diff options
| author | Craig Topper <craig.topper@intel.com> | 2018-06-08 03:24:47 +0000 |
|---|---|---|
| committer | Craig Topper <craig.topper@intel.com> | 2018-06-08 03:24:47 +0000 |
| commit | c38646b3325bacec867358444a9a0fa9e5dca8ba (patch) | |
| tree | 590b6338e651c38227ab84216c150451a69701d7 /lib/Headers/avx512vldqintrin.h | |
| parent | 3abc7713293f9e2b62eea6f9c350ca8c68bd54f6 (diff) | |
[X86] Add subvector insert and extract builtins to enable target feature checking and immediate range checking.
Test changes are due to differences in how we generate undef elements now. We also changed the types used for extractf128_si256/insertf128_si256 to match the signature of the builtin that previously existed which this patch resurrects. This also matches gcc.
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@334261 91177308-0d34-0410-b5e6-96231b3b80d8
Diffstat (limited to 'lib/Headers/avx512vldqintrin.h')
| -rw-r--r-- | lib/Headers/avx512vldqintrin.h | 26 |
1 files changed, 6 insertions, 20 deletions
diff --git a/lib/Headers/avx512vldqintrin.h b/lib/Headers/avx512vldqintrin.h index 60571adb5e..8d6ff3ec8d 100644 --- a/lib/Headers/avx512vldqintrin.h +++ b/lib/Headers/avx512vldqintrin.h @@ -1083,10 +1083,7 @@ _mm256_maskz_broadcast_i64x2 (__mmask8 __M, __m128i __A) } #define _mm256_extractf64x2_pd(A, imm) \ - (__m128d)__builtin_shufflevector((__v4df)(__m256d)(A), \ - (__v4df)_mm256_undefined_pd(), \ - ((imm) & 1) ? 2 : 0, \ - ((imm) & 1) ? 3 : 1) + (__m128d)__builtin_ia32_extractf64x2_256((__v4df)(__m256d)(A), (int)(imm)) #define _mm256_mask_extractf64x2_pd(W, U, A, imm) \ (__m128d)__builtin_ia32_selectpd_128((__mmask8)(U), \ @@ -1099,10 +1096,7 @@ _mm256_maskz_broadcast_i64x2 (__mmask8 __M, __m128i __A) (__v2df)_mm_setzero_pd()) #define _mm256_extracti64x2_epi64(A, imm) \ - (__m128i)__builtin_shufflevector((__v4di)(__m256i)(A), \ - (__v4di)_mm256_undefined_si256(), \ - ((imm) & 1) ? 2 : 0, \ - ((imm) & 1) ? 3 : 1) + (__m128i)__builtin_ia32_extracti64x2_256((__v4di)(__m256i)(A), (int)(imm)) #define _mm256_mask_extracti64x2_epi64(W, U, A, imm) \ (__m128i)__builtin_ia32_selectq_128((__mmask8)(U), \ @@ -1115,12 +1109,8 @@ _mm256_maskz_broadcast_i64x2 (__mmask8 __M, __m128i __A) (__v2di)_mm_setzero_si128()) #define _mm256_insertf64x2(A, B, imm) \ - (__m256d)__builtin_shufflevector((__v4df)(__m256d)(A), \ - (__v4df)_mm256_castpd128_pd256((__m128d)(B)), \ - ((imm) & 0x1) ? 0 : 4, \ - ((imm) & 0x1) ? 1 : 5, \ - ((imm) & 0x1) ? 4 : 2, \ - ((imm) & 0x1) ? 5 : 3) + (__m256d)__builtin_ia32_insertf64x2_256((__v4df)(__m256d)(A), \ + (__v2df)(__m128d)(B), (int)(imm)) #define _mm256_mask_insertf64x2(W, U, A, B, imm) \ (__m256d)__builtin_ia32_selectpd_256((__mmask8)(U), \ @@ -1133,12 +1123,8 @@ _mm256_maskz_broadcast_i64x2 (__mmask8 __M, __m128i __A) (__v4df)_mm256_setzero_pd()) #define _mm256_inserti64x2(A, B, imm) \ - (__m256i)__builtin_shufflevector((__v4di)(__m256i)(A), \ - (__v4di)_mm256_castsi128_si256((__m128i)(B)), \ - ((imm) & 0x1) ? 0 : 4, \ - ((imm) & 0x1) ? 1 : 5, \ - ((imm) & 0x1) ? 4 : 2, \ - ((imm) & 0x1) ? 5 : 3) + (__m256i)__builtin_ia32_inserti64x2_256((__v4di)(__m256i)(A), \ + (__v2di)(__m128i)(B), (int)(imm)) #define _mm256_mask_inserti64x2(W, U, A, B, imm) \ (__m256i)__builtin_ia32_selectq_256((__mmask8)(U), \ |