[AArch64][SME2] Add builtins for FDOT, BFDOT, SUDOT, USDOT, SDOT, UDOT. (#75737)

Add SME2 DOT builtins.

6 files changed, 1755 insertions, 0 deletions

diff --git a/clang/include/clang/Basic/arm_sme.td b/clang/include/clang/Basic/arm_sme.td
index 2a9bc6870bf7..aac3bd486de9 100644
--- a/clang/include/clang/Basic/arm_sme.td
+++ b/clang/include/clang/Basic/arm_sme.td
@@ -351,6 +351,72 @@ let TargetGuard = "sme2" in {
   def SVBMOPA : Inst<"svbmopa_za32[_{d}]_m", "viPPdd", "iUi", MergeNone, "aarch64_sme_bmopa_za32", [IsSharedZA, IsStreaming], [ImmCheck<0, ImmCheck0_3>]>;
 
   def SVBMOPS : Inst<"svbmops_za32[_{d}]_m", "viPPdd", "iUi", MergeNone, "aarch64_sme_bmops_za32", [IsSharedZA, IsStreaming], [ImmCheck<0, ImmCheck0_3>]>;
+
+  // VERTICAL DOT-PRODUCT
+  def SVVDOT_LANE_ZA32_VG1x2_S : Inst<"svvdot_lane_za32[_{d}]_vg1x2", "vm2di", "s", MergeNone, "aarch64_sme_svdot_lane_za32_vg1x2", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_3>]>;
+  def SVVDOT_LANE_ZA32_VG1x4_S : Inst<"svvdot_lane_za32[_{d}]_vg1x4", "vm4di", "c", MergeNone, "aarch64_sme_svdot_lane_za32_vg1x4", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_3>]>;
+  def SVVDOT_LANE_ZA32_VG1x2_U : Inst<"svvdot_lane_za32[_{d}]_vg1x2", "vm2di", "Us", MergeNone, "aarch64_sme_uvdot_lane_za32_vg1x2", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_3>]>;
+  def SVVDOT_LANE_ZA32_VG1x4_U : Inst<"svvdot_lane_za32[_{d}]_vg1x4", "vm4di", "Uc", MergeNone, "aarch64_sme_uvdot_lane_za32_vg1x4", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_3>]>;
+  def SVVDOT_LANE_ZA32_VG1x2_F : Inst<"svvdot_lane_za32[_{d}]_vg1x2", "vm2di", "hb", MergeNone, "aarch64_sme_fvdot_lane_za32_vg1x2", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_3>]>;
+  def SVSUVDOT_LANE_ZA32_VG1x4 : Inst<"svsuvdot_lane_za32[_{d}]_vg1x4", "vm4di", "c", MergeNone, "aarch64_sme_suvdot_lane_za32_vg1x4", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_3>]>;
+  def SVUSVDOT_LANE_ZA32_VG1x4 : Inst<"svusvdot_lane_za32[_{d}]_vg1x4", "vm4di", "Uc", MergeNone, "aarch64_sme_usvdot_lane_za32_vg1x4", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_3>]>;
+
+  // Multi-vector signed & unsigned integer dot-product
+  def SVDOT_MULTI_ZA32_VG1x2_S  : Inst<"svdot_za32[_{d}]_vg1x2", "vm22", "cs", MergeNone, "aarch64_sme_sdot_za32_vg1x2", [IsStreaming, IsSharedZA], []>;
+  def SVDOT_MULTI_ZA32_VG1x4_S  : Inst<"svdot_za32[_{d}]_vg1x4", "vm44", "cs", MergeNone, "aarch64_sme_sdot_za32_vg1x4", [IsStreaming, IsSharedZA], []>;
+  def SVDOT_MULTI_ZA32_VG1x2_U  : Inst<"svdot_za32[_{d}]_vg1x2", "vm22", "UcUs", MergeNone, "aarch64_sme_udot_za32_vg1x2", [IsStreaming, IsSharedZA], []>;
+  def SVDOT_MULTI_ZA32_VG1x4_U  : Inst<"svdot_za32[_{d}]_vg1x4", "vm44", "UcUs", MergeNone, "aarch64_sme_udot_za32_vg1x4", [IsStreaming, IsSharedZA], []>;
+  def SVDOT_SINGLE_ZA32_VG1x2_S : Inst<"svdot[_single]_za32[_{d}]_vg1x2", "vm2d", "cs", MergeNone, "aarch64_sme_sdot_single_za32_vg1x2", [IsStreaming, IsSharedZA], []>;
+  def SVDOT_SINGLE_ZA32_VG1x4_S : Inst<"svdot[_single]_za32[_{d}]_vg1x4", "vm4d", "cs", MergeNone, "aarch64_sme_sdot_single_za32_vg1x4", [IsStreaming, IsSharedZA], []>;
+  def SVDOT_SINGLE_ZA32_VG1x2_U : Inst<"svdot[_single]_za32[_{d}]_vg1x2", "vm2d", "UcUs", MergeNone, "aarch64_sme_udot_single_za32_vg1x2", [IsStreaming, IsSharedZA], []>;
+  def SVDOT_SINGLE_ZA32_VG1x4_U : Inst<"svdot[_single]_za32[_{d}]_vg1x4", "vm4d", "UcUs", MergeNone, "aarch64_sme_udot_single_za32_vg1x4", [IsStreaming, IsSharedZA], []>;
+  def SVDOT_LANE_ZA32_VG1x2_S   : Inst<"svdot_lane_za32[_{d}]_vg1x2", "vm2di", "cs", MergeNone, "aarch64_sme_sdot_lane_za32_vg1x2", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_3>]>;
+  def SVDOT_LANE_ZA32_VG1x4_S   : Inst<"svdot_lane_za32[_{d}]_vg1x4", "vm4di", "cs", MergeNone, "aarch64_sme_sdot_lane_za32_vg1x4", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_3>]>;
+  def SVDOT_LANE_ZA32_VG1x2_U   : Inst<"svdot_lane_za32[_{d}]_vg1x2", "vm2di", "UcUs", MergeNone, "aarch64_sme_udot_lane_za32_vg1x2", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_3>]>;
+  def SVDOT_LANE_ZA32_VG1x4_U   : Inst<"svdot_lane_za32[_{d}]_vg1x4", "vm4di", "UcUs", MergeNone, "aarch64_sme_udot_lane_za32_vg1x4", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_3>]>;
+
+  def SVUSDOT_SINGLE_ZA32_VG1x2 : Inst<"svusdot[_single]_za32[_{d}]_vg1x2", "vm2.dx", "Uc", MergeNone, "aarch64_sme_usdot_single_za32_vg1x2", [IsStreaming, IsSharedZA], []>;
+  def SVUSDOT_SINGLE_ZA32_VG1x4 : Inst<"svusdot[_single]_za32[_{d}]_vg1x4", "vm4.dx", "Uc", MergeNone, "aarch64_sme_usdot_single_za32_vg1x4", [IsStreaming, IsSharedZA], []>;
+  def SVUSDOT_MULTI_ZA32_VG1x2  : Inst<"svusdot_za32[_{d}]_vg1x2", "vm2.d2.x", "Uc", MergeNone, "aarch64_sme_usdot_za32_vg1x2", [IsStreaming, IsSharedZA], []>;
+  def SVUSDOT_MULTI_ZA32_VG1x4  : Inst<"svusdot_za32[_{d}]_vg1x4", "vm4.d4.x", "Uc", MergeNone, "aarch64_sme_usdot_za32_vg1x4", [IsStreaming, IsSharedZA], []>;
+  def SVUSDOT_LANE_ZA32_VG1x2   : Inst<"svusdot_lane_za32[_{d}]_vg1x2", "vm2.dxi", "Uc", MergeNone, "aarch64_sme_usdot_lane_za32_vg1x2", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_3>]>;
+  def SVUSDOT_LANE_ZA32_VG1x4   : Inst<"svusdot_lane_za32[_{d}]_vg1x4", "vm4.dxi", "Uc", MergeNone, "aarch64_sme_usdot_lane_za32_vg1x4", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_3>]>;
+
+  def SVSUDOT_SINGLE_ZA32_VG1x2 : Inst<"svsudot[_single]_za32[_{d}]_vg1x2", "vm2.du", "c", MergeNone, "aarch64_sme_sudot_single_za32_vg1x2", [IsStreaming, IsSharedZA], []>;
+  def SVSUDOT_SINGLE_ZA32_VG1x4 : Inst<"svsudot[_single]_za32[_{d}]_vg1x4", "vm4.du", "c", MergeNone, "aarch64_sme_sudot_single_za32_vg1x4", [IsStreaming, IsSharedZA], []>;
+
+  // Multi-multi sudot builtins are mapped to usdot, with zn & zm operands swapped
+  def SVSUDOT_MULTI_ZA32_VG1x2  : Inst<"svsudot_za32[_{d}]_vg1x2", "vm2.d2.u", "c", MergeNone, "aarch64_sme_usdot_za32_vg1x2", [IsStreaming, IsSharedZA], []>;
+  def SVSUDOT_MULTI_ZA32_VG1x4  : Inst<"svsudot_za32[_{d}]_vg1x4", "vm4.d4.u", "c", MergeNone, "aarch64_sme_usdot_za32_vg1x4", [IsStreaming, IsSharedZA], []>;
+
+  def SVSUDOT_LANE_ZA32_VG1x2   : Inst<"svsudot_lane_za32[_{d}]_vg1x2", "vm2.dui", "c", MergeNone, "aarch64_sme_sudot_lane_za32_vg1x2", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_3>]>;
+  def SVSUDOT_LANE_ZA32_VG1x4   : Inst<"svsudot_lane_za32[_{d}]_vg1x4", "vm4.dui", "c", MergeNone, "aarch64_sme_sudot_lane_za32_vg1x4", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_3>]>;
+
+  // Multi-vector half-precision/BFloat16 floating-point dot-product
+  def SVDOT_MULTI_ZA32_VG1x2_F16  : Inst<"svdot_za32[_{d}]_vg1x2", "vm22", "bh", MergeNone, "aarch64_sme_fdot_za32_vg1x2", [IsStreaming, IsSharedZA], []>;
+  def SVDOT_MULTI_ZA32_VG1x4_F16  : Inst<"svdot_za32[_{d}]_vg1x4", "vm44", "bh", MergeNone, "aarch64_sme_fdot_za32_vg1x4", [IsStreaming, IsSharedZA], []>;
+  def SVDOT_SINGLE_ZA32_VG1x2_F16 : Inst<"svdot[_single]_za32[_{d}]_vg1x2", "vm2d", "bh", MergeNone, "aarch64_sme_fdot_single_za32_vg1x2", [IsStreaming, IsSharedZA], []>;
+  def SVDOT_SINGLE_ZA32_VG1x4_F16 : Inst<"svdot[_single]_za32[_{d}]_vg1x4", "vm4d", "bh", MergeNone, "aarch64_sme_fdot_single_za32_vg1x4", [IsStreaming, IsSharedZA], []>;
+  def SVDOT_LANE_ZA32_VG1x2_F16   : Inst<"svdot_lane_za32[_{d}]_vg1x2", "vm2di", "bh", MergeNone, "aarch64_sme_fdot_lane_za32_vg1x2", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_3>]>;
+  def SVDOT_LANE_ZA32_VG1x4_F16   : Inst<"svdot_lane_za32[_{d}]_vg1x4", "vm4di", "bh", MergeNone, "aarch64_sme_fdot_lane_za32_vg1x4", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_3>]>;
+}
+
+let TargetGuard = "sme2,sme-i16i64" in {
+  def SVVDOT_LANE_ZA64_VG1x4_S : Inst<"svvdot_lane_za64[_{d}]_vg1x4", "vm4di", "s", MergeNone, "aarch64_sme_svdot_lane_za64_vg1x4", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_1>]>;
+  def SVVDOT_LANE_ZA64_VG1x4_U : Inst<"svvdot_lane_za64[_{d}]_vg1x4", "vm4di", "Us", MergeNone, "aarch64_sme_uvdot_lane_za64_vg1x4", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_1>]>;
+
+  def SVDOT_MULTI_ZA64_VG1x2_S16  : Inst<"svdot_za64[_{d}]_vg1x2", "vm22", "s", MergeNone, "aarch64_sme_sdot_za64_vg1x2", [IsStreaming, IsSharedZA], []>;
+  def SVDOT_MULTI_ZA64_VG1x4_S16  : Inst<"svdot_za64[_{d}]_vg1x4", "vm44", "s", MergeNone, "aarch64_sme_sdot_za64_vg1x4", [IsStreaming, IsSharedZA], []>;
+  def SVDOT_MULTI_ZA64_VG1x2_U16  : Inst<"svdot_za64[_{d}]_vg1x2", "vm22", "Us", MergeNone, "aarch64_sme_udot_za64_vg1x2", [IsStreaming, IsSharedZA], []>;
+  def SVDOT_MULTI_ZA64_VG1x4_U16  : Inst<"svdot_za64[_{d}]_vg1x4", "vm44", "Us", MergeNone, "aarch64_sme_udot_za64_vg1x4", [IsStreaming, IsSharedZA], []>;
+  def SVDOT_SINGLE_ZA64_VG1x2_S16 : Inst<"svdot[_single]_za64[_{d}]_vg1x2", "vm2d", "s", MergeNone, "aarch64_sme_sdot_single_za64_vg1x2", [IsStreaming, IsSharedZA], []>;
+  def SVDOT_SINGLE_ZA64_VG1x4_S16 : Inst<"svdot[_single]_za64[_{d}]_vg1x4", "vm4d", "s", MergeNone, "aarch64_sme_sdot_single_za64_vg1x4", [IsStreaming, IsSharedZA], []>;
+  def SVDOT_SINGLE_ZA64_VG1x2_U16 : Inst<"svdot[_single]_za64[_{d}]_vg1x2", "vm2d", "Us", MergeNone, "aarch64_sme_udot_single_za64_vg1x2", [IsStreaming, IsSharedZA], []>;
+  def SVDOT_SINGLE_ZA64_VG1x4_U16 : Inst<"svdot[_single]_za64[_{d}]_vg1x4", "vm4d", "Us", MergeNone, "aarch64_sme_udot_single_za64_vg1x4", [IsStreaming, IsSharedZA], []>;
+  def SVDOT_LANE_ZA64_VG1x2_S16   : Inst<"svdot_lane_za64[_{d}]_vg1x2", "vm2di", "s", MergeNone, "aarch64_sme_sdot_lane_za64_vg1x2", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_1>]>;
+  def SVDOT_LANE_ZA64_VG1x4_S16   : Inst<"svdot_lane_za64[_{d}]_vg1x4", "vm4di", "s", MergeNone, "aarch64_sme_sdot_lane_za64_vg1x4", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_1>]>;
+  def SVDOT_LANE_ZA64_VG1x2_U16   : Inst<"svdot_lane_za64[_{d}]_vg1x2", "vm2di", "Us", MergeNone, "aarch64_sme_udot_lane_za64_vg1x2", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_1>]>;
+  def SVDOT_LANE_ZA64_VG1x4_U16   : Inst<"svdot_lane_za64[_{d}]_vg1x4", "vm4di", "Us", MergeNone, "aarch64_sme_udot_lane_za64_vg1x4", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_1>]>;
 }
 
 // FMLA/FMLS
diff --git a/clang/lib/CodeGen/CGBuiltin.cpp b/clang/lib/CodeGen/CGBuiltin.cpp
index 7bc3b7594c8f..5081062da286 100644
--- a/clang/lib/CodeGen/CGBuiltin.cpp
+++ b/clang/lib/CodeGen/CGBuiltin.cpp
@@ -10328,8 +10328,10 @@ static void swapCommutativeSMEOperands(unsigned BuiltinID,
     MultiVec = 1;
     break;
   case SME::BI__builtin_sme_svsumla_za32_s8_vg4x2:
+  case SME::BI__builtin_sme_svsudot_za32_s8_vg1x2:
     MultiVec = 2;
     break;
+  case SME::BI__builtin_sme_svsudot_za32_s8_vg1x4:
   case SME::BI__builtin_sme_svsumla_za32_s8_vg4x4:
     MultiVec = 4;
     break;
diff --git a/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_fp_dots.c b/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_fp_dots.c
new file mode 100644
index 000000000000..ff4176530710
--- /dev/null
+++ b/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_fp_dots.c
@@ -0,0 +1,297 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
+// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sme2 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s
+// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sme2 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -S -disable-O0-optnone -Werror -Wall -o /dev/null %s
+#include <arm_sme_draft_spec_subject_to_change.h>
+
+#ifdef SVE_OVERLOADED_FORMS
+// A simple used,unused... macro, long enough to represent any SVE builtin.
+#define SVE_ACLE_FUNC(A1,A2_UNUSED,A3,A4_UNUSED,A5) A1##A3##A5
+#else
+#define SVE_ACLE_FUNC(A1,A2,A3,A4,A5) A1##A2##A3##A4##A5
+#endif
+
+//
+// Multi, multi (half)
+// CHECK-LABEL: @test_svdot_multi_za32_vg1x2_f16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZM]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fdot.za32.vg1x2.nxv8f16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z31test_svdot_multi_za32_vg1x2_f16j13svfloat16x2_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZM]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fdot.za32.vg1x2.nxv8f16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_multi_za32_vg1x2_f16(uint32_t slice_base, svfloat16x2_t zn, svfloat16x2_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_za32,,,_f16,_vg1x2)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svdot_multi_za32_vg1x4_f16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 24)
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZM]], i64 8)
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZM]], i64 16)
+// CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZM]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fdot.za32.vg1x4.nxv8f16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]], <vscale x 8 x half> [[TMP4]], <vscale x 8 x half> [[TMP5]], <vscale x 8 x half> [[TMP6]], <vscale x 8 x half> [[TMP7]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z31test_svdot_multi_za32_vg1x4_f16j13svfloat16x4_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZM]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZM]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZM]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fdot.za32.vg1x4.nxv8f16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]], <vscale x 8 x half> [[TMP4]], <vscale x 8 x half> [[TMP5]], <vscale x 8 x half> [[TMP6]], <vscale x 8 x half> [[TMP7]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_multi_za32_vg1x4_f16(uint32_t slice_base, svfloat16x4_t zn, svfloat16x4_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_za32,,,_f16,_vg1x4)(slice_base, zn, zm);
+}
+
+
+//
+// Multi, single (half)
+// CHECK-LABEL: @test_svdot_single_za32_vg1x2_f16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZN]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fdot.single.za32.vg1x2.nxv8f16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z32test_svdot_single_za32_vg1x2_f16j13svfloat16x2_tu13__SVFloat16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fdot.single.za32.vg1x2.nxv8f16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_single_za32_vg1x2_f16(uint32_t slice_base, svfloat16x2_t zn, svfloat16_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_single_za32,,_f16,,_vg1x2)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svdot_single_za32_vg1x4_f16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fdot.single.za32.vg1x4.nxv8f16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]], <vscale x 8 x half> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z32test_svdot_single_za32_vg1x4_f16j13svfloat16x4_tu13__SVFloat16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fdot.single.za32.vg1x4.nxv8f16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]], <vscale x 8 x half> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_single_za32_vg1x4_f16(uint32_t slice_base, svfloat16x4_t zn, svfloat16_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_single_za32,,_f16,,_vg1x4)(slice_base, zn, zm);
+}
+
+
+//
+// Multi, indexed (half)
+// CHECK-LABEL: @test_svdot_lane_za32_vg1x2_f16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZN]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fdot.lane.za32.vg1x2.nxv8f16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[ZM:%.*]], i32 3)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z30test_svdot_lane_za32_vg1x2_f16j13svfloat16x2_tu13__SVFloat16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fdot.lane.za32.vg1x2.nxv8f16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[ZM:%.*]], i32 3)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_lane_za32_vg1x2_f16(uint32_t slice_base, svfloat16x2_t zn, svfloat16_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_lane_za32,,,_f16,_vg1x2)(slice_base, zn, zm, 3);
+}
+
+// CHECK-LABEL: @test_svdot_lane_za32_vg1x4_f16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fdot.lane.za32.vg1x4.nxv8f16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]], <vscale x 8 x half> [[ZM:%.*]], i32 3)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z30test_svdot_lane_za32_vg1x4_f16j13svfloat16x4_tu13__SVFloat16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fdot.lane.za32.vg1x4.nxv8f16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]], <vscale x 8 x half> [[ZM:%.*]], i32 3)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_lane_za32_vg1x4_f16(uint32_t slice_base, svfloat16x4_t zn, svfloat16_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_lane_za32,,,_f16,_vg1x4)(slice_base, zn, zm, 3);
+}
+
+
+//
+// Multi, multi (bfloat)
+// CHECK-LABEL: @test_svdot_multi_za32_vg1x2_bf16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZM]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fdot.za32.vg1x2.nxv8bf16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]], <vscale x 8 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z32test_svdot_multi_za32_vg1x2_bf16j14svbfloat16x2_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZM]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fdot.za32.vg1x2.nxv8bf16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]], <vscale x 8 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_multi_za32_vg1x2_bf16(uint32_t slice_base, svbfloat16x2_t zn, svbfloat16x2_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_za32,,,_bf16,_vg1x2)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svdot_multi_za32_vg1x4_bf16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 24)
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZM]], i64 8)
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZM]], i64 16)
+// CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZM]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fdot.za32.vg1x4.nxv8bf16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]], <vscale x 8 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]], <vscale x 8 x bfloat> [[TMP4]], <vscale x 8 x bfloat> [[TMP5]], <vscale x 8 x bfloat> [[TMP6]], <vscale x 8 x bfloat> [[TMP7]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z32test_svdot_multi_za32_vg1x4_bf16j14svbfloat16x4_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZM]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZM]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZM]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fdot.za32.vg1x4.nxv8bf16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]], <vscale x 8 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]], <vscale x 8 x bfloat> [[TMP4]], <vscale x 8 x bfloat> [[TMP5]], <vscale x 8 x bfloat> [[TMP6]], <vscale x 8 x bfloat> [[TMP7]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_multi_za32_vg1x4_bf16(uint32_t slice_base, svbfloat16x4_t zn, svbfloat16x4_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_za32,,,_bf16,_vg1x4)(slice_base, zn, zm);
+}
+
+
+//
+// Multi, single (bfloat)
+// CHECK-LABEL: @test_svdot_single_za32_vg1x2_bf16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZN]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fdot.single.za32.vg1x2.nxv8bf16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]], <vscale x 8 x bfloat> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z33test_svdot_single_za32_vg1x2_bf16j14svbfloat16x2_tu14__SVBfloat16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fdot.single.za32.vg1x2.nxv8bf16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]], <vscale x 8 x bfloat> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_single_za32_vg1x2_bf16(uint32_t slice_base, svbfloat16x2_t zn, svbfloat16_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_single_za32,,_bf16,,_vg1x2)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svdot_single_za32_vg1x4_bf16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fdot.single.za32.vg1x4.nxv8bf16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]], <vscale x 8 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]], <vscale x 8 x bfloat> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z33test_svdot_single_za32_vg1x4_bf16j14svbfloat16x4_tu14__SVBfloat16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fdot.single.za32.vg1x4.nxv8bf16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]], <vscale x 8 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]], <vscale x 8 x bfloat> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_single_za32_vg1x4_bf16(uint32_t slice_base, svbfloat16x4_t zn, svbfloat16_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_single_za32,,_bf16,,_vg1x4)(slice_base, zn, zm);
+}
+
+
+//
+// Multi, indexed (bfloat)
+// CHECK-LABEL: @test_svdot_lane_za32_vg1x2_bf16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZN]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fdot.lane.za32.vg1x2.nxv8bf16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]], <vscale x 8 x bfloat> [[ZM:%.*]], i32 3)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z31test_svdot_lane_za32_vg1x2_bf16j14svbfloat16x2_tu14__SVBfloat16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fdot.lane.za32.vg1x2.nxv8bf16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]], <vscale x 8 x bfloat> [[ZM:%.*]], i32 3)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_lane_za32_vg1x2_bf16(uint32_t slice_base, svbfloat16x2_t zn, svbfloat16_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_lane_za32,,_bf16,,_vg1x2)(slice_base, zn, zm, 3);
+}
+
+// CHECK-LABEL: @test_svdot_lane_za32_vg1x4_bf16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fdot.lane.za32.vg1x4.nxv8bf16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]], <vscale x 8 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]], <vscale x 8 x bfloat> [[ZM:%.*]], i32 3)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z31test_svdot_lane_za32_vg1x4_bf16j14svbfloat16x4_tu14__SVBfloat16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fdot.lane.za32.vg1x4.nxv8bf16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]], <vscale x 8 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]], <vscale x 8 x bfloat> [[ZM:%.*]], i32 3)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_lane_za32_vg1x4_bf16(uint32_t slice_base, svbfloat16x4_t zn, svbfloat16_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_lane_za32,,_bf16,,_vg1x4)(slice_base, zn, zm, 3);
+}
diff --git a/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_int_dots.c b/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_int_dots.c
new file mode 100644
index 000000000000..0d85071b7fc3
--- /dev/null
+++ b/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_int_dots.c
@@ -0,0 +1,1103 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sme-i16i64 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sme-i16i64 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
+// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sme-i16i64 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s
+// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sme-i16i64 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sme-i16i64 -S -disable-O0-optnone -Werror -Wall -o /dev/null %s
+#include <arm_sme_draft_spec_subject_to_change.h>
+
+#ifdef SVE_OVERLOADED_FORMS
+// A simple used,unused... macro, long enough to represent any SVE builtin.
+#define SVE_ACLE_FUNC(A1,A2_UNUSED,A3,A4_UNUSED,A5) A1##A3##A5
+#else
+#define SVE_ACLE_FUNC(A1,A2,A3,A4,A5) A1##A2##A3##A4##A5
+#endif
+
+//
+// Multi, multi (unsigned)
+// CHECK-LABEL: @test_svdot_multi_za32_vg1x2_u16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.udot.za32.vg1x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z31test_svdot_multi_za32_vg1x2_u16j12svuint16x2_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.udot.za32.vg1x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_multi_za32_vg1x2_u16(uint32_t slice_base, svuint16x2_t zn, svuint16x2_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_za32,,,_u16,_vg1x2)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svdot_multi_za32_vg1x4_u16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 8)
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 16)
+// CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.udot.za32.vg1x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], <vscale x 8 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z31test_svdot_multi_za32_vg1x4_u16j12svuint16x4_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.udot.za32.vg1x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], <vscale x 8 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_multi_za32_vg1x4_u16(uint32_t slice_base, svuint16x4_t zn, svuint16x4_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_za32,,,_u16,_vg1x4)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svdot_multi_za32_vg1x2_u8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZM]], i64 16)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.udot.za32.vg1x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z30test_svdot_multi_za32_vg1x2_u8j11svuint8x2_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZM]], i64 16)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.udot.za32.vg1x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_multi_za32_vg1x2_u8(uint32_t slice_base, svuint8x2_t zn, svuint8x2_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_za32,,,_u8,_vg1x2)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svdot_multi_za32_vg1x4_u8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 16)
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 32)
+// CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 48)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.udot.za32.vg1x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[TMP4]], <vscale x 16 x i8> [[TMP5]], <vscale x 16 x i8> [[TMP6]], <vscale x 16 x i8> [[TMP7]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z30test_svdot_multi_za32_vg1x4_u8j11svuint8x4_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 48)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.udot.za32.vg1x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[TMP4]], <vscale x 16 x i8> [[TMP5]], <vscale x 16 x i8> [[TMP6]], <vscale x 16 x i8> [[TMP7]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_multi_za32_vg1x4_u8(uint32_t slice_base, svuint8x4_t zn, svuint8x4_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_za32,,,_u8,_vg1x4)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svdot_multi_za64_vg1x2_u16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.udot.za64.vg1x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z31test_svdot_multi_za64_vg1x2_u16j12svuint16x2_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.udot.za64.vg1x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_multi_za64_vg1x2_u16(uint32_t slice_base, svuint16x2_t zn, svuint16x2_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_za64,,,_u16,_vg1x2)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svdot_multi_za64_vg1x4_u16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 8)
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 16)
+// CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.udot.za64.vg1x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], <vscale x 8 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z31test_svdot_multi_za64_vg1x4_u16j12svuint16x4_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.udot.za64.vg1x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], <vscale x 8 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_multi_za64_vg1x4_u16(uint32_t slice_base, svuint16x4_t zn, svuint16x4_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_za64,,,_u16,_vg1x4)(slice_base, zn, zm);
+}
+
+
+//
+// Multi, multi (signed)
+// CHECK-LABEL: @test_svdot_multi_za32_vg1x2_s16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.sdot.za32.vg1x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z31test_svdot_multi_za32_vg1x2_s16j11svint16x2_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.sdot.za32.vg1x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_multi_za32_vg1x2_s16(uint32_t slice_base, svint16x2_t zn, svint16x2_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_za32,,,_s16,_vg1x2)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svdot_multi_za32_vg1x4_s16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 8)
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 16)
+// CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.sdot.za32.vg1x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], <vscale x 8 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z31test_svdot_multi_za32_vg1x4_s16j11svint16x4_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.sdot.za32.vg1x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], <vscale x 8 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_multi_za32_vg1x4_s16(uint32_t slice_base, svint16x4_t zn, svint16x4_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_za32,,,_s16,_vg1x4)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svdot_multi_za32_vg1x2_s8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZM]], i64 16)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.sdot.za32.vg1x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z30test_svdot_multi_za32_vg1x2_s8j10svint8x2_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZM]], i64 16)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.sdot.za32.vg1x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_multi_za32_vg1x2_s8(uint32_t slice_base, svint8x2_t zn, svint8x2_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_za32,,,_s8,_vg1x2)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svdot_multi_za32_vg1x4_s8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 16)
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 32)
+// CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 48)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.sdot.za32.vg1x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[TMP4]], <vscale x 16 x i8> [[TMP5]], <vscale x 16 x i8> [[TMP6]], <vscale x 16 x i8> [[TMP7]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z30test_svdot_multi_za32_vg1x4_s8j10svint8x4_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 48)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.sdot.za32.vg1x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[TMP4]], <vscale x 16 x i8> [[TMP5]], <vscale x 16 x i8> [[TMP6]], <vscale x 16 x i8> [[TMP7]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_multi_za32_vg1x4_s8(uint32_t slice_base, svint8x4_t zn, svint8x4_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_za32,,,_s8,_vg1x4)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svdot_multi_za64_vg1x2_s16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.sdot.za64.vg1x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z31test_svdot_multi_za64_vg1x2_s16j11svint16x2_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.sdot.za64.vg1x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_multi_za64_vg1x2_s16(uint32_t slice_base, svint16x2_t zn, svint16x2_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_za64,,,_s16,_vg1x2)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svdot_multi_za64_vg1x4_s16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 8)
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 16)
+// CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.sdot.za64.vg1x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], <vscale x 8 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z31test_svdot_multi_za64_vg1x4_s16j11svint16x4_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.sdot.za64.vg1x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], <vscale x 8 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_multi_za64_vg1x4_s16(uint32_t slice_base, svint16x4_t zn, svint16x4_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_za64,,,_s16,_vg1x4)(slice_base, zn, zm);
+}
+
+
+//
+// Multi, single (unsigned)
+// CHECK-LABEL: @test_svdot_single_za32_vg1x2_u16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.udot.single.za32.vg1x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z32test_svdot_single_za32_vg1x2_u16j12svuint16x2_tu12__SVUint16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.udot.single.za32.vg1x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_single_za32_vg1x2_u16(uint32_t slice_base, svuint16x2_t zn, svuint16_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_single_za32,,_u16,,_vg1x2)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svdot_single_za32_vg1x4_u16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.udot.single.za32.vg1x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z32test_svdot_single_za32_vg1x4_u16j12svuint16x4_tu12__SVUint16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.udot.single.za32.vg1x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_single_za32_vg1x4_u16(uint32_t slice_base, svuint16x4_t zn, svuint16_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_single_za32,,_u16,,_vg1x4)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svdot_single_za32_vg1x2_u8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.udot.single.za32.vg1x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z31test_svdot_single_za32_vg1x2_u8j11svuint8x2_tu11__SVUint8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.udot.single.za32.vg1x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_single_za32_vg1x2_u8(uint32_t slice_base, svuint8x2_t zn, svuint8_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_single_za32,,_u8,,_vg1x2)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svdot_single_za32_vg1x4_u8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.udot.single.za32.vg1x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z31test_svdot_single_za32_vg1x4_u8j11svuint8x4_tu11__SVUint8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.udot.single.za32.vg1x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_single_za32_vg1x4_u8(uint32_t slice_base, svuint8x4_t zn, svuint8_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_single_za32,,_u8,,_vg1x4)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svdot_single_za64_vg1x2_u16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.udot.single.za64.vg1x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z32test_svdot_single_za64_vg1x2_u16j12svuint16x2_tu12__SVUint16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.udot.single.za64.vg1x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_single_za64_vg1x2_u16(uint32_t slice_base, svuint16x2_t zn, svuint16_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_single_za64,,_u16,,_vg1x2)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svdot_single_za64_vg1x4_u16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.udot.single.za64.vg1x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z32test_svdot_single_za64_vg1x4_u16j12svuint16x4_tu12__SVUint16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.udot.single.za64.vg1x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_single_za64_vg1x4_u16(uint32_t slice_base, svuint16x4_t zn, svuint16_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_single_za64,,_u16,,_vg1x4)(slice_base, zn, zm);
+}
+
+
+//
+// Multi, single (signed)
+// CHECK-LABEL: @test_svdot_single_za32_vg1x2_s16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.sdot.single.za32.vg1x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z32test_svdot_single_za32_vg1x2_s16j11svint16x2_tu11__SVInt16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.sdot.single.za32.vg1x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_single_za32_vg1x2_s16(uint32_t slice_base, svint16x2_t zn, svint16_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_single_za32,,_s16,,_vg1x2)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svdot_single_za32_vg1x4_s16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.sdot.single.za32.vg1x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z32test_svdot_single_za32_vg1x4_s16j11svint16x4_tu11__SVInt16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.sdot.single.za32.vg1x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_single_za32_vg1x4_s16(uint32_t slice_base, svint16x4_t zn, svint16_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_single_za32,,_s16,,_vg1x4)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svdot_single_za32_vg1x2_s8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.sdot.single.za32.vg1x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z31test_svdot_single_za32_vg1x2_s8j10svint8x2_tu10__SVInt8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.sdot.single.za32.vg1x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_single_za32_vg1x2_s8(uint32_t slice_base, svint8x2_t zn, svint8_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_single_za32,,_s8,,_vg1x2)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svdot_single_za32_vg1x4_s8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.sdot.single.za32.vg1x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z31test_svdot_single_za32_vg1x4_s8j10svint8x4_tu10__SVInt8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.sdot.single.za32.vg1x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_single_za32_vg1x4_s8(uint32_t slice_base, svint8x4_t zn, svint8_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_single_za32,,_s8,,_vg1x4)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svdot_single_za64_vg1x2_s16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.sdot.single.za64.vg1x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z32test_svdot_single_za64_vg1x2_s16j11svint16x2_tu11__SVInt16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.sdot.single.za64.vg1x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_single_za64_vg1x2_s16(uint32_t slice_base, svint16x2_t zn, svint16_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_single_za64,,_s16,,_vg1x2)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svdot_single_za64_vg1x4_s16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.sdot.single.za64.vg1x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z32test_svdot_single_za64_vg1x4_s16j11svint16x4_tu11__SVInt16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.sdot.single.za64.vg1x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_single_za64_vg1x4_s16(uint32_t slice_base, svint16x4_t zn, svint16_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_single_za64,,_s16,,_vg1x4)(slice_base, zn, zm);
+}
+
+//
+// Multi, indexed (unsigned)
+// CHECK-LABEL: @test_svdot_lane_za32_vg1x2_u16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.udot.lane.za32.vg1x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]], i32 3)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z30test_svdot_lane_za32_vg1x2_u16j12svuint16x2_tu12__SVUint16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.udot.lane.za32.vg1x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]], i32 3)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_lane_za32_vg1x2_u16(uint32_t slice_base, svuint16x2_t zn, svuint16_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_lane_za32,,_u16,,_vg1x2)(slice_base, zn, zm, 3);
+}
+
+// CHECK-LABEL: @test_svdot_lane_za32_vg1x4_u16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.udot.lane.za32.vg1x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]], i32 3)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z30test_svdot_lane_za32_vg1x4_u16j12svuint16x4_tu12__SVUint16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.udot.lane.za32.vg1x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]], i32 3)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_lane_za32_vg1x4_u16(uint32_t slice_base, svuint16x4_t zn, svuint16_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_lane_za32,,_u16,,_vg1x4)(slice_base, zn, zm, 3);
+}
+
+// CHECK-LABEL: @test_svdot_lane_za32_vg1x2_u8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.udot.lane.za32.vg1x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[ZM:%.*]], i32 3)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z29test_svdot_lane_za32_vg1x2_u8j11svuint8x2_tu11__SVUint8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.udot.lane.za32.vg1x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[ZM:%.*]], i32 3)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_lane_za32_vg1x2_u8(uint32_t slice_base, svuint8x2_t zn, svuint8_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_lane_za32,,_u8,,_vg1x2)(slice_base, zn, zm, 3);
+}
+
+// CHECK-LABEL: @test_svdot_lane_za32_vg1x4_u8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.udot.lane.za32.vg1x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]], i32 3)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z29test_svdot_lane_za32_vg1x4_u8j11svuint8x4_tu11__SVUint8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.udot.lane.za32.vg1x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]], i32 3)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_lane_za32_vg1x4_u8(uint32_t slice_base, svuint8x4_t zn, svuint8_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_lane_za32,,_u8,,_vg1x4)(slice_base, zn, zm, 3);
+}
+
+// CHECK-LABEL: @test_svdot_lane_za64_vg1x2_u16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.udot.lane.za64.vg1x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]], i32 1)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z30test_svdot_lane_za64_vg1x2_u16j12svuint16x2_tu12__SVUint16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.udot.lane.za64.vg1x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]], i32 1)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_lane_za64_vg1x2_u16(uint32_t slice_base, svuint16x2_t zn, svuint16_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_lane_za64,,,_u16,_vg1x2)(slice_base, zn, zm, 1);
+}
+
+// CHECK-LABEL: @test_svdot_lane_za64_vg1x4_u16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.udot.lane.za64.vg1x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]], i32 1)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z30test_svdot_lane_za64_vg1x4_u16j12svuint16x4_tu12__SVUint16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.udot.lane.za64.vg1x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]], i32 1)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_lane_za64_vg1x4_u16(uint32_t slice_base, svuint16x4_t zn, svuint16_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_lane_za64,,,_u16,_vg1x4)(slice_base, zn, zm, 1);
+}
+
+
+//
+// Multi, indexed (signed)
+// CHECK-LABEL: @test_svdot_lane_za32_vg1x2_s16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.sdot.lane.za32.vg1x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]], i32 3)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z30test_svdot_lane_za32_vg1x2_s16j11svint16x2_tu11__SVInt16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.sdot.lane.za32.vg1x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]], i32 3)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_lane_za32_vg1x2_s16(uint32_t slice_base, svint16x2_t zn, svint16_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_lane_za32,,,_s16,_vg1x2)(slice_base, zn, zm, 3);
+}
+
+// CHECK-LABEL: @test_svdot_lane_za32_vg1x4_s16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.sdot.lane.za32.vg1x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]], i32 3)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z30test_svdot_lane_za32_vg1x4_s16j11svint16x4_tu11__SVInt16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.sdot.lane.za32.vg1x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]], i32 3)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_lane_za32_vg1x4_s16(uint32_t slice_base, svint16x4_t zn, svint16_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_lane_za32,,,_s16,_vg1x4)(slice_base, zn, zm, 3);
+}
+
+// CHECK-LABEL: @test_svdot_lane_za32_vg1x2_s8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.sdot.lane.za32.vg1x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[ZM:%.*]], i32 3)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z29test_svdot_lane_za32_vg1x2_s8j10svint8x2_tu10__SVInt8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.sdot.lane.za32.vg1x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[ZM:%.*]], i32 3)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_lane_za32_vg1x2_s8(uint32_t slice_base, svint8x2_t zn, svint8_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_lane_za32,,,_s8,_vg1x2)(slice_base, zn, zm, 3);
+}
+
+// CHECK-LABEL: @test_svdot_lane_za32_vg1x4_s8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.sdot.lane.za32.vg1x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]], i32 3)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z29test_svdot_lane_za32_vg1x4_s8j10svint8x4_tu10__SVInt8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.sdot.lane.za32.vg1x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]], i32 3)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_lane_za32_vg1x4_s8(uint32_t slice_base, svint8x4_t zn, svint8_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_lane_za32,,,_s8,_vg1x4)(slice_base, zn, zm, 3);
+}
+
+// CHECK-LABEL: @test_svdot_lane_za64_vg1x2_s16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.sdot.lane.za64.vg1x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]], i32 1)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z30test_svdot_lane_za64_vg1x2_s16j11svint16x2_tu11__SVInt16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.sdot.lane.za64.vg1x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]], i32 1)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_lane_za64_vg1x2_s16(uint32_t slice_base, svint16x2_t zn, svint16_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_lane_za64,,,_s16,_vg1x2)(slice_base, zn, zm, 1);
+}
+
+// CHECK-LABEL: @test_svdot_lane_za64_vg1x4_s16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.sdot.lane.za64.vg1x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]], i32 1)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z30test_svdot_lane_za64_vg1x4_s16j11svint16x4_tu11__SVInt16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.sdot.lane.za64.vg1x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]], i32 1)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svdot_lane_za64_vg1x4_s16(uint32_t slice_base, svint16x4_t zn, svint16_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svdot_lane_za64,,,_s16,_vg1x4)(slice_base, zn, zm, 1);
+}
+
+
+//
+// Multi, multi (unsigned by signed)
+// CHECK-LABEL: @test_svusdot_multi_za32_vg1x2_u8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZM]], i64 16)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.usdot.za32.vg1x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z32test_svusdot_multi_za32_vg1x2_u8j11svuint8x2_t10svint8x2_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZM]], i64 16)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.usdot.za32.vg1x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svusdot_multi_za32_vg1x2_u8(uint32_t slice_base, svuint8x2_t zn, svint8x2_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svusdot_za32,,,_u8,_vg1x2)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svusdot_multi_za32_vg1x4_u8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 16)
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 32)
+// CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 48)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.usdot.za32.vg1x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[TMP4]], <vscale x 16 x i8> [[TMP5]], <vscale x 16 x i8> [[TMP6]], <vscale x 16 x i8> [[TMP7]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z32test_svusdot_multi_za32_vg1x4_u8j11svuint8x4_t10svint8x4_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 48)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.usdot.za32.vg1x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[TMP4]], <vscale x 16 x i8> [[TMP5]], <vscale x 16 x i8> [[TMP6]], <vscale x 16 x i8> [[TMP7]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svusdot_multi_za32_vg1x4_u8(uint32_t slice_base, svuint8x4_t zn, svint8x4_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svusdot_za32,,,_u8,_vg1x4)(slice_base, zn, zm);
+}
+
+
+//
+// Multi, single (unsigned by signed)
+// CHECK-LABEL: @test_svusdot_single_za32_vg1x2_u8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.usdot.single.za32.vg1x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z33test_svusdot_single_za32_vg1x2_u8j11svuint8x2_tu10__SVInt8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.usdot.single.za32.vg1x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svusdot_single_za32_vg1x2_u8(uint32_t slice_base, svuint8x2_t zn, svint8_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svusdot_single_za32,,_u8,,_vg1x2)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svusdot_single_za32_vg1x4_u8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.usdot.single.za32.vg1x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z33test_svusdot_single_za32_vg1x4_u8j11svuint8x4_tu10__SVInt8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.usdot.single.za32.vg1x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svusdot_single_za32_vg1x4_u8(uint32_t slice_base, svuint8x4_t zn, svint8_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svusdot_single_za32,,_u8,,_vg1x4)(slice_base, zn, zm);
+}
+
+//
+// Multi, indexed (unsigned by signed)
+// CHECK-LABEL: @test_svusdot_lane_za32_vg1x2_u8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.usdot.lane.za32.vg1x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[ZM:%.*]], i32 3)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z31test_svusdot_lane_za32_vg1x2_u8j11svuint8x2_tu10__SVInt8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.usdot.lane.za32.vg1x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[ZM:%.*]], i32 3)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svusdot_lane_za32_vg1x2_u8(uint32_t slice_base, svuint8x2_t zn, svint8_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svusdot_lane_za32,,_u8,,_vg1x2)(slice_base, zn, zm, 3);
+}
+
+// CHECK-LABEL: @test_svusdot_lane_za32_vg1x4_u8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.usdot.lane.za32.vg1x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]], i32 3)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z31test_svusdot_lane_za32_vg1x4_u8j11svuint8x4_tu10__SVInt8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.usdot.lane.za32.vg1x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]], i32 3)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svusdot_lane_za32_vg1x4_u8(uint32_t slice_base, svuint8x4_t zn, svint8_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svusdot_lane_za32,,_u8,,_vg1x4)(slice_base, zn, zm, 3);
+}
+
+
+//
+// Multi, single (signed by unsigned)
+// CHECK-LABEL: @test_svsudot_single_za32_vg1x2_s8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.sudot.single.za32.vg1x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z33test_svsudot_single_za32_vg1x2_s8j10svint8x2_tu11__SVUint8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.sudot.single.za32.vg1x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svsudot_single_za32_vg1x2_s8(uint32_t slice_base, svint8x2_t zn, svuint8_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svsudot_single_za32,,_s8,,_vg1x2)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svsudot_single_za32_vg1x4_s8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.sudot.single.za32.vg1x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z33test_svsudot_single_za32_vg1x4_s8j10svint8x4_tu11__SVUint8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.sudot.single.za32.vg1x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svsudot_single_za32_vg1x4_s8(uint32_t slice_base, svint8x4_t zn, svuint8_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svsudot_single_za32,,_s8,,_vg1x4)(slice_base, zn, zm);
+}
+
+//
+// Multi, multi (signed by unsigned)
+// CHECK-LABEL: @test_svsudot_multi_za32_vg1x2_s8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZM]], i64 16)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.usdot.za32.vg1x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z32test_svsudot_multi_za32_vg1x2_s8j10svint8x2_t11svuint8x2_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZM]], i64 16)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.usdot.za32.vg1x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svsudot_multi_za32_vg1x2_s8(uint32_t slice_base, svint8x2_t zn, svuint8x2_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svsudot_za32,,_s8,,_vg1x2)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svsudot_multi_za32_vg1x4_s8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 16)
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 32)
+// CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 48)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.usdot.za32.vg1x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP4]], <vscale x 16 x i8> [[TMP5]], <vscale x 16 x i8> [[TMP6]], <vscale x 16 x i8> [[TMP7]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z32test_svsudot_multi_za32_vg1x4_s8j10svint8x4_t11svuint8x4_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 48)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.usdot.za32.vg1x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP4]], <vscale x 16 x i8> [[TMP5]], <vscale x 16 x i8> [[TMP6]], <vscale x 16 x i8> [[TMP7]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svsudot_multi_za32_vg1x4_s8(uint32_t slice_base, svint8x4_t zn, svuint8x4_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svsudot_za32,,_s8,,_vg1x4)(slice_base, zn, zm);
+}
+
+//
+// Multi, indexed (signed by unsigned)
+// CHECK-LABEL: @test_svsudot_lane_za32_vg1x2_s8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.sudot.lane.za32.vg1x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[ZM:%.*]], i32 3)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z31test_svsudot_lane_za32_vg1x2_s8j10svint8x2_tu11__SVUint8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.sudot.lane.za32.vg1x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[ZM:%.*]], i32 3)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svsudot_lane_za32_vg1x2_s8(uint32_t slice_base, svint8x2_t zn, svuint8_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svsudot_lane_za32,,_s8,,_vg1x2)(slice_base, zn, zm, 3);
+}
+
+// CHECK-LABEL: @test_svsudot_lane_za32_vg1x4_s8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.sudot.lane.za32.vg1x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]], i32 3)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z31test_svsudot_lane_za32_vg1x4_s8j10svint8x4_tu11__SVUint8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.sudot.lane.za32.vg1x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]], i32 3)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svsudot_lane_za32_vg1x4_s8(uint32_t slice_base, svint8x4_t zn, svuint8_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svsudot_lane_za32,,_s8,,_vg1x4)(slice_base, zn, zm, 3);
+}
diff --git a/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_vdot.c b/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_vdot.c
new file mode 100644
index 000000000000..fb313d4cebd7
--- /dev/null
+++ b/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_vdot.c
@@ -0,0 +1,222 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sme-i16i64 -target-feature +sme-f64f64 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sme-i16i64 -target-feature +sme-f64f64 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
+// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sme-i16i64 -target-feature +sme-f64f64 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s
+// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sme-i16i64 -target-feature +sme-f64f64 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sme-i16i64 -S -disable-O0-optnone -Werror -Wall -o /dev/null %s
+#include <arm_sme_draft_spec_subject_to_change.h>
+
+#ifdef SVE_OVERLOADED_FORMS
+// A simple used,unused... macro, long enough to represent any SVE builtin.
+#define SVE_ACLE_FUNC(A1,A2_UNUSED,A3) A1##A3
+#else
+#define SVE_ACLE_FUNC(A1,A2,A3) A1##A2##A3
+#endif
+
+// CHECK-LABEL: @test_svvdot_lane_za32_bf16_vg1x2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZN]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fvdot.lane.za32.vg1x2.nxv8bf16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]], <vscale x 8 x bfloat> [[ZM:%.*]], i32 3)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z32test_svvdot_lane_za32_bf16_vg1x2j14svbfloat16x2_tu14__SVBfloat16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fvdot.lane.za32.vg1x2.nxv8bf16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]], <vscale x 8 x bfloat> [[ZM:%.*]], i32 3)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svvdot_lane_za32_bf16_vg1x2(uint32_t slice_base, svbfloat16x2_t zn, svbfloat16_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svvdot_lane_za32,_bf16,_vg1x2)(slice_base, zn, zm, 3);
+}
+
+// CHECK-LABEL: @test_svvdot_lane_za32_f16_vg1x2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZN]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fvdot.lane.za32.vg1x2.nxv8f16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[ZM:%.*]], i32 3)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z31test_svvdot_lane_za32_f16_vg1x2j13svfloat16x2_tu13__SVFloat16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fvdot.lane.za32.vg1x2.nxv8f16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[ZM:%.*]], i32 3)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svvdot_lane_za32_f16_vg1x2(uint32_t slice_base, svfloat16x2_t zn, svfloat16_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svvdot_lane_za32,_f16,_vg1x2)(slice_base, zn, zm, 3);
+}
+
+// CHECK-LABEL: @test_svvdot_lane_za32_s16_vg1x2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.svdot.lane.za32.vg1x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]], i32 3)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z31test_svvdot_lane_za32_s16_vg1x2j11svint16x2_tu11__SVInt16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.svdot.lane.za32.vg1x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]], i32 3)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svvdot_lane_za32_s16_vg1x2(uint32_t slice_base, svint16x2_t zn, svint16_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svvdot_lane_za32,_s16,_vg1x2)(slice_base, zn, zm, 3);
+}
+
+// CHECK-LABEL: @test_svvdot_lane_za32_u16_vg1x2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.uvdot.lane.za32.vg1x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]], i32 3)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z31test_svvdot_lane_za32_u16_vg1x2j12svuint16x2_tu12__SVUint16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.uvdot.lane.za32.vg1x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]], i32 3)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svvdot_lane_za32_u16_vg1x2(uint32_t slice_base, svuint16x2_t zn, svuint16_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svvdot_lane_za32,_u16,_vg1x2)(slice_base, zn, zm, 3);
+}
+
+// CHECK-LABEL: @test_svvdot_lane_za32_s8_vg1x4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.svdot.lane.za32.vg1x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]], i32 3)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z30test_svvdot_lane_za32_s8_vg1x4j10svint8x4_tu10__SVInt8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.svdot.lane.za32.vg1x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]], i32 3)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svvdot_lane_za32_s8_vg1x4(uint32_t slice_base, svint8x4_t zn, svint8_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svvdot_lane_za32,_s8,_vg1x4)(slice_base, zn, zm, 3);
+}
+
+// CHECK-LABEL: @test_svvdot_lane_za32_u8_vg1x4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.uvdot.lane.za32.vg1x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]], i32 3)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z30test_svvdot_lane_za32_u8_vg1x4j11svuint8x4_tu11__SVUint8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.uvdot.lane.za32.vg1x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]], i32 3)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svvdot_lane_za32_u8_vg1x4(uint32_t slice_base, svuint8x4_t zn, svuint8_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svvdot_lane_za32,_u8,_vg1x4)(slice_base, zn, zm, 3);
+}
+
+// CHECK-LABEL: @test_svvdot_lane_za64_s16_vg1x4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.svdot.lane.za64.vg1x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]], i32 1)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z31test_svvdot_lane_za64_s16_vg1x4j11svint16x4_tu11__SVInt16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.svdot.lane.za64.vg1x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]], i32 1)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svvdot_lane_za64_s16_vg1x4(uint32_t slice_base, svint16x4_t zn, svint16_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svvdot_lane_za64,_s16,_vg1x4)(slice_base, zn, zm, 1);
+}
+
+// CHECK-LABEL: @test_svvdot_lane_za64_u16_vg1x4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.uvdot.lane.za64.vg1x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]], i32 1)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z31test_svvdot_lane_za64_u16_vg1x4j12svuint16x4_tu12__SVUint16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.uvdot.lane.za64.vg1x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]], i32 1)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svvdot_lane_za64_u16_vg1x4(uint32_t slice_base, svuint16x4_t zn, svuint16_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svvdot_lane_za64,_u16,_vg1x4)(slice_base, zn, zm, 1);
+}
+
+
+// CHECK-LABEL: @test_svsuvdot_lane_za32_s8_vg1x4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.suvdot.lane.za32.vg1x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]], i32 3)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z32test_svsuvdot_lane_za32_s8_vg1x4j10svint8x4_tu10__SVInt8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.suvdot.lane.za32.vg1x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]], i32 3)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svsuvdot_lane_za32_s8_vg1x4(uint32_t slice_base, svint8x4_t zn, svint8_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svsuvdot_lane_za32,_s8,_vg1x4)(slice_base, zn, zm, 3);
+}
+
+
+// CHECK-LABEL: @test_svusvdot_lane_za32_u8_vg1x4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.usvdot.lane.za32.vg1x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]], i32 3)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z32test_svusvdot_lane_za32_u8_vg1x4j11svuint8x4_tu11__SVUint8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.usvdot.lane.za32.vg1x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]], i32 3)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svusvdot_lane_za32_u8_vg1x4(uint32_t slice_base, svuint8x4_t zn, svuint8_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svusvdot_lane_za32,_u8,_vg1x4)(slice_base, zn, zm, 3);
+}
+
diff --git a/clang/test/Sema/aarch64-sme2-intrinsics/acle_sme2_imm.cpp b/clang/test/Sema/aarch64-sme2-intrinsics/acle_sme2_imm.cpp
index d07d83a53e46..6a6370bf99b1 100644
--- a/clang/test/Sema/aarch64-sme2-intrinsics/acle_sme2_imm.cpp
+++ b/clang/test/Sema/aarch64-sme2-intrinsics/acle_sme2_imm.cpp
@@ -285,3 +285,68 @@ void test_multiply_add_sub_long(uint32_t base, svint8_t s8, svuint8_t u8,
   svusmla_lane_za32_u8_vg4x2(base, u8x2, s8, 16); // expected-error {{argument value 16 is outside the valid range [0, 15]}}
   svusmla_lane_za32_u8_vg4x4(base, u8x4, s8, 16); // expected-error {{argument value 16 is outside the valid range [0, 15]}}
 }
+
+void test_vertical_dot_product(uint32_t base, svint16x2_t s16x2, svuint16x2_t u16x2,
+                               svint8x4_t s8x4, svuint8x4_t u8x4,
+                               svint16x4_t s16x4, svuint16x4_t u16x4,
+                               svfloat16x2_t f16x2, svbfloat16x2_t bf16x2,
+                               svint16_t s16, svuint16_t u16,
+                               svint8_t s8, svuint8_t u8,
+                               svfloat16_t f16, svbfloat16_t b16) __arm_streaming __arm_shared_za {
+  // Test lane indices.
+  svvdot_lane_za32_s16_vg1x2(base, s16x2, s16, 4);   // expected-error {{argument value 4 is outside the valid range [0, 3]}}
+  svvdot_lane_za32_u16_vg1x2(base, u16x2, u16, 4);   // expected-error {{argument value 4 is outside the valid range [0, 3]}}
+  svvdot_lane_za32_s8_vg1x4(base, s8x4, s8, 4);      // expected-error {{argument value 4 is outside the valid range [0, 3]}}
+  svvdot_lane_za32_u8_vg1x4(base, u8x4, u8, 4);      // expected-error {{argument value 4 is outside the valid range [0, 3]}}
+  svvdot_lane_za64_s16_vg1x4(base, s16x4, s16, 2);   // expected-error {{argument value 2 is outside the valid range [0, 1]}}
+  svvdot_lane_za64_u16_vg1x4(base, u16x4, u16, 2);   // expected-error {{argument value 2 is outside the valid range [0, 1]}}
+  svvdot_lane_za32_f16_vg1x2(base, f16x2, f16, 4);   // expected-error {{argument value 4 is outside the valid range [0, 3]}}
+  svvdot_lane_za32_bf16_vg1x2(base, bf16x2, b16, 4); // expected-error {{argument value 4 is outside the valid range [0, 3]}}
+  svsuvdot_lane_za32_s8_vg1x4(base, s8x4, s8, 4);    // expected-error {{argument value 4 is outside the valid range [0, 3]}}
+  svusvdot_lane_za32_u8_vg1x4(base, u8x4, u8, 4);    // expected-error {{argument value 4 is outside the valid range [0, 3]}}
+}
+
+void test_fdot_za32_bad_lane(uint32_t slice_base, svfloat16_t z_f16,
+                             svfloat16x2_t z_f16x2, svfloat16x4_t z_f16x4,
+                             svbfloat16_t z_bf16, svbfloat16x2_t z_bf16x2,
+                             svbfloat16x4_t z_bf16x4) __arm_streaming __arm_shared_za {
+  // 16-bit float
+  svdot_lane_za32_f16_vg1x2(slice_base, z_f16x2, z_f16, 4); // expected-error {{argument value 4 is outside the valid range [0, 3]}}
+  svdot_lane_za32_f16_vg1x4(slice_base, z_f16x4, z_f16, 4); // expected-error {{argument value 4 is outside the valid range [0, 3]}}
+
+  // 16-bit binary float
+  svdot_lane_za32_bf16_vg1x2(slice_base, z_bf16x2, z_bf16, 4); // expected-error {{argument value 4 is outside the valid range [0, 3]}}
+  svdot_lane_za32_bf16_vg1x4(slice_base, z_bf16x4, z_bf16, 4); // expected-error {{argument value 4 is outside the valid range [0, 3]}}
+}
+
+void test_svdot_multi_za32_bad_lane(uint32_t slice_base, svuint16_t z_u16,
+                                    svuint16x2_t z_u16x2, svuint16x4_t z_u16x4,
+                                    svint16_t z_s16, svint16x2_t z_s16x2,
+                                    svint16x4_t z_s16x4, svuint8_t z_u8,
+                                    svuint8x2_t z_u8x2, svuint8x4_t z_u8x4,
+                                    svint8_t z_s8, svint8x2_t z_s8x2,
+                                    svint8x4_t z_s8x4) __arm_streaming __arm_shared_za {
+  // Multi, indexed (unsigned)
+  svdot_lane_za32_u16_vg1x2(slice_base, z_u16x2, z_u16, 4); // expected-error {{argument value 4 is outside the valid range [0, 3]}}
+  svdot_lane_za32_u16_vg1x4(slice_base, z_u16x4, z_u16, 4); // expected-error {{argument value 4 is outside the valid range [0, 3]}}
+  svdot_lane_za32_u8_vg1x2(slice_base, z_u8x2, z_u8, 4);    // expected-error {{argument value 4 is outside the valid range [0, 3]}}
+  svdot_lane_za32_u8_vg1x4(slice_base, z_u8x4, z_u8, 4);    // expected-error {{argument value 4 is outside the valid range [0, 3]}}
+  svdot_lane_za64_u16_vg1x2(slice_base, z_u16x2, z_u16, 2); // expected-error {{argument value 2 is outside the valid range [0, 1]}}
+  svdot_lane_za64_u16_vg1x4(slice_base, z_u16x4, z_u16, 2); // expected-error {{argument value 2 is outside the valid range [0, 1]}}
+
+  // Multi, indexed (signed)
+  svdot_lane_za32_s16_vg1x2(slice_base, z_s16x2, z_s16, 4); // expected-error {{argument value 4 is outside the valid range [0, 3]}}
+  svdot_lane_za32_s16_vg1x4(slice_base, z_s16x4, z_s16, 4); // expected-error {{argument value 4 is outside the valid range [0, 3]}}
+  svdot_lane_za32_s8_vg1x2(slice_base, z_s8x2, z_s8, 4);    // expected-error {{argument value 4 is outside the valid range [0, 3]}}
+  svdot_lane_za32_s8_vg1x4(slice_base, z_s8x4, z_s8, 4);    // expected-error {{argument value 4 is outside the valid range [0, 3]}}
+  svdot_lane_za64_s16_vg1x2(slice_base, z_s16x2, z_s16, 2); // expected-error {{argument value 2 is outside the valid range [0, 1]}}
+  svdot_lane_za64_s16_vg1x4(slice_base, z_s16x4, z_s16, 2); // expected-error {{argument value 2 is outside the valid range [0, 1]}}
+
+  // Multi, indexed (unsigned by signed)
+  svusdot_lane_za32_u8_vg1x2(slice_base, z_u8x2, z_s8, 4); // expected-error {{argument value 4 is outside the valid range [0, 3]}}
+  svusdot_lane_za32_u8_vg1x4(slice_base, z_u8x4, z_s8, 4); // expected-error {{argument value 4 is outside the valid range [0, 3]}}
+
+  // Multi, indexed (unsigned by signed)
+  svsudot_lane_za32_s8_vg1x2(slice_base, z_s8x2, z_u8, 4); // expected-error {{argument value 4 is outside the valid range [0, 3]}}
+  svsudot_lane_za32_s8_vg1x4(slice_base, z_s8x4, z_u8, 4); // expected-error {{argument value 4 is outside the valid range [0, 3]}}
+}