diff options
| author | Craig Topper <craig.topper@intel.com> | 2017-11-09 04:10:46 +0000 |
|---|---|---|
| committer | Craig Topper <craig.topper@intel.com> | 2017-11-09 04:10:46 +0000 |
| commit | 66cb228856bc2118e00a41f10e0273e08c7f4cab (patch) | |
| tree | 4e05fe72fc6c7fe54ddf63262e7d31bbc797143f /lib/Headers/fmaintrin.h | |
| parent | 559824626977ee1736558af190d9afa6010c1768 (diff) | |
[X86] Rename the VEX scalar fma builtins to end with a '3' to match gcc
I think we need to use different builtins for the FMA4 instructions since those instructions zero the upper bits and FMA3 instructions pass the bits through.
So this moves the existing builtins to be the FMA3 versions. New versions will be added for FMA4.
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@317766 91177308-0d34-0410-b5e6-96231b3b80d8
Diffstat (limited to 'lib/Headers/fmaintrin.h')
| -rw-r--r-- | lib/Headers/fmaintrin.h | 16 |
1 files changed, 8 insertions, 8 deletions
diff --git a/lib/Headers/fmaintrin.h b/lib/Headers/fmaintrin.h index 0e2ef0b171..235931ccb6 100644 --- a/lib/Headers/fmaintrin.h +++ b/lib/Headers/fmaintrin.h @@ -46,13 +46,13 @@ _mm_fmadd_pd(__m128d __A, __m128d __B, __m128d __C) static __inline__ __m128 __DEFAULT_FN_ATTRS _mm_fmadd_ss(__m128 __A, __m128 __B, __m128 __C) { - return (__m128)__builtin_ia32_vfmaddss((__v4sf)__A, (__v4sf)__B, (__v4sf)__C); + return (__m128)__builtin_ia32_vfmaddss3((__v4sf)__A, (__v4sf)__B, (__v4sf)__C); } static __inline__ __m128d __DEFAULT_FN_ATTRS _mm_fmadd_sd(__m128d __A, __m128d __B, __m128d __C) { - return (__m128d)__builtin_ia32_vfmaddsd((__v2df)__A, (__v2df)__B, (__v2df)__C); + return (__m128d)__builtin_ia32_vfmaddsd3((__v2df)__A, (__v2df)__B, (__v2df)__C); } static __inline__ __m128 __DEFAULT_FN_ATTRS @@ -70,13 +70,13 @@ _mm_fmsub_pd(__m128d __A, __m128d __B, __m128d __C) static __inline__ __m128 __DEFAULT_FN_ATTRS _mm_fmsub_ss(__m128 __A, __m128 __B, __m128 __C) { - return (__m128)__builtin_ia32_vfmsubss((__v4sf)__A, (__v4sf)__B, (__v4sf)__C); + return (__m128)__builtin_ia32_vfmsubss3((__v4sf)__A, (__v4sf)__B, (__v4sf)__C); } static __inline__ __m128d __DEFAULT_FN_ATTRS _mm_fmsub_sd(__m128d __A, __m128d __B, __m128d __C) { - return (__m128d)__builtin_ia32_vfmsubsd((__v2df)__A, (__v2df)__B, (__v2df)__C); + return (__m128d)__builtin_ia32_vfmsubsd3((__v2df)__A, (__v2df)__B, (__v2df)__C); } static __inline__ __m128 __DEFAULT_FN_ATTRS @@ -94,13 +94,13 @@ _mm_fnmadd_pd(__m128d __A, __m128d __B, __m128d __C) static __inline__ __m128 __DEFAULT_FN_ATTRS _mm_fnmadd_ss(__m128 __A, __m128 __B, __m128 __C) { - return (__m128)__builtin_ia32_vfnmaddss((__v4sf)__A, (__v4sf)__B, (__v4sf)__C); + return (__m128)__builtin_ia32_vfnmaddss3((__v4sf)__A, (__v4sf)__B, (__v4sf)__C); } static __inline__ __m128d __DEFAULT_FN_ATTRS _mm_fnmadd_sd(__m128d __A, __m128d __B, __m128d __C) { - return (__m128d)__builtin_ia32_vfnmaddsd((__v2df)__A, (__v2df)__B, (__v2df)__C); + return (__m128d)__builtin_ia32_vfnmaddsd3((__v2df)__A, (__v2df)__B, (__v2df)__C); } static __inline__ __m128 __DEFAULT_FN_ATTRS @@ -118,13 +118,13 @@ _mm_fnmsub_pd(__m128d __A, __m128d __B, __m128d __C) static __inline__ __m128 __DEFAULT_FN_ATTRS _mm_fnmsub_ss(__m128 __A, __m128 __B, __m128 __C) { - return (__m128)__builtin_ia32_vfnmsubss((__v4sf)__A, (__v4sf)__B, (__v4sf)__C); + return (__m128)__builtin_ia32_vfnmsubss3((__v4sf)__A, (__v4sf)__B, (__v4sf)__C); } static __inline__ __m128d __DEFAULT_FN_ATTRS _mm_fnmsub_sd(__m128d __A, __m128d __B, __m128d __C) { - return (__m128d)__builtin_ia32_vfnmsubsd((__v2df)__A, (__v2df)__B, (__v2df)__C); + return (__m128d)__builtin_ia32_vfnmsubsd3((__v2df)__A, (__v2df)__B, (__v2df)__C); } static __inline__ __m128 __DEFAULT_FN_ATTRS |