diff options
| author | Craig Topper <craig.topper@intel.com> | 2018-05-14 17:50:40 +0000 |
|---|---|---|
| committer | Craig Topper <craig.topper@intel.com> | 2018-05-14 17:50:40 +0000 |
| commit | ec47066bad48dbb7478ec8d2f35392bd317068e7 (patch) | |
| tree | a8a88fc707a98051b84479b4ab4602f343d79d2f /lib/Headers/avx512vlbwintrin.h | |
| parent | 49c08905862b821364860640b9b9f57d8e4f2cc3 (diff) | |
[X86] Use __builtin_convertvector to replace some of the avx512 truncate builtins.
As long as the destination type is a 256 or 128 bit vector with the same number of elements we can use __builtin_convertvector to directly generate trunc IR instruction which will be handled natively by the backend.
Differential Revision: https://reviews.llvm.org/D46742
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@332266 91177308-0d34-0410-b5e6-96231b3b80d8
Diffstat (limited to 'lib/Headers/avx512vlbwintrin.h')
| -rw-r--r-- | lib/Headers/avx512vlbwintrin.h | 16 |
1 files changed, 7 insertions, 9 deletions
diff --git a/lib/Headers/avx512vlbwintrin.h b/lib/Headers/avx512vlbwintrin.h index e940e2b685..6c5131ccba 100644 --- a/lib/Headers/avx512vlbwintrin.h +++ b/lib/Headers/avx512vlbwintrin.h @@ -1551,23 +1551,21 @@ _mm_mask_cvtusepi16_storeu_epi8 (void * __P, __mmask8 __M, __m128i __A) static __inline__ __m128i __DEFAULT_FN_ATTRS _mm256_cvtepi16_epi8 (__m256i __A) { - return (__m128i) __builtin_ia32_pmovwb256_mask ((__v16hi) __A, - (__v16qi) _mm_setzero_si128(), - (__mmask16) -1); + return (__m128i)__builtin_convertvector((__v16hi) __A, __v16qi); } static __inline__ __m128i __DEFAULT_FN_ATTRS _mm256_mask_cvtepi16_epi8 (__m128i __O, __mmask16 __M, __m256i __A) { - return (__m128i) __builtin_ia32_pmovwb256_mask ((__v16hi) __A, - (__v16qi) __O, - __M); + return (__m128i)__builtin_ia32_selectb_128((__mmask8)__M, + (__v16qi)_mm256_cvtepi16_epi8(__A), + (__v16qi)__O); } static __inline__ __m128i __DEFAULT_FN_ATTRS _mm256_maskz_cvtepi16_epi8 (__mmask16 __M, __m256i __A) { - return (__m128i) __builtin_ia32_pmovwb256_mask ((__v16hi) __A, - (__v16qi) _mm_setzero_si128(), - __M); + return (__m128i)__builtin_ia32_selectb_128((__mmask8)__M, + (__v16qi)_mm256_cvtepi16_epi8(__A), + (__v16qi)_mm_setzero_si128()); } static __inline__ void __DEFAULT_FN_ATTRS |