(i32 [[TMP11]], i32 4, i1 true)
+; IF-EVL-NEXT:    [[TMP13:%.*]] = add i32 [[EVL_BASED_IV]], 0
+; IF-EVL-NEXT:    [[TMP14:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i32 [[TMP13]]
+; IF-EVL-NEXT:    [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[TMP14]], i32 0
+; IF-EVL-NEXT:    [[VP_OP_LOAD:%.*]] = call <vscale x 4 x i32> @llvm.vp.load.nxv4i32.p0(ptr align 4 [[TMP15]], <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer), i32 [[TMP12]])
+; IF-EVL-NEXT:    [[TMP16:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i32 [[TMP13]]
+; IF-EVL-NEXT:    [[TMP17:%.*]] = getelementptr inbounds i32, ptr [[TMP16]], i32 0
+; IF-EVL-NEXT:    call void @llvm.vp.store.nxv4i32.p0(<vscale x 4 x i32> [[VP_OP_LOAD]], ptr align 4 [[TMP17]], <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer), i32 [[TMP12]])
+; IF-EVL-NEXT:    [[INDEX_EVL_NEXT]] = add i32 [[TMP12]], [[EVL_BASED_IV]]
+; IF-EVL-NEXT:    [[IV_NEXT]] = add i32 [[IV]], [[TMP10]]
+; IF-EVL-NEXT:    [[TMP18:%.*]] = icmp eq i32 [[IV_NEXT]], [[N_VEC]]
+; IF-EVL-NEXT:    br i1 [[TMP18]], label [[MIDDLE_BLOCK:%.*]], label [[FOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
+; IF-EVL:       middle.block:
+; IF-EVL-NEXT:    br i1 true, label [[FOR_COND_CLEANUP:%.*]], label [[SCALAR_PH]]
+; IF-EVL:       scalar.ph:
+; IF-EVL-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i32 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY1:%.*]] ]
+; IF-EVL-NEXT:    br label [[FOR_BODY1:%.*]]
+; IF-EVL:       for.body:
+; IF-EVL-NEXT:    [[IV1:%.*]] = phi i32 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT1:%.*]], [[FOR_BODY1]] ]
+; IF-EVL-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[B]], i32 [[IV1]]
+; IF-EVL-NEXT:    [[TMP0:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
+; IF-EVL-NEXT:    [[ARRAYIDX4:%.*]] = getelementptr inbounds i32, ptr [[A]], i32 [[IV1]]
+; IF-EVL-NEXT:    store i32 [[TMP0]], ptr [[ARRAYIDX4]], align 4
+; IF-EVL-NEXT:    [[IV_NEXT1]] = add nuw nsw i32 [[IV1]], 1
+; IF-EVL-NEXT:    [[EXITCOND_NOT:%.*]] = icmp eq i32 [[IV_NEXT1]], [[N]]
+; IF-EVL-NEXT:    br i1 [[EXITCOND_NOT]], label [[FOR_COND_CLEANUP]], label [[FOR_BODY1]], !llvm.loop [[LOOP3:![0-9]+]]
+; IF-EVL:       for.cond.cleanup:
+; IF-EVL-NEXT:    ret void
+;
+; NO-VP-LABEL: @iv32(
+; NO-VP-NEXT:  entry:
+; NO-VP-NEXT:    [[TMP0:%.*]] = call i32 @llvm.vscale.i32()
+; NO-VP-NEXT:    [[TMP10:%.*]] = mul i32 [[TMP0]], 4
+; NO-VP-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i32 [[N:%.*]], [[TMP10]]
+; NO-VP-NEXT:    br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; NO-VP:       vector.ph:
+; NO-VP-NEXT:    [[TMP1:%.*]] = call i32 @llvm.vscale.i32()
+; NO-VP-NEXT:    [[TMP11:%.*]] = mul i32 [[TMP1]], 4
+; NO-VP-NEXT:    [[N_MOD_VF:%.*]] = urem i32 [[N]], [[TMP11]]
+; NO-VP-NEXT:    [[N_VEC:%.*]] = sub i32 [[N]], [[N_MOD_VF]]
+; NO-VP-NEXT:    [[TMP2:%.*]] = call i32 @llvm.vscale.i32()
+; NO-VP-NEXT:    [[TMP12:%.*]] = mul i32 [[TMP2]], 4
+; NO-VP-NEXT:    br label [[VECTOR_BODY:%.*]]
+; NO-VP:       vector.body:
+; NO-VP-NEXT:    [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; NO-VP-NEXT:    [[TMP3:%.*]] = add i32 [[INDEX]], 0
+; NO-VP-NEXT:    [[TMP4:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i32 [[TMP3]]
+; NO-VP-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[TMP4]], i32 0
+; NO-VP-NEXT:    [[WIDE_LOAD:%.*]] = load <vscale x 4 x i32>, ptr [[TMP5]], align 4
+; NO-VP-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i32 [[TMP3]]
+; NO-VP-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[TMP6]], i32 0
+; NO-VP-NEXT:    store <vscale x 4 x i32> [[WIDE_LOAD]], ptr [[TMP7]], align 4
+; NO-VP-NEXT:    [[INDEX_NEXT]] = add nuw i32 [[INDEX]], [[TMP12]]
+; NO-VP-NEXT:    [[TMP8:%.*]] = icmp eq i32 [[INDEX_NEXT]], [[N_VEC]]
+; NO-VP-NEXT:    br i1 [[TMP8]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
+; NO-VP:       middle.block:
+; NO-VP-NEXT:    [[CMP_N:%.*]] = icmp eq i32 [[N]], [[N_VEC]]
+; NO-VP-NEXT:    br i1 [[CMP_N]], label [[FOR_COND_CLEANUP:%.*]], label [[SCALAR_PH]]
+; NO-VP:       scalar.ph:
+; NO-VP-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i32 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; NO-VP-NEXT:    br label [[FOR_BODY:%.*]]
+; NO-VP:       for.body:
+; NO-VP-NEXT:    [[IV:%.*]] = phi i32 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
+; NO-VP-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[B]], i32 [[IV]]
+; NO-VP-NEXT:    [[TMP9:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
+; NO-VP-NEXT:    [[ARRAYIDX4:%.*]] = getelementptr inbounds i32, ptr [[A]], i32 [[IV]]
+; NO-VP-NEXT:    store i32 [[TMP9]], ptr [[ARRAYIDX4]], align 4
+; NO-VP-NEXT:    [[IV_NEXT]] = add nuw nsw i32 [[IV]], 1
+; NO-VP-NEXT:    [[EXITCOND_NOT:%.*]] = icmp eq i32 [[IV_NEXT]], [[N]]
+; NO-VP-NEXT:    br i1 [[EXITCOND_NOT]], label [[FOR_COND_CLEANUP]], label [[FOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
+; NO-VP:       for.cond.cleanup:
+; NO-VP-NEXT:    ret void
+;
+entry:
+  br label %for.body
+
+for.body:
+  %iv = phi i32 [ 0, %entry ], [ %iv.next, %for.body ]
+  %arrayidx = getelementptr inbounds i32, ptr %b, i32 %iv
+  %0 = load i32, ptr %arrayidx, align 4
+  %arrayidx4 = getelementptr inbounds i32, ptr %a, i32 %iv
+  store i32 %0, ptr %arrayidx4, align 4
+  %iv.next = add nuw nsw i32 %iv, 1
+  %exitcond.not = icmp eq i32 %iv.next, %N
+  br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup:
+  ret void
+}
diff --git a/llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-masked-loadstore.ll b/llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-masked-loadstore.ll
new file mode 100644
index 000000000000..203d0c977074
--- /dev/null
+++ b/llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-masked-loadstore.ll
@@ -0,0 +1,132 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
+; RUN: opt -passes=loop-vectorize \
+; RUN: -force-tail-folding-style=data-with-evl \
+; RUN: -prefer-predicate-over-epilogue=predicate-dont-vectorize \
+; RUN: -mtriple=riscv64 -mattr=+v -S < %s | FileCheck %s --check-prefix=IF-EVL
+
+; RUN: opt -passes=loop-vectorize \
+; RUN: -force-tail-folding-style=none \
+; RUN: -prefer-predicate-over-epilogue=predicate-dont-vectorize \
+; RUN: -mtriple=riscv64 -mattr=+v -S < %s | FileCheck %s --check-prefix=NO-VP
+
+define void @masked_loadstore(ptr noalias %a, ptr noalias %b, i64 %n) {
+; IF-EVL-LABEL: @masked_loadstore(
+; IF-EVL-NEXT:  entry:
+; IF-EVL-NEXT:    [[TMP0:%.*]] = sub i64 -1, [[N:%.*]]
+; IF-EVL-NEXT:    [[TMP1:%.*]] = call i64 @llvm.vscale.i64()
+; IF-EVL-NEXT:    [[TMP2:%.*]] = mul i64 [[TMP1]], 4
+; IF-EVL-NEXT:    [[TMP3:%.*]] = icmp ult i64 [[TMP0]], [[TMP2]]
+; IF-EVL-NEXT:    br i1 [[TMP3]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; IF-EVL:       vector.ph:
+; IF-EVL-NEXT:    [[TMP4:%.*]] = call i64 @llvm.vscale.i64()
+; IF-EVL-NEXT:    [[TMP5:%.*]] = mul i64 [[TMP4]], 4
+; IF-EVL-NEXT:    [[TMP6:%.*]] = call i64 @llvm.vscale.i64()
+; IF-EVL-NEXT:    [[TMP7:%.*]] = mul i64 [[TMP6]], 4
+; IF-EVL-NEXT:    [[TMP8:%.*]] = sub i64 [[TMP7]], 1
+; IF-EVL-NEXT:    [[N_RND_UP:%.*]] = add i64 [[N]], [[TMP8]]
+; IF-EVL-NEXT:    [[N_MOD_VF:%.*]] = urem i64 [[N_RND_UP]], [[TMP5]]
+; IF-EVL-NEXT:    [[N_VEC:%.*]] = sub i64 [[N_RND_UP]], [[N_MOD_VF]]
+; IF-EVL-NEXT:    [[TRIP_COUNT_MINUS_1:%.*]] = sub i64 [[N]], 1
+; IF-EVL-NEXT:    [[TMP9:%.*]] = call i64 @llvm.vscale.i64()
+; IF-EVL-NEXT:    [[TMP10:%.*]] = mul i64 [[TMP9]], 4
+; IF-EVL-NEXT:    [[BROADCAST_SPLATINSERT1:%.*]] = insertelement <vscale x 4 x i64> poison, i64 [[TRIP_COUNT_MINUS_1]], i64 0
+; IF-EVL-NEXT:    [[BROADCAST_SPLAT2:%.*]] = shufflevector <vscale x 4 x i64> [[BROADCAST_SPLATINSERT1]], <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer
+; IF-EVL-NEXT:    br label [[VECTOR_BODY:%.*]]
+; IF-EVL:       vector.body:
+; IF-EVL-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; IF-EVL-NEXT:    [[EVL_BASED_IV:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_EVL_NEXT:%.*]], [[VECTOR_BODY]] ]
+; IF-EVL-NEXT:    [[TMP11:%.*]] = sub i64 [[N]], [[EVL_BASED_IV]]
+; IF-EVL-NEXT:    [[TMP12:%.*]] = call i32 @llvm.experimental.get.vector.length.i64(i64 [[TMP11]], i32 4, i1 true)
+; IF-EVL-NEXT:    [[TMP13:%.*]] = add i64 [[EVL_BASED_IV]], 0
+; IF-EVL-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 4 x i64> poison, i64 [[EVL_BASED_IV]], i64 0
+; IF-EVL-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 4 x i64> [[BROADCAST_SPLATINSERT]], <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer
+; IF-EVL-NEXT:    [[TMP14:%.*]] = call <vscale x 4 x i64> @llvm.experimental.stepvector.nxv4i64()
+; IF-EVL-NEXT:    [[TMP15:%.*]] = add <vscale x 4 x i64> zeroinitializer, [[TMP14]]
+; IF-EVL-NEXT:    [[VEC_IV:%.*]] = add <vscale x 4 x i64> [[BROADCAST_SPLAT]], [[TMP15]]
+; IF-EVL-NEXT:    [[TMP16:%.*]] = icmp ule <vscale x 4 x i64> [[VEC_IV]], [[BROADCAST_SPLAT2]]
+; IF-EVL-NEXT:    [[TMP17:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 [[TMP13]]
+; IF-EVL-NEXT:    [[TMP18:%.*]] = getelementptr inbounds i32, ptr [[TMP17]], i32 0
+; IF-EVL-NEXT:    [[VP_OP_LOAD:%.*]] = call <vscale x 4 x i32> @llvm.vp.load.nxv4i32.p0(ptr align 4 [[TMP18]], <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer), i32 [[TMP12]])
+; IF-EVL-NEXT:    [[TMP19:%.*]] = icmp ne <vscale x 4 x i32> [[VP_OP_LOAD]], zeroinitializer
+; IF-EVL-NEXT:    [[TMP20:%.*]] = select <vscale x 4 x i1> [[TMP16]], <vscale x 4 x i1> [[TMP19]], <vscale x 4 x i1> zeroinitializer
+; IF-EVL-NEXT:    [[TMP21:%.*]] = getelementptr i32, ptr [[A:%.*]], i64 [[TMP13]]
+; IF-EVL-NEXT:    [[TMP22:%.*]] = getelementptr i32, ptr [[TMP21]], i32 0
+; IF-EVL-NEXT:    [[VP_OP_LOAD3:%.*]] = call <vscale x 4 x i32> @llvm.vp.load.nxv4i32.p0(ptr align 4 [[TMP22]], <vscale x 4 x i1> [[TMP20]], i32 [[TMP12]])
+; IF-EVL-NEXT:    [[TMP23:%.*]] = add <vscale x 4 x i32> [[VP_OP_LOAD]], [[VP_OP_LOAD3]]
+; IF-EVL-NEXT:    call void @llvm.vp.store.nxv4i32.p0(<vscale x 4 x i32> [[TMP23]], ptr align 4 [[TMP22]], <vscale x 4 x i1> [[TMP20]], i32 [[TMP12]])
+; IF-EVL-NEXT:    [[TMP24:%.*]] = zext i32 [[TMP12]] to i64
+; IF-EVL-NEXT:    [[INDEX_EVL_NEXT]] = add i64 [[TMP24]], [[EVL_BASED_IV]]
+; IF-EVL-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP10]]
+; IF-EVL-NEXT:    [[TMP25:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
+; IF-EVL-NEXT:    br i1 [[TMP25]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
+; IF-EVL:       middle.block:
+; IF-EVL-NEXT:    br i1 true, label [[EXIT:%.*]], label [[SCALAR_PH]]
+; IF-EVL:       scalar.ph:
+; IF-EVL-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; IF-EVL-NEXT:    br label [[FOR_BODY:%.*]]
+; IF-EVL:       for.body:
+; IF-EVL-NEXT:    [[I_011:%.*]] = phi i64 [ [[INC:%.*]], [[FOR_INC:%.*]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
+; IF-EVL-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[I_011]]
+; IF-EVL-NEXT:    [[TMP26:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
+; IF-EVL-NEXT:    [[CMP1:%.*]] = icmp ne i32 [[TMP26]], 0
+; IF-EVL-NEXT:    br i1 [[CMP1]], label [[IF_THEN:%.*]], label [[FOR_INC]]
+; IF-EVL:       if.then:
+; IF-EVL-NEXT:    [[ARRAYIDX3:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[I_011]]
+; IF-EVL-NEXT:    [[TMP27:%.*]] = load i32, ptr [[ARRAYIDX3]], align 4
+; IF-EVL-NEXT:    [[ADD:%.*]] = add i32 [[TMP26]], [[TMP27]]
+; IF-EVL-NEXT:    store i32 [[ADD]], ptr [[ARRAYIDX3]], align 4
+; IF-EVL-NEXT:    br label [[FOR_INC]]
+; IF-EVL:       for.inc:
+; IF-EVL-NEXT:    [[INC]] = add nuw nsw i64 [[I_011]], 1
+; IF-EVL-NEXT:    [[EXITCOND_NOT:%.*]] = icmp eq i64 [[INC]], [[N]]
+; IF-EVL-NEXT:    br i1 [[EXITCOND_NOT]], label [[EXIT]], label [[FOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
+; IF-EVL:       exit:
+; IF-EVL-NEXT:    ret void
+;
+; NO-VP-LABEL: @masked_loadstore(
+; NO-VP-NEXT:  entry:
+; NO-VP-NEXT:    br label [[FOR_BODY:%.*]]
+; NO-VP:       for.body:
+; NO-VP-NEXT:    [[I_011:%.*]] = phi i64 [ [[INC:%.*]], [[FOR_INC:%.*]] ], [ 0, [[ENTRY:%.*]] ]
+; NO-VP-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 [[I_011]]
+; NO-VP-NEXT:    [[TMP0:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
+; NO-VP-NEXT:    [[CMP1:%.*]] = icmp ne i32 [[TMP0]], 0
+; NO-VP-NEXT:    br i1 [[CMP1]], label [[IF_THEN:%.*]], label [[FOR_INC]]
+; NO-VP:       if.then:
+; NO-VP-NEXT:    [[ARRAYIDX3:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[I_011]]
+; NO-VP-NEXT:    [[TMP1:%.*]] = load i32, ptr [[ARRAYIDX3]], align 4
+; NO-VP-NEXT:    [[ADD:%.*]] = add i32 [[TMP0]], [[TMP1]]
+; NO-VP-NEXT:    store i32 [[ADD]], ptr [[ARRAYIDX3]], align 4
+; NO-VP-NEXT:    br label [[FOR_INC]]
+; NO-VP:       for.inc:
+; NO-VP-NEXT:    [[INC]] = add nuw nsw i64 [[I_011]], 1
+; NO-VP-NEXT:    [[EXITCOND_NOT:%.*]] = icmp eq i64 [[INC]], [[N:%.*]]
+; NO-VP-NEXT:    br i1 [[EXITCOND_NOT]], label [[EXIT:%.*]], label [[FOR_BODY]]
+; NO-VP:       exit:
+; NO-VP-NEXT:    ret void
+;
+entry:
+  br label %for.body
+
+for.body:
+  %i.011 = phi i64 [ %inc, %for.inc ], [ 0, %entry ]
+  %arrayidx = getelementptr inbounds i32, ptr %b, i64 %i.011
+  %0 = load i32, ptr %arrayidx, align 4
+  %cmp1 = icmp ne i32 %0, 0
+  br i1 %cmp1, label %if.then, label %for.inc
+
+if.then:
+  %arrayidx3 = getelementptr inbounds i32, ptr %a, i64 %i.011
+  %1 = load i32, ptr %arrayidx3, align 4
+  %add = add i32 %0, %1
+  store i32 %add, ptr %arrayidx3, align 4
+  br label %for.inc
+
+for.inc:
+  %inc = add nuw nsw i64 %i.011, 1
+  %exitcond.not = icmp eq i64 %inc, %n
+  br i1 %exitcond.not, label %exit, label %for.body
+
+exit:
+  ret void
+}
diff --git a/llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-no-masking.ll b/llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-no-masking.ll
new file mode 100644
index 000000000000..1c49fba1370e
--- /dev/null
+++ b/llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-no-masking.ll
@@ -0,0 +1,36 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
+; RUN: opt -passes=loop-vectorize \
+; RUN: -force-tail-folding-style=data-with-evl \
+; RUN: -prefer-predicate-over-epilogue=predicate-dont-vectorize \
+; RUN: -mtriple=riscv64 -mattr=+v -S < %s | FileCheck %s
+
+; RUN: opt -passes=loop-vectorize \
+; RUN: -force-tail-folding-style=none \
+; RUN: -prefer-predicate-over-epilogue=predicate-dont-vectorize \
+; RUN: -mtriple=riscv64 -mattr=+v -S < %s | FileCheck %s
+
+; No need to emit predicated vector code if the vector instructions with masking are not required.
+define i32 @no_masking() {
+; CHECK-LABEL: @no_masking(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    br label [[BODY:%.*]]
+; CHECK:       body:
+; CHECK-NEXT:    [[P:%.*]] = phi i32 [ 1, [[ENTRY:%.*]] ], [ [[INC:%.*]], [[BODY]] ]
+; CHECK-NEXT:    [[INC]] = add i32 [[P]], 1
+; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i32 [[INC]], 0
+; CHECK-NEXT:    br i1 [[CMP]], label [[END:%.*]], label [[BODY]]
+; CHECK:       end:
+; CHECK-NEXT:    ret i32 0
+;
+entry:
+  br label %body
+
+body:
+  %p = phi i32 [ 1, %entry ], [ %inc, %body ]
+  %inc = add i32 %p, 1
+  %cmp = icmp eq i32 %inc, 0
+  br i1 %cmp, label %end, label %body
+
+end:
+  ret i32 0
+}
diff --git a/llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-reverse-load-store.ll b/llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-reverse-load-store.ll
new file mode 100644
index 000000000000..f2222e0a1f93
--- /dev/null
+++ b/llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-reverse-load-store.ll
@@ -0,0 +1,119 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
+; RUN: opt -passes=loop-vectorize \
+; RUN: -force-tail-folding-style=data-with-evl \
+; RUN: -prefer-predicate-over-epilogue=predicate-dont-vectorize \
+; RUN: -mtriple=riscv64 -mattr=+v -S < %s | FileCheck %s --check-prefix=IF-EVL
+
+; RUN: opt -passes=loop-vectorize \
+; RUN: -force-tail-folding-style=none \
+; RUN: -prefer-predicate-over-epilogue=predicate-dont-vectorize \
+; RUN: -mtriple=riscv64 -mattr=+v -S < %s | FileCheck %s --check-prefix=NO-VP
+
+; FIXME: reversed loads/stores are not supported yet with predicated vectorization.
+define void @reverse_load_store(i64 %startval, ptr noalias %ptr, ptr noalias %ptr2) {
+; IF-EVL-LABEL: @reverse_load_store(
+; IF-EVL-NEXT:  entry:
+; IF-EVL-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; IF-EVL:       vector.ph:
+; IF-EVL-NEXT:    [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
+; IF-EVL-NEXT:    [[TMP1:%.*]] = mul i64 [[TMP0]], 4
+; IF-EVL-NEXT:    [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
+; IF-EVL-NEXT:    [[TMP3:%.*]] = mul i64 [[TMP2]], 4
+; IF-EVL-NEXT:    [[TMP4:%.*]] = sub i64 [[TMP3]], 1
+; IF-EVL-NEXT:    [[N_RND_UP:%.*]] = add i64 1024, [[TMP4]]
+; IF-EVL-NEXT:    [[N_MOD_VF:%.*]] = urem i64 [[N_RND_UP]], [[TMP1]]
+; IF-EVL-NEXT:    [[N_VEC:%.*]] = sub i64 [[N_RND_UP]], [[N_MOD_VF]]
+; IF-EVL-NEXT:    [[IND_END:%.*]] = sub i64 [[STARTVAL:%.*]], [[N_VEC]]
+; IF-EVL-NEXT:    [[IND_END1:%.*]] = trunc i64 [[N_VEC]] to i32
+; IF-EVL-NEXT:    [[TMP5:%.*]] = call i64 @llvm.vscale.i64()
+; IF-EVL-NEXT:    [[TMP6:%.*]] = mul i64 [[TMP5]], 4
+; IF-EVL-NEXT:    br label [[VECTOR_BODY:%.*]]
+; IF-EVL:       vector.body:
+; IF-EVL-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; IF-EVL-NEXT:    [[OFFSET_IDX:%.*]] = sub i64 [[STARTVAL]], [[INDEX]]
+; IF-EVL-NEXT:    [[TMP7:%.*]] = add i64 [[OFFSET_IDX]], 0
+; IF-EVL-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 4 x i64> poison, i64 [[INDEX]], i64 0
+; IF-EVL-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 4 x i64> [[BROADCAST_SPLATINSERT]], <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer
+; IF-EVL-NEXT:    [[TMP8:%.*]] = call <vscale x 4 x i64> @llvm.experimental.stepvector.nxv4i64()
+; IF-EVL-NEXT:    [[TMP9:%.*]] = add <vscale x 4 x i64> zeroinitializer, [[TMP8]]
+; IF-EVL-NEXT:    [[VEC_IV:%.*]] = add <vscale x 4 x i64> [[BROADCAST_SPLAT]], [[TMP9]]
+; IF-EVL-NEXT:    [[TMP10:%.*]] = icmp ule <vscale x 4 x i64> [[VEC_IV]], shufflevector (<vscale x 4 x i64> insertelement (<vscale x 4 x i64> poison, i64 1023, i64 0), <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer)
+; IF-EVL-NEXT:    [[TMP11:%.*]] = add i64 [[TMP7]], -1
+; IF-EVL-NEXT:    [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[PTR:%.*]], i64 [[TMP11]]
+; IF-EVL-NEXT:    [[TMP13:%.*]] = call i64 @llvm.vscale.i64()
+; IF-EVL-NEXT:    [[TMP14:%.*]] = mul i64 [[TMP13]], 4
+; IF-EVL-NEXT:    [[TMP15:%.*]] = mul i64 0, [[TMP14]]
+; IF-EVL-NEXT:    [[TMP16:%.*]] = sub i64 1, [[TMP14]]
+; IF-EVL-NEXT:    [[TMP17:%.*]] = getelementptr inbounds i32, ptr [[TMP12]], i64 [[TMP15]]
+; IF-EVL-NEXT:    [[TMP18:%.*]] = getelementptr inbounds i32, ptr [[TMP17]], i64 [[TMP16]]
+; IF-EVL-NEXT:    [[REVERSE:%.*]] = call <vscale x 4 x i1> @llvm.experimental.vector.reverse.nxv4i1(<vscale x 4 x i1> [[TMP10]])
+; IF-EVL-NEXT:    [[WIDE_MASKED_LOAD:%.*]] = call <vscale x 4 x i32> @llvm.masked.load.nxv4i32.p0(ptr [[TMP18]], i32 4, <vscale x 4 x i1> [[REVERSE]], <vscale x 4 x i32> poison)
+; IF-EVL-NEXT:    [[REVERSE3:%.*]] = call <vscale x 4 x i32> @llvm.experimental.vector.reverse.nxv4i32(<vscale x 4 x i32> [[WIDE_MASKED_LOAD]])
+; IF-EVL-NEXT:    [[TMP19:%.*]] = getelementptr inbounds i32, ptr [[PTR2:%.*]], i64 [[TMP11]]
+; IF-EVL-NEXT:    [[TMP20:%.*]] = call i64 @llvm.vscale.i64()
+; IF-EVL-NEXT:    [[TMP21:%.*]] = mul i64 [[TMP20]], 4
+; IF-EVL-NEXT:    [[TMP22:%.*]] = mul i64 0, [[TMP21]]
+; IF-EVL-NEXT:    [[TMP23:%.*]] = sub i64 1, [[TMP21]]
+; IF-EVL-NEXT:    [[TMP24:%.*]] = getelementptr inbounds i32, ptr [[TMP19]], i64 [[TMP22]]
+; IF-EVL-NEXT:    [[TMP25:%.*]] = getelementptr inbounds i32, ptr [[TMP24]], i64 [[TMP23]]
+; IF-EVL-NEXT:    [[REVERSE4:%.*]] = call <vscale x 4 x i1> @llvm.experimental.vector.reverse.nxv4i1(<vscale x 4 x i1> [[TMP10]])
+; IF-EVL-NEXT:    [[REVERSE5:%.*]] = call <vscale x 4 x i32> @llvm.experimental.vector.reverse.nxv4i32(<vscale x 4 x i32> [[REVERSE3]])
+; IF-EVL-NEXT:    call void @llvm.masked.store.nxv4i32.p0(<vscale x 4 x i32> [[REVERSE5]], ptr [[TMP25]], i32 4, <vscale x 4 x i1> [[REVERSE4]])
+; IF-EVL-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP6]]
+; IF-EVL-NEXT:    [[TMP26:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
+; IF-EVL-NEXT:    br i1 [[TMP26]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
+; IF-EVL:       middle.block:
+; IF-EVL-NEXT:    br i1 true, label [[LOOPEND:%.*]], label [[SCALAR_PH]]
+; IF-EVL:       scalar.ph:
+; IF-EVL-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[IND_END]], [[MIDDLE_BLOCK]] ], [ [[STARTVAL]], [[ENTRY:%.*]] ]
+; IF-EVL-NEXT:    [[BC_RESUME_VAL2:%.*]] = phi i32 [ [[IND_END1]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
+; IF-EVL-NEXT:    br label [[FOR_BODY:%.*]]
+; IF-EVL:       for.body:
+; IF-EVL-NEXT:    [[ADD_PHI:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[ADD:%.*]], [[FOR_BODY]] ]
+; IF-EVL-NEXT:    [[I:%.*]] = phi i32 [ [[BC_RESUME_VAL2]], [[SCALAR_PH]] ], [ [[INC:%.*]], [[FOR_BODY]] ]
+; IF-EVL-NEXT:    [[ADD]] = add i64 [[ADD_PHI]], -1
+; IF-EVL-NEXT:    [[GEPL:%.*]] = getelementptr inbounds i32, ptr [[PTR]], i64 [[ADD]]
+; IF-EVL-NEXT:    [[TMP:%.*]] = load i32, ptr [[GEPL]], align 4
+; IF-EVL-NEXT:    [[GEPS:%.*]] = getelementptr inbounds i32, ptr [[PTR2]], i64 [[ADD]]
+; IF-EVL-NEXT:    store i32 [[TMP]], ptr [[GEPS]], align 4
+; IF-EVL-NEXT:    [[INC]] = add i32 [[I]], 1
+; IF-EVL-NEXT:    [[EXITCOND:%.*]] = icmp ne i32 [[INC]], 1024
+; IF-EVL-NEXT:    br i1 [[EXITCOND]], label [[FOR_BODY]], label [[LOOPEND]], !llvm.loop [[LOOP3:![0-9]+]]
+; IF-EVL:       loopend:
+; IF-EVL-NEXT:    ret void
+;
+; NO-VP-LABEL: @reverse_load_store(
+; NO-VP-NEXT:  entry:
+; NO-VP-NEXT:    br label [[FOR_BODY:%.*]]
+; NO-VP:       for.body:
+; NO-VP-NEXT:    [[ADD_PHI:%.*]] = phi i64 [ [[STARTVAL:%.*]], [[ENTRY:%.*]] ], [ [[ADD:%.*]], [[FOR_BODY]] ]
+; NO-VP-NEXT:    [[I:%.*]] = phi i32 [ 0, [[ENTRY]] ], [ [[INC:%.*]], [[FOR_BODY]] ]
+; NO-VP-NEXT:    [[ADD]] = add i64 [[ADD_PHI]], -1
+; NO-VP-NEXT:    [[GEPL:%.*]] = getelementptr inbounds i32, ptr [[PTR:%.*]], i64 [[ADD]]
+; NO-VP-NEXT:    [[TMP:%.*]] = load i32, ptr [[GEPL]], align 4
+; NO-VP-NEXT:    [[GEPS:%.*]] = getelementptr inbounds i32, ptr [[PTR2:%.*]], i64 [[ADD]]
+; NO-VP-NEXT:    store i32 [[TMP]], ptr [[GEPS]], align 4
+; NO-VP-NEXT:    [[INC]] = add i32 [[I]], 1
+; NO-VP-NEXT:    [[EXITCOND:%.*]] = icmp ne i32 [[INC]], 1024
+; NO-VP-NEXT:    br i1 [[EXITCOND]], label [[FOR_BODY]], label [[LOOPEND:%.*]]
+; NO-VP:       loopend:
+; NO-VP-NEXT:    ret void
+;
+entry:
+  br label %for.body
+
+for.body:
+  %add.phi = phi i64 [ %startval, %entry ], [ %add, %for.body ]
+  %i = phi i32 [ 0, %entry ], [ %inc, %for.body ]
+  %add = add i64 %add.phi, -1
+  %gepl = getelementptr inbounds i32, ptr %ptr, i64 %add
+  %tmp = load i32, ptr %gepl, align 4
+  %geps = getelementptr inbounds i32, ptr %ptr2, i64 %add
+  store i32 %tmp, ptr %geps, align 4
+  %inc = add i32 %i, 1
+  %exitcond = icmp ne i32 %inc, 1024
+  br i1 %exitcond, label %for.body, label %loopend
+
+loopend:
+  ret void
+}
diff --git a/llvm/test/Transforms/LoopVectorize/RISCV/vectorize-vp-intrinsics.ll b/llvm/test/Transforms/LoopVectorize/RISCV/vectorize-vp-intrinsics.ll
new file mode 100644
index 000000000000..c69bb17f698a
--- /dev/null
+++ b/llvm/test/Transforms/LoopVectorize/RISCV/vectorize-vp-intrinsics.ll
@@ -0,0 +1,142 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
+; RUN: opt -passes=loop-vectorize \
+; RUN: -force-tail-folding-style=data-with-evl \
+; RUN: -prefer-predicate-over-epilogue=predicate-dont-vectorize \
+; RUN: -mtriple=riscv64 -mattr=+v -S < %s | FileCheck --check-prefix=IF-EVL %s
+
+; RUN: opt -passes=loop-vectorize \
+; RUN: -force-tail-folding-style=none \
+; RUN: -prefer-predicate-over-epilogue=predicate-else-scalar-epilogue \
+; RUN: -mtriple=riscv64 -mattr=+v -S < %s | FileCheck --check-prefix=NO-VP %s
+
+define void @foo(ptr noalias %a, ptr noalias %b, ptr noalias %c, i64 %N) {
+; IF-EVL-LABEL: @foo(
+; IF-EVL-NEXT:  entry:
+; IF-EVL-NEXT:    [[TMP0:%.*]] = sub i64 -1, [[N:%.*]]
+; IF-EVL-NEXT:    [[TMP1:%.*]] = call i64 @llvm.vscale.i64()
+; IF-EVL-NEXT:    [[TMP2:%.*]] = mul i64 [[TMP1]], 4
+; IF-EVL-NEXT:    [[TMP3:%.*]] = icmp ult i64 [[TMP0]], [[TMP2]]
+; IF-EVL-NEXT:    br i1 [[TMP3]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; IF-EVL:       vector.ph:
+; IF-EVL-NEXT:    [[TMP4:%.*]] = call i64 @llvm.vscale.i64()
+; IF-EVL-NEXT:    [[TMP5:%.*]] = mul i64 [[TMP4]], 4
+; IF-EVL-NEXT:    [[TMP6:%.*]] = call i64 @llvm.vscale.i64()
+; IF-EVL-NEXT:    [[TMP7:%.*]] = mul i64 [[TMP6]], 4
+; IF-EVL-NEXT:    [[TMP8:%.*]] = sub i64 [[TMP7]], 1
+; IF-EVL-NEXT:    [[N_RND_UP:%.*]] = add i64 [[N]], [[TMP8]]
+; IF-EVL-NEXT:    [[N_MOD_VF:%.*]] = urem i64 [[N_RND_UP]], [[TMP5]]
+; IF-EVL-NEXT:    [[N_VEC:%.*]] = sub i64 [[N_RND_UP]], [[N_MOD_VF]]
+; IF-EVL-NEXT:    [[TMP9:%.*]] = call i64 @llvm.vscale.i64()
+; IF-EVL-NEXT:    [[TMP10:%.*]] = mul i64 [[TMP9]], 4
+; IF-EVL-NEXT:    br label [[VECTOR_BODY:%.*]]
+; IF-EVL:       vector.body:
+; IF-EVL-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; IF-EVL-NEXT:    [[EVL_BASED_IV:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_EVL_NEXT:%.*]], [[VECTOR_BODY]] ]
+; IF-EVL-NEXT:    [[TMP11:%.*]] = sub i64 [[N]], [[EVL_BASED_IV]]
+; IF-EVL-NEXT:    [[TMP12:%.*]] = call i32 @llvm.experimental.get.vector.length.i64(i64 [[TMP11]], i32 4, i1 true)
+; IF-EVL-NEXT:    [[TMP13:%.*]] = add i64 [[EVL_BASED_IV]], 0
+; IF-EVL-NEXT:    [[TMP14:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 [[TMP13]]
+; IF-EVL-NEXT:    [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[TMP14]], i32 0
+; IF-EVL-NEXT:    [[VP_OP_LOAD:%.*]] = call <vscale x 4 x i32> @llvm.vp.load.nxv4i32.p0(ptr align 4 [[TMP15]], <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer), i32 [[TMP12]])
+; IF-EVL-NEXT:    [[TMP16:%.*]] = getelementptr inbounds i32, ptr [[C:%.*]], i64 [[TMP13]]
+; IF-EVL-NEXT:    [[TMP17:%.*]] = getelementptr inbounds i32, ptr [[TMP16]], i32 0
+; IF-EVL-NEXT:    [[VP_OP_LOAD1:%.*]] = call <vscale x 4 x i32> @llvm.vp.load.nxv4i32.p0(ptr align 4 [[TMP17]], <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer), i32 [[TMP12]])
+; IF-EVL-NEXT:    [[TMP18:%.*]] = add nsw <vscale x 4 x i32> [[VP_OP_LOAD1]], [[VP_OP_LOAD]]
+; IF-EVL-NEXT:    [[TMP19:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP13]]
+; IF-EVL-NEXT:    [[TMP20:%.*]] = getelementptr inbounds i32, ptr [[TMP19]], i32 0
+; IF-EVL-NEXT:    call void @llvm.vp.store.nxv4i32.p0(<vscale x 4 x i32> [[TMP18]], ptr align 4 [[TMP20]], <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer), i32 [[TMP12]])
+; IF-EVL-NEXT:    [[TMP21:%.*]] = zext i32 [[TMP12]] to i64
+; IF-EVL-NEXT:    [[INDEX_EVL_NEXT]] = add i64 [[TMP21]], [[EVL_BASED_IV]]
+; IF-EVL-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP10]]
+; IF-EVL-NEXT:    [[TMP22:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
+; IF-EVL-NEXT:    br i1 [[TMP22]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
+; IF-EVL:       middle.block:
+; IF-EVL-NEXT:    br i1 true, label [[FOR_COND_CLEANUP:%.*]], label [[SCALAR_PH]]
+; IF-EVL:       scalar.ph:
+; IF-EVL-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; IF-EVL-NEXT:    br label [[FOR_BODY:%.*]]
+; IF-EVL:       for.body:
+; IF-EVL-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
+; IF-EVL-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[IV]]
+; IF-EVL-NEXT:    [[TMP23:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
+; IF-EVL-NEXT:    [[ARRAYIDX2:%.*]] = getelementptr inbounds i32, ptr [[C]], i64 [[IV]]
+; IF-EVL-NEXT:    [[TMP24:%.*]] = load i32, ptr [[ARRAYIDX2]], align 4
+; IF-EVL-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP24]], [[TMP23]]
+; IF-EVL-NEXT:    [[ARRAYIDX4:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[IV]]
+; IF-EVL-NEXT:    store i32 [[ADD]], ptr [[ARRAYIDX4]], align 4
+; IF-EVL-NEXT:    [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
+; IF-EVL-NEXT:    [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
+; IF-EVL-NEXT:    br i1 [[EXITCOND_NOT]], label [[FOR_COND_CLEANUP]], label [[FOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
+; IF-EVL:       for.cond.cleanup:
+; IF-EVL-NEXT:    ret void
+;
+; NO-VP-LABEL: @foo(
+; NO-VP-NEXT:  entry:
+; NO-VP-NEXT:    [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
+; NO-VP-NEXT:    [[TMP1:%.*]] = mul i64 [[TMP0]], 4
+; NO-VP-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[N:%.*]], [[TMP1]]
+; NO-VP-NEXT:    br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; NO-VP:       vector.ph:
+; NO-VP-NEXT:    [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
+; NO-VP-NEXT:    [[TMP3:%.*]] = mul i64 [[TMP2]], 4
+; NO-VP-NEXT:    [[N_MOD_VF:%.*]] = urem i64 [[N]], [[TMP3]]
+; NO-VP-NEXT:    [[N_VEC:%.*]] = sub i64 [[N]], [[N_MOD_VF]]
+; NO-VP-NEXT:    [[TMP4:%.*]] = call i64 @llvm.vscale.i64()
+; NO-VP-NEXT:    [[TMP5:%.*]] = mul i64 [[TMP4]], 4
+; NO-VP-NEXT:    br label [[VECTOR_BODY:%.*]]
+; NO-VP:       vector.body:
+; NO-VP-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; NO-VP-NEXT:    [[TMP6:%.*]] = add i64 [[INDEX]], 0
+; NO-VP-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 [[TMP6]]
+; NO-VP-NEXT:    [[TMP8:%.*]] = getelementptr inbounds i32, ptr [[TMP7]], i32 0
+; NO-VP-NEXT:    [[WIDE_LOAD:%.*]] = load <vscale x 4 x i32>, ptr [[TMP8]], align 4
+; NO-VP-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i32, ptr [[C:%.*]], i64 [[TMP6]]
+; NO-VP-NEXT:    [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[TMP9]], i32 0
+; NO-VP-NEXT:    [[WIDE_LOAD1:%.*]] = load <vscale x 4 x i32>, ptr [[TMP10]], align 4
+; NO-VP-NEXT:    [[TMP11:%.*]] = add nsw <vscale x 4 x i32> [[WIDE_LOAD1]], [[WIDE_LOAD]]
+; NO-VP-NEXT:    [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP6]]
+; NO-VP-NEXT:    [[TMP13:%.*]] = getelementptr inbounds i32, ptr [[TMP12]], i32 0
+; NO-VP-NEXT:    store <vscale x 4 x i32> [[TMP11]], ptr [[TMP13]], align 4
+; NO-VP-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP5]]
+; NO-VP-NEXT:    [[TMP14:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
+; NO-VP-NEXT:    br i1 [[TMP14]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
+; NO-VP:       middle.block:
+; NO-VP-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[N]], [[N_VEC]]
+; NO-VP-NEXT:    br i1 [[CMP_N]], label [[FOR_COND_CLEANUP:%.*]], label [[SCALAR_PH]]
+; NO-VP:       scalar.ph:
+; NO-VP-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; NO-VP-NEXT:    br label [[FOR_BODY:%.*]]
+; NO-VP:       for.body:
+; NO-VP-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
+; NO-VP-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[IV]]
+; NO-VP-NEXT:    [[TMP15:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
+; NO-VP-NEXT:    [[ARRAYIDX2:%.*]] = getelementptr inbounds i32, ptr [[C]], i64 [[IV]]
+; NO-VP-NEXT:    [[TMP16:%.*]] = load i32, ptr [[ARRAYIDX2]], align 4
+; NO-VP-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP16]], [[TMP15]]
+; NO-VP-NEXT:    [[ARRAYIDX4:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[IV]]
+; NO-VP-NEXT:    store i32 [[ADD]], ptr [[ARRAYIDX4]], align 4
+; NO-VP-NEXT:    [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
+; NO-VP-NEXT:    [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
+; NO-VP-NEXT:    br i1 [[EXITCOND_NOT]], label [[FOR_COND_CLEANUP]], label [[FOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
+; NO-VP:       for.cond.cleanup:
+; NO-VP-NEXT:    ret void
+;
+entry:
+  br label %for.body
+
+for.body:
+  %iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]
+  %arrayidx = getelementptr inbounds i32, ptr %b, i64 %iv
+  %0 = load i32, ptr %arrayidx, align 4
+  %arrayidx2 = getelementptr inbounds i32, ptr %c, i64 %iv
+  %1 = load i32, ptr %arrayidx2, align 4
+  %add = add nsw i32 %1, %0
+  %arrayidx4 = getelementptr inbounds i32, ptr %a, i64 %iv
+  store i32 %add, ptr %arrayidx4, align 4
+  %iv.next = add nuw nsw i64 %iv, 1
+  %exitcond.not = icmp eq i64 %iv.next, %N
+  br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup:
+  ret void
+}
diff --git a/llvm/test/Transforms/LoopVectorize/RISCV/vplan-vp-intrinsics.ll b/llvm/test/Transforms/LoopVectorize/RISCV/vplan-vp-intrinsics.ll
new file mode 100644
index 000000000000..72b881bd44c7
--- /dev/null
+++ b/llvm/test/Transforms/LoopVectorize/RISCV/vplan-vp-intrinsics.ll
@@ -0,0 +1,134 @@
+; REQUIRES: asserts
+
+; RUN: opt -passes=loop-vectorize -debug-only=loop-vectorize \
+; RUN: -force-tail-folding-style=data-with-evl \
+; RUN: -prefer-predicate-over-epilogue=predicate-dont-vectorize \
+; RUN: -mtriple=riscv64 -mattr=+v -riscv-v-vector-bits-max=128 -disable-output < %s 2>&1 | FileCheck --check-prefixes=IF-EVL,CHECK %s
+
+; RUN: opt -passes=loop-vectorize -debug-only=loop-vectorize \
+; RUN: -force-tail-folding-style=none \
+; RUN: -prefer-predicate-over-epilogue=predicate-else-scalar-epilogue \
+; RUN: -mtriple=riscv64 -mattr=+v -riscv-v-vector-bits-max=128 -disable-output < %s 2>&1 | FileCheck --check-prefixes=NO-VP,CHECK %s
+
+define void @foo(ptr noalias %a, ptr noalias %b, ptr noalias %c, i64 %N) {
+; IF-EVL: VPlan 'Initial VPlan for VF={vscale x 1,vscale x 2,vscale x 4},UF={1}' {
+; IF-EVL-NEXT: Live-in vp<[[VFUF:%[0-9]+]]> = VF * UF
+; IF-EVL-NEXT: Live-in vp<[[VTC:%[0-9]+]]> = vector-trip-count
+; IF-EVL-NEXT: Live-in ir<%N> = original trip-count
+; IF-EVL-EMPTY:
+; IF-EVL:      vector.ph:
+; IF-EVL-NEXT: Successor(s): vector loop
+; IF-EVL-EMPTY:
+; IF-EVL-NEXT: <x1> vector loop: {
+; IF-EVL-NEXT:  vector.body:
+; IF-EVL-NEXT:    EMIT vp<[[IV:%[0-9]+]]> = CANONICAL-INDUCTION
+; IF-EVL-NEXT:    EXPLICIT-VECTOR-LENGTH-BASED-IV-PHI vp<[[EVL_PHI:%[0-9]+]]> = phi ir<0>, vp<[[IV_NEXT:%[0-9]+]]>
+; IF-EVL-NEXT:    EMIT vp<[[EVL:%.+]]> = EXPLICIT-VECTOR-LENGTH vp<[[EVL_PHI]]>, ir<%N>
+; IF-EVL-NEXT:    vp<[[ST:%[0-9]+]]> = SCALAR-STEPS vp<[[EVL_PHI]]>, ir<1>
+; IF-EVL-NEXT:    CLONE ir<[[GEP1:%.+]]> = getelementptr inbounds ir<%b>, vp<[[ST]]>
+; IF-EVL-NEXT:    vp<[[PTR1:%[0-9]+]]> = vector-pointer ir<[[GEP1]]>
+; IF-EVL-NEXT:    WIDEN ir<[[LD1:%.+]]> = load vp<[[PTR1]]>, ir<true>
+; IF-EVL-NEXT:    CLONE ir<[[GEP2:%.+]]> = getelementptr inbounds ir<%c>, vp<[[ST]]>
+; IF-EVL-NEXT:    vp<[[PTR2:%[0-9]+]]> = vector-pointer ir<[[GEP2]]>
+; IF-EVL-NEXT:    WIDEN ir<[[LD2:%.+]]> = load vp<[[PTR2]]>, ir<true>
+; IF-EVL-NEXT:    WIDEN ir<[[ADD:%.+]]> = add nsw ir<[[LD2]]>, ir<[[LD1]]>
+; IF-EVL-NEXT:    CLONE ir<[[GEP3:%.+]]> = getelementptr inbounds ir<%a>, vp<[[ST]]>
+; IF-EVL-NEXT:    vp<[[PTR3:%[0-9]+]]> = vector-pointer ir<[[GEP3]]>
+; IF-EVL-NEXT:    WIDEN store vp<[[PTR3]]>, ir<[[ADD]]>, ir<true>
+; IF-EVL-NEXT:    SCALAR-CAST vp<[[CAST:%[0-9]+]]> = zext vp<[[EVL]]> to i64
+; IF-EVL-NEXT:    EMIT vp<[[IV_NEXT]]> = add vp<[[CAST]]>, vp<[[EVL_PHI]]>
+; IF-EVL-NEXT:    EMIT vp<[[IV_NEXT_EXIT:%[0-9]+]]> = add vp<[[IV]]>, vp<[[VFUF]]>
+; IF-EVL-NEXT:    EMIT branch-on-count  vp<[[IV_NEXT_EXIT]]>, vp<[[VTC]]>
+; IF-EVL-NEXT:  No successors
+; IF-EVL-NEXT: }
+
+; NO-VP: VPlan 'Initial VPlan for VF={vscale x 1,vscale x 2,vscale x 4},UF>=1' {
+; NO-VP-NEXT: Live-in vp<[[VFUF:%[0-9]+]]> = VF * UF
+; NO-VP-NEXT: Live-in vp<[[VTC:%[0-9]+]]> = vector-trip-count
+; NO-VP-NEXT: Live-in ir<%N> = original trip-count
+; NO-VP-EMPTY:
+; NO-VP:      vector.ph:
+; NO-VP-NEXT: Successor(s): vector loop
+; NO-VP-EMPTY:
+; NO-VP-NEXT: <x1> vector loop: {
+; NO-VP-NEXT:  vector.body:
+; NO-VP-NEXT:    EMIT vp<[[IV:%[0-9]+]]> = CANONICAL-INDUCTION
+; NO-VP-NEXT:    vp<[[ST:%[0-9]+]]>    = SCALAR-STEPS vp<[[IV]]>, ir<1>
+; NO-VP-NEXT:    CLONE ir<[[GEP1:%.+]]> = getelementptr inbounds ir<%b>, vp<[[ST]]>
+; NO-VP-NEXT:    vp<[[PTR1:%[0-9]+]]> = vector-pointer ir<[[GEP1]]>
+; NO-VP-NEXT:    WIDEN ir<[[LD1:%.+]]> = load vp<[[PTR1]]>
+; NO-VP-NEXT:    CLONE ir<[[GEP2:%.+]]> = getelementptr inbounds ir<%c>, vp<[[ST]]>
+; NO-VP-NEXT:    vp<[[PTR2:%[0-9]+]]> = vector-pointer ir<[[GEP2]]>
+; NO-VP-NEXT:    WIDEN ir<[[LD2:%.+]]> = load vp<[[PTR2]]>
+; NO-VP-NEXT:    WIDEN ir<[[ADD:%.+]]> = add nsw ir<[[LD2]]>, ir<[[LD1]]>
+; NO-VP-NEXT:    CLONE ir<[[GEP3:%.+]]> = getelementptr inbounds ir<%a>, vp<[[ST]]>
+; NO-VP-NEXT:    vp<[[PTR3:%[0-9]+]]> = vector-pointer ir<[[GEP3]]>
+; NO-VP-NEXT:    WIDEN store vp<[[PTR3]]>, ir<[[ADD]]>
+; NO-VP-NEXT:    EMIT vp<[[IV_NEXT:%[0-9]+]]> = add nuw vp<[[IV]]>, vp<[[VFUF]]>
+; NO-VP-NEXT:    EMIT branch-on-count  vp<[[IV_NEXT]]>, vp<[[VTC]]>
+; NO-VP-NEXT:  No successors
+; NO-VP-NEXT: }
+
+entry:
+  br label %for.body
+
+for.body:
+  %iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]
+  %arrayidx = getelementptr inbounds i32, ptr %b, i64 %iv
+  %0 = load i32, ptr %arrayidx, align 4
+  %arrayidx2 = getelementptr inbounds i32, ptr %c, i64 %iv
+  %1 = load i32, ptr %arrayidx2, align 4
+  %add = add nsw i32 %1, %0
+  %arrayidx4 = getelementptr inbounds i32, ptr %a, i64 %iv
+  store i32 %add, ptr %arrayidx4, align 4
+  %iv.next = add nuw nsw i64 %iv, 1
+  %exitcond.not = icmp eq i64 %iv.next, %N
+  br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup:
+  ret void
+}
+
+define void @safe_dep(ptr %p) {
+; CHECK: VPlan 'Initial VPlan for VF={vscale x 1,vscale x 2},UF>=1' {
+; CHECK-NEXT: Live-in vp<[[VFUF:%[0-9]+]]> = VF * UF
+; CHECK-NEXT: Live-in vp<[[VTC:%[0-9]+]]> = vector-trip-count
+; CHECK-NEXT: Live-in ir<512> = original trip-count
+; CHECK-EMPTY:
+; CHECK:      vector.ph:
+; CHECK-NEXT: Successor(s): vector loop
+; CHECK-EMPTY:
+; CHECK-NEXT: <x1> vector loop: {
+; CHECK-NEXT:  vector.body:
+; CHECK-NEXT:    EMIT vp<[[IV:%[0-9]+]]> = CANONICAL-INDUCTION
+; CHECK-NEXT:    vp<[[ST:%[0-9]+]]>    = SCALAR-STEPS vp<[[IV]]>, ir<1>
+; CHECK-NEXT:    CLONE ir<[[GEP1:%.+]]> = getelementptr ir<%p>, vp<[[ST]]>
+; CHECK-NEXT:    vp<[[PTR1:%[0-9]+]]> = vector-pointer ir<[[GEP1]]>
+; CHECK-NEXT:    WIDEN ir<[[V:%.+]]> = load vp<[[PTR1]]>
+; CHECK-NEXT:    CLONE ir<[[OFFSET:.+]]> = add vp<[[ST]]>, ir<100>
+; CHECK-NEXT:    CLONE ir<[[GEP2:%.+]]> = getelementptr ir<%p>, ir<[[OFFSET]]>
+; CHECK-NEXT:    vp<[[PTR2:%[0-9]+]]> = vector-pointer ir<[[GEP2]]>
+; CHECK-NEXT:    WIDEN store vp<[[PTR2]]>, ir<[[V]]>
+; CHECK-NEXT:    EMIT vp<[[IV_NEXT:%[0-9]+]]> = add nuw vp<[[IV]]>, vp<[[VFUF]]>
+; CHECK-NEXT:    EMIT branch-on-count  vp<[[IV_NEXT]]>, vp<[[VTC]]>
+; CHECK-NEXT:  No successors
+; CHECK-NEXT: }
+
+entry:
+  br label %loop
+
+loop:
+  %iv = phi i64 [0, %entry], [%iv.next, %loop]
+  %a1 = getelementptr i64, ptr %p, i64 %iv
+  %v = load i64, ptr %a1, align 32
+  %offset = add i64 %iv, 100
+  %a2 = getelementptr i64, ptr %p, i64 %offset
+  store i64 %v, ptr %a2, align 32
+  %iv.next = add i64 %iv, 1
+  %cmp = icmp ne i64 %iv, 511
+  br i1 %cmp, label %loop, label %exit
+
+exit:
+  ret void
+}
+
diff --git a/llvm/test/Transforms/LoopVectorize/X86/vectorize-force-tail-with-evl.ll b/llvm/test/Transforms/LoopVectorize/X86/vectorize-force-tail-with-evl.ll
new file mode 100644
index 000000000000..1cf71360adf7
--- /dev/null
+++ b/llvm/test/Transforms/LoopVectorize/X86/vectorize-force-tail-with-evl.ll
@@ -0,0 +1,191 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
+; RUN: opt -passes=loop-vectorize \
+; RUN: -force-tail-folding-style=data-with-evl \
+; RUN: -prefer-predicate-over-epilogue=predicate-dont-vectorize \
+; RUN: -mtriple=x86_64 -mattr=+avx512f -S < %s 2>&1 | FileCheck --check-prefix=IF-EVL %s
+
+; RUN: opt -passes=loop-vectorize \
+; RUN: -force-tail-folding-style=none \
+; RUN: -prefer-predicate-over-epilogue=predicate-else-scalar-epilogue \
+; RUN: -mtriple=x86_64 -mattr=+avx512f -S < %s 2>&1 | FileCheck --check-prefix=NO-VP %s
+
+define void @foo(ptr noalias %a, ptr noalias %b, ptr noalias %c, i64 %N) {
+; IF-EVL-LABEL: @foo(
+; IF-EVL-NEXT:  entry:
+; IF-EVL-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; IF-EVL:       vector.ph:
+; IF-EVL-NEXT:    [[N_RND_UP:%.*]] = add i64 [[N:%.*]], 15
+; IF-EVL-NEXT:    [[N_MOD_VF:%.*]] = urem i64 [[N_RND_UP]], 16
+; IF-EVL-NEXT:    [[N_VEC:%.*]] = sub i64 [[N_RND_UP]], [[N_MOD_VF]]
+; IF-EVL-NEXT:    [[TRIP_COUNT_MINUS_1:%.*]] = sub i64 [[N]], 1
+; IF-EVL-NEXT:    [[BROADCAST_SPLATINSERT1:%.*]] = insertelement <16 x i64> poison, i64 [[TRIP_COUNT_MINUS_1]], i64 0
+; IF-EVL-NEXT:    [[BROADCAST_SPLAT2:%.*]] = shufflevector <16 x i64> [[BROADCAST_SPLATINSERT1]], <16 x i64> poison, <16 x i32> zeroinitializer
+; IF-EVL-NEXT:    br label [[VECTOR_BODY:%.*]]
+; IF-EVL:       vector.body:
+; IF-EVL-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; IF-EVL-NEXT:    [[TMP0:%.*]] = add i64 [[INDEX]], 0
+; IF-EVL-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <16 x i64> poison, i64 [[INDEX]], i64 0
+; IF-EVL-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <16 x i64> [[BROADCAST_SPLATINSERT]], <16 x i64> poison, <16 x i32> zeroinitializer
+; IF-EVL-NEXT:    [[VEC_IV:%.*]] = add <16 x i64> [[BROADCAST_SPLAT]], <i64 0, i64 1, i64 2, i64 3, i64 4, i64 5, i64 6, i64 7, i64 8, i64 9, i64 10, i64 11, i64 12, i64 13, i64 14, i64 15>
+; IF-EVL-NEXT:    [[TMP1:%.*]] = icmp ule <16 x i64> [[VEC_IV]], [[BROADCAST_SPLAT2]]
+; IF-EVL-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 [[TMP0]]
+; IF-EVL-NEXT:    [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i32 0
+; IF-EVL-NEXT:    [[WIDE_MASKED_LOAD:%.*]] = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr [[TMP3]], i32 4, <16 x i1> [[TMP1]], <16 x i32> poison)
+; IF-EVL-NEXT:    [[TMP4:%.*]] = getelementptr inbounds i32, ptr [[C:%.*]], i64 [[TMP0]]
+; IF-EVL-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[TMP4]], i32 0
+; IF-EVL-NEXT:    [[WIDE_MASKED_LOAD3:%.*]] = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr [[TMP5]], i32 4, <16 x i1> [[TMP1]], <16 x i32> poison)
+; IF-EVL-NEXT:    [[TMP6:%.*]] = add nsw <16 x i32> [[WIDE_MASKED_LOAD3]], [[WIDE_MASKED_LOAD]]
+; IF-EVL-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP0]]
+; IF-EVL-NEXT:    [[TMP8:%.*]] = getelementptr inbounds i32, ptr [[TMP7]], i32 0
+; IF-EVL-NEXT:    call void @llvm.masked.store.v16i32.p0(<16 x i32> [[TMP6]], ptr [[TMP8]], i32 4, <16 x i1> [[TMP1]])
+; IF-EVL-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], 16
+; IF-EVL-NEXT:    [[TMP9:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
+; IF-EVL-NEXT:    br i1 [[TMP9]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
+; IF-EVL:       middle.block:
+; IF-EVL-NEXT:    br i1 true, label [[FOR_COND_CLEANUP:%.*]], label [[SCALAR_PH]]
+; IF-EVL:       scalar.ph:
+; IF-EVL-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; IF-EVL-NEXT:    br label [[FOR_BODY:%.*]]
+; IF-EVL:       for.body:
+; IF-EVL-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
+; IF-EVL-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[IV]]
+; IF-EVL-NEXT:    [[TMP10:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
+; IF-EVL-NEXT:    [[ARRAYIDX2:%.*]] = getelementptr inbounds i32, ptr [[C]], i64 [[IV]]
+; IF-EVL-NEXT:    [[TMP11:%.*]] = load i32, ptr [[ARRAYIDX2]], align 4
+; IF-EVL-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP10]]
+; IF-EVL-NEXT:    [[ARRAYIDX4:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[IV]]
+; IF-EVL-NEXT:    store i32 [[ADD]], ptr [[ARRAYIDX4]], align 4
+; IF-EVL-NEXT:    [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
+; IF-EVL-NEXT:    [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
+; IF-EVL-NEXT:    br i1 [[EXITCOND_NOT]], label [[FOR_COND_CLEANUP]], label [[FOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
+; IF-EVL:       for.cond.cleanup:
+; IF-EVL-NEXT:    ret void
+;
+; NO-VP-LABEL: @foo(
+; NO-VP-NEXT:  iter.check:
+; NO-VP-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[N:%.*]], 8
+; NO-VP-NEXT:    br i1 [[MIN_ITERS_CHECK]], label [[VEC_EPILOG_SCALAR_PH:%.*]], label [[VECTOR_MAIN_LOOP_ITER_CHECK:%.*]]
+; NO-VP:       vector.main.loop.iter.check:
+; NO-VP-NEXT:    [[MIN_ITERS_CHECK1:%.*]] = icmp ult i64 [[N]], 64
+; NO-VP-NEXT:    br i1 [[MIN_ITERS_CHECK1]], label [[VEC_EPILOG_PH:%.*]], label [[VECTOR_PH:%.*]]
+; NO-VP:       vector.ph:
+; NO-VP-NEXT:    [[N_MOD_VF:%.*]] = urem i64 [[N]], 64
+; NO-VP-NEXT:    [[N_VEC:%.*]] = sub i64 [[N]], [[N_MOD_VF]]
+; NO-VP-NEXT:    br label [[VECTOR_BODY:%.*]]
+; NO-VP:       vector.body:
+; NO-VP-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; NO-VP-NEXT:    [[TMP0:%.*]] = add i64 [[INDEX]], 0
+; NO-VP-NEXT:    [[TMP1:%.*]] = add i64 [[INDEX]], 16
+; NO-VP-NEXT:    [[TMP2:%.*]] = add i64 [[INDEX]], 32
+; NO-VP-NEXT:    [[TMP3:%.*]] = add i64 [[INDEX]], 48
+; NO-VP-NEXT:    [[TMP4:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 [[TMP0]]
+; NO-VP-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[TMP1]]
+; NO-VP-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[TMP2]]
+; NO-VP-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[TMP3]]
+; NO-VP-NEXT:    [[TMP8:%.*]] = getelementptr inbounds i32, ptr [[TMP4]], i32 0
+; NO-VP-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i32, ptr [[TMP4]], i32 16
+; NO-VP-NEXT:    [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[TMP4]], i32 32
+; NO-VP-NEXT:    [[TMP11:%.*]] = getelementptr inbounds i32, ptr [[TMP4]], i32 48
+; NO-VP-NEXT:    [[WIDE_LOAD:%.*]] = load <16 x i32>, ptr [[TMP8]], align 4
+; NO-VP-NEXT:    [[WIDE_LOAD2:%.*]] = load <16 x i32>, ptr [[TMP9]], align 4
+; NO-VP-NEXT:    [[WIDE_LOAD3:%.*]] = load <16 x i32>, ptr [[TMP10]], align 4
+; NO-VP-NEXT:    [[WIDE_LOAD4:%.*]] = load <16 x i32>, ptr [[TMP11]], align 4
+; NO-VP-NEXT:    [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[C:%.*]], i64 [[TMP0]]
+; NO-VP-NEXT:    [[TMP13:%.*]] = getelementptr inbounds i32, ptr [[C]], i64 [[TMP1]]
+; NO-VP-NEXT:    [[TMP14:%.*]] = getelementptr inbounds i32, ptr [[C]], i64 [[TMP2]]
+; NO-VP-NEXT:    [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[C]], i64 [[TMP3]]
+; NO-VP-NEXT:    [[TMP16:%.*]] = getelementptr inbounds i32, ptr [[TMP12]], i32 0
+; NO-VP-NEXT:    [[TMP17:%.*]] = getelementptr inbounds i32, ptr [[TMP12]], i32 16
+; NO-VP-NEXT:    [[TMP18:%.*]] = getelementptr inbounds i32, ptr [[TMP12]], i32 32
+; NO-VP-NEXT:    [[TMP19:%.*]] = getelementptr inbounds i32, ptr [[TMP12]], i32 48
+; NO-VP-NEXT:    [[WIDE_LOAD5:%.*]] = load <16 x i32>, ptr [[TMP16]], align 4
+; NO-VP-NEXT:    [[WIDE_LOAD6:%.*]] = load <16 x i32>, ptr [[TMP17]], align 4
+; NO-VP-NEXT:    [[WIDE_LOAD7:%.*]] = load <16 x i32>, ptr [[TMP18]], align 4
+; NO-VP-NEXT:    [[WIDE_LOAD8:%.*]] = load <16 x i32>, ptr [[TMP19]], align 4
+; NO-VP-NEXT:    [[TMP20:%.*]] = add nsw <16 x i32> [[WIDE_LOAD5]], [[WIDE_LOAD]]
+; NO-VP-NEXT:    [[TMP21:%.*]] = add nsw <16 x i32> [[WIDE_LOAD6]], [[WIDE_LOAD2]]
+; NO-VP-NEXT:    [[TMP22:%.*]] = add nsw <16 x i32> [[WIDE_LOAD7]], [[WIDE_LOAD3]]
+; NO-VP-NEXT:    [[TMP23:%.*]] = add nsw <16 x i32> [[WIDE_LOAD8]], [[WIDE_LOAD4]]
+; NO-VP-NEXT:    [[TMP24:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP0]]
+; NO-VP-NEXT:    [[TMP25:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP1]]
+; NO-VP-NEXT:    [[TMP26:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP2]]
+; NO-VP-NEXT:    [[TMP27:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP3]]
+; NO-VP-NEXT:    [[TMP28:%.*]] = getelementptr inbounds i32, ptr [[TMP24]], i32 0
+; NO-VP-NEXT:    [[TMP29:%.*]] = getelementptr inbounds i32, ptr [[TMP24]], i32 16
+; NO-VP-NEXT:    [[TMP30:%.*]] = getelementptr inbounds i32, ptr [[TMP24]], i32 32
+; NO-VP-NEXT:    [[TMP31:%.*]] = getelementptr inbounds i32, ptr [[TMP24]], i32 48
+; NO-VP-NEXT:    store <16 x i32> [[TMP20]], ptr [[TMP28]], align 4
+; NO-VP-NEXT:    store <16 x i32> [[TMP21]], ptr [[TMP29]], align 4
+; NO-VP-NEXT:    store <16 x i32> [[TMP22]], ptr [[TMP30]], align 4
+; NO-VP-NEXT:    store <16 x i32> [[TMP23]], ptr [[TMP31]], align 4
+; NO-VP-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 64
+; NO-VP-NEXT:    [[TMP32:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
+; NO-VP-NEXT:    br i1 [[TMP32]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
+; NO-VP:       middle.block:
+; NO-VP-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[N]], [[N_VEC]]
+; NO-VP-NEXT:    br i1 [[CMP_N]], label [[FOR_COND_CLEANUP:%.*]], label [[VEC_EPILOG_ITER_CHECK:%.*]]
+; NO-VP:       vec.epilog.iter.check:
+; NO-VP-NEXT:    [[N_VEC_REMAINING:%.*]] = sub i64 [[N]], [[N_VEC]]
+; NO-VP-NEXT:    [[MIN_EPILOG_ITERS_CHECK:%.*]] = icmp ult i64 [[N_VEC_REMAINING]], 8
+; NO-VP-NEXT:    br i1 [[MIN_EPILOG_ITERS_CHECK]], label [[VEC_EPILOG_SCALAR_PH]], label [[VEC_EPILOG_PH]]
+; NO-VP:       vec.epilog.ph:
+; NO-VP-NEXT:    [[VEC_EPILOG_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[VECTOR_MAIN_LOOP_ITER_CHECK]] ]
+; NO-VP-NEXT:    [[N_MOD_VF9:%.*]] = urem i64 [[N]], 8
+; NO-VP-NEXT:    [[N_VEC10:%.*]] = sub i64 [[N]], [[N_MOD_VF9]]
+; NO-VP-NEXT:    br label [[VEC_EPILOG_VECTOR_BODY:%.*]]
+; NO-VP:       vec.epilog.vector.body:
+; NO-VP-NEXT:    [[INDEX12:%.*]] = phi i64 [ [[VEC_EPILOG_RESUME_VAL]], [[VEC_EPILOG_PH]] ], [ [[INDEX_NEXT15:%.*]], [[VEC_EPILOG_VECTOR_BODY]] ]
+; NO-VP-NEXT:    [[TMP33:%.*]] = add i64 [[INDEX12]], 0
+; NO-VP-NEXT:    [[TMP34:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[TMP33]]
+; NO-VP-NEXT:    [[TMP35:%.*]] = getelementptr inbounds i32, ptr [[TMP34]], i32 0
+; NO-VP-NEXT:    [[WIDE_LOAD13:%.*]] = load <8 x i32>, ptr [[TMP35]], align 4
+; NO-VP-NEXT:    [[TMP36:%.*]] = getelementptr inbounds i32, ptr [[C]], i64 [[TMP33]]
+; NO-VP-NEXT:    [[TMP37:%.*]] = getelementptr inbounds i32, ptr [[TMP36]], i32 0
+; NO-VP-NEXT:    [[WIDE_LOAD14:%.*]] = load <8 x i32>, ptr [[TMP37]], align 4
+; NO-VP-NEXT:    [[TMP38:%.*]] = add nsw <8 x i32> [[WIDE_LOAD14]], [[WIDE_LOAD13]]
+; NO-VP-NEXT:    [[TMP39:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP33]]
+; NO-VP-NEXT:    [[TMP40:%.*]] = getelementptr inbounds i32, ptr [[TMP39]], i32 0
+; NO-VP-NEXT:    store <8 x i32> [[TMP38]], ptr [[TMP40]], align 4
+; NO-VP-NEXT:    [[INDEX_NEXT15]] = add nuw i64 [[INDEX12]], 8
+; NO-VP-NEXT:    [[TMP41:%.*]] = icmp eq i64 [[INDEX_NEXT15]], [[N_VEC10]]
+; NO-VP-NEXT:    br i1 [[TMP41]], label [[VEC_EPILOG_MIDDLE_BLOCK:%.*]], label [[VEC_EPILOG_VECTOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
+; NO-VP:       vec.epilog.middle.block:
+; NO-VP-NEXT:    [[CMP_N11:%.*]] = icmp eq i64 [[N]], [[N_VEC10]]
+; NO-VP-NEXT:    br i1 [[CMP_N11]], label [[FOR_COND_CLEANUP]], label [[VEC_EPILOG_SCALAR_PH]]
+; NO-VP:       vec.epilog.scalar.ph:
+; NO-VP-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC10]], [[VEC_EPILOG_MIDDLE_BLOCK]] ], [ [[N_VEC]], [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[ITER_CHECK:%.*]] ]
+; NO-VP-NEXT:    br label [[FOR_BODY:%.*]]
+; NO-VP:       for.body:
+; NO-VP-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[VEC_EPILOG_SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
+; NO-VP-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[IV]]
+; NO-VP-NEXT:    [[TMP42:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
+; NO-VP-NEXT:    [[ARRAYIDX2:%.*]] = getelementptr inbounds i32, ptr [[C]], i64 [[IV]]
+; NO-VP-NEXT:    [[TMP43:%.*]] = load i32, ptr [[ARRAYIDX2]], align 4
+; NO-VP-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP43]], [[TMP42]]
+; NO-VP-NEXT:    [[ARRAYIDX4:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[IV]]
+; NO-VP-NEXT:    store i32 [[ADD]], ptr [[ARRAYIDX4]], align 4
+; NO-VP-NEXT:    [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
+; NO-VP-NEXT:    [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
+; NO-VP-NEXT:    br i1 [[EXITCOND_NOT]], label [[FOR_COND_CLEANUP]], label [[FOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
+; NO-VP:       for.cond.cleanup:
+; NO-VP-NEXT:    ret void
+;
+entry:
+  br label %for.body
+
+for.body:
+  %iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]
+  %arrayidx = getelementptr inbounds i32, ptr %b, i64 %iv
+  %0 = load i32, ptr %arrayidx, align 4
+  %arrayidx2 = getelementptr inbounds i32, ptr %c, i64 %iv
+  %1 = load i32, ptr %arrayidx2, align 4
+  %add = add nsw i32 %1, %0
+  %arrayidx4 = getelementptr inbounds i32, ptr %a, i64 %iv
+  store i32 %add, ptr %arrayidx4, align 4
+  %iv.next = add nuw nsw i64 %iv, 1
+  %exitcond.not = icmp eq i64 %iv.next, %N
+  br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup:
+  ret void
+}
diff --git a/llvm/test/Transforms/LoopVectorize/X86/vplan-vp-intrinsics.ll b/llvm/test/Transforms/LoopVectorize/X86/vplan-vp-intrinsics.ll
new file mode 100644
index 000000000000..9b49d44141db
--- /dev/null
+++ b/llvm/test/Transforms/LoopVectorize/X86/vplan-vp-intrinsics.ll
@@ -0,0 +1,89 @@
+; REQUIRES: asserts
+
+; RUN: opt -passes=loop-vectorize -debug-only=loop-vectorize -force-vector-width=4 \
+; RUN: -force-tail-folding-style=data-with-evl \
+; RUN: -prefer-predicate-over-epilogue=predicate-dont-vectorize \
+; RUN: -mtriple=x86_64 -mattr=+avx512f -disable-output < %s 2>&1 | FileCheck --check-prefix=IF-EVL %s
+
+; RUN: opt -passes=loop-vectorize -debug-only=loop-vectorize -force-vector-width=4 \
+; RUN: -force-tail-folding-style=none \
+; RUN: -prefer-predicate-over-epilogue=predicate-else-scalar-epilogue \
+; RUN: -mtriple=x86_64 -mattr=+avx512f -disable-output < %s 2>&1 | FileCheck --check-prefix=NO-VP %s
+
+define void @foo(ptr noalias %a, ptr noalias %b, ptr noalias %c, i64 %N) {
+; IF-EVL: VPlan 'Initial VPlan for VF={4},UF>=1' {
+; IF-EVL-NEXT: Live-in vp<[[VFUF:%[0-9]+]]> = VF * UF
+; IF-EVL-NEXT: Live-in vp<[[VTC:%[0-9]+]]> = vector-trip-count
+; IF-EVL-NEXT: Live-in vp<[[BETC:%[0-9]+]]> = backedge-taken count
+; IF-EVL-NEXT: Live-in ir<%N> = original trip-count
+; IF-EVL-EMPTY:
+; IF-EVL:      vector.ph:
+; IF-EVL-NEXT: Successor(s): vector loop
+; IF-EVL-EMPTY:
+; IF-EVL-NEXT: <x1> vector loop: {
+; IF-EVL-NEXT:  vector.body:
+; IF-EVL-NEXT:    EMIT vp<[[IV:%[0-9]+]]> = CANONICAL-INDUCTION
+; IF-EVL-NEXT:    vp<[[ST:%[0-9]+]]>    = SCALAR-STEPS vp<[[IV]]>, ir<1>
+; IF-EVL-NEXT:    EMIT vp<[[VIV:%[0-9]+]]> = WIDEN-CANONICAL-INDUCTION vp<[[IV]]>
+; IF-EVL-NEXT:    EMIT vp<[[MASK:%[0-9]+]]> = icmp ule vp<[[VIV]]>, vp<[[BETC]]>
+; IF-EVL-NEXT:    CLONE ir<[[GEP1:%.+]]> = getelementptr inbounds ir<%b>, vp<[[ST]]>
+; IF-EVL-NEXT:    vp<[[PTR1:%[0-9]+]]> = vector-pointer ir<[[GEP1]]>
+; IF-EVL-NEXT:    WIDEN ir<[[LD1:%.+]]> = load vp<[[PTR1]]>, vp<[[MASK]]>
+; IF-EVL-NEXT:    CLONE ir<[[GEP2:%.+]]> = getelementptr inbounds ir<%c>, vp<[[ST]]>
+; IF-EVL-NEXT:    vp<[[PTR2:%[0-9]+]]> = vector-pointer ir<[[GEP2]]>
+; IF-EVL-NEXT:    WIDEN ir<[[LD2:%.+]]> = load vp<[[PTR2]]>, vp<[[MASK]]>
+; IF-EVL-NEXT:    WIDEN ir<[[ADD:%.+]]> = add nsw ir<[[LD2]]>, ir<[[LD1]]>
+; IF-EVL-NEXT:    CLONE ir<[[GEP3:%.+]]> = getelementptr inbounds ir<%a>, vp<[[ST]]>
+; IF-EVL-NEXT:    vp<[[PTR3:%[0-9]+]]> = vector-pointer ir<[[GEP3]]>
+; IF-EVL-NEXT:    WIDEN store vp<[[PTR3]]>, ir<[[ADD]]>, vp<[[MASK]]>
+; IF-EVL-NEXT:    EMIT vp<[[IV_NEXT:%[0-9]+]]> = add vp<[[IV]]>, vp<[[VFUF]]>
+; IF-EVL-NEXT:    EMIT branch-on-count  vp<[[IV_NEXT]]>, vp<[[VTC]]>
+; IF-EVL-NEXT:  No successors
+; IF-EVL-NEXT: }
+
+; NO-VP: VPlan 'Initial VPlan for VF={4},UF>=1' {
+; NO-VP-NEXT: Live-in vp<[[VFUF:%[0-9]+]]> = VF * UF
+; NO-VP-NEXT: Live-in vp<[[VTC:%[0-9]+]]> = vector-trip-count
+; NO-VP-NEXT: Live-in ir<%N> = original trip-count
+; NO-VP-EMPTY:
+; NO-VP:      vector.ph:
+; NO-VP-NEXT: Successor(s): vector loop
+; NO-VP-EMPTY:
+; NO-VP-NEXT: <x1> vector loop: {
+; NO-VP-NEXT:  vector.body:
+; NO-VP-NEXT:    EMIT vp<[[IV:%[0-9]+]]> = CANONICAL-INDUCTION
+; NO-VP-NEXT:    vp<[[ST:%[0-9]+]]>    = SCALAR-STEPS vp<[[IV]]>, ir<1>
+; NO-VP-NEXT:    CLONE ir<[[GEP1:%.+]]> = getelementptr inbounds ir<%b>, vp<[[ST]]>
+; NO-VP-NEXT:    vp<[[PTR1:%[0-9]+]]> = vector-pointer ir<[[GEP1]]>
+; NO-VP-NEXT:    WIDEN ir<[[LD1:%.+]]> = load vp<[[PTR1]]>
+; NO-VP-NEXT:    CLONE ir<[[GEP2:%.+]]> = getelementptr inbounds ir<%c>, vp<[[ST]]>
+; NO-VP-NEXT:    vp<[[PTR2:%[0-9]+]]> = vector-pointer ir<[[GEP2]]>
+; NO-VP-NEXT:    WIDEN ir<[[LD2:%.+]]> = load vp<[[PTR2]]>
+; NO-VP-NEXT:    WIDEN ir<[[ADD:%.+]]> = add nsw ir<[[LD2]]>, ir<[[LD1]]>
+; NO-VP-NEXT:    CLONE ir<[[GEP3:%.+]]> = getelementptr inbounds ir<%a>, vp<[[ST]]>
+; NO-VP-NEXT:    vp<[[PTR3:%[0-9]+]]> = vector-pointer ir<[[GEP3]]>
+; NO-VP-NEXT:    WIDEN store vp<[[PTR3]]>, ir<[[ADD]]>
+; NO-VP-NEXT:    EMIT vp<[[IV_NEXT:%[0-9]+]]> = add nuw vp<[[IV]]>, vp<[[VFUF]]>
+; NO-VP-NEXT:    EMIT branch-on-count  vp<[[IV_NEXT]]>, vp<[[VTC]]>
+; NO-VP-NEXT:  No successors
+; NO-VP-NEXT: }
+
+entry:
+  br label %for.body
+
+for.body:
+  %iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]
+  %arrayidx = getelementptr inbounds i32, ptr %b, i64 %iv
+  %0 = load i32, ptr %arrayidx, align 4
+  %arrayidx2 = getelementptr inbounds i32, ptr %c, i64 %iv
+  %1 = load i32, ptr %arrayidx2, align 4
+  %add = add nsw i32 %1, %0
+  %arrayidx4 = getelementptr inbounds i32, ptr %a, i64 %iv
+  store i32 %add, ptr %arrayidx4, align 4
+  %iv.next = add nuw nsw i64 %iv, 1
+  %exitcond.not = icmp eq i64 %iv.next, %N
+  br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup:
+  ret void
+}
diff --git a/llvm/test/Transforms/LoopVectorize/epilog-vectorization-any-of-reductions.ll b/llvm/test/Transforms/LoopVectorize/epilog-vectorization-any-of-reductions.ll
index 0b872709ec6c..c721da7597b1 100644
--- a/llvm/test/Transforms/LoopVectorize/epilog-vectorization-any-of-reductions.ll
+++ b/llvm/test/Transforms/LoopVectorize/epilog-vectorization-any-of-reductions.ll
@@ -19,20 +19,20 @@ define i32 @any_of_reduction_epilog(ptr %src, i64 %N) {
 ; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
 ; CHECK:       vector.body:
 ; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP8:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <4 x i1> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP5:%.*]], [[VECTOR_BODY]] ]
 ; CHECK-NEXT:    [[TMP1:%.*]] = add i64 [[INDEX]], 0
 ; CHECK-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[SRC]], i64 [[TMP1]]
 ; CHECK-NEXT:    [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[TMP2]], i32 0
 ; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <4 x i8>, ptr [[TMP3]], align 1
 ; CHECK-NEXT:    [[TMP4:%.*]] = icmp eq <4 x i8> [[WIDE_LOAD]], zeroinitializer
-; CHECK-NEXT:    [[TMP8]] = select <4 x i1> [[TMP4]], <4 x i32> <i32 1, i32 1, i32 1, i32 1>, <4 x i32> [[VEC_PHI]]
+; CHECK-NEXT:    [[TMP5]] = or <4 x i1> [[VEC_PHI]], [[TMP4]]
 ; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
 ; CHECK-NEXT:    [[TMP6:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
 ; CHECK-NEXT:    br i1 [[TMP6]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
 ; CHECK:       middle.block:
-; CHECK-NEXT:    [[TMP5:%.*]] = icmp ne <4 x i32> [[TMP8]], zeroinitializer
 ; CHECK-NEXT:    [[TMP7:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[TMP5]])
-; CHECK-NEXT:    [[RDX_SELECT:%.*]] = select i1 [[TMP7]], i32 1, i32 0
+; CHECK-NEXT:    [[TMP8:%.*]] = freeze i1 [[TMP7]]
+; CHECK-NEXT:    [[RDX_SELECT:%.*]] = select i1 [[TMP8]], i32 1, i32 0
 ; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[TMP0]], [[N_VEC]]
 ; CHECK-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[VEC_EPILOG_ITER_CHECK:%.*]]
 ; CHECK:       vec.epilog.iter.check:
@@ -42,32 +42,33 @@ define i32 @any_of_reduction_epilog(ptr %src, i64 %N) {
 ; CHECK:       vec.epilog.ph:
 ; CHECK-NEXT:    [[BC_MERGE_RDX:%.*]] = phi i32 [ 0, [[VECTOR_MAIN_LOOP_ITER_CHECK]] ], [ [[RDX_SELECT]], [[VEC_EPILOG_ITER_CHECK]] ]
 ; CHECK-NEXT:    [[VEC_EPILOG_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[VECTOR_MAIN_LOOP_ITER_CHECK]] ]
+; CHECK-NEXT:    [[TMP9:%.*]] = icmp ne i32 [[BC_MERGE_RDX]], 0
 ; CHECK-NEXT:    [[N_MOD_VF2:%.*]] = urem i64 [[TMP0]], 4
 ; CHECK-NEXT:    [[N_VEC3:%.*]] = sub i64 [[TMP0]], [[N_MOD_VF2]]
-; CHECK-NEXT:    [[MINMAX_IDENT_SPLATINSERT:%.*]] = insertelement <4 x i32> poison, i32 [[BC_MERGE_RDX]], i64 0
-; CHECK-NEXT:    [[MINMAX_IDENT_SPLAT:%.*]] = shufflevector <4 x i32> [[MINMAX_IDENT_SPLATINSERT]], <4 x i32> poison, <4 x i32> zeroinitializer
+; CHECK-NEXT:    [[MINMAX_IDENT_SPLATINSERT:%.*]] = insertelement <4 x i1> poison, i1 [[TMP9]], i64 0
+; CHECK-NEXT:    [[MINMAX_IDENT_SPLAT:%.*]] = shufflevector <4 x i1> [[MINMAX_IDENT_SPLATINSERT]], <4 x i1> poison, <4 x i32> zeroinitializer
 ; CHECK-NEXT:    br label [[VEC_EPILOG_VECTOR_BODY:%.*]]
 ; CHECK:       vec.epilog.vector.body:
 ; CHECK-NEXT:    [[INDEX5:%.*]] = phi i64 [ [[VEC_EPILOG_RESUME_VAL]], [[VEC_EPILOG_PH]] ], [ [[INDEX_NEXT8:%.*]], [[VEC_EPILOG_VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_PHI6:%.*]] = phi <4 x i32> [ [[MINMAX_IDENT_SPLAT]], [[VEC_EPILOG_PH]] ], [ [[TMP17:%.*]], [[VEC_EPILOG_VECTOR_BODY]] ]
+; CHECK-NEXT:    [[VEC_PHI6:%.*]] = phi <4 x i1> [ [[MINMAX_IDENT_SPLAT]], [[VEC_EPILOG_PH]] ], [ [[TMP14:%.*]], [[VEC_EPILOG_VECTOR_BODY]] ]
 ; CHECK-NEXT:    [[TMP10:%.*]] = add i64 [[INDEX5]], 0
 ; CHECK-NEXT:    [[TMP11:%.*]] = getelementptr inbounds i8, ptr [[SRC]], i64 [[TMP10]]
 ; CHECK-NEXT:    [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[TMP11]], i32 0
 ; CHECK-NEXT:    [[WIDE_LOAD7:%.*]] = load <4 x i8>, ptr [[TMP12]], align 1
 ; CHECK-NEXT:    [[TMP13:%.*]] = icmp eq <4 x i8> [[WIDE_LOAD7]], zeroinitializer
-; CHECK-NEXT:    [[TMP17]] = select <4 x i1> [[TMP13]], <4 x i32> <i32 1, i32 1, i32 1, i32 1>, <4 x i32> [[VEC_PHI6]]
+; CHECK-NEXT:    [[TMP14]] = or <4 x i1> [[VEC_PHI6]], [[TMP13]]
 ; CHECK-NEXT:    [[INDEX_NEXT8]] = add nuw i64 [[INDEX5]], 4
 ; CHECK-NEXT:    [[TMP15:%.*]] = icmp eq i64 [[INDEX_NEXT8]], [[N_VEC3]]
 ; CHECK-NEXT:    br i1 [[TMP15]], label [[VEC_EPILOG_MIDDLE_BLOCK:%.*]], label [[VEC_EPILOG_VECTOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
 ; CHECK:       vec.epilog.middle.block:
-; CHECK-NEXT:    [[TMP14:%.*]] = icmp ne <4 x i32> [[TMP17]], zeroinitializer
 ; CHECK-NEXT:    [[TMP16:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[TMP14]])
-; CHECK-NEXT:    [[RDX_SELECT9:%.*]] = select i1 [[TMP16]], i32 1, i32 0
+; CHECK-NEXT:    [[TMP17:%.*]] = freeze i1 [[TMP16]]
+; CHECK-NEXT:    [[RDX_SELECT9:%.*]] = select i1 [[TMP17]], i32 1, i32 0
 ; CHECK-NEXT:    [[CMP_N4:%.*]] = icmp eq i64 [[TMP0]], [[N_VEC3]]
 ; CHECK-NEXT:    br i1 [[CMP_N4]], label [[EXIT]], label [[VEC_EPILOG_SCALAR_PH]]
 ; CHECK:       vec.epilog.scalar.ph:
 ; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC3]], [[VEC_EPILOG_MIDDLE_BLOCK]] ], [ [[N_VEC]], [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[ITER_CHECK:%.*]] ]
-; CHECK-NEXT:    [[BC_MERGE_RDX10:%.*]] = phi i32 [ 0, [[ITER_CHECK]] ], [ [[RDX_SELECT]], [[VEC_EPILOG_ITER_CHECK]] ], [ [[RDX_SELECT9]], [[VEC_EPILOG_MIDDLE_BLOCK]] ]
+; CHECK-NEXT:    [[BC_MERGE_RDX10:%.*]] = phi i32 [ 0, [[ITER_CHECK]] ], [ 0, [[VEC_EPILOG_ITER_CHECK]] ], [ [[RDX_SELECT9]], [[VEC_EPILOG_MIDDLE_BLOCK]] ]
 ; CHECK-NEXT:    br label [[LOOP:%.*]]
 ; CHECK:       loop:
 ; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[VEC_EPILOG_SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
@@ -101,6 +102,104 @@ exit:
   ret i32 %select
 }
 
+define i32 @any_of_reduction_epilog_arg_as_start_value(ptr %src, i64 %N, i32 %start) {
+; CHECK-LABEL: define i32 @any_of_reduction_epilog_arg_as_start_value(
+; CHECK-SAME: ptr [[SRC:%.*]], i64 [[N:%.*]], i32 [[START:%.*]]) {
+; CHECK-NEXT:  iter.check:
+; CHECK-NEXT:    [[TMP0:%.*]] = add i64 [[N]], 1
+; CHECK-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[TMP0]], 4
+; CHECK-NEXT:    br i1 [[MIN_ITERS_CHECK]], label [[VEC_EPILOG_SCALAR_PH:%.*]], label [[VECTOR_MAIN_LOOP_ITER_CHECK:%.*]]
+; CHECK:       vector.main.loop.iter.check:
+; CHECK-NEXT:    [[MIN_ITERS_CHECK1:%.*]] = icmp ult i64 [[TMP0]], 4
+; CHECK-NEXT:    br i1 [[MIN_ITERS_CHECK1]], label [[VEC_EPILOG_PH:%.*]], label [[VECTOR_PH:%.*]]
+; CHECK:       vector.ph:
+; CHECK-NEXT:    [[N_MOD_VF:%.*]] = urem i64 [[TMP0]], 4
+; CHECK-NEXT:    [[N_VEC:%.*]] = sub i64 [[TMP0]], [[N_MOD_VF]]
+; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
+; CHECK:       vector.body:
+; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <4 x i1> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP5:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT:    [[TMP1:%.*]] = add i64 [[INDEX]], 0
+; CHECK-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[SRC]], i64 [[TMP1]]
+; CHECK-NEXT:    [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[TMP2]], i32 0
+; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <4 x i8>, ptr [[TMP3]], align 1
+; CHECK-NEXT:    [[TMP4:%.*]] = icmp eq <4 x i8> [[WIDE_LOAD]], zeroinitializer
+; CHECK-NEXT:    [[TMP5]] = or <4 x i1> [[VEC_PHI]], [[TMP4]]
+; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; CHECK-NEXT:    [[TMP6:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
+; CHECK-NEXT:    br i1 [[TMP6]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP5:![0-9]+]]
+; CHECK:       middle.block:
+; CHECK-NEXT:    [[TMP7:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[TMP5]])
+; CHECK-NEXT:    [[TMP8:%.*]] = freeze i1 [[TMP7]]
+; CHECK-NEXT:    [[RDX_SELECT:%.*]] = select i1 [[TMP8]], i32 1, i32 [[START]]
+; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[TMP0]], [[N_VEC]]
+; CHECK-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[VEC_EPILOG_ITER_CHECK:%.*]]
+; CHECK:       vec.epilog.iter.check:
+; CHECK-NEXT:    [[N_VEC_REMAINING:%.*]] = sub i64 [[TMP0]], [[N_VEC]]
+; CHECK-NEXT:    [[MIN_EPILOG_ITERS_CHECK:%.*]] = icmp ult i64 [[N_VEC_REMAINING]], 4
+; CHECK-NEXT:    br i1 [[MIN_EPILOG_ITERS_CHECK]], label [[VEC_EPILOG_SCALAR_PH]], label [[VEC_EPILOG_PH]]
+; CHECK:       vec.epilog.ph:
+; CHECK-NEXT:    [[BC_MERGE_RDX:%.*]] = phi i32 [ [[START]], [[VECTOR_MAIN_LOOP_ITER_CHECK]] ], [ [[RDX_SELECT]], [[VEC_EPILOG_ITER_CHECK]] ]
+; CHECK-NEXT:    [[VEC_EPILOG_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[VECTOR_MAIN_LOOP_ITER_CHECK]] ]
+; CHECK-NEXT:    [[TMP9:%.*]] = icmp ne i32 [[BC_MERGE_RDX]], [[START]]
+; CHECK-NEXT:    [[N_MOD_VF2:%.*]] = urem i64 [[TMP0]], 4
+; CHECK-NEXT:    [[N_VEC3:%.*]] = sub i64 [[TMP0]], [[N_MOD_VF2]]
+; CHECK-NEXT:    [[MINMAX_IDENT_SPLATINSERT:%.*]] = insertelement <4 x i1> poison, i1 [[TMP9]], i64 0
+; CHECK-NEXT:    [[MINMAX_IDENT_SPLAT:%.*]] = shufflevector <4 x i1> [[MINMAX_IDENT_SPLATINSERT]], <4 x i1> poison, <4 x i32> zeroinitializer
+; CHECK-NEXT:    br label [[VEC_EPILOG_VECTOR_BODY:%.*]]
+; CHECK:       vec.epilog.vector.body:
+; CHECK-NEXT:    [[INDEX5:%.*]] = phi i64 [ [[VEC_EPILOG_RESUME_VAL]], [[VEC_EPILOG_PH]] ], [ [[INDEX_NEXT8:%.*]], [[VEC_EPILOG_VECTOR_BODY]] ]
+; CHECK-NEXT:    [[VEC_PHI6:%.*]] = phi <4 x i1> [ [[MINMAX_IDENT_SPLAT]], [[VEC_EPILOG_PH]] ], [ [[TMP14:%.*]], [[VEC_EPILOG_VECTOR_BODY]] ]
+; CHECK-NEXT:    [[TMP10:%.*]] = add i64 [[INDEX5]], 0
+; CHECK-NEXT:    [[TMP11:%.*]] = getelementptr inbounds i8, ptr [[SRC]], i64 [[TMP10]]
+; CHECK-NEXT:    [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[TMP11]], i32 0
+; CHECK-NEXT:    [[WIDE_LOAD7:%.*]] = load <4 x i8>, ptr [[TMP12]], align 1
+; CHECK-NEXT:    [[TMP13:%.*]] = icmp eq <4 x i8> [[WIDE_LOAD7]], zeroinitializer
+; CHECK-NEXT:    [[TMP14]] = or <4 x i1> [[VEC_PHI6]], [[TMP13]]
+; CHECK-NEXT:    [[INDEX_NEXT8]] = add nuw i64 [[INDEX5]], 4
+; CHECK-NEXT:    [[TMP15:%.*]] = icmp eq i64 [[INDEX_NEXT8]], [[N_VEC3]]
+; CHECK-NEXT:    br i1 [[TMP15]], label [[VEC_EPILOG_MIDDLE_BLOCK:%.*]], label [[VEC_EPILOG_VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
+; CHECK:       vec.epilog.middle.block:
+; CHECK-NEXT:    [[TMP16:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[TMP14]])
+; CHECK-NEXT:    [[TMP17:%.*]] = freeze i1 [[TMP16]]
+; CHECK-NEXT:    [[RDX_SELECT9:%.*]] = select i1 [[TMP17]], i32 1, i32 [[START]]
+; CHECK-NEXT:    [[CMP_N4:%.*]] = icmp eq i64 [[TMP0]], [[N_VEC3]]
+; CHECK-NEXT:    br i1 [[CMP_N4]], label [[EXIT]], label [[VEC_EPILOG_SCALAR_PH]]
+; CHECK:       vec.epilog.scalar.ph:
+; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC3]], [[VEC_EPILOG_MIDDLE_BLOCK]] ], [ [[N_VEC]], [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[ITER_CHECK:%.*]] ]
+; CHECK-NEXT:    [[BC_MERGE_RDX10:%.*]] = phi i32 [ [[START]], [[ITER_CHECK]] ], [ [[START]], [[VEC_EPILOG_ITER_CHECK]] ], [ [[RDX_SELECT9]], [[VEC_EPILOG_MIDDLE_BLOCK]] ]
+; CHECK-NEXT:    br label [[LOOP:%.*]]
+; CHECK:       loop:
+; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[VEC_EPILOG_SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; CHECK-NEXT:    [[RED:%.*]] = phi i32 [ [[BC_MERGE_RDX10]], [[VEC_EPILOG_SCALAR_PH]] ], [ [[SELECT:%.*]], [[LOOP]] ]
+; CHECK-NEXT:    [[GEP:%.*]] = getelementptr inbounds i8, ptr [[SRC]], i64 [[IV]]
+; CHECK-NEXT:    [[LOAD:%.*]] = load i8, ptr [[GEP]], align 1
+; CHECK-NEXT:    [[ICMP:%.*]] = icmp eq i8 [[LOAD]], 0
+; CHECK-NEXT:    [[SELECT]] = select i1 [[ICMP]], i32 1, i32 [[RED]]
+; CHECK-NEXT:    [[IV_NEXT]] = add i64 [[IV]], 1
+; CHECK-NEXT:    [[ICMP3:%.*]] = icmp eq i64 [[IV]], [[N]]
+; CHECK-NEXT:    br i1 [[ICMP3]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP7:![0-9]+]]
+; CHECK:       exit:
+; CHECK-NEXT:    [[SELECT_LCSSA:%.*]] = phi i32 [ [[SELECT]], [[LOOP]] ], [ [[RDX_SELECT]], [[MIDDLE_BLOCK]] ], [ [[RDX_SELECT9]], [[VEC_EPILOG_MIDDLE_BLOCK]] ]
+; CHECK-NEXT:    ret i32 [[SELECT_LCSSA]]
+;
+entry:
+  br label %loop
+
+loop:
+  %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ]
+  %red = phi i32 [ %start, %entry ], [ %select, %loop ]
+  %gep = getelementptr inbounds i8, ptr %src, i64 %iv
+  %load = load i8, ptr %gep, align 1
+  %icmp = icmp eq i8 %load, 0
+  %select = select i1 %icmp, i32 1, i32 %red
+  %iv.next = add i64 %iv, 1
+  %icmp3 = icmp eq i64 %iv, %N
+  br i1 %icmp3, label %exit, label %loop
+
+exit:
+  ret i32 %select
+}
 
 define i1 @any_of_reduction_i1_epilog(i64 %N, i32 %a) {
 ; CHECK-LABEL: define i1 @any_of_reduction_i1_epilog(
@@ -124,14 +223,15 @@ define i1 @any_of_reduction_i1_epilog(i64 %N, i32 %a) {
 ; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <4 x i1> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP3:%.*]], [[VECTOR_BODY]] ]
 ; CHECK-NEXT:    [[VEC_IND:%.*]] = phi <4 x i32> [ <i32 0, i32 1, i32 2, i32 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
 ; CHECK-NEXT:    [[TMP1:%.*]] = icmp eq <4 x i32> [[VEC_IND]], [[BROADCAST_SPLAT]]
-; CHECK-NEXT:    [[TMP3]] = select <4 x i1> [[TMP1]], <4 x i1> [[VEC_PHI]], <4 x i1> zeroinitializer
+; CHECK-NEXT:    [[TMP2:%.*]] = xor <4 x i1> [[TMP1]], <i1 true, i1 true, i1 true, i1 true>
+; CHECK-NEXT:    [[TMP3]] = or <4 x i1> [[VEC_PHI]], [[TMP2]]
 ; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
 ; CHECK-NEXT:    [[VEC_IND_NEXT]] = add <4 x i32> [[VEC_IND]], <i32 4, i32 4, i32 4, i32 4>
 ; CHECK-NEXT:    [[TMP4:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT:    br i1 [[TMP4]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP5:![0-9]+]]
+; CHECK-NEXT:    br i1 [[TMP4]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP8:![0-9]+]]
 ; CHECK:       middle.block:
-; CHECK-NEXT:    [[RDX_SELECT_CMP:%.*]] = icmp ne <4 x i1> [[TMP3]], zeroinitializer
-; CHECK-NEXT:    [[TMP6:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[RDX_SELECT_CMP]])
+; CHECK-NEXT:    [[TMP5:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[TMP3]])
+; CHECK-NEXT:    [[TMP6:%.*]] = freeze i1 [[TMP5]]
 ; CHECK-NEXT:    [[RDX_SELECT:%.*]] = select i1 [[TMP6]], i1 false, i1 false
 ; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[TMP0]], [[N_VEC]]
 ; CHECK-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[VEC_EPILOG_ITER_CHECK:%.*]]
@@ -144,10 +244,11 @@ define i1 @any_of_reduction_i1_epilog(i64 %N, i32 %a) {
 ; CHECK-NEXT:    [[BC_MERGE_RDX:%.*]] = phi i1 [ false, [[VECTOR_MAIN_LOOP_ITER_CHECK]] ], [ [[RDX_SELECT]], [[VEC_EPILOG_ITER_CHECK]] ]
 ; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i32 [ [[IND_END]], [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[VECTOR_MAIN_LOOP_ITER_CHECK]] ]
 ; CHECK-NEXT:    [[VEC_EPILOG_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[VECTOR_MAIN_LOOP_ITER_CHECK]] ]
+; CHECK-NEXT:    [[TMP7:%.*]] = icmp ne i1 [[BC_MERGE_RDX]], false
 ; CHECK-NEXT:    [[N_MOD_VF2:%.*]] = urem i64 [[TMP0]], 4
 ; CHECK-NEXT:    [[N_VEC3:%.*]] = sub i64 [[TMP0]], [[N_MOD_VF2]]
 ; CHECK-NEXT:    [[IND_END5:%.*]] = trunc i64 [[N_VEC3]] to i32
-; CHECK-NEXT:    [[MINMAX_IDENT_SPLATINSERT:%.*]] = insertelement <4 x i1> poison, i1 [[BC_MERGE_RDX]], i64 0
+; CHECK-NEXT:    [[MINMAX_IDENT_SPLATINSERT:%.*]] = insertelement <4 x i1> poison, i1 [[TMP7]], i64 0
 ; CHECK-NEXT:    [[MINMAX_IDENT_SPLAT:%.*]] = shufflevector <4 x i1> [[MINMAX_IDENT_SPLATINSERT]], <4 x i1> poison, <4 x i32> zeroinitializer
 ; CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <4 x i32> poison, i32 [[BC_RESUME_VAL]], i64 0
 ; CHECK-NEXT:    [[DOTSPLAT:%.*]] = shufflevector <4 x i32> [[DOTSPLATINSERT]], <4 x i32> poison, <4 x i32> zeroinitializer
@@ -160,21 +261,22 @@ define i1 @any_of_reduction_i1_epilog(i64 %N, i32 %a) {
 ; CHECK-NEXT:    [[VEC_PHI10:%.*]] = phi <4 x i1> [ [[MINMAX_IDENT_SPLAT]], [[VEC_EPILOG_PH]] ], [ [[TMP10:%.*]], [[VEC_EPILOG_VECTOR_BODY]] ]
 ; CHECK-NEXT:    [[VEC_IND11:%.*]] = phi <4 x i32> [ [[INDUCTION]], [[VEC_EPILOG_PH]] ], [ [[VEC_IND_NEXT12:%.*]], [[VEC_EPILOG_VECTOR_BODY]] ]
 ; CHECK-NEXT:    [[TMP8:%.*]] = icmp eq <4 x i32> [[VEC_IND11]], [[BROADCAST_SPLAT14]]
-; CHECK-NEXT:    [[TMP10]] = select <4 x i1> [[TMP8]], <4 x i1> [[VEC_PHI10]], <4 x i1> zeroinitializer
+; CHECK-NEXT:    [[TMP9:%.*]] = xor <4 x i1> [[TMP8]], <i1 true, i1 true, i1 true, i1 true>
+; CHECK-NEXT:    [[TMP10]] = or <4 x i1> [[VEC_PHI10]], [[TMP9]]
 ; CHECK-NEXT:    [[INDEX_NEXT15]] = add nuw i64 [[INDEX9]], 4
 ; CHECK-NEXT:    [[VEC_IND_NEXT12]] = add <4 x i32> [[VEC_IND11]], <i32 4, i32 4, i32 4, i32 4>
 ; CHECK-NEXT:    [[TMP11:%.*]] = icmp eq i64 [[INDEX_NEXT15]], [[N_VEC3]]
-; CHECK-NEXT:    br i1 [[TMP11]], label [[VEC_EPILOG_MIDDLE_BLOCK:%.*]], label [[VEC_EPILOG_VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
+; CHECK-NEXT:    br i1 [[TMP11]], label [[VEC_EPILOG_MIDDLE_BLOCK:%.*]], label [[VEC_EPILOG_VECTOR_BODY]], !llvm.loop [[LOOP9:![0-9]+]]
 ; CHECK:       vec.epilog.middle.block:
-; CHECK-NEXT:    [[RDX_SELECT_CMP16:%.*]] = icmp ne <4 x i1> [[TMP10]], zeroinitializer
-; CHECK-NEXT:    [[TMP13:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[RDX_SELECT_CMP16]])
+; CHECK-NEXT:    [[TMP12:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[TMP10]])
+; CHECK-NEXT:    [[TMP13:%.*]] = freeze i1 [[TMP12]]
 ; CHECK-NEXT:    [[RDX_SELECT16:%.*]] = select i1 [[TMP13]], i1 false, i1 false
 ; CHECK-NEXT:    [[CMP_N8:%.*]] = icmp eq i64 [[TMP0]], [[N_VEC3]]
 ; CHECK-NEXT:    br i1 [[CMP_N8]], label [[EXIT]], label [[VEC_EPILOG_SCALAR_PH]]
 ; CHECK:       vec.epilog.scalar.ph:
 ; CHECK-NEXT:    [[BC_RESUME_VAL4:%.*]] = phi i64 [ [[N_VEC3]], [[VEC_EPILOG_MIDDLE_BLOCK]] ], [ [[N_VEC]], [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[ITER_CHECK:%.*]] ]
 ; CHECK-NEXT:    [[BC_RESUME_VAL7:%.*]] = phi i32 [ [[IND_END5]], [[VEC_EPILOG_MIDDLE_BLOCK]] ], [ [[IND_END6]], [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[ITER_CHECK]] ]
-; CHECK-NEXT:    [[BC_MERGE_RDX17:%.*]] = phi i1 [ false, [[ITER_CHECK]] ], [ [[RDX_SELECT]], [[VEC_EPILOG_ITER_CHECK]] ], [ [[RDX_SELECT16]], [[VEC_EPILOG_MIDDLE_BLOCK]] ]
+; CHECK-NEXT:    [[BC_MERGE_RDX17:%.*]] = phi i1 [ false, [[ITER_CHECK]] ], [ false, [[VEC_EPILOG_ITER_CHECK]] ], [ [[RDX_SELECT16]], [[VEC_EPILOG_MIDDLE_BLOCK]] ]
 ; CHECK-NEXT:    br label [[LOOP:%.*]]
 ; CHECK:       loop:
 ; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL4]], [[VEC_EPILOG_SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
@@ -185,7 +287,7 @@ define i1 @any_of_reduction_i1_epilog(i64 %N, i32 %a) {
 ; CHECK-NEXT:    [[IV_NEXT]] = add i64 [[IV]], 1
 ; CHECK-NEXT:    [[IV_2_NEXT]] = add i32 [[IV_2]], 1
 ; CHECK-NEXT:    [[CMP_2:%.*]] = icmp eq i64 [[IV]], [[N]]
-; CHECK-NEXT:    br i1 [[CMP_2]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP7:![0-9]+]]
+; CHECK-NEXT:    br i1 [[CMP_2]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP10:![0-9]+]]
 ; CHECK:       exit:
 ; CHECK-NEXT:    [[SEL_LCSSA:%.*]] = phi i1 [ [[SEL]], [[LOOP]] ], [ [[RDX_SELECT]], [[MIDDLE_BLOCK]] ], [ [[RDX_SELECT16]], [[VEC_EPILOG_MIDDLE_BLOCK]] ]
 ; CHECK-NEXT:    ret i1 [[SEL_LCSSA]]
@@ -219,4 +321,7 @@ exit:
 ; CHECK: [[LOOP5]] = distinct !{[[LOOP5]], [[META1]], [[META2]]}
 ; CHECK: [[LOOP6]] = distinct !{[[LOOP6]], [[META1]], [[META2]]}
 ; CHECK: [[LOOP7]] = distinct !{[[LOOP7]], [[META2]], [[META1]]}
+; CHECK: [[LOOP8]] = distinct !{[[LOOP8]], [[META1]], [[META2]]}
+; CHECK: [[LOOP9]] = distinct !{[[LOOP9]], [[META1]], [[META2]]}
+; CHECK: [[LOOP10]] = distinct !{[[LOOP10]], [[META2]], [[META1]]}
 ;.
diff --git a/llvm/test/Transforms/LoopVectorize/select-cmp-predicated.ll b/llvm/test/Transforms/LoopVectorize/select-cmp-predicated.ll
index e8c75635f2c6..cedc769b811f 100644
--- a/llvm/test/Transforms/LoopVectorize/select-cmp-predicated.ll
+++ b/llvm/test/Transforms/LoopVectorize/select-cmp-predicated.ll
@@ -4,7 +4,7 @@
 define i32 @pred_select_const_i32_from_icmp(ptr noalias nocapture readonly %src1, ptr noalias nocapture readonly %src2, i64 %n) {
 ; CHECK-VF2IC1-LABEL: @pred_select_const_i32_from_icmp(
 ; CHECK-VF2IC1:       vector.body:
-; CHECK-VF2IC1:         [[VEC_PHI:%.*]] = phi <2 x i32> [ zeroinitializer, %vector.ph ], [ [[PREDPHI:%.*]], %pred.load.continue2 ]
+; CHECK-VF2IC1:         [[VEC_PHI:%.*]] = phi <2 x i1> [ zeroinitializer, %vector.ph ], [ [[PREDPHI:%.*]], %pred.load.continue2 ]
 ; CHECK-VF2IC1:         [[WIDE_LOAD:%.*]] = load <2 x i32>, ptr {{%.*}}, align 4
 ; CHECK-VF2IC1-NEXT:    [[TMP4:%.*]] = icmp sgt <2 x i32> [[WIDE_LOAD]], <i32 35, i32 35>
 ; CHECK-VF2IC1-NEXT:    [[TMP5:%.*]] = extractelement <2 x i1> [[TMP4]], i32 0
@@ -26,14 +26,14 @@ define i32 @pred_select_const_i32_from_icmp(ptr noalias nocapture readonly %src1
 ; CHECK-VF2IC1:       pred.load.continue2:
 ; CHECK-VF2IC1-NEXT:    [[TMP15:%.*]] = phi <2 x i32> [ [[TMP9]], %pred.load.continue ], [ [[TMP14]], %pred.load.if1 ]
 ; CHECK-VF2IC1-NEXT:    [[TMP16:%.*]] = icmp eq <2 x i32> [[TMP15]], <i32 2, i32 2>
-; CHECK-VF2IC1-NEXT:    [[TMP17:%.*]] = select <2 x i1> [[TMP16]], <2 x i32> <i32 1, i32 1>, <2 x i32> [[VEC_PHI]]
+; CHECK-VF2IC1-NEXT:    [[TMP17:%.*]] = or <2 x i1> [[VEC_PHI]], [[TMP16]]
 ; CHECK-VF2IC1-NEXT:    [[TMP18:%.*]] = xor <2 x i1> [[TMP4]], <i1 true, i1 true>
-; CHECK-VF2IC1-NEXT:    [[PREDPHI]] = select <2 x i1> [[TMP4]], <2 x i32> [[TMP17]], <2 x i32> [[VEC_PHI]]
+; CHECK-VF2IC1-NEXT:    [[PREDPHI]] = select <2 x i1> [[TMP4]], <2 x i1> [[TMP17]], <2 x i1> [[VEC_PHI]]
 ; CHECK-VF2IC1:         br i1 {{%.*}}, label %middle.block, label %vector.body
 ; CHECK-VF2IC1:       middle.block:
-; CHECK-VF2IC1-NEXT:    [[RDX_SELECT_CMP:%.*]] = icmp ne <2 x i32> [[PREDPHI]], zeroinitializer
-; CHECK-VF2IC1-NEXT:    [[TMP20:%.*]] = call i1 @llvm.vector.reduce.or.v2i1(<2 x i1> [[RDX_SELECT_CMP]])
-; CHECK-VF2IC1-NEXT:    [[RDX_SELECT:%.*]] = select i1 [[TMP20]], i32 1, i32 0
+; CHECK-VF2IC1-NEXT:    [[TMP20:%.*]] = call i1 @llvm.vector.reduce.or.v2i1(<2 x i1> [[PREDPHI]])
+; CHECK-VF2IC1-NEXT:    [[FR_TMP20:%.*]] = freeze i1 [[TMP20]]
+; CHECK-VF2IC1-NEXT:    [[RDX_SELECT:%.*]] = select i1 [[FR_TMP20]], i32 1, i32 0
 ; CHECK-VF2IC1:       scalar.ph:
 ; CHECK-VF2IC1:         [[BC_RESUME_VAL:%.*]] = phi i64 [ {{%.*}}, %middle.block ], [ 0, %entry ]
 ; CHECK-VF2IC1-NEXT:    [[BC_MERGE_RDX:%.*]] = phi i32 [ 0, %entry ], [ [[RDX_SELECT]], %middle.block ]
@@ -56,8 +56,8 @@ define i32 @pred_select_const_i32_from_icmp(ptr noalias nocapture readonly %src1
 ;
 ; CHECK-VF1IC2-LABEL: @pred_select_const_i32_from_icmp(
 ; CHECK-VF1IC2:       vector.body:
-; CHECK-VF1IC2:         [[VEC_PHI:%.*]] = phi i32 [ 0, %vector.ph ], [ [[PREDPHI:%.*]], %pred.load.continue3 ]
-; CHECK-VF1IC2-NEXT:    [[VEC_PHI2:%.*]] = phi i32 [ 0, %vector.ph ], [ [[PREDPHI5:%.*]], %pred.load.continue3 ]
+; CHECK-VF1IC2:         [[VEC_PHI:%.*]] = phi i1 [ false, %vector.ph ], [ [[PREDPHI:%.*]], %pred.load.continue3 ]
+; CHECK-VF1IC2-NEXT:    [[VEC_PHI2:%.*]] = phi i1 [ false, %vector.ph ], [ [[PREDPHI5:%.*]], %pred.load.continue3 ]
 ; CHECK-VF1IC2:         [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[SRC1:%.*]], i64 {{%.*}}
 ; CHECK-VF1IC2-NEXT:    [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[SRC1]], i64 {{%.*}}
 ; CHECK-VF1IC2-NEXT:    [[TMP2:%.*]] = load i32, ptr [[TMP0]], align 4
@@ -80,16 +80,17 @@ define i32 @pred_select_const_i32_from_icmp(ptr noalias nocapture readonly %src1
 ; CHECK-VF1IC2-NEXT:    [[TMP11:%.*]] = phi i32 [ poison, %pred.load.continue ], [ [[TMP10]], %pred.load.if2 ]
 ; CHECK-VF1IC2-NEXT:    [[TMP12:%.*]] = icmp eq i32 [[TMP8]], 2
 ; CHECK-VF1IC2-NEXT:    [[TMP13:%.*]] = icmp eq i32 [[TMP11]], 2
-; CHECK-VF1IC2-NEXT:    [[TMP14:%.*]] = select i1 [[TMP12]], i32 1, i32 [[VEC_PHI]]
-; CHECK-VF1IC2-NEXT:    [[TMP15:%.*]] = select i1 [[TMP13]], i32 1, i32 [[VEC_PHI2]]
+; CHECK-VF1IC2-NEXT:    [[TMP14:%.*]] = or i1 [[VEC_PHI]], [[TMP12]]
+; CHECK-VF1IC2-NEXT:    [[TMP15:%.*]] = or i1 [[VEC_PHI2]], [[TMP13]]
 ; CHECK-VF1IC2-NEXT:    [[TMP16:%.*]] = xor i1 [[TMP4]], true
 ; CHECK-VF1IC2-NEXT:    [[TMP17:%.*]] = xor i1 [[TMP5]], true
-; CHECK-VF1IC2-NEXT:    [[PREDPHI]] = select i1 [[TMP4]], i32 [[TMP14]], i32 [[VEC_PHI]]
-; CHECK-VF1IC2-NEXT:    [[PREDPHI5]] = select i1 [[TMP5]], i32 [[TMP15]], i32 [[VEC_PHI2]]
+; CHECK-VF1IC2-NEXT:    [[PREDPHI]] = select i1 [[TMP4]], i1 [[TMP14]], i1 [[VEC_PHI]]
+; CHECK-VF1IC2-NEXT:    [[PREDPHI5]] = select i1 [[TMP5]], i1 [[TMP15]], i1 [[VEC_PHI2]]
 ; CHECK-VF1IC2:         br i1 {{%.*}}, label %middle.block, label %vector.body
 ; CHECK-VF1IC2:       middle.block:
-; CHECK-VF1IC2-NEXT:    [[RDX_SELECT_CMP:%.*]] = icmp ne i32 [[PREDPHI]], 0
-; CHECK-VF1IC2-NEXT:    [[RDX_SELECT:%.*]] = select i1 [[RDX_SELECT_CMP]], i32 [[PREDPHI]], i32 [[PREDPHI5]]
+; CHECK-VF1IC2-NEXT:    [[OR:%.*]] = or i1 [[PREDPHI5]], [[PREDPHI]]
+; CHECK-VF1IC2-NEXT:    [[FR_OR:%.*]] = freeze i1 [[OR]]
+; CHECK-VF1IC2-NEXT:    [[RDX_SELECT:%.*]] = select i1 [[FR_OR]], i32 1, i32 0
 ; CHECK-VF1IC2:         br i1 {{%.*}}, label %for.end.loopexit, label %scalar.ph
 ; CHECK-VF1IC2:       scalar.ph:
 ; CHECK-VF1IC2-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ {{%.*}}, %middle.block ], [ 0, %entry ]
diff --git a/llvm/test/Transforms/LoopVectorize/select-cmp.ll b/llvm/test/Transforms/LoopVectorize/select-cmp.ll
index c9f2aaef6d5c..993b56a05207 100644
--- a/llvm/test/Transforms/LoopVectorize/select-cmp.ll
+++ b/llvm/test/Transforms/LoopVectorize/select-cmp.ll
@@ -5,45 +5,47 @@
 define i32 @select_const_i32_from_icmp(ptr nocapture readonly %v, i64 %n) {
 ; CHECK-LABEL: @select_const_i32_from_icmp
 ; CHECK-VF4IC1:      vector.body:
-; CHECK-VF4IC1:        [[VEC_PHI:%.*]] = phi <4 x i32> [ <i32 3, i32 3, i32 3, i32 3>, %vector.ph ], [ [[VEC_SEL:%.*]], %vector.body ]
+; CHECK-VF4IC1:        [[VEC_PHI:%.*]] = phi <4 x i1> [ zeroinitializer, %vector.ph ], [ [[VEC_SEL:%.*]], %vector.body ]
 ; CHECK-VF4IC1:        [[VEC_LOAD:%.*]] = load <4 x i32>
 ; CHECK-VF4IC1-NEXT:   [[VEC_ICMP:%.*]] = icmp eq <4 x i32> [[VEC_LOAD]], <i32 3, i32 3, i32 3, i32 3>
-; CHECK-VF4IC1-NEXT:   [[VEC_SEL]] = select <4 x i1> [[VEC_ICMP]], <4 x i32> [[VEC_PHI]], <4 x i32> <i32 7, i32 7, i32 7, i32 7>
+; CHECK-VF4IC1-NEXT:   [[NOT:%.*]] = xor <4 x i1> [[VEC_ICMP]], <i1 true, i1 true, i1 true, i1 true>
+; CHECK-VF4IC1-NEXT:   [[VEC_SEL]] = or <4 x i1> [[VEC_PHI]], [[NOT]]
 ; CHECK-VF4IC1:      middle.block:
-; CHECK-VF4IC1-NEXT:   [[FIN_ICMP:%.*]] = icmp ne <4 x i32> [[VEC_SEL]], <i32 3, i32 3, i32 3, i32 3>
-; CHECK-VF4IC1-NEXT:   [[OR_RDX:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[FIN_ICMP]])
-; CHECK-VF4IC1-NEXT:   {{.*}} = select i1 [[OR_RDX]], i32 7, i32 3
+; CHECK-VF4IC1-NEXT:   [[OR_RDX:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[VEC_SEL]])
+; CHECK-VF4IC1-NEXT:   [[FR_OR_RDX:%.*]] = freeze i1 [[OR_RDX]]
+; CHECK-VF4IC1-NEXT:   {{.*}} = select i1 [[FR_OR_RDX]], i32 7, i32 3
 
 ; CHECK-VF4IC4:      vector.body:
-; CHECK-VF4IC4:        [[VEC_PHI1:%.*]] = phi <4 x i32> [ <i32 3, i32 3, i32 3, i32 3>, %vector.ph ], [ [[VEC_SEL1:%.*]], %vector.body ]
-; CHECK-VF4IC4-NEXT:   [[VEC_PHI2:%.*]] = phi <4 x i32> [ <i32 3, i32 3, i32 3, i32 3>, %vector.ph ], [ [[VEC_SEL2:%.*]], %vector.body ]
-; CHECK-VF4IC4-NEXT:   [[VEC_PHI3:%.*]] = phi <4 x i32> [ <i32 3, i32 3, i32 3, i32 3>, %vector.ph ], [ [[VEC_SEL3:%.*]], %vector.body ]
-; CHECK-VF4IC4-NEXT:   [[VEC_PHI4:%.*]] = phi <4 x i32> [ <i32 3, i32 3, i32 3, i32 3>, %vector.ph ], [ [[VEC_SEL4:%.*]], %vector.body ]
+; CHECK-VF4IC4:        [[VEC_PHI1:%.*]] = phi <4 x i1> [ zeroinitializer, %vector.ph ], [ [[VEC_SEL1:%.*]], %vector.body ]
+; CHECK-VF4IC4-NEXT:   [[VEC_PHI2:%.*]] = phi <4 x i1> [ zeroinitializer, %vector.ph ], [ [[VEC_SEL2:%.*]], %vector.body ]
+; CHECK-VF4IC4-NEXT:   [[VEC_PHI3:%.*]] = phi <4 x i1> [ zeroinitializer, %vector.ph ], [ [[VEC_SEL3:%.*]], %vector.body ]
+; CHECK-VF4IC4-NEXT:   [[VEC_PHI4:%.*]] = phi <4 x i1> [ zeroinitializer, %vector.ph ], [ [[VEC_SEL4:%.*]], %vector.body ]
 ; CHECK-VF4IC4:        [[VEC_ICMP1:%.*]] = icmp eq <4 x i32> {{.*}}, <i32 3, i32 3, i32 3, i32 3>
 ; CHECK-VF4IC4-NEXT:   [[VEC_ICMP2:%.*]] = icmp eq <4 x i32> {{.*}}, <i32 3, i32 3, i32 3, i32 3>
 ; CHECK-VF4IC4-NEXT:   [[VEC_ICMP3:%.*]] = icmp eq <4 x i32> {{.*}}, <i32 3, i32 3, i32 3, i32 3>
 ; CHECK-VF4IC4-NEXT:   [[VEC_ICMP4:%.*]] = icmp eq <4 x i32> {{.*}}, <i32 3, i32 3, i32 3, i32 3>
-; CHECK-VF4IC4-NEXT:   [[VEC_SEL1:%.*]] = select <4 x i1> [[VEC_ICMP1]], <4 x i32> [[VEC_PHI1]], <4 x i32> <i32 7, i32 7, i32 7, i32 7>
-; CHECK-VF4IC4-NEXT:   [[VEC_SEL2:%.*]] = select <4 x i1> [[VEC_ICMP2]], <4 x i32> [[VEC_PHI2]], <4 x i32> <i32 7, i32 7, i32 7, i32 7>
-; CHECK-VF4IC4-NEXT:   [[VEC_SEL3:%.*]] = select <4 x i1> [[VEC_ICMP3]], <4 x i32> [[VEC_PHI3]], <4 x i32> <i32 7, i32 7, i32 7, i32 7>
-; CHECK-VF4IC4-NEXT:   [[VEC_SEL4:%.*]] = select <4 x i1> [[VEC_ICMP4]], <4 x i32> [[VEC_PHI4]], <4 x i32> <i32 7, i32 7, i32 7, i32 7>
+; CHECK-VF4IC4-NEXT:   [[NOT1:%.*]] = xor <4 x i1> [[VEC_ICMP1]], <i1 true, i1 true, i1 true, i1 true>
+; CHECK-VF4IC4-NEXT:   [[NOT2:%.*]] = xor <4 x i1> [[VEC_ICMP2]], <i1 true, i1 true, i1 true, i1 true>
+; CHECK-VF4IC4-NEXT:   [[NOT3:%.*]] = xor <4 x i1> [[VEC_ICMP3]], <i1 true, i1 true, i1 true, i1 true>
+; CHECK-VF4IC4-NEXT:   [[NOT4:%.*]] = xor <4 x i1> [[VEC_ICMP4]], <i1 true, i1 true, i1 true, i1 true>
+; CHECK-VF4IC4-NEXT:   [[VEC_SEL1:%.*]] = or <4 x i1> [[VEC_PHI1]], [[NOT1]]
+; CHECK-VF4IC4-NEXT:   [[VEC_SEL2:%.*]] = or <4 x i1> [[VEC_PHI2]], [[NOT2]]
+; CHECK-VF4IC4-NEXT:   [[VEC_SEL3:%.*]] = or <4 x i1> [[VEC_PHI3]], [[NOT3]]
+; CHECK-VF4IC4-NEXT:   [[VEC_SEL4:%.*]] = or <4 x i1> [[VEC_PHI4]], [[NOT4]]
 ; CHECK-VF4IC4:      middle.block:
-; CHECK-VF4IC4-NEXT:   [[VEC_ICMP5:%.*]] = icmp ne <4 x i32> [[VEC_SEL1]], <i32 3, i32 3, i32 3, i32 3>
-; CHECK-VF4IC4-NEXT:   [[VEC_SEL5:%.*]] = select <4 x i1> [[VEC_ICMP5]], <4 x i32> [[VEC_SEL1]], <4 x i32> [[VEC_SEL2]]
-; CHECK-VF4IC4-NEXT:   [[VEC_ICMP6:%.*]] = icmp ne <4 x i32> [[VEC_SEL5]], <i32 3, i32 3, i32 3, i32 3>
-; CHECK-VF4IC4-NEXT:   [[VEC_SEL6:%.*]] = select <4 x i1> [[VEC_ICMP6]], <4 x i32> [[VEC_SEL5]], <4 x i32> [[VEC_SEL3]]
-; CHECK-VF4IC4-NEXT:   [[VEC_ICMP7:%.*]] = icmp ne <4 x i32> [[VEC_SEL6]], <i32 3, i32 3, i32 3, i32 3>
-; CHECK-VF4IC4-NEXT:   [[VEC_SEL_FIN:%.*]] = select <4 x i1> [[VEC_ICMP7]], <4 x i32> [[VEC_SEL6]], <4 x i32> [[VEC_SEL4]]
-; CHECK-VF4IC4-NEXT:   [[FIN_ICMP:%.*]] = icmp ne <4 x i32> [[VEC_SEL_FIN]], <i32 3, i32 3, i32 3, i32 3>
-; CHECK-VF4IC4-NEXT:   [[OR_RDX:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[FIN_ICMP]])
-; CHECK-VF4IC4-NEXT:   {{.*}} = select i1 [[OR_RDX]], i32 7, i32 3
+; CHECK-VF4IC4-NEXT:   [[VEC_SEL5:%.*]] = or <4 x i1>  [[VEC_SEL2]], [[VEC_SEL1]]
+; CHECK-VF4IC4-NEXT:   [[VEC_SEL6:%.*]] = or <4 x i1> [[VEC_SEL3]], [[VEC_SEL5]]
+; CHECK-VF4IC4-NEXT:   [[VEC_SEL7:%.*]] = or <4 x i1> [[VEC_SEL4]], [[VEC_SEL6]]
+; CHECK-VF4IC4-NEXT:   [[OR_RDX:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[VEC_SEL7]])
+; CHECK-VF4IC4-NEXT:   [[FR_OR_RDX:%.*]] = freeze i1 [[OR_RDX]]
+; CHECK-VF4IC4-NEXT:   {{.*}} = select i1 [[FR_OR_RDX]], i32 7, i32 3
 
 
 ; CHECK-VF1IC4:      vector.body:
-; CHECK-VF1IC4:        [[VEC_PHI1:%.*]] = phi i32 [ 3, %vector.ph ], [ [[VEC_SEL1:%.*]], %vector.body ]
-; CHECK-VF1IC4-NEXT:   [[VEC_PHI2:%.*]] = phi i32 [ 3, %vector.ph ], [ [[VEC_SEL2:%.*]], %vector.body ]
-; CHECK-VF1IC4-NEXT:   [[VEC_PHI3:%.*]] = phi i32 [ 3, %vector.ph ], [ [[VEC_SEL3:%.*]], %vector.body ]
-; CHECK-VF1IC4-NEXT:   [[VEC_PHI4:%.*]] = phi i32 [ 3, %vector.ph ], [ [[VEC_SEL4:%.*]], %vector.body ]
+; CHECK-VF1IC4:        [[VEC_PHI1:%.*]] = phi i1 [ false, %vector.ph ], [ [[VEC_SEL1:%.*]], %vector.body ]
+; CHECK-VF1IC4-NEXT:   [[VEC_PHI2:%.*]] = phi i1 [ false, %vector.ph ], [ [[VEC_SEL2:%.*]], %vector.body ]
+; CHECK-VF1IC4-NEXT:   [[VEC_PHI3:%.*]] = phi i1 [ false, %vector.ph ], [ [[VEC_SEL3:%.*]], %vector.body ]
+; CHECK-VF1IC4-NEXT:   [[VEC_PHI4:%.*]] = phi i1 [ false, %vector.ph ], [ [[VEC_SEL4:%.*]], %vector.body ]
 ; CHECK-VF1IC4:        [[VEC_LOAD1:%.*]] = load i32
 ; CHECK-VF1IC4-NEXT:   [[VEC_LOAD2:%.*]] = load i32
 ; CHECK-VF1IC4-NEXT:   [[VEC_LOAD3:%.*]] = load i32
@@ -52,17 +54,20 @@ define i32 @select_const_i32_from_icmp(ptr nocapture readonly %v, i64 %n) {
 ; CHECK-VF1IC4-NEXT:   [[VEC_ICMP2:%.*]] = icmp eq i32 [[VEC_LOAD2]], 3
 ; CHECK-VF1IC4-NEXT:   [[VEC_ICMP3:%.*]] = icmp eq i32 [[VEC_LOAD3]], 3
 ; CHECK-VF1IC4-NEXT:   [[VEC_ICMP4:%.*]] = icmp eq i32 [[VEC_LOAD4]], 3
-; CHECK-VF1IC4-NEXT:   [[VEC_SEL1]] = select i1 [[VEC_ICMP1]], i32 [[VEC_PHI1]], i32 7
-; CHECK-VF1IC4-NEXT:   [[VEC_SEL2]] = select i1 [[VEC_ICMP2]], i32 [[VEC_PHI2]], i32 7
-; CHECK-VF1IC4-NEXT:   [[VEC_SEL3]] = select i1 [[VEC_ICMP3]], i32 [[VEC_PHI3]], i32 7
-; CHECK-VF1IC4-NEXT:   [[VEC_SEL4]] = select i1 [[VEC_ICMP4]], i32 [[VEC_PHI4]], i32 7
+; CHECK-VF1IC4-NEXT:   [[NOT1:%.*]] = xor i1 [[VEC_ICMP1]], true
+; CHECK-VF1IC4-NEXT:   [[NOT2:%.*]] = xor i1 [[VEC_ICMP2]], true
+; CHECK-VF1IC4-NEXT:   [[NOT3:%.*]] = xor i1 [[VEC_ICMP3]], true
+; CHECK-VF1IC4-NEXT:   [[NOT4:%.*]] = xor i1 [[VEC_ICMP4]], true
+; CHECK-VF1IC4-NEXT:   [[VEC_SEL1:%.*]] = or i1 [[VEC_PHI1]], [[NOT1]]
+; CHECK-VF1IC4-NEXT:   [[VEC_SEL2:%.*]] = or i1 [[VEC_PHI2]], [[NOT2]]
+; CHECK-VF1IC4-NEXT:   [[VEC_SEL3:%.*]] = or i1 [[VEC_PHI3]], [[NOT3]]
+; CHECK-VF1IC4-NEXT:   [[VEC_SEL4:%.*]] = or i1 [[VEC_PHI4]], [[NOT4]]
 ; CHECK-VF1IC4:      middle.block:
-; CHECK-VF1IC4-NEXT:   [[VEC_ICMP4:%.*]] = icmp ne i32 [[VEC_SEL1]], 3
-; CHECK-VF1IC4-NEXT:   [[VEC_SEL5:%.*]] = select i1 [[VEC_ICMP4]], i32 [[VEC_SEL1]], i32 [[VEC_SEL2]]
-; CHECK-VF1IC4-NEXT:   [[VEC_ICMP5:%.*]] = icmp ne i32 [[VEC_SEL5]], 3
-; CHECK-VF1IC4-NEXT:   [[VEC_SEL6:%.*]] = select i1 [[VEC_ICMP5]], i32 [[VEC_SEL5]], i32 [[VEC_SEL3]]
-; CHECK-VF1IC4-NEXT:   [[VEC_ICMP6:%.*]] = icmp ne i32 [[VEC_SEL6]], 3
-; CHECK-VF1IC4-NEXT:   {{.*}} = select i1 [[VEC_ICMP6]], i32 [[VEC_SEL6]], i32 [[VEC_SEL4]]
+; CHECK-VF1IC4-NEXT:   [[VEC_SEL5:%.*]] = or i1 [[VEC_SEL2]], [[VEC_SEL1]]
+; CHECK-VF1IC4-NEXT:   [[VEC_SEL6:%.*]] = or i1 [[VEC_SEL3]], [[VEC_SEL5]]
+; CHECK-VF1IC4-NEXT:   [[OR_RDX:%.*]] = or i1  [[VEC_SEL4]], [[VEC_SEL6]]
+; CHECK-VF1IC4-NEXT:   [[FR_OR_RDX:%.*]] = freeze i1 [[OR_RDX]]
+; CHECK-VF1IC4-NEXT:   {{.*}} = select i1 [[FR_OR_RDX]], i32 7, i32 3
 
 entry:
   br label %for.body
@@ -86,14 +91,14 @@ exit:                                     ; preds = %for.body
 define i32 @select_const_i32_from_icmp2(ptr nocapture readonly %v, i64 %n) {
 ; CHECK-LABEL: @select_const_i32_from_icmp2
 ; CHECK-VF4IC1:      vector.body:
-; CHECK-VF4IC1:        [[VEC_PHI:%.*]] = phi <4 x i32> [ <i32 3, i32 3, i32 3, i32 3>, %vector.ph ], [ [[VEC_SEL:%.*]], %vector.body ]
+; CHECK-VF4IC1:        [[VEC_PHI:%.*]] = phi <4 x i1> [ zeroinitializer, %vector.ph ], [ [[VEC_SEL:%.*]], %vector.body ]
 ; CHECK-VF4IC1:        [[VEC_LOAD:%.*]] = load <4 x i32>
 ; CHECK-VF4IC1-NEXT:   [[VEC_ICMP:%.*]] = icmp eq <4 x i32> [[VEC_LOAD]], <i32 3, i32 3, i32 3, i32 3>
-; CHECK-VF4IC1-NEXT:   [[VEC_SEL]] = select <4 x i1> [[VEC_ICMP]], <4 x i32> <i32 7, i32 7, i32 7, i32 7>, <4 x i32> [[VEC_PHI]]
+; CHECK-VF4IC1-NEXT:   [[VEC_SEL]] = or <4 x i1> [[VEC_PHI]], [[VEC_ICMP]]
 ; CHECK-VF4IC1:      middle.block:
-; CHECK-VF4IC1-NEXT:   [[FIN_ICMP:%.*]] = icmp ne <4 x i32> [[VEC_SEL]], <i32 3, i32 3, i32 3, i32 3>
-; CHECK-VF4IC1-NEXT:   [[OR_RDX:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[FIN_ICMP]])
-; CHECK-VF4IC1-NEXT:   {{.*}} = select i1 [[OR_RDX]], i32 7, i32 3
+; CHECK-VF4IC1-NEXT:   [[OR_RDX:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[VEC_SEL]])
+; CHECK-VF4IC1-NEXT:   [[FR_OR_RDX:%.*]] = freeze i1 [[OR_RDX]]
+; CHECK-VF4IC1-NEXT:   {{.*}} = select i1 [[FR_OR_RDX]], i32 7, i32 3
 
 entry:
   br label %for.body
@@ -117,21 +122,18 @@ exit:                                     ; preds = %for.body
 define i32 @select_i32_from_icmp(ptr nocapture readonly %v, i32 %a, i32 %b, i64 %n) {
 ; CHECK-LABEL: @select_i32_from_icmp
 ; CHECK-VF4IC1:      vector.ph:
-; CHECK-VF4IC1:        [[TMP1:%.*]] = insertelement <4 x i32> poison, i32 %a, i64 0
-; CHECK-VF4IC1-NEXT:   [[SPLAT_OF_A:%.*]] = shufflevector <4 x i32> [[TMP1]], <4 x i32> poison, <4 x i32> zeroinitializer
-; CHECK-VF4IC1-NEXT:   [[TMP2:%.*]] = insertelement <4 x i32> poison, i32 %b, i64 0
-; CHECK-VF4IC1-NEXT:   [[SPLAT_OF_B:%.*]] = shufflevector <4 x i32> [[TMP2]], <4 x i32> poison, <4 x i32> zeroinitializer
+; CHECK-VF4IC1-NOT:    shufflevector <4 x i32>
+; CHECK-VF4IC1-NOT:    shufflevector <4 x i32>
 ; CHECK-VF4IC1:      vector.body:
-; CHECK-VF4IC1:        [[VEC_PHI:%.*]] = phi <4 x i32> [ [[SPLAT_OF_A]], %vector.ph ], [ [[VEC_SEL:%.*]], %vector.body ]
+; CHECK-VF4IC1:        [[VEC_PHI:%.*]] = phi <4 x i1> [ zeroinitializer, %vector.ph ], [ [[VEC_SEL:%.*]], %vector.body ]
 ; CHECK-VF4IC1:        [[VEC_LOAD:%.*]] = load <4 x i32>
 ; CHECK-VF4IC1-NEXT:   [[VEC_ICMP:%.*]] = icmp eq <4 x i32> [[VEC_LOAD]], <i32 3, i32 3, i32 3, i32 3>
-; CHECK-VF4IC1-NEXT:   [[VEC_SEL]] = select <4 x i1> [[VEC_ICMP]], <4 x i32> [[VEC_PHI]], <4 x i32> [[SPLAT_OF_B]]
+; CHECK-VF4IC1-NEXT:   [[NOT:%.*]] = xor <4 x i1> [[VEC_ICMP]], <i1 true, i1 true, i1 true, i1 true>
+; CHECK-VF4IC1-NEXT:   [[VEC_SEL]] = or <4 x i1> [[VEC_PHI]], [[NOT]]
 ; CHECK-VF4IC1:      middle.block:
-; CHECK-VF4IC1-NEXT:   [[FIN_INS:%.*]] = insertelement <4 x i32> poison, i32 %a, i64 0
-; CHECK-VF4IC1-NEXT:   [[FIN_SPLAT:%.*]] = shufflevector <4 x i32> [[FIN_INS]], <4 x i32> poison, <4 x i32> zeroinitializer
-; CHECK-VF4IC1-NEXT:   [[FIN_CMP:%.*]] = icmp ne <4 x i32> [[VEC_SEL]], [[FIN_SPLAT]]
-; CHECK-VF4IC1-NEXT:   [[OR_RDX:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[FIN_CMP]])
-; CHECK-VF4IC1-NEXT:   {{.*}} = select i1 [[OR_RDX]], i32 %b, i32 %a
+; CHECK-VF4IC1-NEXT:   [[OR_RDX:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[VEC_SEL]])
+; CHECK-VF4IC1-NEXT:   [[FR_OR_RDX:%.*]] = freeze i1 [[OR_RDX]]
+; CHECK-VF4IC1-NEXT:   {{.*}} = select i1 [[FR_OR_RDX]], i32 %b, i32 %a
 entry:
   br label %for.body
 
@@ -154,14 +156,15 @@ exit:                                     ; preds = %for.body
 define i32 @select_const_i32_from_fcmp_fast(ptr nocapture readonly %v, i64 %n) {
 ; CHECK-LABEL: @select_const_i32_from_fcmp_fast
 ; CHECK-VF4IC1:      vector.body:
-; CHECK-VF4IC1:        [[VEC_PHI:%.*]] = phi <4 x i32> [ <i32 2, i32 2, i32 2, i32 2>, %vector.ph ], [ [[VEC_SEL:%.*]], %vector.body ]
+; CHECK-VF4IC1:        [[VEC_PHI:%.*]] = phi <4 x i1> [ zeroinitializer, %vector.ph ], [ [[VEC_SEL:%.*]], %vector.body ]
 ; CHECK-VF4IC1:        [[VEC_LOAD:%.*]] = load <4 x float>
 ; CHECK-VF4IC1-NEXT:   [[VEC_FCMP:%.*]] = fcmp fast ueq <4 x float> [[VEC_LOAD]], <float 3.000000e+00, float 3.000000e+00, float 3.000000e+00, float 3.000000e+00>
-; CHECK-VF4IC1-NEXT:   [[VEC_SEL]] = select <4 x i1> [[VEC_FCMP]], <4 x i32> [[VEC_PHI]], <4 x i32> <i32 1, i32 1, i32 1, i32 1>
+; CHECK-VF4IC1-NEXT:   [[NOT:%.*]] = xor <4 x i1> [[VEC_FCMP]], <i1 true, i1 true, i1 true, i1 true>
+; CHECK-VF4IC1-NEXT:   [[VEC_SEL]] = or <4 x i1> [[VEC_PHI]], [[NOT]]
 ; CHECK-VF4IC1:      middle.block:
-; CHECK-VF4IC1-NEXT:   [[FIN_ICMP:%.*]] = icmp ne <4 x i32> [[VEC_SEL]], <i32 2, i32 2, i32 2, i32 2>
-; CHECK-VF4IC1-NEXT:   [[OR_RDX:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[FIN_ICMP]])
-; CHECK-VF4IC1-NEXT:   {{.*}} = select i1 [[OR_RDX]], i32 1, i32 2
+; CHECK-VF4IC1-NEXT:   [[OR_RDX:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[VEC_SEL]])
+; CHECK-VF4IC1-NEXT:   [[FR_OR_RDX:%.*]] = freeze i1 [[OR_RDX]]
+; CHECK-VF4IC1-NEXT:   {{.*}} = select i1 [[FR_OR_RDX]], i32 1, i32 2
 entry:
   br label %for.body
 
@@ -184,14 +187,15 @@ exit:                                     ; preds = %for.body
 define i32 @select_const_i32_from_fcmp(ptr nocapture readonly %v, i64 %n) {
 ; CHECK-LABEL: @select_const_i32_from_fcmp
 ; CHECK-VF4IC1:      vector.body:
-; CHECK-VF4IC1:        [[VEC_PHI:%.*]] = phi <4 x i32> [ <i32 2, i32 2, i32 2, i32 2>, %vector.ph ], [ [[VEC_SEL:%.*]], %vector.body ]
+; CHECK-VF4IC1:        [[VEC_PHI:%.*]] = phi <4 x i1> [ zeroinitializer, %vector.ph ], [ [[VEC_SEL:%.*]], %vector.body ]
 ; CHECK-VF4IC1:        [[VEC_LOAD:%.*]] = load <4 x float>
 ; CHECK-VF4IC1-NEXT:   [[VEC_FCMP:%.*]] = fcmp ueq <4 x float> [[VEC_LOAD]], <float 3.000000e+00, float 3.000000e+00, float 3.000000e+00, float 3.000000e+00>
-; CHECK-VF4IC1-NEXT:   [[VEC_SEL]] = select <4 x i1> [[VEC_FCMP]], <4 x i32> [[VEC_PHI]], <4 x i32> <i32 1, i32 1, i32 1, i32 1>
+; CHECK-VF4IC1-NEXT:   [[NOT:%.*]] = xor <4 x i1> [[VEC_FCMP]], <i1 true, i1 true, i1 true, i1 true>
+; CHECK-VF4IC1-NEXT:   [[VEC_SEL]] = or <4 x i1> [[VEC_PHI]], [[NOT]]
 ; CHECK-VF4IC1:      middle.block:
-; CHECK-VF4IC1-NEXT:   [[FIN_ICMP:%.*]] = icmp ne <4 x i32> [[VEC_SEL]], <i32 2, i32 2, i32 2, i32 2>
-; CHECK-VF4IC1-NEXT:   [[OR_RDX:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[FIN_ICMP]])
-; CHECK-VF4IC1-NEXT:   {{.*}} = select i1 [[OR_RDX]], i32 1, i32 2
+; CHECK-VF4IC1-NEXT:   [[OR_RDX:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[VEC_SEL]])
+; CHECK-VF4IC1-NEXT:   [[FR_OR_RDX:%.*]] = freeze i1 [[OR_RDX]]
+; CHECK-VF4IC1-NEXT:   {{.*}} = select i1 [[FR_OR_RDX]], i32 1, i32 2
 entry:
   br label %for.body
 
@@ -216,18 +220,16 @@ define i32 @select_i32_from_icmp_same_inputs(i32 %a, i32 %b, i64 %n) {
 ; CHECK-VF4IC1:      vector.ph:
 ; CHECK-VF4IC1:        [[TMP1:%.*]] = insertelement <4 x i32> poison, i32 %a, i64 0
 ; CHECK-VF4IC1-NEXT:   [[SPLAT_OF_A:%.*]] = shufflevector <4 x i32> [[TMP1]], <4 x i32> poison, <4 x i32> zeroinitializer
-; CHECK-VF4IC1-NEXT:   [[TMP2:%.*]] = insertelement <4 x i32> poison, i32 %b, i64 0
-; CHECK-VF4IC1-NEXT:   [[SPLAT_OF_B:%.*]] = shufflevector <4 x i32> [[TMP2]], <4 x i32> poison, <4 x i32> zeroinitializer
+; CHECK-VF4IC1-NOT:   [[TMP2:%.*]] = insertelement <4 x i32> poison, i32 %b, i64 0
 ; CHECK-VF4IC1:      vector.body:
-; CHECK-VF4IC1:        [[VEC_PHI:%.*]] = phi <4 x i32> [ [[SPLAT_OF_A]], %vector.ph ], [ [[VEC_SEL:%.*]], %vector.body ]
-; CHECK-VF4IC1:        [[VEC_ICMP:%.*]] = icmp eq <4 x i32> [[VEC_PHI]], <i32 3, i32 3, i32 3, i32 3>
-; CHECK-VF4IC1-NEXT:   [[VEC_SEL]] = select <4 x i1> [[VEC_ICMP]], <4 x i32> [[VEC_PHI]], <4 x i32> [[SPLAT_OF_B]]
+; CHECK-VF4IC1:        [[VEC_PHI:%.*]] = phi <4 x i1> [ zeroinitializer, %vector.ph ], [ [[VEC_SEL:%.*]], %vector.body ]
+; CHECK-VF4IC1:        [[VEC_ICMP:%.*]] = icmp eq <4 x i32> [[SPLAT_OF_A]], <i32 3, i32 3, i32 3, i32 3>
+; CHECK-VF4IC1-NEXT:   [[NOT:%.*]] = xor <4 x i1> [[VEC_ICMP]], <i1 true, i1 true, i1 true, i1 true>
+; CHECK-VF4IC1-NEXT:   [[VEC_SEL]] = or <4 x i1> [[VEC_PHI]], [[NOT]]
 ; CHECK-VF4IC1:      middle.block:
-; CHECK-VF4IC1-NEXT:   [[FIN_INS:%.*]] = insertelement <4 x i32> poison, i32 %a, i64 0
-; CHECK-VF4IC1-NEXT:   [[FIN_SPLAT:%.*]] = shufflevector <4 x i32> [[FIN_INS]], <4 x i32> poison, <4 x i32> zeroinitializer
-; CHECK-VF4IC1-NEXT:   [[FIN_CMP:%.*]] = icmp ne <4 x i32> [[VEC_SEL]], [[FIN_SPLAT]]
-; CHECK-VF4IC1-NEXT:   [[OR_RDX:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[FIN_CMP]])
-; CHECK-VF4IC1-NEXT:   {{.*}} = select i1 [[OR_RDX]], i32 %b, i32 %a
+; CHECK-VF4IC1-NEXT:   [[OR_RDX:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[VEC_SEL]])
+; CHECK-VF4IC1-NEXT:   [[FR_OR_RDX:%.*]] = freeze i1 [[OR_RDX]]
+; CHECK-VF4IC1-NEXT:   {{.*}} = select i1 [[FR_OR_RDX]], i32 %b, i32 %a
 entry:
   br label %for.body
 
diff --git a/llvm/test/Transforms/LoopVectorize/select-reduction-start-value-may-be-undef-or-poison.ll b/llvm/test/Transforms/LoopVectorize/select-reduction-start-value-may-be-undef-or-poison.ll
index 16ab45415b5c..55e61158a79c 100644
--- a/llvm/test/Transforms/LoopVectorize/select-reduction-start-value-may-be-undef-or-poison.ll
+++ b/llvm/test/Transforms/LoopVectorize/select-reduction-start-value-may-be-undef-or-poison.ll
@@ -8,26 +8,25 @@ define i64 @pr62565_incoming_value_known_undef(i64 %a, ptr %src) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
 ; CHECK:       vector.ph:
-; CHECK-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <2 x i64> poison, i64 [[A]], i64 0
-; CHECK-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <2 x i64> [[BROADCAST_SPLATINSERT]], <2 x i64> poison, <2 x i32> zeroinitializer
 ; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
 ; CHECK:       vector.body:
 ; CHECK-NEXT:    [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <2 x i64> [ undef, [[VECTOR_PH]] ], [ [[TMP4:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <2 x i1> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP4:%.*]], [[VECTOR_BODY]] ]
 ; CHECK-NEXT:    [[OFFSET_IDX:%.*]] = add i32 1, [[INDEX]]
 ; CHECK-NEXT:    [[TMP0:%.*]] = add i32 [[OFFSET_IDX]], 0
 ; CHECK-NEXT:    [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[SRC]], i32 [[TMP0]]
 ; CHECK-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i32 0
 ; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <2 x i32>, ptr [[TMP2]], align 4
 ; CHECK-NEXT:    [[TMP3:%.*]] = icmp eq <2 x i32> [[WIDE_LOAD]], <i32 1, i32 1>
-; CHECK-NEXT:    [[TMP4]] = select <2 x i1> [[TMP3]], <2 x i64> [[VEC_PHI]], <2 x i64> [[BROADCAST_SPLAT]]
+; CHECK-NEXT:    [[NOT:%*]] = xor <2 x i1> [[TMP3]], <i1 true, i1 true>
+; CHECK-NEXT:    [[TMP4]] = or <2 x i1> [[VEC_PHI]], [[NOT]]
 ; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 2
 ; CHECK-NEXT:    [[TMP5:%.*]] = icmp eq i32 [[INDEX_NEXT]], 32
 ; CHECK-NEXT:    br i1 [[TMP5]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
 ; CHECK:       middle.block:
-; CHECK-NEXT:    [[RDX_SELECT_CMP:%.*]] = icmp ne <2 x i64> [[TMP4]], undef
-; CHECK-NEXT:    [[TMP6:%.*]] = call i1 @llvm.vector.reduce.or.v2i1(<2 x i1> [[RDX_SELECT_CMP]])
-; CHECK-NEXT:    [[RDX_SELECT:%.*]] = select i1 [[TMP6]], i64 [[A]], i64 undef
+; CHECK-NEXT:    [[TMP6:%.*]] = call i1 @llvm.vector.reduce.or.v2i1(<2 x i1> [[TMP4]])
+; CHECK-NEXT:    [[FR_TMP6:%.*]] = freeze i1 [[TMP6]]
+; CHECK-NEXT:    [[RDX_SELECT:%.*]] = select i1 [[FR_TMP6]], i64 [[A]], i64 undef
 ; CHECK-NEXT:    br i1 true, label [[EXIT:%.*]], label [[SCALAR_PH]]
 ; CHECK:       scalar.ph:
 ; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i32 [ 33, [[MIDDLE_BLOCK]] ], [ 1, [[ENTRY:%.*]] ]
@@ -72,26 +71,25 @@ define i64 @pr62565_incoming_value_known_poison(i64 %a, ptr %src) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
 ; CHECK:       vector.ph:
-; CHECK-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <2 x i64> poison, i64 [[A]], i64 0
-; CHECK-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <2 x i64> [[BROADCAST_SPLATINSERT]], <2 x i64> poison, <2 x i32> zeroinitializer
 ; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
 ; CHECK:       vector.body:
 ; CHECK-NEXT:    [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <2 x i64> [ poison, [[VECTOR_PH]] ], [ [[TMP4:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <2 x i1> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP4:%.*]], [[VECTOR_BODY]] ]
 ; CHECK-NEXT:    [[OFFSET_IDX:%.*]] = add i32 1, [[INDEX]]
 ; CHECK-NEXT:    [[TMP0:%.*]] = add i32 [[OFFSET_IDX]], 0
 ; CHECK-NEXT:    [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[SRC]], i32 [[TMP0]]
 ; CHECK-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i32 0
 ; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <2 x i32>, ptr [[TMP2]], align 4
 ; CHECK-NEXT:    [[TMP3:%.*]] = icmp eq <2 x i32> [[WIDE_LOAD]], <i32 1, i32 1>
-; CHECK-NEXT:    [[TMP4]] = select <2 x i1> [[TMP3]], <2 x i64> [[VEC_PHI]], <2 x i64> [[BROADCAST_SPLAT]]
+; CHECK-NEXT:    [[NOT:%.*]] = xor <2 x i1> [[TMP3]], <i1 true, i1 true>
+; CHECK-NEXT:    [[TMP4]] = or <2 x i1> [[VEC_PHI]], [[NOT]]
 ; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 2
 ; CHECK-NEXT:    [[TMP5:%.*]] = icmp eq i32 [[INDEX_NEXT]], 32
 ; CHECK-NEXT:    br i1 [[TMP5]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
 ; CHECK:       middle.block:
-; CHECK-NEXT:    [[RDX_SELECT_CMP:%.*]] = icmp ne <2 x i64> [[TMP4]], poison
-; CHECK-NEXT:    [[TMP6:%.*]] = call i1 @llvm.vector.reduce.or.v2i1(<2 x i1> [[RDX_SELECT_CMP]])
-; CHECK-NEXT:    [[RDX_SELECT:%.*]] = select i1 [[TMP6]], i64 [[A]], i64 poison
+; CHECK-NEXT:    [[TMP6:%.*]] = call i1 @llvm.vector.reduce.or.v2i1(<2 x i1> [[TMP4]])
+; CHECK-NEXT:    [[FR_TMP6:%.*]] = freeze i1 [[TMP6]]
+; CHECK-NEXT:    [[RDX_SELECT:%.*]] = select i1 [[FR_TMP6]], i64 [[A]], i64 poison
 ; CHECK-NEXT:    br i1 true, label [[EXIT:%.*]], label [[SCALAR_PH]]
 ; CHECK:       scalar.ph:
 ; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i32 [ 33, [[MIDDLE_BLOCK]] ], [ 1, [[ENTRY:%.*]] ]
@@ -136,30 +134,25 @@ define i64 @pr62565_incoming_value_may_be_poison(i64 %a, ptr %src, i64 %start) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
 ; CHECK:       vector.ph:
-; CHECK-NEXT:    [[MINMAX_IDENT_SPLATINSERT:%.*]] = insertelement <2 x i64> poison, i64 [[START]], i64 0
-; CHECK-NEXT:    [[MINMAX_IDENT_SPLAT:%.*]] = shufflevector <2 x i64> [[MINMAX_IDENT_SPLATINSERT]], <2 x i64> poison, <2 x i32> zeroinitializer
-; CHECK-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <2 x i64> poison, i64 [[A]], i64 0
-; CHECK-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <2 x i64> [[BROADCAST_SPLATINSERT]], <2 x i64> poison, <2 x i32> zeroinitializer
 ; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
 ; CHECK:       vector.body:
 ; CHECK-NEXT:    [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <2 x i64> [ [[MINMAX_IDENT_SPLAT]], [[VECTOR_PH]] ], [ [[TMP4:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <2 x i1> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP4:%.*]], [[VECTOR_BODY]] ]
 ; CHECK-NEXT:    [[OFFSET_IDX:%.*]] = add i32 1, [[INDEX]]
 ; CHECK-NEXT:    [[TMP0:%.*]] = add i32 [[OFFSET_IDX]], 0
 ; CHECK-NEXT:    [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[SRC]], i32 [[TMP0]]
 ; CHECK-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i32 0
 ; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <2 x i32>, ptr [[TMP2]], align 4
 ; CHECK-NEXT:    [[TMP3:%.*]] = icmp eq <2 x i32> [[WIDE_LOAD]], <i32 1, i32 1>
-; CHECK-NEXT:    [[TMP4]] = select <2 x i1> [[TMP3]], <2 x i64> [[VEC_PHI]], <2 x i64> [[BROADCAST_SPLAT]]
+; CHECK-NEXT:    [[NOT:%.*]] = xor <2 x i1> [[TMP3]], <i1 true, i1 true>
+; CHECK-NEXT:    [[TMP4]] = or <2 x i1> [[VEC_PHI]], [[NOT]]
 ; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 2
 ; CHECK-NEXT:    [[TMP5:%.*]] = icmp eq i32 [[INDEX_NEXT]], 32
 ; CHECK-NEXT:    br i1 [[TMP5]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
 ; CHECK:       middle.block:
-; CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <2 x i64> poison, i64 [[START]], i64 0
-; CHECK-NEXT:    [[DOTSPLAT:%.*]] = shufflevector <2 x i64> [[DOTSPLATINSERT]], <2 x i64> poison, <2 x i32> zeroinitializer
-; CHECK-NEXT:    [[RDX_SELECT_CMP:%.*]] = icmp ne <2 x i64> [[TMP4]], [[DOTSPLAT]]
-; CHECK-NEXT:    [[TMP6:%.*]] = call i1 @llvm.vector.reduce.or.v2i1(<2 x i1> [[RDX_SELECT_CMP]])
-; CHECK-NEXT:    [[RDX_SELECT:%.*]] = select i1 [[TMP6]], i64 [[A]], i64 [[START]]
+; CHECK-NEXT:    [[TMP6:%.*]] = call i1 @llvm.vector.reduce.or.v2i1(<2 x i1> [[TMP4]])
+; CHECK-NEXT:    [[FR_TMP6:%.*]] = freeze i1 [[TMP6]]
+; CHECK-NEXT:    [[RDX_SELECT:%.*]] = select i1 [[FR_TMP6]], i64 [[A]], i64 [[START]]
 ; CHECK-NEXT:    br i1 true, label [[EXIT:%.*]], label [[SCALAR_PH]]
 ; CHECK:       scalar.ph:
 ; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i32 [ 33, [[MIDDLE_BLOCK]] ], [ 1, [[ENTRY:%.*]] ]
diff --git a/llvm/test/Transforms/LoopVectorize/vectorize-force-tail-with-evl.ll b/llvm/test/Transforms/LoopVectorize/vectorize-force-tail-with-evl.ll
new file mode 100644
index 000000000000..a90b38c6a960
--- /dev/null
+++ b/llvm/test/Transforms/LoopVectorize/vectorize-force-tail-with-evl.ll
@@ -0,0 +1,101 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
+; RUN: opt -passes=loop-vectorize \
+; RUN: -force-tail-folding-style=data-with-evl \
+; RUN: -prefer-predicate-over-epilogue=predicate-dont-vectorize \
+; RUN: -force-target-supports-scalable-vectors -scalable-vectorization=on -S < %s | FileCheck --check-prefix=IF-EVL %s
+
+; RUN: opt -passes=loop-vectorize \
+; RUN: -force-tail-folding-style=none \
+; RUN: -prefer-predicate-over-epilogue=predicate-else-scalar-epilogue -force-vector-width=4 \
+; RUN: -force-target-supports-scalable-vectors -scalable-vectorization=on -S < %s | FileCheck --check-prefix=NO-VP %s
+
+; The target does not support predicated vectorization.
+define void @foo(ptr noalias %a, ptr noalias %b, ptr noalias %c, i64 %N) {
+; IF-EVL-LABEL: @foo(
+; IF-EVL-NEXT:  entry:
+; IF-EVL-NEXT:    br label [[FOR_BODY:%.*]]
+; IF-EVL:       for.body:
+; IF-EVL-NEXT:    [[IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
+; IF-EVL-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 [[IV]]
+; IF-EVL-NEXT:    [[TMP0:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
+; IF-EVL-NEXT:    [[ARRAYIDX2:%.*]] = getelementptr inbounds i32, ptr [[C:%.*]], i64 [[IV]]
+; IF-EVL-NEXT:    [[TMP1:%.*]] = load i32, ptr [[ARRAYIDX2]], align 4
+; IF-EVL-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP1]], [[TMP0]]
+; IF-EVL-NEXT:    [[ARRAYIDX4:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[IV]]
+; IF-EVL-NEXT:    store i32 [[ADD]], ptr [[ARRAYIDX4]], align 4
+; IF-EVL-NEXT:    [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
+; IF-EVL-NEXT:    [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], [[N:%.*]]
+; IF-EVL-NEXT:    br i1 [[EXITCOND_NOT]], label [[FOR_COND_CLEANUP:%.*]], label [[FOR_BODY]]
+; IF-EVL:       for.cond.cleanup:
+; IF-EVL-NEXT:    ret void
+;
+; NO-VP-LABEL: @foo(
+; NO-VP-NEXT:  entry:
+; NO-VP-NEXT:    [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
+; NO-VP-NEXT:    [[TMP8:%.*]] = mul i64 [[TMP0]], 4
+; NO-VP-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[N:%.*]], [[TMP8]]
+; NO-VP-NEXT:    br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; NO-VP:       vector.ph:
+; NO-VP-NEXT:    [[TMP1:%.*]] = call i64 @llvm.vscale.i64()
+; NO-VP-NEXT:    [[TMP14:%.*]] = mul i64 [[TMP1]], 4
+; NO-VP-NEXT:    [[N_MOD_VF:%.*]] = urem i64 [[N]], [[TMP14]]
+; NO-VP-NEXT:    [[N_VEC:%.*]] = sub i64 [[N]], [[N_MOD_VF]]
+; NO-VP-NEXT:    [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
+; NO-VP-NEXT:    [[TMP15:%.*]] = mul i64 [[TMP2]], 4
+; NO-VP-NEXT:    br label [[VECTOR_BODY:%.*]]
+; NO-VP:       vector.body:
+; NO-VP-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; NO-VP-NEXT:    [[TMP3:%.*]] = add i64 [[INDEX]], 0
+; NO-VP-NEXT:    [[TMP4:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 [[TMP3]]
+; NO-VP-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[TMP4]], i32 0
+; NO-VP-NEXT:    [[WIDE_LOAD:%.*]] = load <vscale x 4 x i32>, ptr [[TMP5]], align 4
+; NO-VP-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[C:%.*]], i64 [[TMP3]]
+; NO-VP-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[TMP6]], i32 0
+; NO-VP-NEXT:    [[WIDE_LOAD1:%.*]] = load <vscale x 4 x i32>, ptr [[TMP7]], align 4
+; NO-VP-NEXT:    [[TMP16:%.*]] = add nsw <vscale x 4 x i32> [[WIDE_LOAD1]], [[WIDE_LOAD]]
+; NO-VP-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP3]]
+; NO-VP-NEXT:    [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[TMP9]], i32 0
+; NO-VP-NEXT:    store <vscale x 4 x i32> [[TMP16]], ptr [[TMP10]], align 4
+; NO-VP-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP15]]
+; NO-VP-NEXT:    [[TMP11:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
+; NO-VP-NEXT:    br i1 [[TMP11]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
+; NO-VP:       middle.block:
+; NO-VP-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[N]], [[N_VEC]]
+; NO-VP-NEXT:    br i1 [[CMP_N]], label [[FOR_COND_CLEANUP:%.*]], label [[SCALAR_PH]]
+; NO-VP:       scalar.ph:
+; NO-VP-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; NO-VP-NEXT:    br label [[FOR_BODY:%.*]]
+; NO-VP:       for.body:
+; NO-VP-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
+; NO-VP-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[IV]]
+; NO-VP-NEXT:    [[TMP12:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
+; NO-VP-NEXT:    [[ARRAYIDX2:%.*]] = getelementptr inbounds i32, ptr [[C]], i64 [[IV]]
+; NO-VP-NEXT:    [[TMP13:%.*]] = load i32, ptr [[ARRAYIDX2]], align 4
+; NO-VP-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP13]], [[TMP12]]
+; NO-VP-NEXT:    [[ARRAYIDX4:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[IV]]
+; NO-VP-NEXT:    store i32 [[ADD]], ptr [[ARRAYIDX4]], align 4
+; NO-VP-NEXT:    [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
+; NO-VP-NEXT:    [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
+; NO-VP-NEXT:    br i1 [[EXITCOND_NOT]], label [[FOR_COND_CLEANUP]], label [[FOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
+; NO-VP:       for.cond.cleanup:
+; NO-VP-NEXT:    ret void
+;
+entry:
+  br label %for.body
+
+for.body:
+  %iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]
+  %arrayidx = getelementptr inbounds i32, ptr %b, i64 %iv
+  %0 = load i32, ptr %arrayidx, align 4
+  %arrayidx2 = getelementptr inbounds i32, ptr %c, i64 %iv
+  %1 = load i32, ptr %arrayidx2, align 4
+  %add = add nsw i32 %1, %0
+  %arrayidx4 = getelementptr inbounds i32, ptr %a, i64 %iv
+  store i32 %add, ptr %arrayidx4, align 4
+  %iv.next = add nuw nsw i64 %iv, 1
+  %exitcond.not = icmp eq i64 %iv.next, %N
+  br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup:
+  ret void
+}
diff --git a/llvm/test/Transforms/LoopVectorize/vplan-force-tail-with-evl.ll b/llvm/test/Transforms/LoopVectorize/vplan-force-tail-with-evl.ll
new file mode 100644
index 000000000000..f510d47d06e3
--- /dev/null
+++ b/llvm/test/Transforms/LoopVectorize/vplan-force-tail-with-evl.ll
@@ -0,0 +1,37 @@
+; REQUIRES: asserts
+
+; RUN: opt -passes=loop-vectorize -debug-only=loop-vectorize \
+; RUN: -force-tail-folding-style=data-with-evl -force-vector-width=4 \
+; RUN: -force-target-supports-scalable-vectors -scalable-vectorization=on \
+; RUN: -disable-output < %s 2>&1 | FileCheck --check-prefixes=NO-VP %s
+
+; RUN: opt -passes=loop-vectorize -debug-only=loop-vectorize \
+; RUN: -force-tail-folding-style=none \
+; RUN: -prefer-predicate-over-epilogue=predicate-dont-vectorize \
+; RUN: -force-target-supports-scalable-vectors -scalable-vectorization=on \
+; RUN: -disable-output < %s 2>&1 | FileCheck --check-prefixes=NO-VP %s
+
+; The target does not support predicated vectorization.
+define void @foo(ptr noalias %a, ptr noalias %b, ptr noalias %c, i64 %N) {
+; NO-VP-NOT: EXPLICIT-VECTOR-LENGTH-BASED-IV-PHI
+
+entry:
+  br label %for.body
+
+for.body:
+  %iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]
+  %arrayidx = getelementptr inbounds i32, ptr %b, i64 %iv
+  %0 = load i32, ptr %arrayidx, align 4
+  %arrayidx2 = getelementptr inbounds i32, ptr %c, i64 %iv
+  %1 = load i32, ptr %arrayidx2, align 4
+  %add = add nsw i32 %1, %0
+  %arrayidx4 = getelementptr inbounds i32, ptr %a, i64 %iv
+  store i32 %add, ptr %arrayidx4, align 4
+  %iv.next = add nuw nsw i64 %iv, 1
+  %exitcond.not = icmp eq i64 %iv.next, %N
+  br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup:
+  ret void
+}
+
diff --git a/llvm/test/Transforms/SLPVectorizer/AArch64/scalable-external-user.ll b/llvm/test/Transforms/SLPVectorizer/AArch64/scalable-external-user.ll
new file mode 100644
index 000000000000..5df6b8581dd7
--- /dev/null
+++ b/llvm/test/Transforms/SLPVectorizer/AArch64/scalable-external-user.ll
@@ -0,0 +1,28 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
+; RUN: opt < %s -passes=slp-vectorizer -S | FileCheck %s
+
+target datalayout = "e-m:e-i8:8:32-i16:16:32-i64:64-i128:128-n32:64-S128"
+target triple = "aarch64-unknown-linux-gnu"
+
+; Protect against a crash with scalable vector users
+
+define i1 @crash(i32 %a, i32 %b) {
+; CHECK-LABEL: @crash(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[CONV_I446:%.*]] = sext i32 [[A:%.*]] to i64
+; CHECK-NEXT:    [[CMP_I618870_NOT_NOT:%.*]] = icmp ult i64 0, [[CONV_I446]]
+; CHECK-NEXT:    [[CONV_I401:%.*]] = sext i32 [[B:%.*]] to i64
+; CHECK-NEXT:    [[CMP_I407876_NOT_NOT:%.*]] = icmp ult i64 0, [[CONV_I401]]
+; CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.whilelo.nxv2i1.i64(i64 0, i64 [[CONV_I401]])
+; CHECK-NEXT:    [[R:%.*]] = select i1 [[CMP_I618870_NOT_NOT]], i1 [[CMP_I407876_NOT_NOT]], i1 false
+; CHECK-NEXT:    ret i1 [[R]]
+;
+entry:
+  %conv.i446 = sext i32 %a to i64
+  %cmp.i618870.not.not = icmp ult i64 0, %conv.i446
+  %conv.i401 = sext i32 %b to i64
+  %cmp.i407876.not.not = icmp ult i64 0, %conv.i401
+  %0 = tail call <vscale x 2 x i1> @llvm.aarch64.sve.whilelo.nxv2i1.i64(i64 0, i64 %conv.i401)
+  %r = select i1 %cmp.i618870.not.not, i1 %cmp.i407876.not.not, i1 0
+  ret i1 %r
+}
diff --git a/llvm/test/Transforms/RemoveTraps/remove-traps.ll b/llvm/test/Transforms/lower-builtin-allow-check.ll
index 4853149f955b..a9f4e200c42c 100644
--- a/llvm/test/Transforms/RemoveTraps/remove-traps.ll
+++ b/llvm/test/Transforms/lower-builtin-allow-check.ll
@@ -1,18 +1,21 @@
 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 4
-; RUN: opt < %s -passes='function(remove-traps)' -S | FileCheck %s --check-prefixes=NOPROFILE
-; RUN: opt < %s -passes='function(remove-traps)' -remove-traps-random-rate=1 -S | FileCheck %s --check-prefixes=ALL
-; RUN: opt < %s -passes='require<profile-summary>,function(remove-traps)' -remove-traps-percentile-cutoff-hot=990000 -S | FileCheck %s --check-prefixes=HOT99
-; RUN: opt < %s -passes='require<profile-summary>,function(remove-traps)' -remove-traps-percentile-cutoff-hot=700000 -S | FileCheck %s --check-prefixes=HOT70
+; RUN: opt < %s -passes='function(lower-allow-check)' -S | FileCheck %s --check-prefixes=NOPROFILE
+; RUN: opt < %s -passes='function(lower-allow-check)' -lower-allow-check-random-rate=1 -S | FileCheck %s --check-prefixes=ALL
+; RUN: opt < %s -passes='require<profile-summary>,function(lower-allow-check)' -lower-allow-check-percentile-cutoff-hot=990000 -S | FileCheck %s --check-prefixes=HOT99
+; RUN: opt < %s -passes='require<profile-summary>,function(lower-allow-check)' -lower-allow-check-percentile-cutoff-hot=700000 -S | FileCheck %s --check-prefixes=HOT70
 
 target triple = "x86_64-pc-linux-gnu"
 
 declare void @llvm.ubsantrap(i8 immarg)
+declare i1 @llvm.allow.ubsan.check(i8 immarg)
 
 define dso_local noundef i32 @simple(ptr noundef readonly %0) {
 ; NOPROFILE-LABEL: define dso_local noundef i32 @simple(
 ; NOPROFILE-SAME: ptr noundef readonly [[TMP0:%.*]]) {
 ; NOPROFILE-NEXT:    [[TMP2:%.*]] = icmp eq ptr [[TMP0]], null
-; NOPROFILE-NEXT:    br i1 [[TMP2]], label [[TMP3:%.*]], label [[TMP4:%.*]]
+; NOPROFILE-NEXT:    [[HOT:%.*]] = xor i1 true, true
+; NOPROFILE-NEXT:    [[TMP6:%.*]] = or i1 [[TMP2]], [[HOT]]
+; NOPROFILE-NEXT:    br i1 [[TMP6]], label [[TMP3:%.*]], label [[TMP4:%.*]]
 ; NOPROFILE:       3:
 ; NOPROFILE-NEXT:    tail call void @llvm.ubsantrap(i8 22)
 ; NOPROFILE-NEXT:    unreachable
@@ -23,8 +26,11 @@ define dso_local noundef i32 @simple(ptr noundef readonly %0) {
 ; ALL-LABEL: define dso_local noundef i32 @simple(
 ; ALL-SAME: ptr noundef readonly [[TMP0:%.*]]) {
 ; ALL-NEXT:    [[TMP2:%.*]] = icmp eq ptr [[TMP0]], null
-; ALL-NEXT:    br i1 [[TMP2]], label [[TMP3:%.*]], label [[TMP4:%.*]]
+; ALL-NEXT:    [[HOT:%.*]] = xor i1 false, true
+; ALL-NEXT:    [[TMP6:%.*]] = or i1 [[TMP2]], [[HOT]]
+; ALL-NEXT:    br i1 [[TMP6]], label [[TMP3:%.*]], label [[TMP4:%.*]]
 ; ALL:       3:
+; ALL-NEXT:    tail call void @llvm.ubsantrap(i8 22)
 ; ALL-NEXT:    unreachable
 ; ALL:       4:
 ; ALL-NEXT:    [[TMP5:%.*]] = load i32, ptr [[TMP0]], align 4
@@ -33,7 +39,9 @@ define dso_local noundef i32 @simple(ptr noundef readonly %0) {
 ; HOT99-LABEL: define dso_local noundef i32 @simple(
 ; HOT99-SAME: ptr noundef readonly [[TMP0:%.*]]) {
 ; HOT99-NEXT:    [[TMP2:%.*]] = icmp eq ptr [[TMP0]], null
-; HOT99-NEXT:    br i1 [[TMP2]], label [[TMP3:%.*]], label [[TMP4:%.*]]
+; HOT99-NEXT:    [[HOT:%.*]] = xor i1 true, true
+; HOT99-NEXT:    [[TMP6:%.*]] = or i1 [[TMP2]], [[HOT]]
+; HOT99-NEXT:    br i1 [[TMP6]], label [[TMP3:%.*]], label [[TMP4:%.*]]
 ; HOT99:       3:
 ; HOT99-NEXT:    tail call void @llvm.ubsantrap(i8 22)
 ; HOT99-NEXT:    unreachable
@@ -44,7 +52,9 @@ define dso_local noundef i32 @simple(ptr noundef readonly %0) {
 ; HOT70-LABEL: define dso_local noundef i32 @simple(
 ; HOT70-SAME: ptr noundef readonly [[TMP0:%.*]]) {
 ; HOT70-NEXT:    [[TMP2:%.*]] = icmp eq ptr [[TMP0]], null
-; HOT70-NEXT:    br i1 [[TMP2]], label [[TMP3:%.*]], label [[TMP4:%.*]]
+; HOT70-NEXT:    [[HOT:%.*]] = xor i1 true, true
+; HOT70-NEXT:    [[TMP6:%.*]] = or i1 [[TMP2]], [[HOT]]
+; HOT70-NEXT:    br i1 [[TMP6]], label [[TMP3:%.*]], label [[TMP4:%.*]]
 ; HOT70:       3:
 ; HOT70-NEXT:    tail call void @llvm.ubsantrap(i8 22)
 ; HOT70-NEXT:    unreachable
@@ -52,7 +62,10 @@ define dso_local noundef i32 @simple(ptr noundef readonly %0) {
 ; HOT70-NEXT:    [[TMP5:%.*]] = load i32, ptr [[TMP0]], align 4
 ; HOT70-NEXT:    ret i32 [[TMP5]]
 ;
-  %2 = icmp eq ptr %0, null
+  %chk = icmp eq ptr %0, null
+  %allow = call i1 @llvm.allow.ubsan.check(i8 22)
+  %hot = xor i1 %allow, true
+  %2 = or i1 %chk, %hot
   br i1 %2, label %3, label %4
 
 3:
@@ -69,7 +82,9 @@ define dso_local noundef i32 @hot(ptr noundef readonly %0) !prof !36 {
 ; NOPROFILE-LABEL: define dso_local noundef i32 @hot(
 ; NOPROFILE-SAME: ptr noundef readonly [[TMP0:%.*]]) !prof [[PROF16:![0-9]+]] {
 ; NOPROFILE-NEXT:    [[TMP2:%.*]] = icmp eq ptr [[TMP0]], null
-; NOPROFILE-NEXT:    br i1 [[TMP2]], label [[TMP3:%.*]], label [[TMP4:%.*]]
+; NOPROFILE-NEXT:    [[HOT:%.*]] = xor i1 true, true
+; NOPROFILE-NEXT:    [[TMP6:%.*]] = or i1 [[TMP2]], [[HOT]]
+; NOPROFILE-NEXT:    br i1 [[TMP6]], label [[TMP3:%.*]], label [[TMP4:%.*]]
 ; NOPROFILE:       3:
 ; NOPROFILE-NEXT:    tail call void @llvm.ubsantrap(i8 22)
 ; NOPROFILE-NEXT:    unreachable
@@ -80,8 +95,11 @@ define dso_local noundef i32 @hot(ptr noundef readonly %0) !prof !36 {
 ; ALL-LABEL: define dso_local noundef i32 @hot(
 ; ALL-SAME: ptr noundef readonly [[TMP0:%.*]]) !prof [[PROF16:![0-9]+]] {
 ; ALL-NEXT:    [[TMP2:%.*]] = icmp eq ptr [[TMP0]], null
-; ALL-NEXT:    br i1 [[TMP2]], label [[TMP3:%.*]], label [[TMP4:%.*]]
+; ALL-NEXT:    [[HOT:%.*]] = xor i1 false, true
+; ALL-NEXT:    [[TMP6:%.*]] = or i1 [[TMP2]], [[HOT]]
+; ALL-NEXT:    br i1 [[TMP6]], label [[TMP3:%.*]], label [[TMP4:%.*]]
 ; ALL:       3:
+; ALL-NEXT:    tail call void @llvm.ubsantrap(i8 22)
 ; ALL-NEXT:    unreachable
 ; ALL:       4:
 ; ALL-NEXT:    [[TMP5:%.*]] = load i32, ptr [[TMP0]], align 4
@@ -90,8 +108,11 @@ define dso_local noundef i32 @hot(ptr noundef readonly %0) !prof !36 {
 ; HOT99-LABEL: define dso_local noundef i32 @hot(
 ; HOT99-SAME: ptr noundef readonly [[TMP0:%.*]]) !prof [[PROF16:![0-9]+]] {
 ; HOT99-NEXT:    [[TMP2:%.*]] = icmp eq ptr [[TMP0]], null
-; HOT99-NEXT:    br i1 [[TMP2]], label [[TMP3:%.*]], label [[TMP4:%.*]]
+; HOT99-NEXT:    [[HOT:%.*]] = xor i1 false, true
+; HOT99-NEXT:    [[TMP6:%.*]] = or i1 [[TMP2]], [[HOT]]
+; HOT99-NEXT:    br i1 [[TMP6]], label [[TMP3:%.*]], label [[TMP4:%.*]]
 ; HOT99:       3:
+; HOT99-NEXT:    tail call void @llvm.ubsantrap(i8 22)
 ; HOT99-NEXT:    unreachable
 ; HOT99:       4:
 ; HOT99-NEXT:    [[TMP5:%.*]] = load i32, ptr [[TMP0]], align 4
@@ -100,7 +121,9 @@ define dso_local noundef i32 @hot(ptr noundef readonly %0) !prof !36 {
 ; HOT70-LABEL: define dso_local noundef i32 @hot(
 ; HOT70-SAME: ptr noundef readonly [[TMP0:%.*]]) !prof [[PROF16:![0-9]+]] {
 ; HOT70-NEXT:    [[TMP2:%.*]] = icmp eq ptr [[TMP0]], null
-; HOT70-NEXT:    br i1 [[TMP2]], label [[TMP3:%.*]], label [[TMP4:%.*]]
+; HOT70-NEXT:    [[HOT:%.*]] = xor i1 true, true
+; HOT70-NEXT:    [[TMP6:%.*]] = or i1 [[TMP2]], [[HOT]]
+; HOT70-NEXT:    br i1 [[TMP6]], label [[TMP3:%.*]], label [[TMP4:%.*]]
 ; HOT70:       3:
 ; HOT70-NEXT:    tail call void @llvm.ubsantrap(i8 22)
 ; HOT70-NEXT:    unreachable
@@ -108,7 +131,10 @@ define dso_local noundef i32 @hot(ptr noundef readonly %0) !prof !36 {
 ; HOT70-NEXT:    [[TMP5:%.*]] = load i32, ptr [[TMP0]], align 4
 ; HOT70-NEXT:    ret i32 [[TMP5]]
 ;
-  %2 = icmp eq ptr %0, null
+  %chk = icmp eq ptr %0, null
+  %allow = call i1 @llvm.allow.ubsan.check(i8 22)
+  %hot = xor i1 %allow, true
+  %2 = or i1 %chk, %hot
   br i1 %2, label %3, label %4
 
 3:
@@ -124,7 +150,9 @@ define dso_local noundef i32 @veryHot(ptr noundef readonly %0) !prof !39 {
 ; NOPROFILE-LABEL: define dso_local noundef i32 @veryHot(
 ; NOPROFILE-SAME: ptr noundef readonly [[TMP0:%.*]]) !prof [[PROF17:![0-9]+]] {
 ; NOPROFILE-NEXT:    [[TMP2:%.*]] = icmp eq ptr [[TMP0]], null
-; NOPROFILE-NEXT:    br i1 [[TMP2]], label [[TMP3:%.*]], label [[TMP4:%.*]]
+; NOPROFILE-NEXT:    [[HOT:%.*]] = xor i1 true, true
+; NOPROFILE-NEXT:    [[TMP6:%.*]] = or i1 [[TMP2]], [[HOT]]
+; NOPROFILE-NEXT:    br i1 [[TMP6]], label [[TMP3:%.*]], label [[TMP4:%.*]]
 ; NOPROFILE:       3:
 ; NOPROFILE-NEXT:    tail call void @llvm.ubsantrap(i8 22)
 ; NOPROFILE-NEXT:    unreachable
@@ -135,8 +163,11 @@ define dso_local noundef i32 @veryHot(ptr noundef readonly %0) !prof !39 {
 ; ALL-LABEL: define dso_local noundef i32 @veryHot(
 ; ALL-SAME: ptr noundef readonly [[TMP0:%.*]]) !prof [[PROF17:![0-9]+]] {
 ; ALL-NEXT:    [[TMP2:%.*]] = icmp eq ptr [[TMP0]], null
-; ALL-NEXT:    br i1 [[TMP2]], label [[TMP3:%.*]], label [[TMP4:%.*]]
+; ALL-NEXT:    [[HOT:%.*]] = xor i1 false, true
+; ALL-NEXT:    [[TMP6:%.*]] = or i1 [[TMP2]], [[HOT]]
+; ALL-NEXT:    br i1 [[TMP6]], label [[TMP3:%.*]], label [[TMP4:%.*]]
 ; ALL:       3:
+; ALL-NEXT:    tail call void @llvm.ubsantrap(i8 22)
 ; ALL-NEXT:    unreachable
 ; ALL:       4:
 ; ALL-NEXT:    [[TMP5:%.*]] = load i32, ptr [[TMP0]], align 4
@@ -145,8 +176,11 @@ define dso_local noundef i32 @veryHot(ptr noundef readonly %0) !prof !39 {
 ; HOT99-LABEL: define dso_local noundef i32 @veryHot(
 ; HOT99-SAME: ptr noundef readonly [[TMP0:%.*]]) !prof [[PROF17:![0-9]+]] {
 ; HOT99-NEXT:    [[TMP2:%.*]] = icmp eq ptr [[TMP0]], null
-; HOT99-NEXT:    br i1 [[TMP2]], label [[TMP3:%.*]], label [[TMP4:%.*]]
+; HOT99-NEXT:    [[HOT:%.*]] = xor i1 false, true
+; HOT99-NEXT:    [[TMP6:%.*]] = or i1 [[TMP2]], [[HOT]]
+; HOT99-NEXT:    br i1 [[TMP6]], label [[TMP3:%.*]], label [[TMP4:%.*]]
 ; HOT99:       3:
+; HOT99-NEXT:    tail call void @llvm.ubsantrap(i8 22)
 ; HOT99-NEXT:    unreachable
 ; HOT99:       4:
 ; HOT99-NEXT:    [[TMP5:%.*]] = load i32, ptr [[TMP0]], align 4
@@ -155,14 +189,20 @@ define dso_local noundef i32 @veryHot(ptr noundef readonly %0) !prof !39 {
 ; HOT70-LABEL: define dso_local noundef i32 @veryHot(
 ; HOT70-SAME: ptr noundef readonly [[TMP0:%.*]]) !prof [[PROF17:![0-9]+]] {
 ; HOT70-NEXT:    [[TMP2:%.*]] = icmp eq ptr [[TMP0]], null
-; HOT70-NEXT:    br i1 [[TMP2]], label [[TMP3:%.*]], label [[TMP4:%.*]]
+; HOT70-NEXT:    [[HOT:%.*]] = xor i1 false, true
+; HOT70-NEXT:    [[TMP6:%.*]] = or i1 [[TMP2]], [[HOT]]
+; HOT70-NEXT:    br i1 [[TMP6]], label [[TMP3:%.*]], label [[TMP4:%.*]]
 ; HOT70:       3:
+; HOT70-NEXT:    tail call void @llvm.ubsantrap(i8 22)
 ; HOT70-NEXT:    unreachable
 ; HOT70:       4:
 ; HOT70-NEXT:    [[TMP5:%.*]] = load i32, ptr [[TMP0]], align 4
 ; HOT70-NEXT:    ret i32 [[TMP5]]
 ;
-  %2 = icmp eq ptr %0, null
+  %chk = icmp eq ptr %0, null
+  %allow = call i1 @llvm.allow.ubsan.check(i8 22)
+  %hot = xor i1 %allow, true
+  %2 = or i1 %chk, %hot
   br i1 %2, label %3, label %4
 
 3:
@@ -182,7 +222,9 @@ define dso_local noundef i32 @branchColdFnHot(i32 noundef %0, ptr noundef readon
 ; NOPROFILE-NEXT:    br i1 [[TMP3]], label [[TMP9:%.*]], label [[TMP4:%.*]], !prof [[PROF18:![0-9]+]]
 ; NOPROFILE:       4:
 ; NOPROFILE-NEXT:    [[TMP5:%.*]] = icmp eq ptr [[TMP1]], null
-; NOPROFILE-NEXT:    br i1 [[TMP5]], label [[TMP6:%.*]], label [[TMP7:%.*]]
+; NOPROFILE-NEXT:    [[HOT:%.*]] = xor i1 true, true
+; NOPROFILE-NEXT:    [[TMP11:%.*]] = or i1 [[TMP5]], [[HOT]]
+; NOPROFILE-NEXT:    br i1 [[TMP11]], label [[TMP6:%.*]], label [[TMP7:%.*]]
 ; NOPROFILE:       6:
 ; NOPROFILE-NEXT:    tail call void @llvm.ubsantrap(i8 22)
 ; NOPROFILE-NEXT:    unreachable
@@ -199,8 +241,11 @@ define dso_local noundef i32 @branchColdFnHot(i32 noundef %0, ptr noundef readon
 ; ALL-NEXT:    br i1 [[TMP3]], label [[TMP9:%.*]], label [[TMP4:%.*]], !prof [[PROF18:![0-9]+]]
 ; ALL:       4:
 ; ALL-NEXT:    [[TMP5:%.*]] = icmp eq ptr [[TMP1]], null
-; ALL-NEXT:    br i1 [[TMP5]], label [[TMP6:%.*]], label [[TMP7:%.*]]
+; ALL-NEXT:    [[HOT:%.*]] = xor i1 false, true
+; ALL-NEXT:    [[TMP11:%.*]] = or i1 [[TMP5]], [[HOT]]
+; ALL-NEXT:    br i1 [[TMP11]], label [[TMP6:%.*]], label [[TMP7:%.*]]
 ; ALL:       6:
+; ALL-NEXT:    tail call void @llvm.ubsantrap(i8 22)
 ; ALL-NEXT:    unreachable
 ; ALL:       7:
 ; ALL-NEXT:    [[TMP8:%.*]] = load i32, ptr [[TMP1]], align 4
@@ -215,7 +260,9 @@ define dso_local noundef i32 @branchColdFnHot(i32 noundef %0, ptr noundef readon
 ; HOT99-NEXT:    br i1 [[TMP3]], label [[TMP9:%.*]], label [[TMP4:%.*]], !prof [[PROF18:![0-9]+]]
 ; HOT99:       4:
 ; HOT99-NEXT:    [[TMP5:%.*]] = icmp eq ptr [[TMP1]], null
-; HOT99-NEXT:    br i1 [[TMP5]], label [[TMP6:%.*]], label [[TMP7:%.*]]
+; HOT99-NEXT:    [[HOT:%.*]] = xor i1 true, true
+; HOT99-NEXT:    [[TMP11:%.*]] = or i1 [[TMP5]], [[HOT]]
+; HOT99-NEXT:    br i1 [[TMP11]], label [[TMP6:%.*]], label [[TMP7:%.*]]
 ; HOT99:       6:
 ; HOT99-NEXT:    tail call void @llvm.ubsantrap(i8 22)
 ; HOT99-NEXT:    unreachable
@@ -232,7 +279,9 @@ define dso_local noundef i32 @branchColdFnHot(i32 noundef %0, ptr noundef readon
 ; HOT70-NEXT:    br i1 [[TMP3]], label [[TMP9:%.*]], label [[TMP4:%.*]], !prof [[PROF18:![0-9]+]]
 ; HOT70:       4:
 ; HOT70-NEXT:    [[TMP5:%.*]] = icmp eq ptr [[TMP1]], null
-; HOT70-NEXT:    br i1 [[TMP5]], label [[TMP6:%.*]], label [[TMP7:%.*]]
+; HOT70-NEXT:    [[HOT:%.*]] = xor i1 true, true
+; HOT70-NEXT:    [[TMP11:%.*]] = or i1 [[TMP5]], [[HOT]]
+; HOT70-NEXT:    br i1 [[TMP11]], label [[TMP6:%.*]], label [[TMP7:%.*]]
 ; HOT70:       6:
 ; HOT70-NEXT:    tail call void @llvm.ubsantrap(i8 22)
 ; HOT70-NEXT:    unreachable
@@ -247,7 +296,10 @@ define dso_local noundef i32 @branchColdFnHot(i32 noundef %0, ptr noundef readon
   br i1 %3, label %9, label %4, !prof !38
 
 4:
-  %5 = icmp eq ptr %1, null
+  %chk = icmp eq ptr %1, null
+  %allow = call i1 @llvm.allow.ubsan.check(i8 22)
+  %hot = xor i1 %allow, true
+  %5 = or i1 %chk, %hot
   br i1 %5, label %6, label %7
 
 6:
@@ -270,7 +322,9 @@ define dso_local noundef i32 @branchHotFnCold(i32 noundef %0, ptr noundef readon
 ; NOPROFILE-NEXT:    br i1 [[TMP3]], label [[TMP9:%.*]], label [[TMP4:%.*]], !prof [[PROF19:![0-9]+]]
 ; NOPROFILE:       4:
 ; NOPROFILE-NEXT:    [[TMP5:%.*]] = icmp eq ptr [[TMP1]], null
-; NOPROFILE-NEXT:    br i1 [[TMP5]], label [[TMP6:%.*]], label [[TMP7:%.*]]
+; NOPROFILE-NEXT:    [[HOT:%.*]] = xor i1 true, true
+; NOPROFILE-NEXT:    [[TMP11:%.*]] = or i1 [[TMP5]], [[HOT]]
+; NOPROFILE-NEXT:    br i1 [[TMP11]], label [[TMP6:%.*]], label [[TMP7:%.*]]
 ; NOPROFILE:       6:
 ; NOPROFILE-NEXT:    tail call void @llvm.ubsantrap(i8 22)
 ; NOPROFILE-NEXT:    unreachable
@@ -287,8 +341,11 @@ define dso_local noundef i32 @branchHotFnCold(i32 noundef %0, ptr noundef readon
 ; ALL-NEXT:    br i1 [[TMP3]], label [[TMP9:%.*]], label [[TMP4:%.*]], !prof [[PROF19:![0-9]+]]
 ; ALL:       4:
 ; ALL-NEXT:    [[TMP5:%.*]] = icmp eq ptr [[TMP1]], null
-; ALL-NEXT:    br i1 [[TMP5]], label [[TMP6:%.*]], label [[TMP7:%.*]]
+; ALL-NEXT:    [[HOT:%.*]] = xor i1 false, true
+; ALL-NEXT:    [[TMP11:%.*]] = or i1 [[TMP5]], [[HOT]]
+; ALL-NEXT:    br i1 [[TMP11]], label [[TMP6:%.*]], label [[TMP7:%.*]]
 ; ALL:       6:
+; ALL-NEXT:    tail call void @llvm.ubsantrap(i8 22)
 ; ALL-NEXT:    unreachable
 ; ALL:       7:
 ; ALL-NEXT:    [[TMP8:%.*]] = load i32, ptr [[TMP1]], align 4
@@ -303,8 +360,11 @@ define dso_local noundef i32 @branchHotFnCold(i32 noundef %0, ptr noundef readon
 ; HOT99-NEXT:    br i1 [[TMP3]], label [[TMP9:%.*]], label [[TMP4:%.*]], !prof [[PROF19:![0-9]+]]
 ; HOT99:       4:
 ; HOT99-NEXT:    [[TMP5:%.*]] = icmp eq ptr [[TMP1]], null
-; HOT99-NEXT:    br i1 [[TMP5]], label [[TMP6:%.*]], label [[TMP7:%.*]]
+; HOT99-NEXT:    [[HOT:%.*]] = xor i1 false, true
+; HOT99-NEXT:    [[TMP11:%.*]] = or i1 [[TMP5]], [[HOT]]
+; HOT99-NEXT:    br i1 [[TMP11]], label [[TMP6:%.*]], label [[TMP7:%.*]]
 ; HOT99:       6:
+; HOT99-NEXT:    tail call void @llvm.ubsantrap(i8 22)
 ; HOT99-NEXT:    unreachable
 ; HOT99:       7:
 ; HOT99-NEXT:    [[TMP8:%.*]] = load i32, ptr [[TMP1]], align 4
@@ -319,7 +379,9 @@ define dso_local noundef i32 @branchHotFnCold(i32 noundef %0, ptr noundef readon
 ; HOT70-NEXT:    br i1 [[TMP3]], label [[TMP9:%.*]], label [[TMP4:%.*]], !prof [[PROF19:![0-9]+]]
 ; HOT70:       4:
 ; HOT70-NEXT:    [[TMP5:%.*]] = icmp eq ptr [[TMP1]], null
-; HOT70-NEXT:    br i1 [[TMP5]], label [[TMP6:%.*]], label [[TMP7:%.*]]
+; HOT70-NEXT:    [[HOT:%.*]] = xor i1 true, true
+; HOT70-NEXT:    [[TMP11:%.*]] = or i1 [[TMP5]], [[HOT]]
+; HOT70-NEXT:    br i1 [[TMP11]], label [[TMP6:%.*]], label [[TMP7:%.*]]
 ; HOT70:       6:
 ; HOT70-NEXT:    tail call void @llvm.ubsantrap(i8 22)
 ; HOT70-NEXT:    unreachable
@@ -334,7 +396,10 @@ define dso_local noundef i32 @branchHotFnCold(i32 noundef %0, ptr noundef readon
   br i1 %3, label %9, label %4, !prof !37
 
 4:
-  %5 = icmp eq ptr %1, null
+  %chk = icmp eq ptr %1, null
+  %allow = call i1 @llvm.allow.ubsan.check(i8 22)
+  %hot = xor i1 %allow, true
+  %5 = or i1 %chk, %hot
   br i1 %5, label %6, label %7
 
 6:
diff --git a/llvm/test/tools/llvm-objcopy/ELF/compress-sections-within-segment.s b/llvm/test/tools/llvm-objcopy/ELF/compress-sections-within-segment.s
deleted file mode 100644
index 064ffcadc12e..000000000000
--- a/llvm/test/tools/llvm-objcopy/ELF/compress-sections-within-segment.s
+++ /dev/null
@@ -1,38 +0,0 @@
-## Disallow (de)compression for sections within a segment as they are
-## effectively immutable.
-# RUN: rm -rf %t && mkdir %t && cd %t
-# RUN: yaml2obj %s -o a
-# RUN: not llvm-objcopy a /dev/null --compress-sections .text=zlib 2>&1 | FileCheck %s --implicit-check-not=error:
-
-# CHECK: error: 'a': section '.text' within a segment cannot be (de)compressed
-
-# RUN: not llvm-objcopy a /dev/null --compress-sections foo=none 2>&1 | FileCheck %s --check-prefix=CHECK2 --implicit-check-not=error:
-
-# CHECK2: error: 'a': section 'foo' within a segment cannot be (de)compressed
-
-## There is an error even if 'foo' is already compressed with zlib.
-# RUN: not llvm-objcopy a /dev/null --compress-sections foo=zlib 2>&1 | FileCheck %s --check-prefix=CHECK3 --implicit-check-not=error:
-
-# CHECK3: error: 'a': section 'foo' within a segment cannot be (de)compressed
-
---- !ELF
-FileHeader:
-  Class:      ELFCLASS64
-  Data:       ELFDATA2LSB
-  Type:       ET_EXEC
-  Machine:    EM_X86_64
-ProgramHeaders:
-  - Type:     PT_LOAD
-    FirstSec: .text
-    LastSec:  foo
-    Align:    0x1000
-    Offset:   0x1000
-Sections:
-  - Name:     .text
-    Type:     SHT_PROGBITS
-    Offset:   0x1000
-    Content:  C3
-  - Name:     foo
-    Type:     SHT_PROGBITS
-    Flags:    [ SHF_COMPRESSED ]
-    Content:  010000000000000040000000000000000100000000000000789cd36280002d3269002f800151
diff --git a/llvm/test/tools/llvm-objcopy/ELF/compress-sections.s b/llvm/test/tools/llvm-objcopy/ELF/compress-sections.s
deleted file mode 100644
index e6fa86068a1a..000000000000
--- a/llvm/test/tools/llvm-objcopy/ELF/compress-sections.s
+++ /dev/null
@@ -1,128 +0,0 @@
-# REQUIRES: x86-registered-target, zlib, zstd
-
-# RUN: rm -rf %t && mkdir %t && cd %t
-# RUN: llvm-mc -filetype=obj -triple=x86_64 %s -o a.o
-## '*0=none' wins because it is the last. '*0' sections are decompressed (if originally compressed) or kept unchanged (if uncompressed).
-## No section is named 'nomatch'. The third option is a no-op.
-# RUN: llvm-objcopy a.o out --compress-sections='*0=zlib' --compress-sections '*0=none' --compress-sections 'nomatch=none' 2>&1 | count 0
-# RUN: llvm-readelf -S out | FileCheck %s --check-prefix=CHECK1
-
-# CHECK1:      Name           Type          Address     Off      Size     ES Flg Lk Inf Al
-# CHECK1:      .text          PROGBITS [[#%x,TEXT:]]    [[#%x,]] [[#%x,]] 00 AX   0   0  4
-# CHECK1:      foo0           PROGBITS [[#%x,FOO0:]]    [[#%x,]] [[#%x,]] 00 A    0   0  8
-# CHECK1-NEXT: .relafoo0      RELA     [[#%x,]]         [[#%x,]] [[#%x,]] 18  I  11   3  8
-# CHECK1-NEXT: foo1           PROGBITS [[#%x,FOO1:]]    [[#%x,]] [[#%x,]] 00 A    0   0  8
-# CHECK1-NEXT: .relafoo1      RELA     [[#%x,]]         [[#%x,]] [[#%x,]] 18  I  11   5  8
-# CHECK1:      nonalloc0      PROGBITS 0000000000000000 [[#%x,]] [[#%x,]] 00      0   0  8
-# CHECK1-NEXT: .relanonalloc0 RELA     [[#%x,]]         [[#%x,]] [[#%x,]] 18  I  11   7  8
-# CHECK1-NEXT: nonalloc1      PROGBITS 0000000000000000 [[#%x,]] [[#%x,]] 00      0   0  8
-# CHECK1-NEXT: .debug_str     PROGBITS 0000000000000000 [[#%x,]] [[#%x,]] 01 MS   0   0  1
-
-## Mixing zlib and zstd.
-# RUN: llvm-objcopy a.o out2 --compress-sections '*c0=zlib' --compress-sections .debug_str=zstd
-# RUN: llvm-readelf -Sr -x nonalloc0 -x .debug_str out2 2>&1 | FileCheck %s --check-prefix=CHECK2
-# RUN: llvm-readelf -z -x nonalloc0 -x .debug_str out2 | FileCheck %s --check-prefix=CHECK2DE
-
-# CHECK2:      Name           Type          Address     Off      Size     ES Flg Lk Inf Al
-# CHECK2:      .text          PROGBITS [[#%x,TEXT:]]    [[#%x,]] [[#%x,]] 00 AX   0   0  4
-# CHECK2:      foo0           PROGBITS [[#%x,FOO0:]]    [[#%x,]] [[#%x,]] 00 A    0   0  8
-# CHECK2-NEXT: .relafoo0      RELA     [[#%x,]]         [[#%x,]] [[#%x,]] 18  I  11   3  8
-# CHECK2-NEXT: foo1           PROGBITS [[#%x,FOO1:]]    [[#%x,]] [[#%x,]] 00 A    0   0  8
-# CHECK2-NEXT: .relafoo1      RELA     [[#%x,]]         [[#%x,]] [[#%x,]] 18  I  11   5  8
-# CHECK2:      nonalloc0      PROGBITS 0000000000000000 [[#%x,]] [[#%x,]] 00   C  0   0  8
-# CHECK2-NEXT: .relanonalloc0 RELA     [[#%x,]]         [[#%x,]] [[#%x,]] 18  IC 11   7  8
-# CHECK2-NEXT: nonalloc1      PROGBITS 0000000000000000 [[#%x,]] [[#%x,]] 00      0   0  8
-# CHECK2-NEXT: .debug_str     PROGBITS 0000000000000000 [[#%x,]] [[#%x,]] 01 MSC  0   0  8
-
-## llvm-readelf -r doesn't support SHF_COMPRESSED SHT_RELA.
-# CHECK2: warning: {{.*}}: unable to read relocations from SHT_RELA section with index 8: section [index 8] has an invalid sh_size ([[#]]) which is not a multiple of its sh_entsize (24)
-
-# CHECK2:      Hex dump of section 'nonalloc0':
-## zlib with ch_size=0x10
-# CHECK2-NEXT: 01000000 00000000 10000000 00000000
-# CHECK2-NEXT: 08000000 00000000 {{.*}}
-# CHECK2:      Hex dump of section '.debug_str':
-## zstd with ch_size=0x38
-# CHECK2-NEXT: 02000000 00000000 38000000 00000000
-# CHECK2-NEXT: 01000000 00000000 {{.*}}
-
-# CHECK2DE:       Hex dump of section 'nonalloc0':
-# CHECK2DE-NEXT:  0x00000000 00000000 00000000 00000000 00000000 ................
-# CHECK2DE-EMPTY:
-# CHECK2DE-NEXT:  Hex dump of section '.debug_str':
-# CHECK2DE-NEXT:  0x00000000 41414141 41414141 41414141 41414141 AAAAAAAAAAAAAAAA
-
-## --decompress-debug-sections takes precedence, even if it is before --compress-sections.
-# RUN: llvm-objcopy a.o out3 --decompress-debug-sections --compress-sections .debug_str=zstd
-# RUN: llvm-readelf -S out3 | FileCheck %s --check-prefix=CHECK3
-
-# CHECK3:      .debug_str PROGBITS 0000000000000000 [[#%x,]] [[#%x,]] 01 MS   0   0  1
-
-# RUN: llvm-objcopy a.o out4 --compress-sections '*0=zlib'
-# RUN: llvm-readelf -S out4 | FileCheck %s --check-prefix=CHECK4
-
-# CHECK4:      Name           Type          Address     Off      Size     ES Flg Lk Inf Al
-# CHECK4:      .text          PROGBITS [[#%x,TEXT:]]    [[#%x,]] [[#%x,]] 00  AX  0   0  4
-# CHECK4:      foo0           PROGBITS [[#%x,FOO0:]]    [[#%x,]] [[#%x,]] 00  AC  0   0  8
-# CHECK4-NEXT: .relafoo0      RELA     [[#%x,]]         [[#%x,]] [[#%x,]] 18  IC 11   3  8
-# CHECK4-NEXT: foo1           PROGBITS [[#%x,FOO1:]]    [[#%x,]] [[#%x,]] 00   A  0   0  8
-# CHECK4-NEXT: .relafoo1      RELA     [[#%x,]]         [[#%x,]] [[#%x,]] 18   I 11   5  8
-# CHECK4:      nonalloc0      PROGBITS 0000000000000000 [[#%x,]] [[#%x,]] 00   C  0   0  8
-# CHECK4-NEXT: .relanonalloc0 RELA     [[#%x,]]         [[#%x,]] [[#%x,]] 18  IC 11   7  8
-# CHECK4-NEXT: nonalloc1      PROGBITS 0000000000000000 [[#%x,]] [[#%x,]] 00      0   0  8
-# CHECK4-NEXT: .debug_str     PROGBITS 0000000000000000 [[#%x,]] [[#%x,]] 01  MS  0   0  1
-
-## If a section is already compressed, compression request for another format is ignored.
-# RUN: llvm-objcopy a.o out5 --compress-sections 'nonalloc0=zlib'
-# RUN: llvm-readelf -x nonalloc0 out5 | FileCheck %s --check-prefix=CHECK5
-# RUN: llvm-objcopy out5 out5a --compress-sections 'nonalloc0=zstd'
-# RUN: cmp out5 out5a
-
-# CHECK5:      Hex dump of section 'nonalloc0':
-## zlib with ch_size=0x10
-# CHECK5-NEXT: 01000000 00000000 10000000 00000000
-# CHECK5-NEXT: 08000000 00000000 {{.*}}
-
-# RUN: not llvm-objcopy --compress-sections=foo a.o out 2>&1 | \
-# RUN:   FileCheck %s --check-prefix=ERR1 --implicit-check-not=error:
-# ERR1:      error: --compress-sections: parse error, not 'section-glob=[none|zlib|zstd]'
-
-# RUN: llvm-objcopy --compress-sections 'a[=zlib' a.o out 2>&1 | \
-# RUN:   FileCheck %s --check-prefix=ERR2 --implicit-check-not=error:
-# ERR2:      warning: invalid glob pattern, unmatched '['
-
-# RUN: not llvm-objcopy a.o out --compress-sections='.debug*=zlib-gabi' --compress-sections='.debug*=' 2>&1 | \
-# RUN:   FileCheck -check-prefix=ERR3 %s
-# ERR3:      error: invalid or unsupported --compress-sections format: .debug*=zlib-gabi
-
-# RUN: not llvm-objcopy a.o out --compress-sections='!.debug*=zlib' 2>&1 | \
-# RUN:   FileCheck -check-prefix=ERR4 %s
-# ERR4:      error: --compress-sections: negative pattern is unsupported
-
-.globl _start
-_start:
-  ret
-
-.section foo0,"a"
-.balign 8
-.quad .text-.
-.quad .text-.
-.section foo1,"a"
-.balign 8
-.quad .text-.
-.quad .text-.
-.section nonalloc0,""
-.balign 8
-.quad .text+1
-.quad .text+2
-sym0:
-.section nonalloc1,""
-.balign 8
-.quad 42
-sym1:
-
-.section .debug_str,"MS",@progbits,1
-.Linfo_string0:
-  .asciz "AAAAAAAAAAAAAAAAAAAAAAAAAAA"
-.Linfo_string1:
-  .asciz "BBBBBBBBBBBBBBBBBBBBBBBBBBB"
diff --git a/llvm/test/tools/llvm-objcopy/ELF/decompress-sections.test b/llvm/test/tools/llvm-objcopy/ELF/decompress-sections.test
index d9f4f3809c4d..4258ddbe66a3 100644
--- a/llvm/test/tools/llvm-objcopy/ELF/decompress-sections.test
+++ b/llvm/test/tools/llvm-objcopy/ELF/decompress-sections.test
@@ -4,8 +4,6 @@
 # RUN: yaml2obj %s -o %t
 # RUN: llvm-objcopy --decompress-debug-sections %t %t.de
 # RUN: llvm-readelf -S %t.de | FileCheck %s
-# RUN: llvm-objcopy --compress-sections '*nonalloc=none' --compress-sections .debugx=none %t %t.1.de
-# RUN: cmp %t.de %t.1.de
 
 # CHECK:        Name              Type            Address          Off      Size     ES Flg Lk Inf Al
 # CHECK:        .debug_alloc      PROGBITS        0000000000000000 [[#%x,]] [[#%x,]] 00  AC  0   0  0
@@ -13,33 +11,6 @@
 # CHECK-NEXT:   .debugx           PROGBITS        0000000000000000 [[#%x,]] [[#%x,]] 00      0   0  1
 # CHECK-NEXT:   nodebug           PROGBITS        0000000000000000 [[#%x,]] [[#%x,]] 00   C  0   0  0
 
-# RUN: llvm-objcopy --compress-sections '.debug*=none' %t %t2.de
-# RUN: llvm-readelf -S -x .debug_alloc -x .debug_nonalloc -x .debugx %t2.de | FileCheck %s --check-prefix=CHECK2
-
-# CHECK2:        Name              Type            Address          Off      Size     ES Flg Lk Inf Al
-# CHECK2:        .debug_alloc      PROGBITS        0000000000000000 [[#%x,]] [[#%x,]] 00  A   0   0  1
-# CHECK2-NEXT:   .debug_nonalloc   PROGBITS        0000000000000000 [[#%x,]] [[#%x,]] 00      0   0  1
-# CHECK2-NEXT:   .debugx           PROGBITS        0000000000000000 [[#%x,]] [[#%x,]] 00      0   0  1
-# CHECK2-NEXT:   nodebug           PROGBITS        0000000000000000 [[#%x,]] [[#%x,]] 00   C  0   0  0
-
-# CHECK2:       Hex dump of section '.debug_alloc':
-# CHECK2-NEXT:  0x00000000 2a000000 00000000 2a000000 00000000 *.......*.......
-# CHECK2-NEXT:  0x00000010 2a000000 00000000 2a000000 00000000 *.......*.......
-# CHECK2-NEXT:  0x00000020 2a000000 00000000 2a000000 00000000 *.......*.......
-# CHECK2-NEXT:  0x00000030 2a000000 00000000 2a000000 00000000 *.......*.......
-# CHECK2-EMPTY:
-# CHECK2:       Hex dump of section '.debug_nonalloc':
-# CHECK2-NEXT:  0x00000000 2a000000 00000000 2a000000 00000000 *.......*.......
-# CHECK2-NEXT:  0x00000010 2a000000 00000000 2a000000 00000000 *.......*.......
-# CHECK2-NEXT:  0x00000020 2a000000 00000000 2a000000 00000000 *.......*.......
-# CHECK2-NEXT:  0x00000030 2a000000 00000000 2a000000 00000000 *.......*.......
-# CHECK2-EMPTY:
-# CHECK2-NEXT:  Hex dump of section '.debugx':
-# CHECK2-NEXT:  0x00000000 2a000000 00000000 2a000000 00000000 *.......*.......
-# CHECK2-NEXT:  0x00000010 2a000000 00000000 2a000000 00000000 *.......*.......
-# CHECK2-NEXT:  0x00000020 2a000000 00000000 2a000000 00000000 *.......*.......
-# CHECK2-NEXT:  0x00000030 2a000000 00000000 2a000000 00000000 *.......*.......
-
 --- !ELF
 FileHeader:
   Class:   ELFCLASS64
diff --git a/llvm/test/tools/llvm-ranlib/help-message.test b/llvm/test/tools/llvm-ranlib/help-message.test
index 8d8824ac4612..97212aa035b3 100644
--- a/llvm/test/tools/llvm-ranlib/help-message.test
+++ b/llvm/test/tools/llvm-ranlib/help-message.test
@@ -5,13 +5,22 @@
 # RUN: llvm-ranlib -help | FileCheck %s --check-prefix=HELP
 # RUN: llvm-ranlib --help | FileCheck %s --check-prefix=HELP
 # RUN: llvm-ranlib --version | FileCheck %s --check-prefix=VERSION
-# RUN: llvm-ranlib -version | FileCheck %s --check-prefix=VERSION
-# RUN: llvm-ranlib -v | FileCheck %s --check-prefix=VERSION
+# RUN: llvm-ranlib -V | FileCheck %s --check-prefix=VERSION
 
 ## Also check combined options (first -h/-v flag wins)
 # RUN: llvm-ranlib -Dh | FileCheck %s --check-prefix=HELP
-# RUN: llvm-ranlib -Dvh | FileCheck %s --check-prefix=VERSION
-# RUN: llvm-ranlib -Dhv | FileCheck %s --check-prefix=HELP
+# RUN: llvm-ranlib -DVh | FileCheck %s --check-prefix=VERSION
+# RUN: llvm-ranlib -DhV | FileCheck %s --check-prefix=HELP
 
 # HELP: USAGE: llvm-ranlib
 # VERSION: version
+
+## -v enables verbose output in BSD ranlib and GNU ar but is another alias
+## for --version in GNU ranlib. Reject -v.
+# RUN: not llvm-ranlib -v 2>&1 | FileCheck %s --check-prefix=ERR1
+# RUN: not llvm-ranlib -version 2>&1 | FileCheck %s --check-prefix=ERR2
+# RUN: not llvm-ranlib -Dvh 2>&1 | FileCheck %s --check-prefix=ERR3
+
+# ERR1: error: Invalid option: '-v'
+# ERR2: error: Invalid option: '-version'
+# ERR3: error: Invalid option: '-vh'
diff --git a/llvm/tools/llvm-ar/llvm-ar.cpp b/llvm/tools/llvm-ar/llvm-ar.cpp
index 294b8531b08f..3b842b76d5c8 100644
--- a/llvm/tools/llvm-ar/llvm-ar.cpp
+++ b/llvm/tools/llvm-ar/llvm-ar.cpp
@@ -65,7 +65,7 @@ static void printRanLibHelp(StringRef ToolName) {
          << "USAGE: " + ToolName + " archive...\n\n"
          << "OPTIONS:\n"
          << "  -h --help             - Display available options\n"
-         << "  -v --version          - Display the version of this program\n"
+         << "  -V --version          - Display the version of this program\n"
          << "  -D                    - Use zero for timestamps and uids/gids "
             "(default)\n"
          << "  -U                    - Use actual timestamps and uids/gids\n"
@@ -1439,7 +1439,7 @@ static int ranlib_main(int argc, char **argv) {
         } else if (arg.front() == 'h') {
           printHelpMessage();
           return 0;
-        } else if (arg.front() == 'v') {
+        } else if (arg.front() == 'V') {
           cl::PrintVersionMessage();
           return 0;
         } else if (arg.front() == 'X') {
diff --git a/llvm/tools/llvm-link/llvm-link.cpp b/llvm/tools/llvm-link/llvm-link.cpp
index 9049cb5e8580..8f80cc26f316 100644
--- a/llvm/tools/llvm-link/llvm-link.cpp
+++ b/llvm/tools/llvm-link/llvm-link.cpp
@@ -135,6 +135,7 @@ static cl::opt<bool> TryUseNewDbgInfoFormat(
     cl::init(false));
 
 extern cl::opt<bool> UseNewDbgInfoFormat;
+extern cl::opt<cl::boolOrDefault> PreserveInputDbgFormat;
 
 extern cl::opt<cl::boolOrDefault> LoadBitcodeIntoNewDbgInfoFormat;
 
@@ -492,6 +493,10 @@ int main(int argc, char **argv) {
     // Turn the new debug-info format on.
     UseNewDbgInfoFormat = true;
   }
+  // Since llvm-link collects multiple IR modules together, for simplicity's
+  // sake we disable the "PreserveInputDbgFormat" flag to enforce a single
+  // debug info format.
+  PreserveInputDbgFormat = cl::boolOrDefault::BOU_FALSE;
 
   LLVMContext Context;
   Context.setDiagnosticHandler(std::make_unique<LLVMLinkDiagnosticHandler>(),
diff --git a/llvm/tools/llvm-lto/llvm-lto.cpp b/llvm/tools/llvm-lto/llvm-lto.cpp
index 3c452b650cee..f310097eec63 100644
--- a/llvm/tools/llvm-lto/llvm-lto.cpp
+++ b/llvm/tools/llvm-lto/llvm-lto.cpp
@@ -271,6 +271,7 @@ static cl::opt<bool> TryUseNewDbgInfoFormat(
 
 extern cl::opt<bool> UseNewDbgInfoFormat;
 extern cl::opt<cl::boolOrDefault> LoadBitcodeIntoNewDbgInfoFormat;
+extern cl::opt<cl::boolOrDefault> PreserveInputDbgFormat;
 
 namespace {
 
@@ -954,6 +955,10 @@ int main(int argc, char **argv) {
     // Turn the new debug-info format on.
     UseNewDbgInfoFormat = true;
   }
+  // Since llvm-lto collects multiple IR modules together, for simplicity's sake
+  // we disable the "PreserveInputDbgFormat" flag to enforce a single debug info
+  // format.
+  PreserveInputDbgFormat = cl::boolOrDefault::BOU_FALSE;
 
   if (OptLevel < '0' || OptLevel > '3')
     error("optimization level must be between 0 and 3");
diff --git a/llvm/tools/llvm-lto2/llvm-lto2.cpp b/llvm/tools/llvm-lto2/llvm-lto2.cpp
index f222d02bd7ce..faed9ff9939b 100644
--- a/llvm/tools/llvm-lto2/llvm-lto2.cpp
+++ b/llvm/tools/llvm-lto2/llvm-lto2.cpp
@@ -194,6 +194,7 @@ static cl::opt<bool> TryUseNewDbgInfoFormat(
 
 extern cl::opt<bool> UseNewDbgInfoFormat;
 extern cl::opt<cl::boolOrDefault> LoadBitcodeIntoNewDbgInfoFormat;
+extern cl::opt<cl::boolOrDefault> PreserveInputDbgFormat;
 
 static void check(Error E, std::string Msg) {
   if (!E)
@@ -239,6 +240,10 @@ static int run(int argc, char **argv) {
     // Turn the new debug-info format on.
     UseNewDbgInfoFormat = true;
   }
+  // Since llvm-lto2 collects multiple IR modules together, for simplicity's
+  // sake we disable the "PreserveInputDbgFormat" flag to enforce a single debug
+  // info format.
+  PreserveInputDbgFormat = cl::boolOrDefault::BOU_FALSE;
 
   // FIXME: Workaround PR30396 which means that a symbol can appear
   // more than once if it is defined in module-level assembly and
diff --git a/llvm/tools/llvm-objcopy/ObjcopyOptions.cpp b/llvm/tools/llvm-objcopy/ObjcopyOptions.cpp
index 70e85460d3df..7269c51a08d6 100644
--- a/llvm/tools/llvm-objcopy/ObjcopyOptions.cpp
+++ b/llvm/tools/llvm-objcopy/ObjcopyOptions.cpp
@@ -736,42 +736,6 @@ objcopy::parseObjcopyOptions(ArrayRef<const char *> RawArgsArr,
       return createStringError(errc::invalid_argument, Reason);
   }
 
-  for (const auto *A : InputArgs.filtered(OBJCOPY_compress_sections)) {
-    SmallVector<StringRef, 0> Fields;
-    StringRef(A->getValue()).split(Fields, '=');
-    if (Fields.size() != 2 || Fields[1].empty()) {
-      return createStringError(
-          errc::invalid_argument,
-          A->getSpelling() +
-              ": parse error, not 'section-glob=[none|zlib|zstd]'");
-    }
-
-    auto Type = StringSwitch<DebugCompressionType>(Fields[1])
-                    .Case("zlib", DebugCompressionType::Zlib)
-                    .Case("zstd", DebugCompressionType::Zstd)
-                    .Default(DebugCompressionType::None);
-    if (Type == DebugCompressionType::None && Fields[1] != "none") {
-      return createStringError(
-          errc::invalid_argument,
-          "invalid or unsupported --compress-sections format: %s",
-          A->getValue());
-    }
-
-    auto &P = Config.compressSections.emplace_back();
-    P.second = Type;
-    auto Matcher =
-        NameOrPattern::create(Fields[0], SectionMatchStyle, ErrorCallback);
-    // =none allows overriding a previous =zlib or =zstd. Reject negative
-    // patterns, which would be confusing.
-    if (Matcher && !Matcher->isPositiveMatch()) {
-      return createStringError(
-          errc::invalid_argument,
-          "--compress-sections: negative pattern is unsupported");
-    }
-    if (Error E = P.first.addMatcher(std::move(Matcher)))
-      return std::move(E);
-  }
-
   Config.AddGnuDebugLink = InputArgs.getLastArgValue(OBJCOPY_add_gnu_debuglink);
   // The gnu_debuglink's target is expected to not change or else its CRC would
   // become invalidated and get rejected. We can avoid recalculating the
diff --git a/llvm/tools/llvm-objcopy/ObjcopyOpts.td b/llvm/tools/llvm-objcopy/ObjcopyOpts.td
index 4bc80eba05f8..be02616e8c68 100644
--- a/llvm/tools/llvm-objcopy/ObjcopyOpts.td
+++ b/llvm/tools/llvm-objcopy/ObjcopyOpts.td
@@ -35,12 +35,6 @@ def : Flag<["--"], "compress-debug-sections">, Alias<compress_debug_sections>,
       AliasArgs<["zlib"]>;
 def decompress_debug_sections : Flag<["--"], "decompress-debug-sections">,
                                 HelpText<"Decompress DWARF debug sections">;
-defm compress_sections
-    : Eq<"compress-sections",
-         "Compress or decompress sections using specified format. Supported "
-         "formats: zlib, zstd. Specify 'none' for decompression">,
-      MetaVarName<"<section-glob>=<format>">;
-
 defm split_dwo
     : Eq<"split-dwo", "Equivalent to --extract-dwo and <dwo-file> as the output file and no other options, "
                       "and then --strip-dwo on the input file">,
diff --git a/llvm/unittests/ADT/CMakeLists.txt b/llvm/unittests/ADT/CMakeLists.txt
index 12d7325036bf..17c5c9d1c59c 100644
--- a/llvm/unittests/ADT/CMakeLists.txt
+++ b/llvm/unittests/ADT/CMakeLists.txt
@@ -44,6 +44,7 @@ add_llvm_unittest(ADTTests
   ImmutableMapTest.cpp
   ImmutableSetTest.cpp
   IntEqClassesTest.cpp
+  Interleave.cpp
   IntervalMapTest.cpp
   IntervalTreeTest.cpp
   IntrusiveRefCntPtrTest.cpp
diff --git a/llvm/unittests/ADT/Interleave.cpp b/llvm/unittests/ADT/Interleave.cpp
new file mode 100644
index 000000000000..bc1ab1fae725
--- /dev/null
+++ b/llvm/unittests/ADT/Interleave.cpp
@@ -0,0 +1,42 @@
+//===- unittests/ADT/Interleave.cpp - Interleave unit tests ---------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#include "llvm/ADT/STLExtras.h"
+#include "llvm/ADT/SmallVector.h"
+#include "llvm/Support/raw_ostream.h"
+
+#include "gtest/gtest.h"
+
+using namespace llvm;
+
+namespace {
+
+TEST(InterleaveTest, Interleave) {
+  std::string Str;
+  raw_string_ostream OS(Str);
+
+  // Check that interleave works on a SmallVector.
+  SmallVector<const char *> Doodles = {"golden", "berna", "labra"};
+  interleave(
+      Doodles, OS, [&](const char *Name) { OS << Name << "doodle"; }, ", ");
+
+  EXPECT_EQ(OS.str(), "goldendoodle, bernadoodle, labradoodle");
+}
+
+TEST(InterleaveTest, InterleaveComma) {
+  std::string Str;
+  raw_string_ostream OS(Str);
+
+  // Check that interleaveComma uses ADL to find begin/end on an array.
+  const StringRef LongDogs[] = {"dachshund", "doxie", "dackel", "teckel"};
+  interleaveComma(LongDogs, OS);
+
+  EXPECT_EQ(OS.str(), "dachshund, doxie, dackel, teckel");
+}
+
+} // anonymous namespace
diff --git a/llvm/unittests/Option/OptionMarshallingTest.cpp b/llvm/unittests/Option/OptionMarshallingTest.cpp
index 339d825c2016..0464e27d5248 100644
--- a/llvm/unittests/Option/OptionMarshallingTest.cpp
+++ b/llvm/unittests/Option/OptionMarshallingTest.cpp
@@ -19,9 +19,9 @@ struct OptionWithMarshallingInfo {
 static const OptionWithMarshallingInfo MarshallingTable[] = {
 #define OPTION_WITH_MARSHALLING(                                               \
     PREFIX_TYPE, PREFIXED_NAME, ID, KIND, GROUP, ALIAS, ALIASARGS, FLAGS,      \
-    VISIBILITY, PARAM, HELPTEXT, METAVAR, VALUES, SHOULD_PARSE, ALWAYS_EMIT,   \
-    KEYPATH, DEFAULT_VALUE, IMPLIED_CHECK, IMPLIED_VALUE, NORMALIZER,          \
-    DENORMALIZER, MERGER, EXTRACTOR, TABLE_INDEX)                              \
+    VISIBILITY, PARAM, HELPTEXT, HELPTEXTSFORVARIANTS, METAVAR, VALUES,        \
+    SHOULD_PARSE, ALWAYS_EMIT, KEYPATH, DEFAULT_VALUE, IMPLIED_CHECK,          \
+    IMPLIED_VALUE, NORMALIZER, DENORMALIZER, MERGER, EXTRACTOR, TABLE_INDEX)   \
   {PREFIXED_NAME, #KEYPATH, #IMPLIED_CHECK, #IMPLIED_VALUE},
 #include "Opts.inc"
 #undef OPTION_WITH_MARSHALLING
diff --git a/llvm/utils/TableGen/Common/CodeGenDAGPatterns.cpp b/llvm/utils/TableGen/Common/CodeGenDAGPatterns.cpp
index 076d0427a859..7a5d2be3ae95 100644
--- a/llvm/utils/TableGen/Common/CodeGenDAGPatterns.cpp
+++ b/llvm/utils/TableGen/Common/CodeGenDAGPatterns.cpp
@@ -3858,8 +3858,10 @@ void CodeGenDAGPatterns::parseInstructionPattern(CodeGenInstruction &CGI,
   for (unsigned i = NumResults, e = CGI.Operands.size(); i != e; ++i) {
     CGIOperandList::OperandInfo &Op = CGI.Operands[i];
     const std::string &OpName = Op.Name;
-    if (OpName.empty())
+    if (OpName.empty()) {
       I.error("Operand #" + Twine(i) + " in operands list has no name!");
+      continue;
+    }
 
     if (!InstInputs.count(OpName)) {
       // If this is an operand with a DefaultOps set filled in, we can ignore
@@ -3872,16 +3874,19 @@ void CodeGenDAGPatterns::parseInstructionPattern(CodeGenInstruction &CGI,
       }
       I.error("Operand $" + OpName +
               " does not appear in the instruction pattern");
+      continue;
     }
     TreePatternNodePtr InVal = InstInputs[OpName];
     InstInputs.erase(OpName); // It occurred, remove from map.
 
     if (InVal->isLeaf() && isa<DefInit>(InVal->getLeafValue())) {
       Record *InRec = cast<DefInit>(InVal->getLeafValue())->getDef();
-      if (!checkOperandClass(Op, InRec))
+      if (!checkOperandClass(Op, InRec)) {
         I.error("Operand $" + OpName +
                 "'s register class disagrees"
                 " between the operand and pattern");
+        continue;
+      }
     }
     Operands.push_back(Op.Rec);
 
diff --git a/llvm/utils/TableGen/OptParserEmitter.cpp b/llvm/utils/TableGen/OptParserEmitter.cpp
index 6334af53f88f..4fb1ca18ac11 100644
--- a/llvm/utils/TableGen/OptParserEmitter.cpp
+++ b/llvm/utils/TableGen/OptParserEmitter.cpp
@@ -191,6 +191,62 @@ static MarshallingInfo createMarshallingInfo(const Record &R) {
   return Ret;
 }
 
+static void EmitHelpTextsForVariants(
+    raw_ostream &OS, std::vector<std::pair<std::vector<std::string>, StringRef>>
+                         HelpTextsForVariants) {
+  // OptTable must be constexpr so it uses std::arrays with these capacities.
+  const unsigned MaxVisibilityPerHelp = 2;
+  const unsigned MaxVisibilityHelp = 1;
+
+  assert(HelpTextsForVariants.size() <= MaxVisibilityHelp &&
+         "Too many help text variants to store in "
+         "OptTable::HelpTextsForVariants");
+
+  // This function must initialise any unused elements of those arrays.
+  for (auto [Visibilities, _] : HelpTextsForVariants)
+    while (Visibilities.size() < MaxVisibilityPerHelp)
+      Visibilities.push_back("0");
+
+  while (HelpTextsForVariants.size() < MaxVisibilityHelp)
+    HelpTextsForVariants.push_back(
+        {std::vector<std::string>(MaxVisibilityPerHelp, "0"), ""});
+
+  OS << ", (std::array<std::pair<std::array<unsigned, " << MaxVisibilityPerHelp
+     << ">, const char*>, " << MaxVisibilityHelp << ">{{ ";
+
+  auto VisibilityHelpEnd = HelpTextsForVariants.cend();
+  for (auto VisibilityHelp = HelpTextsForVariants.cbegin();
+       VisibilityHelp != VisibilityHelpEnd; ++VisibilityHelp) {
+    auto [Visibilities, Help] = *VisibilityHelp;
+
+    assert(Visibilities.size() <= MaxVisibilityPerHelp &&
+           "Too many visibilities to store in an "
+           "OptTable::HelpTextsForVariants entry");
+    OS << "std::make_pair(std::array<unsigned, " << MaxVisibilityPerHelp
+       << ">{{";
+
+    auto VisibilityEnd = Visibilities.cend();
+    for (auto Visibility = Visibilities.cbegin(); Visibility != VisibilityEnd;
+         ++Visibility) {
+      OS << *Visibility;
+      if (std::next(Visibility) != VisibilityEnd)
+        OS << ", ";
+    }
+
+    OS << "}}, ";
+
+    if (Help.size())
+      write_cstring(OS, Help);
+    else
+      OS << "nullptr";
+    OS << ")";
+
+    if (std::next(VisibilityHelp) != VisibilityHelpEnd)
+      OS << ", ";
+  }
+  OS << " }})";
+}
+
 /// OptParserEmitter - This tablegen backend takes an input .td file
 /// describing a list of options and emits a data structure for parsing and
 /// working with those options when given an input command line.
@@ -312,6 +368,9 @@ static void EmitOptParser(RecordKeeper &Records, raw_ostream &OS) {
     } else
       OS << ", nullptr";
 
+    // Not using Visibility specific text for group help.
+    EmitHelpTextsForVariants(OS, {});
+
     // The option meta-variable name (unused).
     OS << ", nullptr";
 
@@ -410,6 +469,22 @@ static void EmitOptParser(RecordKeeper &Records, raw_ostream &OS) {
     } else
       OS << ", nullptr";
 
+    std::vector<std::pair<std::vector<std::string>, StringRef>>
+        HelpTextsForVariants;
+    for (Record *VisibilityHelp :
+         R.getValueAsListOfDefs("HelpTextsForVariants")) {
+      ArrayRef<Init *> Visibilities =
+          VisibilityHelp->getValueAsListInit("Visibilities")->getValues();
+
+      std::vector<std::string> VisibilityNames;
+      for (Init *Visibility : Visibilities)
+        VisibilityNames.push_back(Visibility->getAsUnquotedString());
+
+      HelpTextsForVariants.push_back(std::make_pair(
+          VisibilityNames, VisibilityHelp->getValueAsString("Text")));
+    }
+    EmitHelpTextsForVariants(OS, HelpTextsForVariants);
+
     // The option meta-variable name.
     OS << ", ";
     if (!isa<UnsetInit>(R.getValueInit("MetaVarName")))
diff --git a/llvm/utils/gn/secondary/llvm/lib/Transforms/Instrumentation/BUILD.gn b/llvm/utils/gn/secondary/llvm/lib/Transforms/Instrumentation/BUILD.gn
index 131308db2aa5..d79b5efe69eb 100644
--- a/llvm/utils/gn/secondary/llvm/lib/Transforms/Instrumentation/BUILD.gn
+++ b/llvm/utils/gn/secondary/llvm/lib/Transforms/Instrumentation/BUILD.gn
@@ -23,13 +23,13 @@ static_library("Instrumentation") {
     "InstrProfiling.cpp",
     "Instrumentation.cpp",
     "KCFI.cpp",
+    "LowerAllowCheckPass.cpp",
     "MemProfiler.cpp",
     "MemorySanitizer.cpp",
     "PGOForceFunctionAttrs.cpp",
     "PGOInstrumentation.cpp",
     "PGOMemOPSizeOpt.cpp",
     "PoisonChecking.cpp",
-    "RemoveTrapsPass.cpp",
     "SanitizerBinaryMetadata.cpp",
     "SanitizerCoverage.cpp",
     "ThreadSanitizer.cpp",
diff --git a/mlir/cmake/modules/MLIRCheckHardwareFeatures.cmake b/mlir/cmake/modules/MLIRCheckHardwareFeatures.cmake
new file mode 100644
index 000000000000..7bc13287cd65
--- /dev/null
+++ b/mlir/cmake/modules/MLIRCheckHardwareFeatures.cmake
@@ -0,0 +1,100 @@
+# A collection of helper CMake functions to detect hardware capabilities. At
+# the moment these are used when configuring MLIR integration tests.
+
+# Checks whether the specified hardware capability is supported by the host
+# Linux system. This is implemented by checking auxiliary vector feature
+# provided by the Linux kernel.
+#
+# check_hwcap(
+#   hwcap_spec
+#   output_var
+# )
+#
+# hwcap_spec - HWCAP value to check - these are defined in hwcap.h in the Linux
+#              kernel.
+#
+# output_var - Output variable to use to save the results (TRUE for supported,
+#              FALSE for not supported).
+#
+# EXAMPLES:
+#
+# check_hwcap("HWCAP2_SME" SME_EMULATOR_REQUIRED)
+#
+function(check_hwcap hwcap_spec output)
+    set(hwcap_test_src
+      [====[
+      #include <asm/hwcap.h>
+      #include <sys/auxv.h>
+
+      int main(void)
+      {
+          long hwcaps = getauxval(AT_<HWCAP_VEC>);
+          return (hwcaps & <HWCAP_SPEC>) != 0;
+      }
+      ]====]
+    )
+
+    # Extract from $hwcap_spec whether this is AT_HWCAP or AT_HWCAP2
+    string(FIND ${hwcap_spec} "_" wsloc)
+    string(SUBSTRING ${hwcap_spec} 0 ${wsloc} hwcap_vec)
+
+    string(REPLACE "<HWCAP_VEC>" ${hwcap_vec} hwcap_test_src "${hwcap_test_src}")
+    string(REPLACE "<HWCAP_SPEC>" ${hwcap_spec} hwcap_test_src "${hwcap_test_src}")
+
+    set(hwcap_test_file ${CMAKE_BINARY_DIR}/temp/hwcap_check.c)
+    file(WRITE ${hwcap_test_file} "${hwcap_test_src}")
+
+    # Compile _and_ run
+    try_run(
+        test_run_result test_compile_result
+        "${CMAKE_BINARY_DIR}"
+        "${hwcap_test_file}"
+    )
+    # Compilation will fail if hwcap_spec is not defined - this usually means
+    # that your Linux kernel is too old.
+    if(${test_compile_result} AND (DEFINED test_run_result))
+      message(STATUS "Checking whether ${hwcap_spec} is supported by the host system: ${test_run_result}")
+      set(${output} ${test_run_result} PARENT_SCOPE)
+    else()
+      message(STATUS "Checking whether ${hwcap_spec} is supported by the host system: FALSE")
+    endif()
+endfunction(check_hwcap)
+
+# For the given group of e2e tests (defined by the `mlir_e2e_tests` flag),
+# checks whether an emulator is required. If yes, verifies that the
+# corresponding CMake var pointing to an emulator (`emulator_exec`) has been
+# set.
+#
+# check_emulator(
+#   mlir_e2e_tests
+#   hwcap_spec
+#   emulator_exec
+# )
+#
+# mlir_e2e_tests  - MLIR CMake variables corresponding to the group of e2e tests
+#                   to check
+# hwcap_spec      - HWCAP value to check. This should correspond to the hardware
+#                   capabilities required by the tests to be checked. Possible
+#                   values are defined in hwcap.h in the Linux kernel.
+# emulator_exec   - variable the defines the emulator (ought to be set if
+#                   required, can be empty otherwise).
+#
+# EXAMPLES:
+#
+#  check_emulator(MLIR_RUN_ARM_SVE_TESTS "HWCAP_SVE" ARM_EMULATOR_EXECUTABLE)
+#
+function(check_emulator mlir_e2e_tests hwcap_spec emulator_exec)
+  if (NOT ${mlir_e2e_tests})
+    return()
+  endif()
+
+  check_hwcap(${hwcap_spec} emulator_not_required)
+  if (${emulator_not_required})
+    return()
+  endif()
+
+  if (${emulator_exec} STREQUAL "")
+    message(FATAL_ERROR "${mlir_e2e_tests} requires an emulator, but ${emulator_exec} is not set")
+  endif()
+
+endfunction()
diff --git a/mlir/docs/DefiningDialects/AttributesAndTypes.md b/mlir/docs/DefiningDialects/AttributesAndTypes.md
index 950acb842022..d6941c0b681f 100644
--- a/mlir/docs/DefiningDialects/AttributesAndTypes.md
+++ b/mlir/docs/DefiningDialects/AttributesAndTypes.md
@@ -14,8 +14,8 @@ from the [MLIR LangRef](../LangRef.md).
 
 Attributes are the mechanism for specifying constant data on operations in
 places where a variable is never allowed - e.g. the comparison predicate of a
-[`arith.cmpi` operation](../Dialects/ArithOps.md#arithcmpi-mlirarithcmpiop), or
-the underlying value of a [`arith.constant` operation](../Dialects/ArithOps.md#arithconstant-mlirarithconstantop).
+[`arith.cmpi` operation](../Dialects/ArithOps.md/#arithcmpi-arithcmpiop), or
+the underlying value of a [`arith.constant` operation](../Dialects/ArithOps.md/#arithconstant-arithconstantop).
 Each operation has an attribute dictionary, which associates a set of attribute
 names to attribute values.
 
@@ -24,7 +24,7 @@ names to attribute values.
 Every SSA value, such as operation results or block arguments, in MLIR has a type
 defined by the type system. MLIR has an open type system with no fixed list of types,
 and there are no restrictions on the abstractions they represent. For example, take
-the following [Arithmetic AddI operation](../Dialects/ArithOps.md#arithaddi-mlirarithaddiop):
+the following [Arithmetic AddI operation](../Dialects/ArithOps.md/#arithaddi-arithaddiop):
 
 ```mlir
   %result = arith.addi %lhs, %rhs : i64
@@ -32,7 +32,7 @@ the following [Arithmetic AddI operation](../Dialects/ArithOps.md#arithaddi-mlir
 
 It takes two input SSA values (`%lhs` and `%rhs`), and returns a single SSA
 value (`%result`). The inputs and outputs of this operation are of type `i64`,
-which is an instance of the [Builtin IntegerType](../Dialects/Builtin.md#integertype).
+which is an instance of the [Builtin IntegerType](../Dialects/Builtin.md/#integertype).
 
 ## Attributes and Types
 
diff --git a/mlir/docs/DefiningDialects/_index.md b/mlir/docs/DefiningDialects/_index.md
index 5a3993508fce..83070b616d0e 100644
--- a/mlir/docs/DefiningDialects/_index.md
+++ b/mlir/docs/DefiningDialects/_index.md
@@ -10,8 +10,8 @@ Before diving into how to define these constructs, below is a quick refresher
 from the [MLIR LangRef](../LangRef.md).
 
 Dialects are the mechanism by which to engage with and extend the MLIR
-ecosystem. They allow for defining new [attributes](../LangRef.md#attributes),
-[operations](../LangRef.md#operations), and [types](../LangRef.md#type-system).
+ecosystem. They allow for defining new [attributes](../LangRef.md/#attributes),
+[operations](../LangRef.md/#operations), and [types](../LangRef.md/#type-system).
 Dialects are used to model a variety of different abstractions; from traditional
 [arithmetic](../Dialects/ArithOps.md) to
 [pattern rewrites](../Dialects/PDLOps.md); and is one of the most fundamental
@@ -196,7 +196,7 @@ only the declaration of the destructor is generated for the Dialect class.
 
 ### Discardable Attribute Verification
 
-As described by the [MLIR Language Reference](../LangRef.md#attributes),
+As described by the [MLIR Language Reference](../LangRef.md/#attributes),
 *discardable attribute* are a type of attribute that has its semantics defined
 by the dialect whose name prefixes that of the attribute. For example, if an
 operation has an attribute named `gpu.contained_module`, the `gpu` dialect
@@ -270,7 +270,7 @@ void *MyDialect::getRegisteredInterfaceForOp(TypeID typeID, StringAttr opName);
 ```
 
 For a more detail description of the expected usages of this hook, view the detailed 
-[interface documentation](../Interfaces.md#dialect-fallback-for-opinterface).
+[interface documentation](../Interfaces.md/#dialect-fallback-for-opinterface).
 
 ### Default Attribute/Type Parsers and Printers 
 
diff --git a/mlir/docs/Dialects/ArmSME.md b/mlir/docs/Dialects/ArmSME.md
index 7326150bcd11..ce0a76ed60f1 100644
--- a/mlir/docs/Dialects/ArmSME.md
+++ b/mlir/docs/Dialects/ArmSME.md
@@ -6,7 +6,7 @@ This dialect defines custom and LLVM IR intrinsic operations that are used to
 target Arm Scalable Matrix Extension. Through the available conversion and
 ArmSME passes you can, for example, lower a
 [linalg.matmul](https://mlir.llvm.org/docs/Dialects/Linalg/#linalgmatmul-linalgmatmulop)
-opereation to Arm SME
+operation to Arm SME
 [FMOPA](https://developer.arm.com/documentation/ddi0602/2023-03/SME-Instructions/FMOPA--widening---Half-precision-floating-point-sum-of-outer-products-and-accumulate-)
 (floating-point outer product) operations. See one of the in-tree end-to-end
 integration tests for reference:
@@ -14,6 +14,14 @@ integration tests for reference:
 * [Linalg/CPU/ArmSME/matmul.mlir](https://github.com/llvm/llvm-project/blob/main/mlir/test/Integration/Dialect/Linalg/CPU/ArmSME/matmul.mlir)
 * [Vector/CPU/ArmSME/test-outerproduct-f64.mlir](https://github.com/llvm/llvm-project/blob/main/mlir/test/Integration/Dialect/Vector/CPU/ArmSME/test-outerproduct-f64.mlir)
 
+In order to run ArmSME integration tests, include these flags in the CMake
+invocation when configuring LLVM and MLIR:
+```bash
+  -DMLIR_INCLUDE_INTEGRATION_TESTS=On
+  -DMLIR_RUN_ARM_SME_TESTS=On
+  -DARM_EMULATOR_EXECUTABLE=<path-to-emulator> 
+```
+
 These tests are run "post-commit" by the
 [clang-aarch64-sve-vla](https://lab.llvm.org/buildbot/#/builders/197) LLVM
 BuildBot worker.
diff --git a/mlir/docs/Dialects/Linalg/_index.md b/mlir/docs/Dialects/Linalg/_index.md
index c8fa0374334e..fbd1a451dc09 100644
--- a/mlir/docs/Dialects/Linalg/_index.md
+++ b/mlir/docs/Dialects/Linalg/_index.md
@@ -279,7 +279,7 @@ loops but previous experience shows that the abstraction generalizes.
 
 A `linalg.generic` op has a compute payload that is fully generic thanks to the
 use of
-[Regions](https://github.com/llvm/llvm-project/blob/58265ad42a90ae8905be6a447cb42e53529a54a0/mlir/docs/LangRef.md#regions).
+[Regions](https://github.com/llvm/llvm-project/blob/58265ad42a90ae8905be6a447cb42e53529a54a0/mlir/docs/LangRef.md/#regions).
 
 The region takes as arguments the scalar elemental types of the tensor or buffer
 operands of the `linalg.generic`. For flexibility and ability to match library
diff --git a/mlir/docs/Dialects/Transform.md b/mlir/docs/Dialects/Transform.md
index 768de9561b95..02f4a766ff42 100644
--- a/mlir/docs/Dialects/Transform.md
+++ b/mlir/docs/Dialects/Transform.md
@@ -84,11 +84,11 @@ try to be explicitly typed when possible.
 
 The transform IR values have transform IR types, which should implement exactly one of:
 
-  * [TransformHandleTypeInterface](Transform.md#transformhandletypeinterface-transformhandletypeinterface),
+  * [TransformHandleTypeInterface](#transformhandletypeinterface-transformhandletypeinterface),
 
-  * [TransformValueHandleTypeInterface](Transform.md#transformvaluehandletypeinterface-transformvaluehandletypeinterface),
+  * [TransformValueHandleTypeInterface](#transformvaluehandletypeinterface-transformvaluehandletypeinterface),
 
-  * [TransformParamTypeInterface](Transform.md##transformparamtypeinterface-transformparamtypeinterface).
+  * [TransformParamTypeInterface](#transformparamtypeinterface-transformparamtypeinterface).
 
 The goal of these type interfaces, beyond providing a common base for accepted
 types, is to verify the properties of the associated objects. For example, a
diff --git a/mlir/docs/LangRef.md b/mlir/docs/LangRef.md
index 8efc88815b88..21cfdc78a6a4 100644
--- a/mlir/docs/LangRef.md
+++ b/mlir/docs/LangRef.md
@@ -761,7 +761,7 @@ attribute-value ::= attribute-alias | dialect-attribute | builtin-attribute
 
 Attributes are the mechanism for specifying constant data on operations in
 places where a variable is never allowed - e.g. the comparison predicate of a
-[`cmpi` operation](Dialects/ArithOps.md#arithcmpi-mlirarithcmpiop). Each operation has an
+[`cmpi` operation](Dialects/ArithOps.md/#arithcmpi-arithcmpiop). Each operation has an
 attribute dictionary, which associates a set of attribute names to attribute
 values. MLIR's builtin dialect provides a rich set of
 [builtin attribute values](#builtin-attribute-values) out of the box (such as
diff --git a/mlir/docs/Rationale/SideEffectsAndSpeculation.md b/mlir/docs/Rationale/SideEffectsAndSpeculation.md
index 0cd5ea59822a..8b08b757531b 100644
--- a/mlir/docs/Rationale/SideEffectsAndSpeculation.md
+++ b/mlir/docs/Rationale/SideEffectsAndSpeculation.md
@@ -4,8 +4,8 @@ This document outlines how MLIR models side effects and how speculation works in
 MLIR.
 
 This rationale only applies to operations used in
-[CFG regions](../LangRef.md#control-flow-and-ssacfg-regions). Side effect
-modeling in [graph regions](../LangRef.md#graph-regions) is TBD.
+[CFG regions](../LangRef.md/#control-flow-and-ssacfg-regions). Side effect
+modeling in [graph regions](../LangRef.md/#graph-regions) is TBD.
 
 [TOC]
 
diff --git a/mlir/docs/TargetLLVMIR.md b/mlir/docs/TargetLLVMIR.md
index df281f65768e..96a4589eb80e 100644
--- a/mlir/docs/TargetLLVMIR.md
+++ b/mlir/docs/TargetLLVMIR.md
@@ -41,7 +41,7 @@ they use or produce `memref`-typed values.
 
 The process relies on the [Dialect Conversion](DialectConversion.md)
 infrastructure and, in particular, on the
-[materialization](DialectConversion.md#type-conversion) hooks of `TypeConverter`
+[materialization](DialectConversion.md/#type-conversion) hooks of `TypeConverter`
 to support progressive lowering by injecting `unrealized_conversion_cast`
 operations between converted and unconverted operations. After multiple partial
 conversions to the LLVM dialect are performed, the cast operations that became
@@ -58,7 +58,7 @@ same type converter.
 
 #### LLVM Dialect-compatible Types
 
-The types [compatible](Dialects/LLVM.md#built-in-type-compatibility) with the
+The types [compatible](Dialects/LLVM.md/#built-in-type-compatibility) with the
 LLVM dialect are kept as is.
 
 #### Complex Type
@@ -188,7 +188,7 @@ Function types are converted to LLVM dialect function types as follows:
     arguments to allow for specifying metadata such as aliasing information on
     individual pointers;
 -   the conversion of `memref`-typed arguments is subject to
-    [calling conventions](TargetLLVMIR.md#calling-conventions).
+    [calling conventions](#calling-conventions).
 -   if a function type has boolean attribute `func.varargs` being set, the
     converted LLVM function will be variadic.
 
@@ -364,7 +364,7 @@ llvm.func @bar() {
 
 The default calling convention converts `memref`-typed function arguments to
 LLVM dialect literal structs
-[defined above](TargetLLVMIR.md#ranked-memref-types) before unbundling them into
+[defined above](#ranked-memref-types) before unbundling them into
 individual scalar arguments.
 
 Examples:
diff --git a/mlir/docs/Tutorials/Toy/Ch-4.md b/mlir/docs/Tutorials/Toy/Ch-4.md
index ae10dc4a0113..b753ee7a5332 100644
--- a/mlir/docs/Tutorials/Toy/Ch-4.md
+++ b/mlir/docs/Tutorials/Toy/Ch-4.md
@@ -383,7 +383,7 @@ void MulOp::inferShapes() { getResult().setType(getLhs().getType()); }
 At this point, each of the necessary Toy operations provide a mechanism by which
 to infer their output shapes. The ShapeInferencePass will operate on functions:
 it will run on each function in isolation. MLIR also supports general
-[OperationPasses](../../PassManagement.md#operation-pass) that run on any
+[OperationPasses](../../PassManagement.md/#operation-pass) that run on any
 isolated operation, but here our module only contains functions, so there is no
 need to generalize to all operations.
 
diff --git a/mlir/docs/Tutorials/transform/Ch4.md b/mlir/docs/Tutorials/transform/Ch4.md
index c3159eb991d1..81263e10b098 100644
--- a/mlir/docs/Tutorials/transform/Ch4.md
+++ b/mlir/docs/Tutorials/transform/Ch4.md
@@ -27,7 +27,7 @@ directly within the transform dialect.
 ## Simple match
 
 Let us reconsider the “fully connected layer” example from [Chapter
-1](Ch1.md#chaining-transformations-with-handles), reproduced below for
+1](Ch1.md/#chaining-transformations-with-handles), reproduced below for
 convenience.
 
 
diff --git a/mlir/include/mlir/Dialect/Arith/IR/ArithOps.td b/mlir/include/mlir/Dialect/Arith/IR/ArithOps.td
index ead19c69a083..4e4c6fd60177 100644
--- a/mlir/include/mlir/Dialect/Arith/IR/ArithOps.td
+++ b/mlir/include/mlir/Dialect/Arith/IR/ArithOps.td
@@ -343,7 +343,9 @@ def Arith_SubIOp : Arith_IntBinaryOpWithOverflowFlags<"subi"> {
 // MulIOp
 //===----------------------------------------------------------------------===//
 
-def Arith_MulIOp : Arith_IntBinaryOpWithOverflowFlags<"muli", [Commutative]> {
+def Arith_MulIOp : Arith_IntBinaryOpWithOverflowFlags<"muli",
+  [Commutative, DeclareOpInterfaceMethods<OpAsmOpInterface, ["getAsmResultNames"]>]
+> {
   let summary = [{
     Integer multiplication operation.
   }];
diff --git a/mlir/include/mlir/Dialect/LLVMIR/Transforms/TypeConsistency.h b/mlir/include/mlir/Dialect/LLVMIR/Transforms/TypeConsistency.h
index cacb241bfd7a..a4bb380b99b8 100644
--- a/mlir/include/mlir/Dialect/LLVMIR/Transforms/TypeConsistency.h
+++ b/mlir/include/mlir/Dialect/LLVMIR/Transforms/TypeConsistency.h
@@ -56,17 +56,6 @@ public:
                                 PatternRewriter &rewrite) const override;
 };
 
-/// Transforms type-inconsistent stores, aka stores where the type hint of
-/// the address contradicts the value stored, by inserting a bitcast if
-/// possible.
-class BitcastStores : public OpRewritePattern<StoreOp> {
-public:
-  using OpRewritePattern::OpRewritePattern;
-
-  LogicalResult matchAndRewrite(StoreOp store,
-                                PatternRewriter &rewriter) const override;
-};
-
 /// Splits GEPs with more than two indices into multiple GEPs with exactly
 /// two indices. The created GEPs are then guaranteed to index into only
 /// one aggregate at a time.
diff --git a/mlir/lib/Dialect/Arith/IR/ArithOps.cpp b/mlir/lib/Dialect/Arith/IR/ArithOps.cpp
index ce0602c853e3..1d68a4f7292b 100644
--- a/mlir/lib/Dialect/Arith/IR/ArithOps.cpp
+++ b/mlir/lib/Dialect/Arith/IR/ArithOps.cpp
@@ -423,6 +423,33 @@ OpFoldResult arith::MulIOp::fold(FoldAdaptor adaptor) {
       [](const APInt &a, const APInt &b) { return a * b; });
 }
 
+void arith::MulIOp::getAsmResultNames(
+    function_ref<void(Value, StringRef)> setNameFn) {
+  if (!isa<IndexType>(getType()))
+    return;
+
+  // Match vector.vscale by name to avoid depending on the vector dialect (which
+  // is a circular dependency).
+  auto isVscale = [](Operation *op) {
+    return op && op->getName().getStringRef() == "vector.vscale";
+  };
+
+  IntegerAttr baseValue;
+  auto isVscaleExpr = [&](Value a, Value b) {
+    return matchPattern(a, m_Constant(&baseValue)) &&
+           isVscale(b.getDefiningOp());
+  };
+
+  if (!isVscaleExpr(getLhs(), getRhs()) && !isVscaleExpr(getRhs(), getLhs()))
+    return;
+
+  // Name `base * vscale` or `vscale * base` as `c<base_value>_vscale`.
+  SmallString<32> specialNameBuffer;
+  llvm::raw_svector_ostream specialName(specialNameBuffer);
+  specialName << 'c' << baseValue.getInt() << "_vscale";
+  setNameFn(getResult(), specialName.str());
+}
+
 void arith::MulIOp::getCanonicalizationPatterns(RewritePatternSet &patterns,
                                                 MLIRContext *context) {
   patterns.add<MulIMulIConstant>(context);
diff --git a/mlir/lib/Dialect/LLVMIR/IR/LLVMDialect.cpp b/mlir/lib/Dialect/LLVMIR/IR/LLVMDialect.cpp
index e5c19a916392..f90240a67dcc 100644
--- a/mlir/lib/Dialect/LLVMIR/IR/LLVMDialect.cpp
+++ b/mlir/lib/Dialect/LLVMIR/IR/LLVMDialect.cpp
@@ -2761,17 +2761,27 @@ LogicalResult SExtOp::verify() { return verifyExtOp<SExtOp>(*this); }
 // Folder and verifier for LLVM::BitcastOp
 //===----------------------------------------------------------------------===//
 
-OpFoldResult LLVM::BitcastOp::fold(FoldAdaptor adaptor) {
-  // bitcast(x : T0, T0) -> x
-  if (getArg().getType() == getType())
-    return getArg();
-  // bitcast(bitcast(x : T0, T1), T0) -> x
-  if (auto prev = getArg().getDefiningOp<BitcastOp>())
-    if (prev.getArg().getType() == getType())
+/// Folds a cast op that can be chained.
+template <typename T>
+static Value foldChainableCast(T castOp, typename T::FoldAdaptor adaptor) {
+  // cast(x : T0, T0) -> x
+  if (castOp.getArg().getType() == castOp.getType())
+    return castOp.getArg();
+  if (auto prev = castOp.getArg().template getDefiningOp<T>()) {
+    // cast(cast(x : T0, T1), T0) -> x
+    if (prev.getArg().getType() == castOp.getType())
       return prev.getArg();
+    // cast(cast(x : T0, T1), T2) -> cast(x: T0, T2)
+    castOp.getArgMutable().set(prev.getArg());
+    return Value{castOp};
+  }
   return {};
 }
 
+OpFoldResult LLVM::BitcastOp::fold(FoldAdaptor adaptor) {
+  return foldChainableCast(*this, adaptor);
+}
+
 LogicalResult LLVM::BitcastOp::verify() {
   auto resultType = llvm::dyn_cast<LLVMPointerType>(
       extractVectorElementType(getResult().getType()));
@@ -2811,14 +2821,7 @@ LogicalResult LLVM::BitcastOp::verify() {
 //===----------------------------------------------------------------------===//
 
 OpFoldResult LLVM::AddrSpaceCastOp::fold(FoldAdaptor adaptor) {
-  // addrcast(x : T0, T0) -> x
-  if (getArg().getType() == getType())
-    return getArg();
-  // addrcast(addrcast(x : T0, T1), T0) -> x
-  if (auto prev = getArg().getDefiningOp<AddrSpaceCastOp>())
-    if (prev.getArg().getType() == getType())
-      return prev.getArg();
-  return {};
+  return foldChainableCast(*this, adaptor);
 }
 
 //===----------------------------------------------------------------------===//
diff --git a/mlir/lib/Dialect/LLVMIR/IR/LLVMMemorySlot.cpp b/mlir/lib/Dialect/LLVMIR/IR/LLVMMemorySlot.cpp
index 563755297c57..c7ca0b4a5843 100644
--- a/mlir/lib/Dialect/LLVMIR/IR/LLVMMemorySlot.cpp
+++ b/mlir/lib/Dialect/LLVMIR/IR/LLVMMemorySlot.cpp
@@ -122,8 +122,37 @@ bool LLVM::StoreOp::storesTo(const MemorySlot &slot) {
   return getAddr() == slot.ptr;
 }
 
+/// Checks that two types are the same or can be cast into one another.
+static bool areCastCompatible(const DataLayout &layout, Type lhs, Type rhs) {
+  return lhs == rhs || (!isa<LLVM::LLVMStructType, LLVM::LLVMArrayType>(lhs) &&
+                        !isa<LLVM::LLVMStructType, LLVM::LLVMArrayType>(rhs) &&
+                        layout.getTypeSize(lhs) == layout.getTypeSize(rhs));
+}
+
+/// Constructs operations that convert `inputValue` into a new value of type
+/// `targetType`. Assumes that this conversion is possible.
+static Value createConversionSequence(RewriterBase &rewriter, Location loc,
+                                      Value inputValue, Type targetType) {
+  if (inputValue.getType() == targetType)
+    return inputValue;
+
+  if (!isa<LLVM::LLVMPointerType>(targetType) &&
+      !isa<LLVM::LLVMPointerType>(inputValue.getType()))
+    return rewriter.createOrFold<LLVM::BitcastOp>(loc, targetType, inputValue);
+
+  if (!isa<LLVM::LLVMPointerType>(targetType))
+    return rewriter.createOrFold<LLVM::PtrToIntOp>(loc, targetType, inputValue);
+
+  if (!isa<LLVM::LLVMPointerType>(inputValue.getType()))
+    return rewriter.createOrFold<LLVM::IntToPtrOp>(loc, targetType, inputValue);
+
+  return rewriter.createOrFold<LLVM::AddrSpaceCastOp>(loc, targetType,
+                                                      inputValue);
+}
+
 Value LLVM::StoreOp::getStored(const MemorySlot &slot, RewriterBase &rewriter) {
-  return getValue();
+  return createConversionSequence(rewriter, getLoc(), getValue(),
+                                  slot.elemType);
 }
 
 bool LLVM::LoadOp::canUsesBeRemoved(
@@ -138,7 +167,8 @@ bool LLVM::LoadOp::canUsesBeRemoved(
   // be removed (provided it loads the exact stored value and is not
   // volatile).
   return blockingUse == slot.ptr && getAddr() == slot.ptr &&
-         getResult().getType() == slot.elemType && !getVolatile_();
+         areCastCompatible(dataLayout, getResult().getType(), slot.elemType) &&
+         !getVolatile_();
 }
 
 DeletionKind LLVM::LoadOp::removeBlockingUses(
@@ -146,7 +176,9 @@ DeletionKind LLVM::LoadOp::removeBlockingUses(
     RewriterBase &rewriter, Value reachingDefinition) {
   // `canUsesBeRemoved` checked this blocking use must be the loaded slot
   // pointer.
-  rewriter.replaceAllUsesWith(getResult(), reachingDefinition);
+  Value newResult = createConversionSequence(
+      rewriter, getLoc(), reachingDefinition, getResult().getType());
+  rewriter.replaceAllUsesWith(getResult(), newResult);
   return DeletionKind::Delete;
 }
 
@@ -161,7 +193,8 @@ bool LLVM::StoreOp::canUsesBeRemoved(
   // fine, provided we are currently promoting its target value. Don't allow a
   // store OF the slot pointer, only INTO the slot pointer.
   return blockingUse == slot.ptr && getAddr() == slot.ptr &&
-         getValue() != slot.ptr && getValue().getType() == slot.elemType &&
+         getValue() != slot.ptr &&
+         areCastCompatible(dataLayout, slot.elemType, getValue().getType()) &&
          !getVolatile_();
 }
 
diff --git a/mlir/lib/Dialect/LLVMIR/Transforms/TypeConsistency.cpp b/mlir/lib/Dialect/LLVMIR/Transforms/TypeConsistency.cpp
index 3d700fe94e3b..b264e9ff9283 100644
--- a/mlir/lib/Dialect/LLVMIR/Transforms/TypeConsistency.cpp
+++ b/mlir/lib/Dialect/LLVMIR/Transforms/TypeConsistency.cpp
@@ -42,13 +42,6 @@ static Type isElementTypeInconsistent(Value addr, Type expectedType) {
   return elemType;
 }
 
-/// Checks that two types are the same or can be bitcast into one another.
-static bool areBitcastCompatible(DataLayout &layout, Type lhs, Type rhs) {
-  return lhs == rhs || (!isa<LLVMStructType, LLVMArrayType>(lhs) &&
-                        !isa<LLVMStructType, LLVMArrayType>(rhs) &&
-                        layout.getTypeSize(lhs) == layout.getTypeSize(rhs));
-}
-
 //===----------------------------------------------------------------------===//
 // CanonicalizeAlignedGep
 //===----------------------------------------------------------------------===//
@@ -518,26 +511,6 @@ LogicalResult SplitStores::matchAndRewrite(StoreOp store,
   return success();
 }
 
-LogicalResult BitcastStores::matchAndRewrite(StoreOp store,
-                                             PatternRewriter &rewriter) const {
-  Type sourceType = store.getValue().getType();
-  Type typeHint = isElementTypeInconsistent(store.getAddr(), sourceType);
-  if (!typeHint) {
-    // Nothing to do, since it is already consistent.
-    return failure();
-  }
-
-  auto dataLayout = DataLayout::closest(store);
-  if (!areBitcastCompatible(dataLayout, typeHint, sourceType))
-    return failure();
-
-  auto bitcastOp =
-      rewriter.create<BitcastOp>(store.getLoc(), typeHint, store.getValue());
-  rewriter.modifyOpInPlace(store,
-                           [&] { store.getValueMutable().assign(bitcastOp); });
-  return success();
-}
-
 LogicalResult SplitGEP::matchAndRewrite(GEPOp gepOp,
                                         PatternRewriter &rewriter) const {
   FailureOr<Type> typeHint = getRequiredConsistentGEPType(gepOp);
@@ -588,7 +561,6 @@ struct LLVMTypeConsistencyPass
     RewritePatternSet rewritePatterns(&getContext());
     rewritePatterns.add<CanonicalizeAlignedGep>(&getContext());
     rewritePatterns.add<SplitStores>(&getContext(), maxVectorSplitSize);
-    rewritePatterns.add<BitcastStores>(&getContext());
     rewritePatterns.add<SplitGEP>(&getContext());
     FrozenRewritePatternSet frozen(std::move(rewritePatterns));
 
diff --git a/mlir/lib/Target/LLVMIR/Dialect/OpenMP/OpenMPToLLVMIRTranslation.cpp b/mlir/lib/Target/LLVMIR/Dialect/OpenMP/OpenMPToLLVMIRTranslation.cpp
index 08ec57803aff..a59677c02fc3 100644
--- a/mlir/lib/Target/LLVMIR/Dialect/OpenMP/OpenMPToLLVMIRTranslation.cpp
+++ b/mlir/lib/Target/LLVMIR/Dialect/OpenMP/OpenMPToLLVMIRTranslation.cpp
@@ -3116,6 +3116,174 @@ convertDeclareTargetAttr(Operation *op, mlir::omp::DeclareTargetAttr attribute,
   return success();
 }
 
+// Returns true if the operation is inside a TargetOp or
+// is part of a declare target function.
+static bool isTargetDeviceOp(Operation *op) {
+  // Assumes no reverse offloading
+  if (op->getParentOfType<omp::TargetOp>())
+    return true;
+
+  if (auto parentFn = op->getParentOfType<LLVM::LLVMFuncOp>())
+    if (auto declareTargetIface =
+            llvm::dyn_cast<mlir::omp::DeclareTargetInterface>(
+                parentFn.getOperation()))
+      if (declareTargetIface.isDeclareTarget() &&
+          declareTargetIface.getDeclareTargetDeviceType() !=
+              mlir::omp::DeclareTargetDeviceType::host)
+        return true;
+
+  return false;
+}
+
+/// Given an OpenMP MLIR operation, create the corresponding LLVM IR
+/// (including OpenMP runtime calls).
+static LogicalResult
+convertHostOrTargetOperation(Operation *op, llvm::IRBuilderBase &builder,
+                             LLVM::ModuleTranslation &moduleTranslation) {
+
+  llvm::OpenMPIRBuilder *ompBuilder = moduleTranslation.getOpenMPBuilder();
+
+  return llvm::TypeSwitch<Operation *, LogicalResult>(op)
+      .Case([&](omp::BarrierOp) {
+        ompBuilder->createBarrier(builder.saveIP(), llvm::omp::OMPD_barrier);
+        return success();
+      })
+      .Case([&](omp::TaskwaitOp) {
+        ompBuilder->createTaskwait(builder.saveIP());
+        return success();
+      })
+      .Case([&](omp::TaskyieldOp) {
+        ompBuilder->createTaskyield(builder.saveIP());
+        return success();
+      })
+      .Case([&](omp::FlushOp) {
+        // No support in Openmp runtime function (__kmpc_flush) to accept
+        // the argument list.
+        // OpenMP standard states the following:
+        //  "An implementation may implement a flush with a list by ignoring
+        //   the list, and treating it the same as a flush without a list."
+        //
+        // The argument list is discarded so that, flush with a list is treated
+        // same as a flush without a list.
+        ompBuilder->createFlush(builder.saveIP());
+        return success();
+      })
+      .Case([&](omp::ParallelOp op) {
+        return convertOmpParallel(op, builder, moduleTranslation);
+      })
+      .Case([&](omp::ReductionOp reductionOp) {
+        return convertOmpReductionOp(reductionOp, builder, moduleTranslation);
+      })
+      .Case([&](omp::MasterOp) {
+        return convertOmpMaster(*op, builder, moduleTranslation);
+      })
+      .Case([&](omp::CriticalOp) {
+        return convertOmpCritical(*op, builder, moduleTranslation);
+      })
+      .Case([&](omp::OrderedRegionOp) {
+        return convertOmpOrderedRegion(*op, builder, moduleTranslation);
+      })
+      .Case([&](omp::OrderedOp) {
+        return convertOmpOrdered(*op, builder, moduleTranslation);
+      })
+      .Case([&](omp::WsloopOp) {
+        return convertOmpWsloop(*op, builder, moduleTranslation);
+      })
+      .Case([&](omp::SimdLoopOp) {
+        return convertOmpSimdLoop(*op, builder, moduleTranslation);
+      })
+      .Case([&](omp::AtomicReadOp) {
+        return convertOmpAtomicRead(*op, builder, moduleTranslation);
+      })
+      .Case([&](omp::AtomicWriteOp) {
+        return convertOmpAtomicWrite(*op, builder, moduleTranslation);
+      })
+      .Case([&](omp::AtomicUpdateOp op) {
+        return convertOmpAtomicUpdate(op, builder, moduleTranslation);
+      })
+      .Case([&](omp::AtomicCaptureOp op) {
+        return convertOmpAtomicCapture(op, builder, moduleTranslation);
+      })
+      .Case([&](omp::SectionsOp) {
+        return convertOmpSections(*op, builder, moduleTranslation);
+      })
+      .Case([&](omp::SingleOp op) {
+        return convertOmpSingle(op, builder, moduleTranslation);
+      })
+      .Case([&](omp::TeamsOp op) {
+        return convertOmpTeams(op, builder, moduleTranslation);
+      })
+      .Case([&](omp::TaskOp op) {
+        return convertOmpTaskOp(op, builder, moduleTranslation);
+      })
+      .Case([&](omp::TaskgroupOp op) {
+        return convertOmpTaskgroupOp(op, builder, moduleTranslation);
+      })
+      .Case<omp::YieldOp, omp::TerminatorOp, omp::DeclareReductionOp,
+            omp::CriticalDeclareOp>([](auto op) {
+        // `yield` and `terminator` can be just omitted. The block structure
+        // was created in the region that handles their parent operation.
+        // `declare_reduction` will be used by reductions and is not
+        // converted directly, skip it.
+        // `critical.declare` is only used to declare names of critical
+        // sections which will be used by `critical` ops and hence can be
+        // ignored for lowering. The OpenMP IRBuilder will create unique
+        // name for critical section names.
+        return success();
+      })
+      .Case([&](omp::ThreadprivateOp) {
+        return convertOmpThreadprivate(*op, builder, moduleTranslation);
+      })
+      .Case<omp::TargetDataOp, omp::TargetEnterDataOp, omp::TargetExitDataOp,
+            omp::TargetUpdateOp>([&](auto op) {
+        return convertOmpTargetData(op, builder, moduleTranslation);
+      })
+      .Case([&](omp::TargetOp) {
+        return convertOmpTarget(*op, builder, moduleTranslation);
+      })
+      .Case<omp::MapInfoOp, omp::MapBoundsOp, omp::PrivateClauseOp>(
+          [&](auto op) {
+            // No-op, should be handled by relevant owning operations e.g.
+            // TargetOp, TargetEnterDataOp, TargetExitDataOp, TargetDataOp etc.
+            // and then discarded
+            return success();
+          })
+      .Default([&](Operation *inst) {
+        return inst->emitError("unsupported OpenMP operation: ")
+               << inst->getName();
+      });
+}
+
+static LogicalResult
+convertTargetDeviceOp(Operation *op, llvm::IRBuilderBase &builder,
+                      LLVM::ModuleTranslation &moduleTranslation) {
+  return convertHostOrTargetOperation(op, builder, moduleTranslation);
+}
+
+static LogicalResult
+convertTargetOpsInNest(Operation *op, llvm::IRBuilderBase &builder,
+                       LLVM::ModuleTranslation &moduleTranslation) {
+  if (isa<omp::TargetOp>(op))
+    return convertOmpTarget(*op, builder, moduleTranslation);
+  if (isa<omp::TargetDataOp>(op))
+    return convertOmpTargetData(op, builder, moduleTranslation);
+  bool interrupted =
+      op->walk<WalkOrder::PreOrder>([&](Operation *oper) {
+          if (isa<omp::TargetOp>(oper)) {
+            if (failed(convertOmpTarget(*oper, builder, moduleTranslation)))
+              return WalkResult::interrupt();
+            return WalkResult::skip();
+          }
+          if (isa<omp::TargetDataOp>(oper)) {
+            if (failed(convertOmpTargetData(oper, builder, moduleTranslation)))
+              return WalkResult::interrupt();
+            return WalkResult::skip();
+          }
+          return WalkResult::advance();
+        }).wasInterrupted();
+  return failure(interrupted);
+}
+
 namespace {
 
 /// Implementation of the dialect interface that converts operations belonging
@@ -3131,8 +3299,8 @@ public:
   convertOperation(Operation *op, llvm::IRBuilderBase &builder,
                    LLVM::ModuleTranslation &moduleTranslation) const final;
 
-  /// Given an OpenMP MLIR attribute, create the corresponding LLVM-IR, runtime
-  /// calls, or operation amendments
+  /// Given an OpenMP MLIR attribute, create the corresponding LLVM-IR,
+  /// runtime calls, or operation amendments
   LogicalResult
   amendOperation(Operation *op, ArrayRef<llvm::Instruction *> instructions,
                  NamedAttribute attribute,
@@ -3237,116 +3405,15 @@ LogicalResult OpenMPDialectLLVMIRTranslationInterface::convertOperation(
     LLVM::ModuleTranslation &moduleTranslation) const {
 
   llvm::OpenMPIRBuilder *ompBuilder = moduleTranslation.getOpenMPBuilder();
+  if (ompBuilder->Config.isTargetDevice()) {
+    if (isTargetDeviceOp(op)) {
+      return convertTargetDeviceOp(op, builder, moduleTranslation);
+    } else {
+      return convertTargetOpsInNest(op, builder, moduleTranslation);
+    }
+  }
 
-  return llvm::TypeSwitch<Operation *, LogicalResult>(op)
-      .Case([&](omp::BarrierOp) {
-        ompBuilder->createBarrier(builder.saveIP(), llvm::omp::OMPD_barrier);
-        return success();
-      })
-      .Case([&](omp::TaskwaitOp) {
-        ompBuilder->createTaskwait(builder.saveIP());
-        return success();
-      })
-      .Case([&](omp::TaskyieldOp) {
-        ompBuilder->createTaskyield(builder.saveIP());
-        return success();
-      })
-      .Case([&](omp::FlushOp) {
-        // No support in Openmp runtime function (__kmpc_flush) to accept
-        // the argument list.
-        // OpenMP standard states the following:
-        //  "An implementation may implement a flush with a list by ignoring
-        //   the list, and treating it the same as a flush without a list."
-        //
-        // The argument list is discarded so that, flush with a list is treated
-        // same as a flush without a list.
-        ompBuilder->createFlush(builder.saveIP());
-        return success();
-      })
-      .Case([&](omp::ParallelOp op) {
-        return convertOmpParallel(op, builder, moduleTranslation);
-      })
-      .Case([&](omp::ReductionOp reductionOp) {
-        return convertOmpReductionOp(reductionOp, builder, moduleTranslation);
-      })
-      .Case([&](omp::MasterOp) {
-        return convertOmpMaster(*op, builder, moduleTranslation);
-      })
-      .Case([&](omp::CriticalOp) {
-        return convertOmpCritical(*op, builder, moduleTranslation);
-      })
-      .Case([&](omp::OrderedRegionOp) {
-        return convertOmpOrderedRegion(*op, builder, moduleTranslation);
-      })
-      .Case([&](omp::OrderedOp) {
-        return convertOmpOrdered(*op, builder, moduleTranslation);
-      })
-      .Case([&](omp::WsloopOp) {
-        return convertOmpWsloop(*op, builder, moduleTranslation);
-      })
-      .Case([&](omp::SimdLoopOp) {
-        return convertOmpSimdLoop(*op, builder, moduleTranslation);
-      })
-      .Case([&](omp::AtomicReadOp) {
-        return convertOmpAtomicRead(*op, builder, moduleTranslation);
-      })
-      .Case([&](omp::AtomicWriteOp) {
-        return convertOmpAtomicWrite(*op, builder, moduleTranslation);
-      })
-      .Case([&](omp::AtomicUpdateOp op) {
-        return convertOmpAtomicUpdate(op, builder, moduleTranslation);
-      })
-      .Case([&](omp::AtomicCaptureOp op) {
-        return convertOmpAtomicCapture(op, builder, moduleTranslation);
-      })
-      .Case([&](omp::SectionsOp) {
-        return convertOmpSections(*op, builder, moduleTranslation);
-      })
-      .Case([&](omp::SingleOp op) {
-        return convertOmpSingle(op, builder, moduleTranslation);
-      })
-      .Case([&](omp::TeamsOp op) {
-        return convertOmpTeams(op, builder, moduleTranslation);
-      })
-      .Case([&](omp::TaskOp op) {
-        return convertOmpTaskOp(op, builder, moduleTranslation);
-      })
-      .Case([&](omp::TaskgroupOp op) {
-        return convertOmpTaskgroupOp(op, builder, moduleTranslation);
-      })
-      .Case<omp::YieldOp, omp::TerminatorOp, omp::DeclareReductionOp,
-            omp::CriticalDeclareOp>([](auto op) {
-        // `yield` and `terminator` can be just omitted. The block structure
-        // was created in the region that handles their parent operation.
-        // `declare_reduction` will be used by reductions and is not
-        // converted directly, skip it.
-        // `critical.declare` is only used to declare names of critical
-        // sections which will be used by `critical` ops and hence can be
-        // ignored for lowering. The OpenMP IRBuilder will create unique
-        // name for critical section names.
-        return success();
-      })
-      .Case([&](omp::ThreadprivateOp) {
-        return convertOmpThreadprivate(*op, builder, moduleTranslation);
-      })
-      .Case<omp::TargetDataOp, omp::TargetEnterDataOp, omp::TargetExitDataOp,
-            omp::TargetUpdateOp>([&](auto op) {
-        return convertOmpTargetData(op, builder, moduleTranslation);
-      })
-      .Case([&](omp::TargetOp) {
-        return convertOmpTarget(*op, builder, moduleTranslation);
-      })
-      .Case<omp::MapInfoOp, omp::MapBoundsOp, omp::PrivateClauseOp>(
-          [&](auto op) {
-            // No-op, should be handled by relevant owning operations e.g.
-            // TargetOp, TargetEnterDataOp, TargetExitDataOp, TargetDataOp etc.
-            // and then discarded
-            return success();
-          })
-      .Default([&](Operation *inst) {
-        return inst->emitError("unsupported OpenMP operation: ")
-               << inst->getName();
-      });
+  return convertHostOrTargetOperation(op, builder, moduleTranslation);
 }
 
 void mlir::registerOpenMPDialectTranslation(DialectRegistry &registry) {
diff --git a/mlir/test/CMakeLists.txt b/mlir/test/CMakeLists.txt
index baf07ea1f010..5319a9cb33e0 100644
--- a/mlir/test/CMakeLists.txt
+++ b/mlir/test/CMakeLists.txt
@@ -1,3 +1,5 @@
+include(MLIRCheckHardwareFeatures)
+
 add_subdirectory(CAPI)
 add_subdirectory(lib)
 
@@ -39,6 +41,10 @@ if (MLIR_INCLUDE_INTEGRATION_TESTS)
   option(MLIR_RUN_ARM_SVE_TESTS "Run Arm SVE tests.")
   option(MLIR_RUN_ARM_SME_TESTS "Run Arm SME tests.")
 
+  # Check whether an emulator is required - if yes then make sure that it's
+  # been set.
+  check_emulator(MLIR_RUN_ARM_SVE_TESTS "HWCAP_SVE" ARM_EMULATOR_EXECUTABLE)
+  check_emulator(MLIR_RUN_ARM_SME_TESTS "HWCAP2_SME" ARM_EMULATOR_EXECUTABLE)
 
   # The native target may not be enabled when cross compiling, raise an error.
   if(NOT MLIR_ENABLE_EXECUTION_ENGINE)
diff --git a/mlir/test/Dialect/Arith/vscale_constants.mlir b/mlir/test/Dialect/Arith/vscale_constants.mlir
new file mode 100644
index 000000000000..324766f49980
--- /dev/null
+++ b/mlir/test/Dialect/Arith/vscale_constants.mlir
@@ -0,0 +1,14 @@
+// RUN: mlir-opt %s | FileCheck %s
+
+// Note: This test is checking value names (so deliberately is not using a regex match).
+
+func.func @test_vscale_constant_names() {
+  %vscale = vector.vscale
+  %c8 = arith.constant 8 : index
+  // CHECK: %c8_vscale = arith.muli
+  %0 = arith.muli %vscale, %c8 : index
+  %c10 = arith.constant 10 : index
+  // CHECK: %c10_vscale = arith.muli
+  %1 = arith.muli %c10, %vscale : index
+  return
+}
diff --git a/mlir/test/Dialect/LLVMIR/canonicalize.mlir b/mlir/test/Dialect/LLVMIR/canonicalize.mlir
index 5e26fa37b681..6b265bbbdbfb 100644
--- a/mlir/test/Dialect/LLVMIR/canonicalize.mlir
+++ b/mlir/test/Dialect/LLVMIR/canonicalize.mlir
@@ -8,6 +8,8 @@ llvm.func @fold_icmp_eq(%arg0 : i32) -> i1 {
   llvm.return %0 : i1
 }
 
+// -----
+
 // CHECK-LABEL: @fold_icmp_ne
 llvm.func @fold_icmp_ne(%arg0 : vector<2xi32>) -> vector<2xi1> {
   // CHECK: %[[C0:.*]] = llvm.mlir.constant(dense<false> : vector<2xi1>) : vector<2xi1>
@@ -16,6 +18,8 @@ llvm.func @fold_icmp_ne(%arg0 : vector<2xi32>) -> vector<2xi1> {
   llvm.return %0 : vector<2xi1>
 }
 
+// -----
+
 // CHECK-LABEL: @fold_icmp_alloca
 llvm.func @fold_icmp_alloca() -> i1 {
   // CHECK: %[[C0:.*]] = llvm.mlir.constant(true) : i1
@@ -83,16 +87,18 @@ llvm.func @fold_unrelated_extractvalue(%arr: !llvm.array<4 x f32>) -> f32 {
 // -----
 
 // CHECK-LABEL: fold_bitcast
-// CHECK-SAME: %[[a0:arg[0-9]+]]
-// CHECK-NEXT: llvm.return %[[a0]]
+// CHECK-SAME: %[[ARG:[[:alnum:]]+]]
+// CHECK-NEXT: llvm.return %[[ARG]]
 llvm.func @fold_bitcast(%x : !llvm.ptr) -> !llvm.ptr {
   %c = llvm.bitcast %x : !llvm.ptr to !llvm.ptr
   llvm.return %c : !llvm.ptr
 }
 
+// -----
+
 // CHECK-LABEL: fold_bitcast2
-// CHECK-SAME: %[[a0:arg[0-9]+]]
-// CHECK-NEXT: llvm.return %[[a0]]
+// CHECK-SAME: %[[ARG:[[:alnum:]]+]]
+// CHECK-NEXT: llvm.return %[[ARG]]
 llvm.func @fold_bitcast2(%x : i32) -> i32 {
   %c = llvm.bitcast %x : i32 to f32
   %d = llvm.bitcast %c : f32 to i32
@@ -101,17 +107,31 @@ llvm.func @fold_bitcast2(%x : i32) -> i32 {
 
 // -----
 
+// CHECK-LABEL: fold_bitcast_chain
+// CHECK-SAME: %[[ARG:[[:alnum:]]+]]
+llvm.func @fold_bitcast_chain(%x : i32) -> vector<2xi16> {
+  %c = llvm.bitcast %x : i32 to f32
+  %d = llvm.bitcast %c : f32 to vector<2xi16>
+  // CHECK: %[[BITCAST:.*]] = llvm.bitcast %[[ARG]] : i32 to vector<2xi16>
+  // CHECK: llvm.return %[[BITCAST]]
+  llvm.return %d : vector<2xi16>
+}
+
+// -----
+
 // CHECK-LABEL: fold_addrcast
-// CHECK-SAME: %[[a0:arg[0-9]+]]
-// CHECK-NEXT: llvm.return %[[a0]]
+// CHECK-SAME: %[[ARG:[[:alnum:]]+]]
+// CHECK-NEXT: llvm.return %[[ARG]]
 llvm.func @fold_addrcast(%x : !llvm.ptr) -> !llvm.ptr {
   %c = llvm.addrspacecast %x : !llvm.ptr to !llvm.ptr
   llvm.return %c : !llvm.ptr
 }
 
+// -----
+
 // CHECK-LABEL: fold_addrcast2
-// CHECK-SAME: %[[a0:arg[0-9]+]]
-// CHECK-NEXT: llvm.return %[[a0]]
+// CHECK-SAME: %[[ARG:[[:alnum:]]+]]
+// CHECK-NEXT: llvm.return %[[ARG]]
 llvm.func @fold_addrcast2(%x : !llvm.ptr) -> !llvm.ptr {
   %c = llvm.addrspacecast %x : !llvm.ptr to !llvm.ptr<5>
   %d = llvm.addrspacecast %c : !llvm.ptr<5> to !llvm.ptr
@@ -120,18 +140,32 @@ llvm.func @fold_addrcast2(%x : !llvm.ptr) -> !llvm.ptr {
 
 // -----
 
+// CHECK-LABEL: fold_addrcast_chain
+// CHECK-SAME: %[[ARG:[[:alnum:]]+]]
+llvm.func @fold_addrcast_chain(%x : !llvm.ptr) -> !llvm.ptr<2> {
+  %c = llvm.addrspacecast %x : !llvm.ptr to !llvm.ptr<1>
+  %d = llvm.addrspacecast %c : !llvm.ptr<1> to !llvm.ptr<2>
+  // CHECK: %[[ADDRCAST:.*]] = llvm.addrspacecast %[[ARG]] : !llvm.ptr to !llvm.ptr<2>
+  // CHECK: llvm.return %[[ADDRCAST]]
+  llvm.return %d : !llvm.ptr<2>
+}
+
+// -----
+
 // CHECK-LABEL: fold_gep
-// CHECK-SAME: %[[a0:arg[0-9]+]]
-// CHECK-NEXT: llvm.return %[[a0]]
+// CHECK-SAME: %[[ARG:[[:alnum:]]+]]
+// CHECK-NEXT: llvm.return %[[ARG]]
 llvm.func @fold_gep(%x : !llvm.ptr) -> !llvm.ptr {
   %c0 = arith.constant 0 : i32
   %c = llvm.getelementptr %x[%c0] : (!llvm.ptr, i32) -> !llvm.ptr, i8
   llvm.return %c : !llvm.ptr
 }
 
+// -----
+
 // CHECK-LABEL: fold_gep_neg
-// CHECK-SAME: %[[a0:arg[0-9]+]]
-// CHECK-NEXT: %[[RES:.*]] = llvm.getelementptr inbounds %[[a0]][0, 1]
+// CHECK-SAME: %[[ARG:[[:alnum:]]+]]
+// CHECK-NEXT: %[[RES:.*]] = llvm.getelementptr inbounds %[[ARG]][0, 1]
 // CHECK-NEXT: llvm.return %[[RES]]
 llvm.func @fold_gep_neg(%x : !llvm.ptr) -> !llvm.ptr {
   %c0 = arith.constant 0 : i32
@@ -139,9 +173,11 @@ llvm.func @fold_gep_neg(%x : !llvm.ptr) -> !llvm.ptr {
   llvm.return %0 : !llvm.ptr
 }
 
+// -----
+
 // CHECK-LABEL: fold_gep_canon
-// CHECK-SAME: %[[a0:arg[0-9]+]]
-// CHECK-NEXT: %[[RES:.*]] = llvm.getelementptr %[[a0]][2]
+// CHECK-SAME: %[[ARG:[[:alnum:]]+]]
+// CHECK-NEXT: %[[RES:.*]] = llvm.getelementptr %[[ARG]][2]
 // CHECK-NEXT: llvm.return %[[RES]]
 llvm.func @fold_gep_canon(%x : !llvm.ptr) -> !llvm.ptr {
   %c2 = arith.constant 2 : i32
@@ -175,6 +211,8 @@ llvm.func @load_dce(%x : !llvm.ptr) {
   llvm.return
 }
 
+// -----
+
 llvm.mlir.global external @fp() : !llvm.ptr
 
 // CHECK-LABEL: addr_dce
@@ -184,6 +222,8 @@ llvm.func @addr_dce(%x : !llvm.ptr) {
   llvm.return
 }
 
+// -----
+
 // CHECK-LABEL: alloca_dce
 // CHECK-NEXT: llvm.return
 llvm.func @alloca_dce() {
diff --git a/mlir/test/Dialect/LLVMIR/mem2reg.mlir b/mlir/test/Dialect/LLVMIR/mem2reg.mlir
index 90e56c1166ed..fa5d842302d0 100644
--- a/mlir/test/Dialect/LLVMIR/mem2reg.mlir
+++ b/mlir/test/Dialect/LLVMIR/mem2reg.mlir
@@ -697,3 +697,249 @@ llvm.func @transitive_reaching_def() -> !llvm.ptr {
   %3 = llvm.load %1 {alignment = 8 : i64} : !llvm.ptr -> !llvm.ptr
   llvm.return %3 : !llvm.ptr
 }
+
+// -----
+
+// CHECK-LABEL: @load_int_from_float
+llvm.func @load_int_from_float() -> i32 {
+  %0 = llvm.mlir.constant(1 : i32) : i32
+  // CHECK-NOT: llvm.alloca
+  %1 = llvm.alloca %0 x f32 {alignment = 4 : i64} : (i32) -> !llvm.ptr
+  %2 = llvm.load %1 {alignment = 4 : i64} : !llvm.ptr -> i32
+  // CHECK: %[[UNDEF:.*]] = llvm.mlir.undef
+  // CHECK: %[[BITCAST:.*]] = llvm.bitcast %[[UNDEF]] : f32 to i32
+  // CHECK: llvm.return %[[BITCAST:.*]]
+  llvm.return %2 : i32
+}
+
+// -----
+
+// CHECK-LABEL: @load_float_from_int
+llvm.func @load_float_from_int() -> f32 {
+  %0 = llvm.mlir.constant(1 : i32) : i32
+  // CHECK-NOT: llvm.alloca
+  %1 = llvm.alloca %0 x i32 {alignment = 4 : i64} : (i32) -> !llvm.ptr
+  %2 = llvm.load %1 {alignment = 4 : i64} : !llvm.ptr -> f32
+  // CHECK: %[[UNDEF:.*]] = llvm.mlir.undef
+  // CHECK: %[[BITCAST:.*]] = llvm.bitcast %[[UNDEF]] : i32 to f32
+  // CHECK: llvm.return %[[BITCAST:.*]]
+  llvm.return %2 : f32
+}
+
+// -----
+
+// CHECK-LABEL: @load_int_from_vector
+llvm.func @load_int_from_vector() -> i32 {
+  %0 = llvm.mlir.constant(1 : i32) : i32
+  // CHECK-NOT: llvm.alloca
+  %1 = llvm.alloca %0 x vector<2xi16> : (i32) -> !llvm.ptr
+  %2 = llvm.load %1 {alignment = 4 : i64} : !llvm.ptr -> i32
+  // CHECK: %[[UNDEF:.*]] = llvm.mlir.undef
+  // CHECK: %[[BITCAST:.*]] = llvm.bitcast %[[UNDEF]] : vector<2xi16> to i32
+  // CHECK: llvm.return %[[BITCAST:.*]]
+  llvm.return %2 : i32
+}
+
+// -----
+
+// LLVM arrays cannot be bitcasted, so the following cannot be promoted.
+
+// CHECK-LABEL: @load_int_from_array
+llvm.func @load_int_from_array() -> i32 {
+  %0 = llvm.mlir.constant(1 : i32) : i32
+  // CHECK: llvm.alloca
+  %1 = llvm.alloca %0 x !llvm.array<2 x i16> : (i32) -> !llvm.ptr
+  %2 = llvm.load %1 {alignment = 4 : i64} : !llvm.ptr -> i32
+  // CHECK-NOT: llvm.bitcast
+  llvm.return %2 : i32
+}
+
+// -----
+
+// CHECK-LABEL: @store_int_to_float
+// CHECK-SAME: %[[ARG:.*]]: i32
+llvm.func @store_int_to_float(%arg: i32) -> i32 {
+  %0 = llvm.mlir.constant(1 : i32) : i32
+  // CHECK-NOT: llvm.alloca
+  %1 = llvm.alloca %0 x f32 {alignment = 4 : i64} : (i32) -> !llvm.ptr
+  llvm.store %arg, %1 {alignment = 4 : i64} : i32, !llvm.ptr
+  %2 = llvm.load %1 {alignment = 4 : i64} : !llvm.ptr -> i32
+  // CHECK: llvm.return %[[ARG]]
+  llvm.return %2 : i32
+}
+
+// -----
+
+// CHECK-LABEL: @store_float_to_int
+// CHECK-SAME: %[[ARG:.*]]: f32
+llvm.func @store_float_to_int(%arg: f32) -> i32 {
+  %0 = llvm.mlir.constant(1 : i32) : i32
+  // CHECK-NOT: llvm.alloca
+  %1 = llvm.alloca %0 x i32 {alignment = 4 : i64} : (i32) -> !llvm.ptr
+  llvm.store %arg, %1 {alignment = 4 : i64} : f32, !llvm.ptr
+  %2 = llvm.load %1 {alignment = 4 : i64} : !llvm.ptr -> i32
+  // CHECK: %[[BITCAST:.*]] = llvm.bitcast %[[ARG]] : f32 to i32
+  // CHECK: llvm.return %[[BITCAST]]
+  llvm.return %2 : i32
+}
+
+// -----
+
+// CHECK-LABEL: @store_int_to_vector
+// CHECK-SAME: %[[ARG:.*]]: i32
+llvm.func @store_int_to_vector(%arg: i32) -> vector<4xi8> {
+  %0 = llvm.mlir.constant(1 : i32) : i32
+  // CHECK-NOT: llvm.alloca
+  %1 = llvm.alloca %0 x vector<2xi16> {alignment = 4 : i64} : (i32) -> !llvm.ptr
+  llvm.store %arg, %1 {alignment = 4 : i64} : i32, !llvm.ptr
+  %2 = llvm.load %1 {alignment = 4 : i64} : !llvm.ptr -> vector<4xi8>
+  // CHECK: %[[BITCAST:.*]] = llvm.bitcast %[[ARG]] : i32 to vector<4xi8>
+  // CHECK: llvm.return %[[BITCAST]]
+  llvm.return %2 : vector<4xi8>
+}
+
+// -----
+
+// CHECK-LABEL: @load_ptr_from_int
+llvm.func @load_ptr_from_int() -> !llvm.ptr {
+  %0 = llvm.mlir.constant(1 : i32) : i32
+  // CHECK-NOT: llvm.alloca
+  %1 = llvm.alloca %0 x i64 {alignment = 4 : i64} : (i32) -> !llvm.ptr
+  %2 = llvm.load %1 {alignment = 4 : i64} : !llvm.ptr -> !llvm.ptr
+  // CHECK: %[[UNDEF:.*]] = llvm.mlir.undef
+  // CHECK: %[[CAST:.*]] = llvm.inttoptr %[[UNDEF]] : i64 to !llvm.ptr
+  // CHECK: llvm.return %[[CAST:.*]]
+  llvm.return %2 : !llvm.ptr
+}
+
+// -----
+
+// CHECK-LABEL: @load_int_from_ptr
+llvm.func @load_int_from_ptr() -> i64 {
+  %0 = llvm.mlir.constant(1 : i32) : i32
+  // CHECK-NOT: llvm.alloca
+  %1 = llvm.alloca %0 x !llvm.ptr {alignment = 4 : i64} : (i32) -> !llvm.ptr
+  %2 = llvm.load %1 {alignment = 4 : i64} : !llvm.ptr -> i64
+  // CHECK: %[[UNDEF:.*]] = llvm.mlir.undef
+  // CHECK: %[[CAST:.*]] = llvm.ptrtoint %[[UNDEF]] : !llvm.ptr to i64
+  // CHECK: llvm.return %[[CAST:.*]]
+  llvm.return %2 : i64
+}
+
+// -----
+
+// CHECK-LABEL: @load_ptr_addrspace_cast
+llvm.func @load_ptr_addrspace_cast() -> !llvm.ptr<2> {
+  %0 = llvm.mlir.constant(1 : i32) : i32
+  // CHECK-NOT: llvm.alloca
+  %1 = llvm.alloca %0 x !llvm.ptr<1> {alignment = 4 : i64} : (i32) -> !llvm.ptr
+  %2 = llvm.load %1 {alignment = 4 : i64} : !llvm.ptr -> !llvm.ptr<2>
+  // CHECK: %[[UNDEF:.*]] = llvm.mlir.undef
+  // CHECK: %[[CAST:.*]] = llvm.addrspacecast %[[UNDEF]] : !llvm.ptr<1> to !llvm.ptr<2>
+  // CHECK: llvm.return %[[CAST:.*]]
+  llvm.return %2 : !llvm.ptr<2>
+}
+
+// -----
+
+// CHECK-LABEL: @stores_with_different_types
+// CHECK-SAME: %[[ARG0:.*]]: i64
+// CHECK-SAME: %[[ARG1:.*]]: f64
+llvm.func @stores_with_different_types(%arg0: i64, %arg1: f64, %cond: i1) -> f64 {
+  %0 = llvm.mlir.constant(1 : i32) : i32
+  // CHECK-NOT: llvm.alloca
+  %1 = llvm.alloca %0 x i64 {alignment = 4 : i64} : (i32) -> !llvm.ptr
+  llvm.cond_br %cond, ^bb1, ^bb2
+^bb1:
+  llvm.store %arg0, %1 {alignment = 4 : i64} : i64, !llvm.ptr
+  // CHECK: llvm.br ^[[BB3:.*]](%[[ARG0]]
+  llvm.br ^bb3
+^bb2:
+  llvm.store %arg1, %1 {alignment = 4 : i64} : f64, !llvm.ptr
+  // CHECK: %[[BITCAST:.*]] = llvm.bitcast %[[ARG1]] : f64 to i64
+  // CHECK: llvm.br ^[[BB3]](%[[BITCAST]]
+  llvm.br ^bb3
+// CHECK: ^[[BB3]](%[[BLOCK_ARG:.*]]: i64)
+^bb3:
+  %2 = llvm.load %1 {alignment = 4 : i64} : !llvm.ptr -> f64
+  // CHECK: %[[BITCAST:.*]] = llvm.bitcast %[[BLOCK_ARG]] : i64 to f64
+  // CHECK: llvm.return %[[BITCAST]]
+  llvm.return %2 : f64
+}
+
+// -----
+
+// Verifies that stores with smaller bitsize inputs are not replaced. A trivial
+// implementation will be incorrect due to endianness considerations.
+
+// CHECK-LABEL: @stores_with_different_type_sizes
+llvm.func @stores_with_different_type_sizes(%arg0: i64, %arg1: f32, %cond: i1) -> f64 {
+  %0 = llvm.mlir.constant(1 : i32) : i32
+  // CHECK: llvm.alloca
+  %1 = llvm.alloca %0 x i64 {alignment = 4 : i64} : (i32) -> !llvm.ptr
+  llvm.cond_br %cond, ^bb1, ^bb2
+^bb1:
+  llvm.store %arg0, %1 {alignment = 4 : i64} : i64, !llvm.ptr
+  llvm.br ^bb3
+^bb2:
+  llvm.store %arg1, %1 {alignment = 4 : i64} : f32, !llvm.ptr
+  llvm.br ^bb3
+^bb3:
+  %2 = llvm.load %1 {alignment = 4 : i64} : !llvm.ptr -> f64
+  llvm.return %2 : f64
+}
+
+// -----
+
+// CHECK-LABEL: @load_smaller_int
+llvm.func @load_smaller_int() -> i16 {
+  %0 = llvm.mlir.constant(1 : i32) : i32
+  // CHECK: llvm.alloca
+  %1 = llvm.alloca %0 x i32 {alignment = 4 : i64} : (i32) -> !llvm.ptr
+  %2 = llvm.load %1 {alignment = 4 : i64} : !llvm.ptr -> i16
+  llvm.return %2 : i16
+}
+
+// -----
+
+// CHECK-LABEL: @load_different_type_smaller
+llvm.func @load_different_type_smaller() -> f32 {
+  %0 = llvm.mlir.constant(1 : i32) : i32
+  // CHECK: llvm.alloca
+  %1 = llvm.alloca %0 x i64 {alignment = 8 : i64} : (i32) -> !llvm.ptr
+  %2 = llvm.load %1 {alignment = 4 : i64} : !llvm.ptr -> f32
+  llvm.return %2 : f32
+}
+
+// -----
+
+// This alloca is too small for the load, still, mem2reg should not touch it.
+
+// CHECK-LABEL: @impossible_load
+llvm.func @impossible_load() -> f64 {
+  %0 = llvm.mlir.constant(1 : i32) : i32
+  // CHECK: llvm.alloca
+  %1 = llvm.alloca %0 x i32 {alignment = 4 : i64} : (i32) -> !llvm.ptr
+  %2 = llvm.load %1 {alignment = 4 : i64} : !llvm.ptr -> f64
+  llvm.return %2 : f64
+}
+
+// -----
+
+// Verifies that mem2reg does not introduce address space casts of pointers
+// with different bitsize.
+
+module attributes { dlti.dl_spec = #dlti.dl_spec<
+  #dlti.dl_entry<!llvm.ptr<1>, dense<[32, 64, 64]> : vector<3xi64>>,
+  #dlti.dl_entry<!llvm.ptr<2>, dense<[64, 64, 64]> : vector<3xi64>>
+>} {
+
+  // CHECK-LABEL: @load_ptr_addrspace_cast_different_size
+  llvm.func @load_ptr_addrspace_cast_different_size() -> !llvm.ptr<2> {
+    %0 = llvm.mlir.constant(1 : i32) : i32
+    // CHECK: llvm.alloca
+    %1 = llvm.alloca %0 x !llvm.ptr<1> {alignment = 4 : i64} : (i32) -> !llvm.ptr
+    %2 = llvm.load %1 {alignment = 4 : i64} : !llvm.ptr -> !llvm.ptr<2>
+    llvm.return %2 : !llvm.ptr<2>
+  }
+}
diff --git a/mlir/test/Dialect/LLVMIR/type-consistency.mlir b/mlir/test/Dialect/LLVMIR/type-consistency.mlir
index a6176142f174..c9c1355d16df 100644
--- a/mlir/test/Dialect/LLVMIR/type-consistency.mlir
+++ b/mlir/test/Dialect/LLVMIR/type-consistency.mlir
@@ -157,8 +157,7 @@ llvm.func @coalesced_store_floats(%arg: i64) {
   // CHECK: %[[SHR:.*]] = llvm.lshr %[[ARG]], %[[CST32]] : i64
   // CHECK: %[[TRUNC:.*]] = llvm.trunc %[[SHR]] : i64 to i32
   // CHECK: %[[GEP:.*]] = llvm.getelementptr %[[ALLOCA]][0, 1] : (!llvm.ptr)  -> !llvm.ptr, !llvm.struct<"foo", (f32, f32)>
-  // CHECK: %[[BIT_CAST:.*]] = llvm.bitcast %[[TRUNC]] : i32 to f32
-  // CHECK: llvm.store %[[BIT_CAST]], %[[GEP]]
+  // CHECK: llvm.store %[[TRUNC]], %[[GEP]]
   llvm.store %arg, %1 : i64, !llvm.ptr
   // CHECK-NOT: llvm.store %[[ARG]], %[[ALLOCA]]
   llvm.return
@@ -327,21 +326,6 @@ llvm.func @vector_write_split_struct(%arg: vector<2xi64>) {
 
 // -----
 
-// CHECK-LABEL: llvm.func @bitcast_insertion
-// CHECK-SAME: %[[ARG:.*]]: i32
-llvm.func @bitcast_insertion(%arg: i32) {
-  %0 = llvm.mlir.constant(1 : i32) : i32
-  // CHECK: %[[ALLOCA:.*]] = llvm.alloca %{{.*}} x f32
-  %1 = llvm.alloca %0 x f32 : (i32) -> !llvm.ptr
-  // CHECK: %[[BIT_CAST:.*]] = llvm.bitcast %[[ARG]] : i32 to f32
-  // CHECK: llvm.store %[[BIT_CAST]], %[[ALLOCA]]
-  llvm.store %arg, %1 : i32, !llvm.ptr
-  // CHECK-NOT: llvm.store %[[ARG]], %[[ALLOCA]]
-  llvm.return
-}
-
-// -----
-
 // CHECK-LABEL: llvm.func @gep_split
 // CHECK-SAME: %[[ARG:.*]]: i64
 llvm.func @gep_split(%arg: i64) {
diff --git a/mlir/test/Target/LLVMIR/omptarget-parallel-wsloop.mlir b/mlir/test/Target/LLVMIR/omptarget-parallel-wsloop.mlir
index 8ab50f05f071..b0fe642238f1 100644
--- a/mlir/test/Target/LLVMIR/omptarget-parallel-wsloop.mlir
+++ b/mlir/test/Target/LLVMIR/omptarget-parallel-wsloop.mlir
@@ -4,10 +4,10 @@
 // for nested omp do loop inside omp target region
 
 module attributes {dlti.dl_spec = #dlti.dl_spec<#dlti.dl_entry<"dlti.alloca_memory_space", 5 : ui32>>, llvm.data_layout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7:8", llvm.target_triple = "amdgcn-amd-amdhsa", omp.is_gpu = true, omp.is_target_device = true } {
-  llvm.func @target_parallel_wsloop(%arg0: !llvm.ptr) attributes {
+  llvm.func @target_parallel_wsloop(%arg0: !llvm.ptr) attributes {omp.declare_target = #omp.declaretarget<device_type = (any), capture_clause = (to)>,
     target_cpu = "gfx90a",
-    target_features = #llvm.target_features<["+gfx9-insts", "+wavefrontsize64"]>
-  } {
+    target_features = #llvm.target_features<["+gfx9-insts", "+wavefrontsize64"]>}
+   {
     omp.parallel {
       %loop_ub = llvm.mlir.constant(9 : i32) : i32
       %loop_lb = llvm.mlir.constant(0 : i32) : i32
diff --git a/mlir/test/Target/LLVMIR/omptarget-target-inside-task.mlir b/mlir/test/Target/LLVMIR/omptarget-target-inside-task.mlir
new file mode 100644
index 000000000000..3d18e608d857
--- /dev/null
+++ b/mlir/test/Target/LLVMIR/omptarget-target-inside-task.mlir
@@ -0,0 +1,41 @@
+// RUN: mlir-translate -mlir-to-llvmir %s | FileCheck %s
+
+module attributes {omp.is_target_device = true, omp.is_gpu = true} {
+  llvm.func @omp_target_region_() {
+    %0 = llvm.mlir.constant(20 : i32) : i32
+    %1 = llvm.mlir.constant(10 : i32) : i32
+    %2 = llvm.mlir.constant(1 : i64) : i64
+    %3 = llvm.alloca %2 x i32 {bindc_name = "a", in_type = i32, operandSegmentSizes = array<i32: 0, 0>, uniq_name = "_QFomp_target_regionEa"} : (i64) -> !llvm.ptr
+    %4 = llvm.mlir.constant(1 : i64) : i64
+    %5 = llvm.alloca %4 x i32 {bindc_name = "b", in_type = i32, operandSegmentSizes = array<i32: 0, 0>, uniq_name = "_QFomp_target_regionEb"} : (i64) -> !llvm.ptr
+    %6 = llvm.mlir.constant(1 : i64) : i64
+    %7 = llvm.alloca %6 x i32 {bindc_name = "c", in_type = i32, operandSegmentSizes = array<i32: 0, 0>, uniq_name = "_QFomp_target_regionEc"} : (i64) -> !llvm.ptr
+    llvm.store %1, %3 : i32, !llvm.ptr
+    llvm.store %0, %5 : i32, !llvm.ptr
+    omp.task {
+        %map1 = omp.map.info var_ptr(%3 : !llvm.ptr, i32)   map_clauses(tofrom) capture(ByRef) -> !llvm.ptr {name = ""}
+        %map2 = omp.map.info var_ptr(%5 : !llvm.ptr, i32)   map_clauses(tofrom) capture(ByRef) -> !llvm.ptr {name = ""}
+        %map3 = omp.map.info var_ptr(%7 : !llvm.ptr, i32)   map_clauses(tofrom) capture(ByRef) -> !llvm.ptr {name = ""}
+      omp.target map_entries(%map1 -> %arg0, %map2 -> %arg1, %map3 -> %arg2 : !llvm.ptr, !llvm.ptr, !llvm.ptr) {
+      ^bb0(%arg0: !llvm.ptr, %arg1: !llvm.ptr, %arg2: !llvm.ptr):
+        %8 = llvm.load %arg0 : !llvm.ptr -> i32
+        %9 = llvm.load %arg1 : !llvm.ptr -> i32
+        %10 = llvm.add %8, %9  : i32
+        llvm.store %10, %arg2 : i32, !llvm.ptr
+        omp.terminator
+      }
+      omp.terminator
+    }
+   llvm.return
+  }
+
+  llvm.func @omp_target_no_map() {
+    omp.target {
+      omp.terminator
+    }
+    llvm.return
+  }
+}
+
+// CHECK: define weak_odr protected void @__omp_offloading_{{.*}}_{{.*}}_omp_target_region__l19
+// CHECK: ret void
diff --git a/mlir/test/Target/LLVMIR/omptarget-teams-llvm.mlir b/mlir/test/Target/LLVMIR/omptarget-teams-llvm.mlir
index 96cced7a1d58..c5f89eb2c327 100644
--- a/mlir/test/Target/LLVMIR/omptarget-teams-llvm.mlir
+++ b/mlir/test/Target/LLVMIR/omptarget-teams-llvm.mlir
@@ -5,7 +5,7 @@
 
 module attributes {omp.is_target_device = true} {
   llvm.func @foo(i32)
-  llvm.func @omp_target_teams_shared_simple(%arg0 : i32)  {
+  llvm.func @omp_target_teams_shared_simple(%arg0 : i32)  attributes {omp.declare_target = #omp.declaretarget<device_type = (any), capture_clause = (to)>} {
     omp.teams {
       llvm.call @foo(%arg0) : (i32) -> ()
       omp.terminator
diff --git a/mlir/test/Target/LLVMIR/omptarget-wsloop-collapsed.mlir b/mlir/test/Target/LLVMIR/omptarget-wsloop-collapsed.mlir
index e246c551886c..0d77423abcb4 100644
--- a/mlir/test/Target/LLVMIR/omptarget-wsloop-collapsed.mlir
+++ b/mlir/test/Target/LLVMIR/omptarget-wsloop-collapsed.mlir
@@ -4,7 +4,7 @@
 // for nested omp do loop with collapse clause inside omp target region
 
 module attributes {dlti.dl_spec = #dlti.dl_spec<#dlti.dl_entry<"dlti.alloca_memory_space", 5 : ui32>>, llvm.data_layout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7:8", llvm.target_triple = "amdgcn-amd-amdhsa", omp.is_gpu = true, omp.is_target_device = true } {
-  llvm.func @target_collapsed_wsloop(%arg0: !llvm.ptr) {
+  llvm.func @target_collapsed_wsloop(%arg0: !llvm.ptr) attributes {omp.declare_target = #omp.declaretarget<device_type = (any), capture_clause = (to)>} {
     %loop_ub = llvm.mlir.constant(99 : i32) : i32
     %loop_lb = llvm.mlir.constant(0 : i32) : i32
     %loop_step = llvm.mlir.constant(1 : index) : i32
diff --git a/mlir/test/Target/LLVMIR/omptarget-wsloop.mlir b/mlir/test/Target/LLVMIR/omptarget-wsloop.mlir
index 220eb85b3483..0f3f503dfa53 100644
--- a/mlir/test/Target/LLVMIR/omptarget-wsloop.mlir
+++ b/mlir/test/Target/LLVMIR/omptarget-wsloop.mlir
@@ -4,7 +4,7 @@
 // for nested omp do loop inside omp target region
 
 module attributes {dlti.dl_spec = #dlti.dl_spec<#dlti.dl_entry<"dlti.alloca_memory_space", 5 : ui32>>, llvm.data_layout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7:8", llvm.target_triple = "amdgcn-amd-amdhsa", omp.is_gpu = true, omp.is_target_device = true } {
-  llvm.func @target_wsloop(%arg0: !llvm.ptr ){
+  llvm.func @target_wsloop(%arg0: !llvm.ptr ) attributes {omp.declare_target = #omp.declaretarget<device_type = (any), capture_clause = (to)>} {
       %loop_ub = llvm.mlir.constant(9 : i32) : i32
       %loop_lb = llvm.mlir.constant(0 : i32) : i32
       %loop_step = llvm.mlir.constant(1 : i32) : i32
@@ -16,7 +16,7 @@ module attributes {dlti.dl_spec = #dlti.dl_spec<#dlti.dl_entry<"dlti.alloca_memo
     llvm.return
   }
 
-  llvm.func @target_empty_wsloop(){
+  llvm.func @target_empty_wsloop() attributes {omp.declare_target = #omp.declaretarget<device_type = (any), capture_clause = (to)>} {
       %loop_ub = llvm.mlir.constant(9 : i32) : i32
       %loop_lb = llvm.mlir.constant(0 : i32) : i32
       %loop_step = llvm.mlir.constant(1 : i32) : i32
diff --git a/mlir/test/Target/LLVMIR/openmp-data-target-device.mlir b/mlir/test/Target/LLVMIR/openmp-data-target-device.mlir
new file mode 100644
index 000000000000..d41429a6de06
--- /dev/null
+++ b/mlir/test/Target/LLVMIR/openmp-data-target-device.mlir
@@ -0,0 +1,61 @@
+// RUN: mlir-translate -mlir-to-llvmir %s | FileCheck %s
+
+// This tests checks that a target op inside a data op
+// We are only interested in ensuring that the -mlir-to-llmvir pass doesn't crash.
+// CHECK: {{.*}} = add i32 {{.*}}, 1
+module attributes { } {
+  llvm.mlir.global weak_odr hidden local_unnamed_addr constant @__oclc_ABI_version(400 : i32) {addr_space = 4 : i32} : i32
+  llvm.func @_QQmain() attributes {fir.bindc_name = "main", omp.declare_target = #omp.declaretarget<device_type = (host), capture_clause = (to)>} {
+    %0 = llvm.mlir.constant(99 : index) : i64
+    %1 = llvm.mlir.constant(0 : index) : i64
+    %2 = llvm.mlir.constant(1 : index) : i64
+    %3 = llvm.mlir.constant(100 : index) : i64
+    %4 = llvm.mlir.constant(1 : i64) : i64
+    %5 = llvm.alloca %4 x i32 {bindc_name = "array_length"} : (i64) -> !llvm.ptr<5>
+    %6 = llvm.addrspacecast %5 : !llvm.ptr<5> to !llvm.ptr
+    %7 = llvm.mlir.constant(1 : i64) : i64
+    %8 = llvm.alloca %7 x i32 {bindc_name = "index_"} : (i64) -> !llvm.ptr<5>
+    %9 = llvm.addrspacecast %8 : !llvm.ptr<5> to !llvm.ptr
+    %10 = llvm.mlir.addressof @_QFEint_array : !llvm.ptr
+    %11 = omp.map.bounds lower_bound(%1 : i64) upper_bound(%0 : i64) extent(%3 : i64) stride(%2 : i64) start_idx(%2 : i64)
+    %12 = omp.map.info var_ptr(%10 : !llvm.ptr, !llvm.array<100 x i32>) map_clauses(from) capture(ByRef) bounds(%11) -> !llvm.ptr {name = "int_array"}
+    omp.target_data map_entries(%12 : !llvm.ptr) {
+      %13 = omp.map.info var_ptr(%10 : !llvm.ptr, !llvm.array<100 x i32>) map_clauses(from) capture(ByRef) bounds(%11) -> !llvm.ptr {name = "int_array"}
+      %14 = omp.map.info var_ptr(%9 : !llvm.ptr, i32) map_clauses(implicit, exit_release_or_enter_alloc) capture(ByCopy) -> !llvm.ptr {name = "index_"}
+      omp.target map_entries(%13 -> %arg0, %14 -> %arg1 : !llvm.ptr, !llvm.ptr) {
+      ^bb0(%arg0: !llvm.ptr, %arg1: !llvm.ptr):
+        %15 = llvm.mlir.constant(100 : i32) : i32
+        %16 = llvm.mlir.constant(1 : i32) : i32
+        %17 = llvm.mlir.constant(100 : index) : i64
+        omp.parallel {
+          %18 = llvm.mlir.constant(1 : i64) : i64
+          %19 = llvm.alloca %18 x i32 {pinned} : (i64) -> !llvm.ptr<5>
+          %20 = llvm.addrspacecast %19 : !llvm.ptr<5> to !llvm.ptr
+          omp.wsloop  for  (%arg2) : i32 = (%16) to (%15) inclusive step (%16) {
+            llvm.store %arg2, %20 : i32, !llvm.ptr
+            %21 = llvm.load %20 : !llvm.ptr -> i32
+            %22 = llvm.sext %21 : i32 to i64
+            %23 = llvm.mlir.constant(1 : i64) : i64
+            %24 = llvm.mlir.constant(0 : i64) : i64
+            %25 = llvm.sub %22, %23 overflow<nsw>  : i64
+            %26 = llvm.mul %25, %23 overflow<nsw>  : i64
+            %27 = llvm.mul %26, %23 overflow<nsw>  : i64
+            %28 = llvm.add %27, %24 overflow<nsw>  : i64
+            %29 = llvm.mul %23, %17 overflow<nsw>  : i64
+            %30 = llvm.getelementptr %arg0[%28] : (!llvm.ptr, i64) -> !llvm.ptr, i32
+            llvm.store %21, %30 : i32, !llvm.ptr
+            omp.yield
+          }
+          omp.terminator
+        }
+        omp.terminator
+      }
+      omp.terminator
+    }
+    llvm.return
+  }
+  llvm.mlir.global internal @_QFEint_array() {addr_space = 0 : i32} : !llvm.array<100 x i32> {
+    %0 = llvm.mlir.zero : !llvm.array<100 x i32>
+    llvm.return %0 : !llvm.array<100 x i32>
+  }
+}
diff --git a/mlir/test/Target/LLVMIR/openmp-task-target-device.mlir b/mlir/test/Target/LLVMIR/openmp-task-target-device.mlir
new file mode 100644
index 000000000000..b4c848beef69
--- /dev/null
+++ b/mlir/test/Target/LLVMIR/openmp-task-target-device.mlir
@@ -0,0 +1,27 @@
+// RUN: mlir-translate -mlir-to-llvmir %s | FileCheck %s
+
+// This tests the fix for https://github.com/llvm/llvm-project/issues/84606
+// We are only interested in ensuring that the -mlir-to-llmvir pass doesn't crash.
+// CHECK: {{.*}} = add i32 {{.*}}, 5
+module attributes {omp.is_target_device = true } {
+  llvm.func @_QQmain() attributes {fir.bindc_name = "main", omp.declare_target = #omp.declaretarget<device_type = (host), capture_clause = (to)>} {
+    %0 = llvm.mlir.constant(0 : i32) : i32
+    %1 = llvm.mlir.constant(1 : i64) : i64
+    %2 = llvm.alloca %1 x i32 {bindc_name = "a"} : (i64) -> !llvm.ptr<5>
+    %3 = llvm.addrspacecast %2 : !llvm.ptr<5> to !llvm.ptr
+    omp.task {
+      llvm.store %0, %3 : i32, !llvm.ptr
+      omp.terminator
+    }
+    %4 = omp.map.info var_ptr(%3 : !llvm.ptr, i32) map_clauses(tofrom) capture(ByRef) -> !llvm.ptr {name = "a"}
+    omp.target map_entries(%4 -> %arg0 : !llvm.ptr) {
+    ^bb0(%arg0: !llvm.ptr):
+      %5 = llvm.mlir.constant(5 : i32) : i32
+      %6 = llvm.load %arg0  : !llvm.ptr -> i32
+      %7 = llvm.add %6, %5  : i32
+      llvm.store %7, %arg0  : i32, !llvm.ptr
+      omp.terminator
+    }
+    llvm.return
+  }
+}
diff --git a/mlir/tools/mlir-tblgen/OpFormatGen.cpp b/mlir/tools/mlir-tblgen/OpFormatGen.cpp
index 1ffac059f198..c8e0476d45b9 100644
--- a/mlir/tools/mlir-tblgen/OpFormatGen.cpp
+++ b/mlir/tools/mlir-tblgen/OpFormatGen.cpp
@@ -508,7 +508,7 @@ const char *const optionalOperandParserCode = R"(
 )";
 const char *const operandParserCode = R"(
   {0}OperandsLoc = parser.getCurrentLocation();
-  if (parser.parseOperand({0}RawOperands[0]))
+  if (parser.parseOperand({0}RawOperand))
     return ::mlir::failure();
 )";
 /// The code snippet used to generate a parser call for a VariadicOfVariadic
@@ -564,11 +564,11 @@ const char *const typeParserCode = R"(
     {0} type;
     if (parser.parseCustomTypeWithFallback(type))
       return ::mlir::failure();
-    {1}RawTypes[0] = type;
+    {1}RawType = type;
   }
 )";
 const char *const qualifiedTypeParserCode = R"(
-  if (parser.parseType({1}RawTypes[0]))
+  if (parser.parseType({1}RawType))
     return ::mlir::failure();
 )";
 
@@ -842,9 +842,9 @@ static void genElementParserStorage(FormatElement *element, const Operator &op,
       }
     } else {
       body << "  ::mlir::OpAsmParser::UnresolvedOperand " << name
-           << "RawOperands[1];\n"
+           << "RawOperand{};\n"
            << "  ::llvm::ArrayRef<::mlir::OpAsmParser::UnresolvedOperand> "
-           << name << "Operands(" << name << "RawOperands);";
+           << name << "Operands(&" << name << "RawOperand, 1);";
     }
     body << llvm::formatv("  ::llvm::SMLoc {0}OperandsLoc;\n"
                           "  (void){0}OperandsLoc;\n",
@@ -879,10 +879,11 @@ static void genElementParserStorage(FormatElement *element, const Operator &op,
     if (lengthKind != ArgumentLengthKind::Single)
       body << "  ::llvm::SmallVector<::mlir::Type, 1> " << name << "Types;\n";
     else
-      body << llvm::formatv("  ::mlir::Type {0}RawTypes[1];\n", name)
-           << llvm::formatv(
-                  "  ::llvm::ArrayRef<::mlir::Type> {0}Types({0}RawTypes);\n",
-                  name);
+      body
+          << llvm::formatv("  ::mlir::Type {0}RawType{{};\n", name)
+          << llvm::formatv(
+                 "  ::llvm::ArrayRef<::mlir::Type> {0}Types(&{0}RawType, 1);\n",
+                 name);
   } else if (auto *dir = dyn_cast<FunctionalTypeDirective>(element)) {
     ArgumentLengthKind ignored;
     body << "  ::llvm::ArrayRef<::mlir::Type> "
@@ -910,7 +911,7 @@ static void genCustomParameterParser(FormatElement *param, MethodBody &body) {
     else if (lengthKind == ArgumentLengthKind::Optional)
       body << llvm::formatv("{0}Operand", name);
     else
-      body << formatv("{0}RawOperands[0]", name);
+      body << formatv("{0}RawOperand", name);
 
   } else if (auto *region = dyn_cast<RegionVariable>(param)) {
     StringRef name = region->getVar()->name;
@@ -939,7 +940,7 @@ static void genCustomParameterParser(FormatElement *param, MethodBody &body) {
     else if (lengthKind == ArgumentLengthKind::Optional)
       body << llvm::formatv("{0}Type", listName);
     else
-      body << formatv("{0}RawTypes[0]", listName);
+      body << formatv("{0}RawType", listName);
 
   } else if (auto *string = dyn_cast<StringElement>(param)) {
     FmtContext ctx;
diff --git a/utils/bazel/llvm-project-overlay/clang/BUILD.bazel b/utils/bazel/llvm-project-overlay/clang/BUILD.bazel
index 14f2b453a766..54cecfaeb20c 100644
--- a/utils/bazel/llvm-project-overlay/clang/BUILD.bazel
+++ b/utils/bazel/llvm-project-overlay/clang/BUILD.bazel
@@ -2,7 +2,6 @@
 # See https://llvm.org/LICENSE.txt for license information.
 # SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 
-load("@bazel_skylib//rules:expand_template.bzl", "expand_template")
 load(
     "//:vars.bzl",
     "LLVM_VERSION",
@@ -13,6 +12,7 @@ load(
 load("//:workspace_root.bzl", "workspace_root")
 load("//llvm:binary_alias.bzl", "binary_alias")
 load("//llvm:cc_plugin_library.bzl", "cc_plugin_library")
+load("//llvm:driver.bzl", "llvm_driver_cc_binary")
 load("//llvm:tblgen.bzl", "gentbl")
 
 package(
@@ -2320,20 +2320,9 @@ cc_binary(
     ],
 )
 
-expand_template(
-    name = "clang_main",
-    out = "clang-driver.cpp",
-    substitutions = {
-        "@TOOL_NAME@": "clang",
-    },
-    template = "//llvm:cmake/modules/llvm-driver-template.cpp.in",
-)
-
 cc_library(
     name = "clang-driver",
-    srcs = glob([
-        "tools/driver/*.cpp",
-    ]) + ["clang-driver.cpp"],
+    srcs = glob(["tools/driver/*.cpp"]),
     copts = [
         # Disable stack frame size checks in the driver because
         # clang::ensureStackAddressSpace allocates a large array on the stack.
@@ -2371,13 +2360,10 @@ cc_library(
     ],
 )
 
-cc_binary(
+llvm_driver_cc_binary(
     name = "clang",
-    srcs = [],
     stamp = 0,
-    deps = [
-        ":clang-driver",
-    ],
+    deps = [":clang-driver"],
 )
 
 cc_binary(
@@ -2622,19 +2608,9 @@ gentbl(
     td_srcs = ["//llvm:include/llvm/Option/OptParser.td"],
 )
 
-expand_template(
-    name = "clang-scan-deps-main",
-    out = "clang-scan-deps-driver.cpp",
-    substitutions = {
-        "@TOOL_NAME@": "clang_scan_deps",
-    },
-    template = "//llvm:cmake/modules/llvm-driver-template.cpp.in",
-)
-
-cc_binary(
-    name = "clang-scan-deps",
-    srcs = glob(["tools/clang-scan-deps/*.cpp"]) + ["clang-scan-deps-driver.cpp"],
-    stamp = 0,
+cc_library(
+    name = "clang-scan-deps-lib",
+    srcs = glob(["tools/clang-scan-deps/*.cpp"]),
     deps = [
         ":ScanDepsTableGen",
         ":driver",
@@ -2646,6 +2622,12 @@ cc_binary(
     ],
 )
 
+llvm_driver_cc_binary(
+    name = "clang-scan-deps",
+    stamp = 0,
+    deps = [":clang-scan-deps-lib"],
+)
+
 cc_library(
     name = "extract_api",
     srcs = glob([
diff --git a/utils/bazel/llvm-project-overlay/compiler-rt/BUILD.bazel b/utils/bazel/llvm-project-overlay/compiler-rt/BUILD.bazel
index 577e6c033b4e..9a4e103be01e 100644
--- a/utils/bazel/llvm-project-overlay/compiler-rt/BUILD.bazel
+++ b/utils/bazel/llvm-project-overlay/compiler-rt/BUILD.bazel
@@ -54,3 +54,57 @@ cc_library(
         ":config",
     ],
 )
+
+cc_library(
+    name = "orc_rt_common_headers",
+    hdrs = [
+        "lib/orc/adt.h",
+        "lib/orc/bitmask_enum.h",
+        "lib/orc/common.h",
+        "lib/orc/compiler.h",
+        "lib/orc/debug.h",
+        "lib/orc/endianness.h",
+        "lib/orc/error.h",
+        "lib/orc/executor_address.h",
+        "lib/orc/executor_symbol_def.h",
+        "lib/orc/extensible_rtti.h",
+        "lib/orc/interval_map.h",
+        "lib/orc/interval_set.h",
+        "lib/orc/simple_packed_serialization.h",
+        "lib/orc/stl_extras.h",
+        "lib/orc/string_pool.h",
+        "lib/orc/wrapper_function_utils.h",
+    ],
+    strip_include_prefix = "lib/orc",
+)
+
+cc_library(
+    name = "orc_rt",
+    srcs = [
+        "lib/orc/debug.cpp",
+        "lib/orc/dlfcn_wrapper.cpp",
+        "lib/orc/extensible_rtti.cpp",
+        "lib/orc/log_error_to_stderr.cpp",
+        "lib/orc/run_program_wrapper.cpp",
+    ] + select({
+        "@platforms//os:macos": [
+            "lib/orc/macho_platform.cpp",
+            "lib/orc/macho_platform.h",
+            "lib/orc/macho_tlv.arm64.S",
+            "lib/orc/macho_tlv.x86-64.S",
+        ],
+        "@platforms//os:linux": [
+            "lib/orc/elfnix_platform.cpp",
+            "lib/orc/elfnix_platform.h",
+            "lib/orc/elfnix_tls.aarch64.S",
+            "lib/orc/elfnix_tls.ppc64.S",
+            "lib/orc/elfnix_tls.x86-64.S",
+        ],
+    }),
+    hdrs = glob(["include/orc_rt/*.h"]),
+    includes = ["include"],
+    linkstatic = True,
+    deps = [
+        ":orc_rt_common_headers",
+    ],
+)
diff --git a/utils/bazel/llvm-project-overlay/compiler-rt/lib/orc/tests/unit/BUILD.bazel b/utils/bazel/llvm-project-overlay/compiler-rt/lib/orc/tests/unit/BUILD.bazel
new file mode 100644
index 000000000000..520bf2ab397a
--- /dev/null
+++ b/utils/bazel/llvm-project-overlay/compiler-rt/lib/orc/tests/unit/BUILD.bazel
@@ -0,0 +1,28 @@
+# This file is licensed under the Apache License v2.0 with LLVM Exceptions.
+# See https://llvm.org/LICENSE.txt for license information.
+# SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+
+package(
+    default_visibility = ["//visibility:public"],
+    features = ["layering_check"],
+)
+
+licenses(["notice"])
+
+[
+    cc_test(
+        name = "{}".format(file.split(".")[0]),
+        srcs = [file] + glob(["*.h"]),
+        deps = [
+            "//compiler-rt:orc_rt",
+            "//compiler-rt:orc_rt_common_headers",
+            "//third-party/unittest:gtest",
+            "//third-party/unittest:gtest_main",
+        ],
+    )
+    for file in glob(
+        ["*.cpp"],
+        # TODO: Broken with older libc++ versions, currently unused anyways
+        exclude = ["interval_set_test.cpp"],
+    )
+]
diff --git a/utils/bazel/llvm-project-overlay/libc/BUILD.bazel b/utils/bazel/llvm-project-overlay/libc/BUILD.bazel
index b8cd290a41a2..42af5cbe8f35 100644
--- a/utils/bazel/llvm-project-overlay/libc/BUILD.bazel
+++ b/utils/bazel/llvm-project-overlay/libc/BUILD.bazel
@@ -94,6 +94,10 @@ libc_support_library(
 libc_support_library(
     name = "llvm_libc_types_float128",
     hdrs = ["include/llvm-libc-types/float128.h"],
+    target_compatible_with = select({
+        "@platforms//os:linux": [],
+        "//conditions:default": ["@platforms//:incompatible"],
+    }),
     deps = [":llvm_libc_macros_float_macros"],
 )
 
@@ -972,11 +976,17 @@ libc_support_library(
 libc_support_library(
     name = "__support_osutil_syscall",
     hdrs = ["src/__support/OSUtil/syscall.h"],
-    textual_hdrs = [
-        "src/__support/OSUtil/linux/syscall.h",
-        "src/__support/OSUtil/linux/aarch64/syscall.h",
-        "src/__support/OSUtil/linux/x86_64/syscall.h",
-    ],
+    textual_hdrs = select({
+        "@platforms//os:macos": [
+            "src/__support/OSUtil/darwin/syscall.h",
+            "src/__support/OSUtil/darwin/arm/syscall.h",
+        ],
+        "@platforms//os:linux": [
+            "src/__support/OSUtil/linux/syscall.h",
+            "src/__support/OSUtil/linux/aarch64/syscall.h",
+            "src/__support/OSUtil/linux/x86_64/syscall.h",
+        ],
+    }),
     deps = [
         ":__support_common",
         ":__support_cpp_bit",
@@ -986,9 +996,10 @@ libc_support_library(
 libc_support_library(
     name = "__support_osutil_io",
     hdrs = ["src/__support/OSUtil/io.h"],
-    textual_hdrs = [
-        "src/__support/OSUtil/linux/io.h",
-    ],
+    textual_hdrs = select({
+        "@platforms//os:macos": ["src/__support/OSUtil/darwin/io.h"],
+        "@platforms//os:linux": ["src/__support/OSUtil/linux/io.h"],
+    }),
     deps = [
         ":__support_common",
         ":__support_cpp_string_view",
@@ -1001,6 +1012,10 @@ libc_support_library(
     name = "__support_osutil_quick_exit",
     srcs = ["src/__support/OSUtil/linux/quick_exit.cpp"],
     hdrs = ["src/__support/OSUtil/quick_exit.h"],
+    target_compatible_with = select({
+        "@platforms//os:linux": [],
+        "//conditions:default": ["@platforms//:incompatible"],
+    }),
     deps = [
         ":__support_osutil_syscall",
     ],
@@ -1019,6 +1034,10 @@ libc_support_library(
         "src/__support/StringUtil/error_to_string.h",
         "src/__support/StringUtil/signal_to_string.h",
     ],
+    target_compatible_with = select({
+        "@platforms//os:linux": [],
+        "//conditions:default": ["@platforms//:incompatible"],
+    }),
     deps = [
         ":__support_cpp_array",
         ":__support_cpp_span",
@@ -1036,6 +1055,10 @@ libc_support_library(
         "src/__support/threads/mutex.h",
         "src/__support/threads/mutex_common.h",
     ],
+    target_compatible_with = select({
+        "@platforms//os:linux": [],
+        "//conditions:default": ["@platforms//:incompatible"],
+    }),
     textual_hdrs = [
         "src/__support/threads/linux/mutex.h",
         "src/__support/threads/linux/futex_word.h",
@@ -1695,13 +1718,19 @@ libc_math_function(
 )
 
 libc_math_function(name = "fabs")
+
 libc_math_function(name = "fabsf")
+
 libc_math_function(name = "fabsl")
+
 libc_math_function(name = "fabsf128")
 
 libc_math_function(name = "fdim")
+
 libc_math_function(name = "fdimf")
+
 libc_math_function(name = "fdiml")
+
 libc_math_function(name = "fdimf128")
 
 libc_math_function(
@@ -1729,7 +1758,6 @@ libc_math_function(
 
 libc_math_function(name = "ceilf128")
 
-
 libc_math_function(
     name = "floor",
     specializations = [
@@ -1747,11 +1775,15 @@ libc_math_function(
 )
 
 libc_math_function(name = "floorl")
+
 libc_math_function(name = "floorf128")
 
 libc_math_function(name = "ldexp")
+
 libc_math_function(name = "ldexpf")
+
 libc_math_function(name = "ldexpl")
+
 libc_math_function(name = "ldexpf128")
 
 libc_math_function(
@@ -1771,6 +1803,7 @@ libc_math_function(
 )
 
 libc_math_function(name = "truncl")
+
 libc_math_function(name = "truncf128")
 
 libc_math_function(
@@ -1790,6 +1823,7 @@ libc_math_function(
 )
 
 libc_math_function(name = "roundl")
+
 libc_math_function(name = "roundf128")
 
 libc_math_function(
@@ -1807,8 +1841,11 @@ libc_math_function(
 )
 
 libc_math_function(name = "frexp")
+
 libc_math_function(name = "frexpf")
+
 libc_math_function(name = "frexpl")
+
 libc_math_function(name = "frexpf128")
 
 libc_math_function(name = "hypot")
@@ -1821,31 +1858,47 @@ libc_math_function(
 )
 
 libc_math_function(name = "logb")
+
 libc_math_function(name = "logbf")
+
 libc_math_function(name = "logbl")
+
 libc_math_function(name = "logbf128")
 
 libc_math_function(name = "modf")
+
 libc_math_function(name = "modff")
+
 libc_math_function(name = "modfl")
+
 libc_math_function(name = "modff128")
 
 libc_math_function(name = "remquo")
+
 libc_math_function(name = "remquof")
+
 libc_math_function(name = "remquol")
 
 libc_math_function(name = "remainder")
+
 libc_math_function(name = "remainderf")
+
 libc_math_function(name = "remainderl")
 
 libc_math_function(name = "fmin")
+
 libc_math_function(name = "fminf")
+
 libc_math_function(name = "fminl")
+
 libc_math_function(name = "fminf128")
 
 libc_math_function(name = "fmax")
+
 libc_math_function(name = "fmaxf")
+
 libc_math_function(name = "fmaxl")
+
 libc_math_function(name = "fmaxf128")
 
 libc_math_function(
@@ -1928,38 +1981,64 @@ libc_math_function(
 )
 
 libc_math_function(name = "copysign")
+
 libc_math_function(name = "copysignf")
+
 libc_math_function(name = "copysignl")
-libc_math_function(name = "copysignf128")
+
+libc_math_function(
+    name = "copysignf128",
+    additional_deps = [
+        ":llvm_libc_types_float128",
+    ],
+)
 
 libc_math_function(name = "ilogb")
+
 libc_math_function(name = "ilogbf")
+
 libc_math_function(name = "ilogbl")
+
 libc_math_function(name = "ilogbf128")
 
 libc_math_function(name = "rint")
+
 libc_math_function(name = "rintf")
+
 libc_math_function(name = "rintl")
+
 libc_math_function(name = "rintf128")
 
 libc_math_function(name = "lrint")
+
 libc_math_function(name = "lrintf")
+
 libc_math_function(name = "lrintl")
+
 libc_math_function(name = "lrintf128")
 
 libc_math_function(name = "llrint")
+
 libc_math_function(name = "llrintf")
+
 libc_math_function(name = "llrintl")
+
 libc_math_function(name = "llrintf128")
 
 libc_math_function(name = "lround")
+
 libc_math_function(name = "lroundf")
+
 libc_math_function(name = "lroundl")
+
 libc_math_function(name = "lroundf128")
 
 libc_math_function(name = "llround")
+
 libc_math_function(name = "llroundf")
+
 libc_math_function(name = "llroundl")
+
 libc_math_function(name = "llroundf128")
 
 libc_math_function(
@@ -1995,20 +2074,29 @@ libc_math_function(
 )
 
 libc_math_function(name = "nearbyint")
+
 libc_math_function(name = "nearbyintf")
+
 libc_math_function(name = "nearbyintl")
 
 libc_math_function(name = "nextafter")
+
 libc_math_function(name = "nextafterf")
+
 libc_math_function(name = "nextafterl")
+
 libc_math_function(name = "nextafterf128")
 
 libc_math_function(name = "nexttoward")
+
 libc_math_function(name = "nexttowardf")
+
 libc_math_function(name = "nexttowardl")
 
 libc_math_function(name = "scalbn")
+
 libc_math_function(name = "scalbnf")
+
 libc_math_function(name = "scalbnl")
 
 ############################## inttypes targets ##############################
@@ -2590,6 +2678,10 @@ libc_function(
     name = "open",
     srcs = ["src/fcntl/linux/open.cpp"],
     hdrs = ["src/fcntl/open.h"],
+    target_compatible_with = select({
+        "@platforms//os:linux": [],
+        "//conditions:default": ["@platforms//:incompatible"],
+    }),
     deps = [
         ":__support_common",
         ":__support_osutil_syscall",
@@ -2601,6 +2693,10 @@ libc_function(
     name = "openat",
     srcs = ["src/fcntl/linux/openat.cpp"],
     hdrs = ["src/fcntl/openat.h"],
+    target_compatible_with = select({
+        "@platforms//os:linux": [],
+        "//conditions:default": ["@platforms//:incompatible"],
+    }),
     deps = [
         ":__support_common",
         ":__support_osutil_syscall",
@@ -2680,6 +2776,10 @@ libc_function(
     name = "dup3",
     srcs = ["src/unistd/linux/dup3.cpp"],
     hdrs = ["src/unistd/dup3.h"],
+    target_compatible_with = select({
+        "@platforms//os:linux": [],
+        "//conditions:default": ["@platforms//:incompatible"],
+    }),
     deps = [
         ":__support_common",
         ":__support_osutil_syscall",
@@ -2849,6 +2949,10 @@ libc_function(
     name = "pread",
     srcs = ["src/unistd/linux/pread.cpp"],
     hdrs = ["src/unistd/pread.h"],
+    target_compatible_with = select({
+        "@platforms//os:linux": [],
+        "//conditions:default": ["@platforms//:incompatible"],
+    }),
     deps = [
         ":__support_common",
         ":__support_osutil_syscall",
@@ -2860,6 +2964,10 @@ libc_function(
     name = "pwrite",
     srcs = ["src/unistd/linux/pwrite.cpp"],
     hdrs = ["src/unistd/pwrite.h"],
+    target_compatible_with = select({
+        "@platforms//os:linux": [],
+        "//conditions:default": ["@platforms//:incompatible"],
+    }),
     deps = [
         ":__support_common",
         ":__support_osutil_syscall",
@@ -3263,6 +3371,10 @@ libc_function(
     name = "rename",
     srcs = ["src/stdio/linux/rename.cpp"],
     hdrs = ["src/stdio/rename.h"],
+    target_compatible_with = select({
+        "@platforms//os:linux": [],
+        "//conditions:default": ["@platforms//:incompatible"],
+    }),
     deps = [
         ":__support_common",
         ":__support_osutil_syscall",
@@ -3301,6 +3413,10 @@ libc_function(
     name = "epoll_wait",
     srcs = ["src/sys/epoll/linux/epoll_wait.cpp"],
     hdrs = ["src/sys/epoll/epoll_wait.h"],
+    target_compatible_with = select({
+        "@platforms//os:linux": [],
+        "//conditions:default": ["@platforms//:incompatible"],
+    }),
     weak = True,
     deps = [
         ":__support_osutil_syscall",
@@ -3312,6 +3428,10 @@ libc_function(
     name = "epoll_pwait",
     srcs = ["src/sys/epoll/linux/epoll_pwait.cpp"],
     hdrs = ["src/sys/epoll/epoll_pwait.h"],
+    target_compatible_with = select({
+        "@platforms//os:linux": [],
+        "//conditions:default": ["@platforms//:incompatible"],
+    }),
     weak = True,
     deps = [
         ":__support_osutil_syscall",
diff --git a/utils/bazel/llvm-project-overlay/lld/BUILD.bazel b/utils/bazel/llvm-project-overlay/lld/BUILD.bazel
index 5a494a13acea..2ccd6fcae635 100644
--- a/utils/bazel/llvm-project-overlay/lld/BUILD.bazel
+++ b/utils/bazel/llvm-project-overlay/lld/BUILD.bazel
@@ -2,12 +2,12 @@
 # See https://llvm.org/LICENSE.txt for license information.
 # SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 
-load("@bazel_skylib//rules:expand_template.bzl", "expand_template")
 load(
     "//:vars.bzl",
     "LLVM_VERSION",
 )
 load("//llvm:binary_alias.bzl", "binary_alias")
+load("//llvm:driver.bzl", "llvm_driver_cc_binary")
 load("//llvm:tblgen.bzl", "gentbl")
 
 package(
@@ -282,20 +282,9 @@ cc_library(
     ],
 )
 
-expand_template(
-    name = "lld_main",
-    out = "lld-driver.cpp",
-    substitutions = {
-        "@TOOL_NAME@": "lld",
-    },
-    template = "//llvm:cmake/modules/llvm-driver-template.cpp.in",
-)
-
-cc_binary(
-    name = "lld",
-    srcs = glob([
-        "tools/lld/*.cpp",
-    ]) + ["lld-driver.cpp"],
+cc_library(
+    name = "lld-lib",
+    srcs = glob(["tools/lld/*.cpp"]),
     deps = [
         ":COFF",
         ":Common",
@@ -308,6 +297,11 @@ cc_binary(
     ],
 )
 
+llvm_driver_cc_binary(
+    name = "lld",
+    deps = [":lld-lib"],
+)
+
 # These are the required names for lld running under different environs.
 #
 # Unix/Linux require that the binary be named "ld.lld".
diff --git a/utils/bazel/llvm-project-overlay/llvm/BUILD.bazel b/utils/bazel/llvm-project-overlay/llvm/BUILD.bazel
index 9cfcb7d3838e..af9dc26abec4 100644
--- a/utils/bazel/llvm-project-overlay/llvm/BUILD.bazel
+++ b/utils/bazel/llvm-project-overlay/llvm/BUILD.bazel
@@ -3214,23 +3214,13 @@ gentbl(
     td_srcs = ["include/llvm/Option/OptParser.td"],
 )
 
-expand_template(
-    name = "dsymutil_main",
-    out = "dsymutil-driver.cpp",
-    substitutions = {
-        "@TOOL_NAME@": "dsymutil",
-    },
-    template = "cmake/modules/llvm-driver-template.cpp.in",
-)
-
-cc_binary(
-    name = "dsymutil",
+cc_library(
+    name = "dsymutil-lib",
     srcs = glob([
         "tools/dsymutil/*.cpp",
         "tools/dsymutil/*.h",
-    ]) + ["dsymutil-driver.cpp"],
+    ]),
     copts = llvm_copts,
-    stamp = 0,
     deps = [
         ":AllTargetsCodeGens",
         ":BinaryFormat",
@@ -3253,6 +3243,12 @@ cc_binary(
     ],
 )
 
+llvm_driver_cc_binary(
+    name = "dsymutil",
+    stamp = 0,
+    deps = [":dsymutil-lib"],
+)
+
 cc_binary(
     name = "llc",
     srcs = glob([
@@ -3532,22 +3528,10 @@ gentbl(
     td_srcs = ["include/llvm/Option/OptParser.td"],
 )
 
-expand_template(
-    name = "cxxfilt_main",
-    out = "llvm-cxxfilt-driver.cpp",
-    substitutions = {
-        "@TOOL_NAME@": "llvm_cxxfilt",
-    },
-    template = "cmake/modules/llvm-driver-template.cpp.in",
-)
-
-cc_binary(
-    name = "llvm-cxxfilt",
-    srcs = glob([
-        "tools/llvm-cxxfilt/*.cpp",
-    ]) + ["llvm-cxxfilt-driver.cpp"],
+cc_library(
+    name = "llvm-cxxfilt-lib",
+    srcs = glob(["tools/llvm-cxxfilt/*.cpp"]),
     copts = llvm_copts,
-    stamp = 0,
     deps = [
         ":CxxfiltOptsTableGen",
         ":Demangle",
@@ -3557,6 +3541,12 @@ cc_binary(
     ],
 )
 
+llvm_driver_cc_binary(
+    name = "llvm-cxxfilt",
+    stamp = 0,
+    deps = [":llvm-cxxfilt-lib"],
+)
+
 cc_binary(
     name = "llvm-debuginfo-analyzer",
     srcs = glob([
@@ -3674,22 +3664,10 @@ gentbl(
     td_srcs = ["include/llvm/Option/OptParser.td"],
 )
 
-expand_template(
-    name = "dwp_main",
-    out = "llvm-dwp-driver.cpp",
-    substitutions = {
-        "@TOOL_NAME@": "llvm_dwp",
-    },
-    template = "cmake/modules/llvm-driver-template.cpp.in",
-)
-
-cc_binary(
-    name = "llvm-dwp",
-    srcs = glob([
-        "tools/llvm-dwp/*.cpp",
-    ]) + ["llvm-dwp-driver.cpp"],
+cc_library(
+    name = "llvm-dwp-lib",
+    srcs = glob(["tools/llvm-dwp/*.cpp"]),
     copts = llvm_copts,
-    stamp = 0,
     deps = [
         ":AllTargetsCodeGens",
         ":DWP",
@@ -3700,6 +3678,12 @@ cc_binary(
     ],
 )
 
+llvm_driver_cc_binary(
+    name = "llvm-dwp",
+    stamp = 0,
+    deps = [":llvm-dwp-lib"],
+)
+
 cc_binary(
     name = "llvm-exegesis",
     srcs = [
@@ -3754,22 +3738,10 @@ gentbl(
     td_srcs = ["include/llvm/Option/OptParser.td"],
 )
 
-expand_template(
-    name = "gsymutil_main",
-    out = "llvm-gsymutil-driver.cpp",
-    substitutions = {
-        "@TOOL_NAME@": "llvm_gsymutil",
-    },
-    template = "cmake/modules/llvm-driver-template.cpp.in",
-)
-
-cc_binary(
-    name = "llvm-gsymutil",
-    srcs = glob([
-        "tools/llvm-gsymutil/*.cpp",
-    ]) + ["llvm-gsymutil-driver.cpp"],
+cc_library(
+    name = "llvm-gsymutil-lib",
+    srcs = glob(["tools/llvm-gsymutil/*.cpp"]),
     copts = llvm_copts,
-    stamp = 0,
     deps = [
         ":AllTargetsCodeGens",
         ":DebugInfo",
@@ -3785,6 +3757,12 @@ cc_binary(
     ],
 )
 
+llvm_driver_cc_binary(
+    name = "llvm-gsymutil",
+    stamp = 0,
+    deps = [":llvm-gsymutil-lib"],
+)
+
 gentbl(
     name = "IfsOptionsTableGen",
     strip_include_prefix = "tools/llvm-ifs",
@@ -3797,23 +3775,13 @@ gentbl(
     td_srcs = ["include/llvm/Option/OptParser.td"],
 )
 
-expand_template(
-    name = "ifs_main",
-    out = "llvm-ifs-driver.cpp",
-    substitutions = {
-        "@TOOL_NAME@": "llvm_ifs",
-    },
-    template = "cmake/modules/llvm-driver-template.cpp.in",
-)
-
-cc_binary(
-    name = "llvm-ifs",
+cc_library(
+    name = "llvm-ifs-lib",
     srcs = glob([
         "tools/llvm-ifs/*.cpp",
         "tools/llvm-ifs/*.h",
-    ]) + ["llvm-ifs-driver.cpp"],
+    ]),
     copts = llvm_copts,
-    stamp = 0,
     deps = [
         ":BinaryFormat",
         ":IfsOptionsTableGen",
@@ -3826,6 +3794,12 @@ cc_binary(
     ],
 )
 
+llvm_driver_cc_binary(
+    name = "llvm-ifs",
+    stamp = 0,
+    deps = [":llvm-ifs-lib"],
+)
+
 cc_binary(
     name = "llvm-jitlink",
     srcs = glob([
@@ -3884,23 +3858,13 @@ gentbl(
     td_srcs = ["include/llvm/Option/OptParser.td"],
 )
 
-expand_template(
-    name = "libtool-darwin_main",
-    out = "llvm-libtool-darwin-driver.cpp",
-    substitutions = {
-        "@TOOL_NAME@": "llvm_libtool_darwin",
-    },
-    template = "cmake/modules/llvm-driver-template.cpp.in",
-)
-
-cc_binary(
-    name = "llvm-libtool-darwin",
+cc_library(
+    name = "llvm-libtool-darwin-lib",
     srcs = glob([
         "tools/llvm-libtool-darwin/*.cpp",
         "tools/llvm-libtool-darwin/*.h",
-    ]) + ["llvm-libtool-darwin-driver.cpp"],
+    ]),
     copts = llvm_copts,
-    stamp = 0,
     deps = [
         ":AllTargetsAsmParsers",
         ":AllTargetsCodeGens",
@@ -3914,6 +3878,12 @@ cc_binary(
     ],
 )
 
+llvm_driver_cc_binary(
+    name = "llvm-libtool-darwin",
+    stamp = 0,
+    deps = [":llvm-libtool-darwin-lib"],
+)
+
 cc_binary(
     name = "llvm-link",
     srcs = glob([
@@ -3949,22 +3919,10 @@ gentbl(
     td_srcs = ["include/llvm/Option/OptParser.td"],
 )
 
-expand_template(
-    name = "lipo_main",
-    out = "llvm-lipo-driver.cpp",
-    substitutions = {
-        "@TOOL_NAME@": "llvm_lipo",
-    },
-    template = "cmake/modules/llvm-driver-template.cpp.in",
-)
-
-cc_binary(
-    name = "llvm-lipo",
-    srcs = [
-        "tools/llvm-lipo/llvm-lipo.cpp",
-    ] + ["llvm-lipo-driver.cpp"],
+cc_library(
+    name = "llvm-lipo-lib",
+    srcs = ["tools/llvm-lipo/llvm-lipo.cpp"],
     copts = llvm_copts,
-    stamp = 0,
     deps = [
         ":AllTargetsAsmParsers",
         ":BinaryFormat",
@@ -3978,6 +3936,12 @@ cc_binary(
     ],
 )
 
+llvm_driver_cc_binary(
+    name = "llvm-lipo",
+    stamp = 0,
+    deps = [":llvm-lipo-lib"],
+)
+
 cc_binary(
     name = "llvm-lto",
     srcs = glob([
@@ -4077,23 +4041,13 @@ gentbl(
     td_srcs = ["include/llvm/Option/OptParser.td"],
 )
 
-expand_template(
-    name = "ml_main",
-    out = "llvm-ml-driver.cpp",
-    substitutions = {
-        "@TOOL_NAME@": "llvm_ml",
-    },
-    template = "cmake/modules/llvm-driver-template.cpp.in",
-)
-
-cc_binary(
-    name = "llvm-ml",
+cc_library(
+    name = "llvm-ml-lib",
     srcs = glob([
         "tools/llvm-ml/*.cpp",
         "tools/llvm-ml/*.h",
-    ]) + ["llvm-ml-driver.cpp"],
+    ]),
     copts = llvm_copts,
-    stamp = 0,
     deps = [
         ":AllTargetsAsmParsers",
         ":AllTargetsCodeGens",
@@ -4108,6 +4062,12 @@ cc_binary(
     ],
 )
 
+llvm_driver_cc_binary(
+    name = "llvm-ml",
+    stamp = 0,
+    deps = [":llvm-ml-lib"],
+)
+
 cc_binary(
     name = "llvm-modextract",
     srcs = glob([
@@ -4136,22 +4096,10 @@ gentbl(
     td_srcs = ["include/llvm/Option/OptParser.td"],
 )
 
-expand_template(
-    name = "mt_main",
-    out = "llvm-mt-driver.cpp",
-    substitutions = {
-        "@TOOL_NAME@": "llvm_mt",
-    },
-    template = "cmake/modules/llvm-driver-template.cpp.in",
-)
-
-cc_binary(
-    name = "llvm-mt",
-    srcs = glob([
-        "tools/llvm-mt/*.cpp",
-    ]) + ["llvm-mt-driver.cpp"],
+cc_library(
+    name = "llvm-mt-lib",
+    srcs = glob(["tools/llvm-mt/*.cpp"]),
     copts = llvm_copts,
-    stamp = 0,
     deps = [
         ":MtTableGen",
         ":Option",
@@ -4160,6 +4108,12 @@ cc_binary(
     ],
 )
 
+llvm_driver_cc_binary(
+    name = "llvm-mt",
+    stamp = 0,
+    deps = [":llvm-mt-lib"],
+)
+
 gentbl(
     name = "NmOptsTableGen",
     strip_include_prefix = "tools/llvm-nm",
@@ -4271,23 +4225,13 @@ cc_binary(
     ],
 )
 
-expand_template(
-    name = "objcopy_main",
-    out = "llvm-objcopy-driver.cpp",
-    substitutions = {
-        "@TOOL_NAME@": "llvm_objcopy",
-    },
-    template = "cmake/modules/llvm-driver-template.cpp.in",
-)
-
-cc_binary(
-    name = "llvm-objcopy",
+cc_library(
+    name = "llvm-objcopy-lib",
     srcs = glob([
         "tools/llvm-objcopy/*.cpp",
         "tools/llvm-objcopy/*.h",
-    ]) + ["llvm-objcopy-driver.cpp"],
+    ]),
     copts = llvm_copts,
-    stamp = 0,
     deps = [
         ":BinaryFormat",
         ":MC",
@@ -4305,6 +4249,12 @@ cc_binary(
     ],
 )
 
+llvm_driver_cc_binary(
+    name = "llvm-objcopy",
+    stamp = 0,
+    deps = [":llvm-objcopy-lib"],
+)
+
 binary_alias(
     name = "llvm-strip",
     binary = ":llvm-objcopy",
@@ -4320,23 +4270,13 @@ binary_alias(
     binary = ":llvm-objcopy",
 )
 
-expand_template(
-    name = "objdump_main",
-    out = "llvm-objdump-driver.cpp",
-    substitutions = {
-        "@TOOL_NAME@": "llvm_objdump",
-    },
-    template = "cmake/modules/llvm-driver-template.cpp.in",
-)
-
-cc_binary(
-    name = "llvm-objdump",
+cc_library(
+    name = "llvm-objdump-lib",
     srcs = glob([
         "tools/llvm-objdump/*.cpp",
         "tools/llvm-objdump/*.h",
-    ]) + ["llvm-objdump-driver.cpp"],
+    ]),
     copts = llvm_copts,
-    stamp = 0,
     deps = [
         ":AllTargetsAsmParsers",
         ":AllTargetsCodeGens",
@@ -4360,6 +4300,12 @@ cc_binary(
     ],
 )
 
+llvm_driver_cc_binary(
+    name = "llvm-objdump",
+    stamp = 0,
+    deps = [":llvm-objdump-lib"],
+)
+
 gentbl(
     name = "ObjdumpOptsTableGen",
     strip_include_prefix = "tools/llvm-objdump",
@@ -4425,22 +4371,10 @@ cc_binary(
     ],
 )
 
-expand_template(
-    name = "profdata_main",
-    out = "llvm-profdata-driver.cpp",
-    substitutions = {
-        "@TOOL_NAME@": "llvm_profdata",
-    },
-    template = "cmake/modules/llvm-driver-template.cpp.in",
-)
-
-cc_binary(
-    name = "llvm-profdata",
-    srcs = glob([
-        "tools/llvm-profdata/*.cpp",
-    ]) + ["llvm-profdata-driver.cpp"],
+cc_library(
+    name = "llvm-profdata-lib",
+    srcs = glob(["tools/llvm-profdata/*.cpp"]),
     copts = llvm_copts,
-    stamp = 0,
     deps = [
         ":Core",
         ":Object",
@@ -4449,6 +4383,12 @@ cc_binary(
     ],
 )
 
+llvm_driver_cc_binary(
+    name = "llvm-profdata",
+    stamp = 0,
+    deps = [":llvm-profdata-lib"],
+)
+
 cc_binary(
     name = "llvm-profgen",
     srcs = glob([
@@ -4504,23 +4444,13 @@ cc_library(
     textual_hdrs = glob(["tools/llvm-rc/*.def"]),
 )
 
-expand_template(
-    name = "rc_main",
-    out = "llvm-rc-driver.cpp",
-    substitutions = {
-        "@TOOL_NAME@": "llvm_rc",
-    },
-    template = "cmake/modules/llvm-driver-template.cpp.in",
-)
-
-cc_binary(
-    name = "llvm-rc",
+cc_library(
+    name = "llvm-rc-lib",
     srcs = glob([
         "tools/llvm-rc/*.cpp",
         "tools/llvm-rc/*.h",
-    ]) + ["llvm-rc-driver.cpp"],
+    ]),
     copts = llvm_copts,
-    stamp = 0,
     deps = [
         ":Object",
         ":Option",
@@ -4533,6 +4463,12 @@ cc_binary(
     ],
 )
 
+llvm_driver_cc_binary(
+    name = "llvm-rc",
+    stamp = 0,
+    deps = [":llvm-rc-lib"],
+)
+
 binary_alias(
     name = "llvm-windres",
     binary = ":llvm-rc",
@@ -4550,23 +4486,13 @@ gentbl(
     td_srcs = ["include/llvm/Option/OptParser.td"],
 )
 
-expand_template(
-    name = "readobj_main",
-    out = "llvm-readobj-driver.cpp",
-    substitutions = {
-        "@TOOL_NAME@": "llvm_readobj",
-    },
-    template = "cmake/modules/llvm-driver-template.cpp.in",
-)
-
-cc_binary(
-    name = "llvm-readobj",
+cc_library(
+    name = "llvm-readobj-lib",
     srcs = glob([
         "tools/llvm-readobj/*.cpp",
         "tools/llvm-readobj/*.h",
-    ]) + ["llvm-readobj-driver.cpp"],
+    ]),
     copts = llvm_copts,
-    stamp = 0,
     deps = [
         ":AllTargetsCodeGens",
         ":BinaryFormat",
@@ -4582,6 +4508,12 @@ cc_binary(
     ],
 )
 
+llvm_driver_cc_binary(
+    name = "llvm-readobj",
+    stamp = 0,
+    deps = [":llvm-readobj-lib"],
+)
+
 # Create an 'llvm-readelf' named binary from the 'llvm-readobj' tool.
 binary_alias(
     name = "llvm-readelf",
@@ -4723,22 +4655,10 @@ gentbl(
     td_srcs = ["include/llvm/Option/OptParser.td"],
 )
 
-expand_template(
-    name = "symbolizer_main",
-    out = "llvm-symbolizer-driver.cpp",
-    substitutions = {
-        "@TOOL_NAME@": "llvm_symbolizer",
-    },
-    template = "cmake/modules/llvm-driver-template.cpp.in",
-)
-
-cc_binary(
-    name = "llvm-symbolizer",
-    srcs = glob([
-        "tools/llvm-symbolizer/*.cpp",
-    ]) + ["llvm-symbolizer-driver.cpp"],
+cc_library(
+    name = "llvm-symbolizer-lib",
+    srcs = glob(["tools/llvm-symbolizer/*.cpp"]),
     copts = llvm_copts,
-    stamp = 0,
     deps = [
         ":DebugInfoDWARF",
         ":DebugInfoPDB",
@@ -4752,6 +4672,12 @@ cc_binary(
     ],
 )
 
+llvm_driver_cc_binary(
+    name = "llvm-symbolizer",
+    stamp = 0,
+    deps = [":llvm-symbolizer-lib"],
+)
+
 binary_alias(
     name = "llvm-addr2line",
     binary = ":llvm-symbolizer",
@@ -4843,22 +4769,10 @@ gentbl(
     td_srcs = ["include/llvm/Option/OptParser.td"],
 )
 
-expand_template(
-    name = "sancov_main",
-    out = "sancov-driver.cpp",
-    substitutions = {
-        "@TOOL_NAME@": "sancov",
-    },
-    template = "cmake/modules/llvm-driver-template.cpp.in",
-)
-
-cc_binary(
-    name = "sancov",
-    srcs = glob([
-        "tools/sancov/*.cpp",
-    ]) + ["sancov-driver.cpp"],
+cc_library(
+    name = "sancov-lib",
+    srcs = glob(["tools/sancov/*.cpp"]),
     copts = llvm_copts,
-    stamp = 0,
     deps = [
         ":AllTargetsCodeGens",
         ":AllTargetsDisassemblers",
@@ -4874,6 +4788,12 @@ cc_binary(
     ],
 )
 
+llvm_driver_cc_binary(
+    name = "sancov",
+    stamp = 0,
+    deps = [":sancov-lib"],
+)
+
 cc_binary(
     name = "sanstats",
     srcs = glob([
diff --git a/utils/bazel/llvm-project-overlay/llvm/driver.bzl b/utils/bazel/llvm-project-overlay/llvm/driver.bzl
index bd0d26d64f48..888626d7cf84 100644
--- a/utils/bazel/llvm-project-overlay/llvm/driver.bzl
+++ b/utils/bazel/llvm-project-overlay/llvm/driver.bzl
@@ -8,31 +8,29 @@ load("@bazel_skylib//rules:common_settings.bzl", "BuildSettingInfo")
 load("@bazel_skylib//rules:expand_template.bzl", "expand_template")
 
 # Mapping from every tool to the cc_library that implements the tool's entrypoint.
-# TODO: uncomment the remaining targets after splitting them
-# into separate library/binary targets.
 _TOOLS = {
-    # "clang-scan-deps": "//clang:clang-scan-deps-lib",
-    # "clang": "//clang:clang-driver",
-    # "dsymutil": "//llvm:dsymutil-lib",
-    # "lld": "//lld:lld-lib",
+    "clang-scan-deps": "//clang:clang-scan-deps-lib",
+    "clang": "//clang:clang-driver",
+    "dsymutil": "//llvm:dsymutil-lib",
+    "lld": "//lld:lld-lib",
     "llvm-ar": "//llvm:llvm-ar-lib",
-    # "llvm-cxxfilt": "//llvm:llvm-cxxfilt-lib",
-    # "llvm-dwp": "//llvm:llvm-dwp-lib",
-    # "llvm-gsymutil": "//llvm:llvm-gsymutil-lib",
-    # "llvm-ifs": "//llvm:llvm-ifs-lib",
-    # "llvm-libtool-darwin": "//llvm:llvm-libtool-darwin-lib",
-    # "llvm-lipo": "//llvm:llvm-lipo-lib",
-    # "llvm-ml": "//llvm:llvm-ml-lib",
-    # "llvm-mt": "//llvm:llvm-mt-lib",
+    "llvm-cxxfilt": "//llvm:llvm-cxxfilt-lib",
+    "llvm-dwp": "//llvm:llvm-dwp-lib",
+    "llvm-gsymutil": "//llvm:llvm-gsymutil-lib",
+    "llvm-ifs": "//llvm:llvm-ifs-lib",
+    "llvm-libtool-darwin": "//llvm:llvm-libtool-darwin-lib",
+    "llvm-lipo": "//llvm:llvm-lipo-lib",
+    "llvm-ml": "//llvm:llvm-ml-lib",
+    "llvm-mt": "//llvm:llvm-mt-lib",
     "llvm-nm": "//llvm:llvm-nm-lib",
-    # "llvm-objcopy": "//llvm:llvm-objcopy-lib",
-    # "llvm-objdump": "//llvm:llvm-objdump-lib",
-    # "llvm-profdata": "//llvm:llvm-profdata-lib",
-    # "llvm-rc": "//llvm:llvm-rc-lib",
-    # "llvm-readobj": "//llvm:llvm-readobj-lib",
+    "llvm-objcopy": "//llvm:llvm-objcopy-lib",
+    "llvm-objdump": "//llvm:llvm-objdump-lib",
+    "llvm-profdata": "//llvm:llvm-profdata-lib",
+    "llvm-rc": "//llvm:llvm-rc-lib",
+    "llvm-readobj": "//llvm:llvm-readobj-lib",
     "llvm-size": "//llvm:llvm-size-lib",
-    # "llvm-symbolizer": "//llvm:llvm-symbolizer-lib",
-    # "sancov": "//llvm:sancov-lib",
+    "llvm-symbolizer": "//llvm:llvm-symbolizer-lib",
+    "sancov": "//llvm:sancov-lib",
 }
 
 # Tools automatically get their own name as an alias, but there may be additional