diff options
| author | Craig Topper <craig.topper@intel.com> | 2018-09-26 17:01:44 +0000 |
|---|---|---|
| committer | Craig Topper <craig.topper@intel.com> | 2018-09-26 17:01:44 +0000 |
| commit | 26346fbcb9bb9fe7a6374e84eb994e8743e65327 (patch) | |
| tree | 2a9067e82d5b0b301240a1a8ec97403b9ccc2f3e /include/clang/Basic/BuiltinsX86.def | |
| parent | f10f8c29df7bf378dcde4d3341709145d82c0156 (diff) | |
[X86] For lzcnt/tzcnt intrinsics use cttz/ctlz intrinsics with zero_undef flag set to false.
Previously we used a select and the zero_undef=true intrinsic. In -O2 this pattern will get optimized to zero_undef=false. But in -O0 this optimization won't happen. This results in a compare and cmov being wrapped around a tzcnt/lzcnt instruction.
By using the zero_undef=false intrinsic directly without the select, we can improve the -O0 codegen to just an lzcnt/tzcnt instruction.
Differential Revision: https://reviews.llvm.org/D52392
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@343126 91177308-0d34-0410-b5e6-96231b3b80d8
Diffstat (limited to 'include/clang/Basic/BuiltinsX86.def')
| -rw-r--r-- | include/clang/Basic/BuiltinsX86.def | 6 |
1 files changed, 6 insertions, 0 deletions
diff --git a/include/clang/Basic/BuiltinsX86.def b/include/clang/Basic/BuiltinsX86.def index db81f58e08..45d366341c 100644 --- a/include/clang/Basic/BuiltinsX86.def +++ b/include/clang/Basic/BuiltinsX86.def @@ -727,8 +727,14 @@ TARGET_BUILTIN(__builtin_ia32_subborrow_u32, "UcUcUiUiUi*", "n", "") TARGET_BUILTIN(__builtin_ia32_rdseed16_step, "UiUs*", "n", "rdseed") TARGET_BUILTIN(__builtin_ia32_rdseed32_step, "UiUi*", "n", "rdseed") +// LZCNT +TARGET_BUILTIN(__builtin_ia32_lzcnt_u16, "UsUs", "nc", "lzcnt") +TARGET_BUILTIN(__builtin_ia32_lzcnt_u32, "UiUi", "nc", "lzcnt") + // BMI TARGET_BUILTIN(__builtin_ia32_bextr_u32, "UiUiUi", "nc", "bmi") +TARGET_BUILTIN(__builtin_ia32_tzcnt_u16, "UsUs", "nc", "") +TARGET_BUILTIN(__builtin_ia32_tzcnt_u32, "UiUi", "nc", "") // BMI2 TARGET_BUILTIN(__builtin_ia32_bzhi_si, "UiUiUi", "nc", "bmi2") |