diff options
Diffstat (limited to 'contrib/llvm/tools/clang/lib/Headers/avx2intrin.h')
| -rw-r--r-- | contrib/llvm/tools/clang/lib/Headers/avx2intrin.h | 451 |
1 files changed, 205 insertions, 246 deletions
diff --git a/contrib/llvm/tools/clang/lib/Headers/avx2intrin.h b/contrib/llvm/tools/clang/lib/Headers/avx2intrin.h index d8b6b0a..f786572 100644 --- a/contrib/llvm/tools/clang/lib/Headers/avx2intrin.h +++ b/contrib/llvm/tools/clang/lib/Headers/avx2intrin.h @@ -29,7 +29,7 @@ #define __AVX2INTRIN_H /* Define the default attributes for the functions in this file. */ -#define __DEFAULT_FN_ATTRS __attribute__((__always_inline__, __nodebug__)) +#define __DEFAULT_FN_ATTRS __attribute__((__always_inline__, __nodebug__, __target__("avx2"))) /* SSE4 Multiple Packed Sums of Absolute Difference. */ #define _mm256_mpsadbw_epu8(X, Y, M) __builtin_ia32_mpsadbw256((X), (Y), (M)) @@ -124,10 +124,9 @@ _mm256_adds_epu16(__m256i __a, __m256i __b) return (__m256i)__builtin_ia32_paddusw256((__v16hi)__a, (__v16hi)__b); } -#define _mm256_alignr_epi8(a, b, n) __extension__ ({ \ - __m256i __a = (a); \ - __m256i __b = (b); \ - (__m256i)__builtin_ia32_palignr256((__v32qi)__a, (__v32qi)__b, (n)); }) +#define _mm256_alignr_epi8(a, b, n) __extension__ ({ \ + (__m256i)__builtin_ia32_palignr256((__v32qi)(__m256i)(a), \ + (__v32qi)(__m256i)(b), (n)); }) static __inline__ __m256i __DEFAULT_FN_ATTRS _mm256_and_si256(__m256i __a, __m256i __b) @@ -160,20 +159,19 @@ _mm256_blendv_epi8(__m256i __V1, __m256i __V2, __m256i __M) (__v32qi)__M); } -#define _mm256_blend_epi16(V1, V2, M) __extension__ ({ \ - __m256i __V1 = (V1); \ - __m256i __V2 = (V2); \ - (__m256i)__builtin_shufflevector((__v16hi)__V1, (__v16hi)__V2, \ - (((M) & 0x01) ? 16 : 0), \ - (((M) & 0x02) ? 17 : 1), \ - (((M) & 0x04) ? 18 : 2), \ - (((M) & 0x08) ? 19 : 3), \ - (((M) & 0x10) ? 20 : 4), \ - (((M) & 0x20) ? 21 : 5), \ - (((M) & 0x40) ? 22 : 6), \ - (((M) & 0x80) ? 23 : 7), \ - (((M) & 0x01) ? 24 : 8), \ - (((M) & 0x02) ? 25 : 9), \ +#define _mm256_blend_epi16(V1, V2, M) __extension__ ({ \ + (__m256i)__builtin_shufflevector((__v16hi)(__m256i)(V1), \ + (__v16hi)(__m256i)(V2), \ + (((M) & 0x01) ? 16 : 0), \ + (((M) & 0x02) ? 17 : 1), \ + (((M) & 0x04) ? 18 : 2), \ + (((M) & 0x08) ? 19 : 3), \ + (((M) & 0x10) ? 20 : 4), \ + (((M) & 0x20) ? 21 : 5), \ + (((M) & 0x40) ? 22 : 6), \ + (((M) & 0x80) ? 23 : 7), \ + (((M) & 0x01) ? 24 : 8), \ + (((M) & 0x02) ? 25 : 9), \ (((M) & 0x04) ? 26 : 10), \ (((M) & 0x08) ? 27 : 11), \ (((M) & 0x10) ? 28 : 12), \ @@ -208,7 +206,9 @@ _mm256_cmpeq_epi64(__m256i __a, __m256i __b) static __inline__ __m256i __DEFAULT_FN_ATTRS _mm256_cmpgt_epi8(__m256i __a, __m256i __b) { - return (__m256i)((__v32qi)__a > (__v32qi)__b); + /* This function always performs a signed comparison, but __v32qi is a char + which may be signed or unsigned, so use __v32qs. */ + return (__m256i)((__v32qs)__a > (__v32qs)__b); } static __inline__ __m256i __DEFAULT_FN_ATTRS @@ -488,8 +488,8 @@ _mm256_shuffle_epi8(__m256i __a, __m256i __b) } #define _mm256_shuffle_epi32(a, imm) __extension__ ({ \ - __m256i __a = (a); \ - (__m256i)__builtin_shufflevector((__v8si)__a, (__v8si)_mm256_set1_epi32(0), \ + (__m256i)__builtin_shufflevector((__v8si)(__m256i)(a), \ + (__v8si)_mm256_setzero_si256(), \ (imm) & 0x3, ((imm) & 0xc) >> 2, \ ((imm) & 0x30) >> 4, ((imm) & 0xc0) >> 6, \ 4 + (((imm) & 0x03) >> 0), \ @@ -498,8 +498,8 @@ _mm256_shuffle_epi8(__m256i __a, __m256i __b) 4 + (((imm) & 0xc0) >> 6)); }) #define _mm256_shufflehi_epi16(a, imm) __extension__ ({ \ - __m256i __a = (a); \ - (__m256i)__builtin_shufflevector((__v16hi)__a, (__v16hi)_mm256_set1_epi16(0), \ + (__m256i)__builtin_shufflevector((__v16hi)(__m256i)(a), \ + (__v16hi)_mm256_setzero_si256(), \ 0, 1, 2, 3, \ 4 + (((imm) & 0x03) >> 0), \ 4 + (((imm) & 0x0c) >> 2), \ @@ -512,8 +512,8 @@ _mm256_shuffle_epi8(__m256i __a, __m256i __b) 12 + (((imm) & 0xc0) >> 6)); }) #define _mm256_shufflelo_epi16(a, imm) __extension__ ({ \ - __m256i __a = (a); \ - (__m256i)__builtin_shufflevector((__v16hi)__a, (__v16hi)_mm256_set1_epi16(0), \ + (__m256i)__builtin_shufflevector((__v16hi)(__m256i)(a), \ + (__v16hi)_mm256_setzero_si256(), \ (imm) & 0x3,((imm) & 0xc) >> 2, \ ((imm) & 0x30) >> 4, ((imm) & 0xc0) >> 6, \ 4, 5, 6, 7, \ @@ -542,8 +542,7 @@ _mm256_sign_epi32(__m256i __a, __m256i __b) } #define _mm256_slli_si256(a, count) __extension__ ({ \ - __m256i __a = (a); \ - (__m256i)__builtin_ia32_pslldqi256(__a, (count)*8); }) + (__m256i)__builtin_ia32_pslldqi256((__m256i)(a), (count)*8); }) #define _mm256_bslli_epi128(a, count) _mm256_slli_si256((a), (count)) @@ -608,8 +607,7 @@ _mm256_sra_epi32(__m256i __a, __m128i __count) } #define _mm256_srli_si256(a, count) __extension__ ({ \ - __m256i __a = (a); \ - (__m256i)__builtin_ia32_psrldqi256(__a, (count)*8); }) + (__m256i)__builtin_ia32_psrldqi256((__m256i)(a), (count)*8); }) #define _mm256_bsrli_epi128(a, count) _mm256_srli_si256((a), (count)) @@ -752,15 +750,15 @@ _mm256_xor_si256(__m256i __a, __m256i __b) } static __inline__ __m256i __DEFAULT_FN_ATTRS -_mm256_stream_load_si256(__m256i *__V) +_mm256_stream_load_si256(__m256i const *__V) { - return (__m256i)__builtin_ia32_movntdqa256((__v4di *)__V); + return (__m256i)__builtin_ia32_movntdqa256((const __v4di *)__V); } static __inline__ __m128 __DEFAULT_FN_ATTRS _mm_broadcastss_ps(__m128 __X) { - return (__m128)__builtin_ia32_vbroadcastss_ps((__v4sf)__X); + return (__m128)__builtin_shufflevector((__v4sf)__X, (__v4sf)__X, 0, 0, 0, 0); } static __inline__ __m128d __DEFAULT_FN_ATTRS @@ -772,13 +770,13 @@ _mm_broadcastsd_pd(__m128d __a) static __inline__ __m256 __DEFAULT_FN_ATTRS _mm256_broadcastss_ps(__m128 __X) { - return (__m256)__builtin_ia32_vbroadcastss_ps256((__v4sf)__X); + return (__m256)__builtin_shufflevector((__v4sf)__X, (__v4sf)__X, 0, 0, 0, 0, 0, 0, 0, 0); } static __inline__ __m256d __DEFAULT_FN_ATTRS _mm256_broadcastsd_pd(__m128d __X) { - return (__m256d)__builtin_ia32_vbroadcastsd_pd256((__v2df)__X); + return (__m256d)__builtin_shufflevector((__v2df)__X, (__v2df)__X, 0, 0, 0, 0); } static __inline__ __m256i __DEFAULT_FN_ATTRS @@ -788,18 +786,16 @@ _mm256_broadcastsi128_si256(__m128i __X) } #define _mm_blend_epi32(V1, V2, M) __extension__ ({ \ - __m128i __V1 = (V1); \ - __m128i __V2 = (V2); \ - (__m128i)__builtin_shufflevector((__v4si)__V1, (__v4si)__V2, \ + (__m128i)__builtin_shufflevector((__v4si)(__m128i)(V1), \ + (__v4si)(__m128i)(V2), \ (((M) & 0x01) ? 4 : 0), \ (((M) & 0x02) ? 5 : 1), \ (((M) & 0x04) ? 6 : 2), \ (((M) & 0x08) ? 7 : 3)); }) #define _mm256_blend_epi32(V1, V2, M) __extension__ ({ \ - __m256i __V1 = (V1); \ - __m256i __V2 = (V2); \ - (__m256i)__builtin_shufflevector((__v8si)__V1, (__v8si)__V2, \ + (__m256i)__builtin_shufflevector((__v8si)(__m256i)(V1), \ + (__v8si)(__m256i)(V2), \ (((M) & 0x01) ? 8 : 0), \ (((M) & 0x02) ? 9 : 1), \ (((M) & 0x04) ? 10 : 2), \ @@ -812,50 +808,50 @@ _mm256_broadcastsi128_si256(__m128i __X) static __inline__ __m256i __DEFAULT_FN_ATTRS _mm256_broadcastb_epi8(__m128i __X) { - return (__m256i)__builtin_ia32_pbroadcastb256((__v16qi)__X); + return (__m256i)__builtin_shufflevector((__v16qi)__X, (__v16qi)__X, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0); } static __inline__ __m256i __DEFAULT_FN_ATTRS _mm256_broadcastw_epi16(__m128i __X) { - return (__m256i)__builtin_ia32_pbroadcastw256((__v8hi)__X); + return (__m256i)__builtin_shufflevector((__v8hi)__X, (__v8hi)__X, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0); } static __inline__ __m256i __DEFAULT_FN_ATTRS _mm256_broadcastd_epi32(__m128i __X) { - return (__m256i)__builtin_ia32_pbroadcastd256((__v4si)__X); + return (__m256i)__builtin_shufflevector((__v4si)__X, (__v4si)__X, 0, 0, 0, 0, 0, 0, 0, 0); } static __inline__ __m256i __DEFAULT_FN_ATTRS _mm256_broadcastq_epi64(__m128i __X) { - return (__m256i)__builtin_ia32_pbroadcastq256(__X); + return (__m256i)__builtin_shufflevector(__X, __X, 0, 0, 0, 0); } static __inline__ __m128i __DEFAULT_FN_ATTRS _mm_broadcastb_epi8(__m128i __X) { - return (__m128i)__builtin_ia32_pbroadcastb128((__v16qi)__X); + return (__m128i)__builtin_shufflevector((__v16qi)__X, (__v16qi)__X, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0); } static __inline__ __m128i __DEFAULT_FN_ATTRS _mm_broadcastw_epi16(__m128i __X) { - return (__m128i)__builtin_ia32_pbroadcastw128((__v8hi)__X); + return (__m128i)__builtin_shufflevector((__v8hi)__X, (__v8hi)__X, 0, 0, 0, 0, 0, 0, 0, 0); } static __inline__ __m128i __DEFAULT_FN_ATTRS _mm_broadcastd_epi32(__m128i __X) { - return (__m128i)__builtin_ia32_pbroadcastd128((__v4si)__X); + return (__m128i)__builtin_shufflevector((__v4si)__X, (__v4si)__X, 0, 0, 0, 0); } static __inline__ __m128i __DEFAULT_FN_ATTRS _mm_broadcastq_epi64(__m128i __X) { - return (__m128i)__builtin_ia32_pbroadcastq128(__X); + return (__m128i)__builtin_shufflevector(__X, __X, 0, 0); } static __inline__ __m256i __DEFAULT_FN_ATTRS @@ -865,43 +861,39 @@ _mm256_permutevar8x32_epi32(__m256i __a, __m256i __b) } #define _mm256_permute4x64_pd(V, M) __extension__ ({ \ - __m256d __V = (V); \ - (__m256d)__builtin_shufflevector((__v4df)__V, (__v4df) _mm256_setzero_pd(), \ + (__m256d)__builtin_shufflevector((__v4df)(__m256d)(V), \ + (__v4df)_mm256_setzero_pd(), \ (M) & 0x3, ((M) & 0xc) >> 2, \ ((M) & 0x30) >> 4, ((M) & 0xc0) >> 6); }) static __inline__ __m256 __DEFAULT_FN_ATTRS -_mm256_permutevar8x32_ps(__m256 __a, __m256 __b) +_mm256_permutevar8x32_ps(__m256 __a, __m256i __b) { - return (__m256)__builtin_ia32_permvarsf256((__v8sf)__a, (__v8sf)__b); + return (__m256)__builtin_ia32_permvarsf256((__v8sf)__a, (__v8si)__b); } #define _mm256_permute4x64_epi64(V, M) __extension__ ({ \ - __m256i __V = (V); \ - (__m256i)__builtin_shufflevector((__v4di)__V, (__v4di) _mm256_setzero_si256(), \ + (__m256i)__builtin_shufflevector((__v4di)(__m256i)(V), \ + (__v4di)_mm256_setzero_si256(), \ (M) & 0x3, ((M) & 0xc) >> 2, \ ((M) & 0x30) >> 4, ((M) & 0xc0) >> 6); }) #define _mm256_permute2x128_si256(V1, V2, M) __extension__ ({ \ - __m256i __V1 = (V1); \ - __m256i __V2 = (V2); \ - (__m256i)__builtin_ia32_permti256(__V1, __V2, (M)); }) + (__m256i)__builtin_ia32_permti256((__m256i)(V1), (__m256i)(V2), (M)); }) #define _mm256_extracti128_si256(V, M) __extension__ ({ \ - (__m128i)__builtin_shufflevector( \ - (__v4di)(V), \ - (__v4di)(_mm256_setzero_si256()), \ - (((M) & 1) ? 2 : 0), \ - (((M) & 1) ? 3 : 1) );}) + (__m128i)__builtin_shufflevector((__v4di)(__m256i)(V), \ + (__v4di)_mm256_setzero_si256(), \ + (((M) & 1) ? 2 : 0), \ + (((M) & 1) ? 3 : 1) ); }) #define _mm256_inserti128_si256(V1, V2, M) __extension__ ({ \ - (__m256i)__builtin_shufflevector( \ - (__v4di)(V1), \ - (__v4di)_mm256_castsi128_si256((__m128i)(V2)), \ - (((M) & 1) ? 0 : 4), \ - (((M) & 1) ? 1 : 5), \ - (((M) & 1) ? 4 : 2), \ - (((M) & 1) ? 5 : 3) );}) + (__m256i)__builtin_shufflevector((__v4di)(__m256i)(V1), \ + (__v4di)_mm256_castsi128_si256((__m128i)(V2)), \ + (((M) & 1) ? 0 : 4), \ + (((M) & 1) ? 1 : 5), \ + (((M) & 1) ? 4 : 2), \ + (((M) & 1) ? 5 : 3) ); }) static __inline__ __m256i __DEFAULT_FN_ATTRS _mm256_maskload_epi32(int const *__X, __m256i __M) @@ -1012,244 +1004,211 @@ _mm_srlv_epi64(__m128i __X, __m128i __Y) } #define _mm_mask_i32gather_pd(a, m, i, mask, s) __extension__ ({ \ - __m128d __a = (a); \ - double const *__m = (m); \ - __m128i __i = (i); \ - __m128d __mask = (mask); \ - (__m128d)__builtin_ia32_gatherd_pd((__v2df)__a, (const __v2df *)__m, \ - (__v4si)__i, (__v2df)__mask, (s)); }) + (__m128d)__builtin_ia32_gatherd_pd((__v2df)(__m128i)(a), \ + (double const *)(m), \ + (__v4si)(__m128i)(i), \ + (__v2df)(__m128d)(mask), (s)); }) #define _mm256_mask_i32gather_pd(a, m, i, mask, s) __extension__ ({ \ - __m256d __a = (a); \ - double const *__m = (m); \ - __m128i __i = (i); \ - __m256d __mask = (mask); \ - (__m256d)__builtin_ia32_gatherd_pd256((__v4df)__a, (const __v4df *)__m, \ - (__v4si)__i, (__v4df)__mask, (s)); }) + (__m256d)__builtin_ia32_gatherd_pd256((__v4df)(__m256d)(a), \ + (double const *)(m), \ + (__v4si)(__m128i)(i), \ + (__v4df)(__m256d)(mask), (s)); }) #define _mm_mask_i64gather_pd(a, m, i, mask, s) __extension__ ({ \ - __m128d __a = (a); \ - double const *__m = (m); \ - __m128i __i = (i); \ - __m128d __mask = (mask); \ - (__m128d)__builtin_ia32_gatherq_pd((__v2df)__a, (const __v2df *)__m, \ - (__v2di)__i, (__v2df)__mask, (s)); }) + (__m128d)__builtin_ia32_gatherq_pd((__v2df)(__m128d)(a), \ + (double const *)(m), \ + (__v2di)(__m128i)(i), \ + (__v2df)(__m128d)(mask), (s)); }) #define _mm256_mask_i64gather_pd(a, m, i, mask, s) __extension__ ({ \ - __m256d __a = (a); \ - double const *__m = (m); \ - __m256i __i = (i); \ - __m256d __mask = (mask); \ - (__m256d)__builtin_ia32_gatherq_pd256((__v4df)__a, (const __v4df *)__m, \ - (__v4di)__i, (__v4df)__mask, (s)); }) + (__m256d)__builtin_ia32_gatherq_pd256((__v4df)(__m256d)(a), \ + (double const *)(m), \ + (__v4di)(__m256i)(i), \ + (__v4df)(__m256d)(mask), (s)); }) #define _mm_mask_i32gather_ps(a, m, i, mask, s) __extension__ ({ \ - __m128 __a = (a); \ - float const *__m = (m); \ - __m128i __i = (i); \ - __m128 __mask = (mask); \ - (__m128)__builtin_ia32_gatherd_ps((__v4sf)__a, (const __v4sf *)__m, \ - (__v4si)__i, (__v4sf)__mask, (s)); }) + (__m128)__builtin_ia32_gatherd_ps((__v4sf)(__m128)(a), \ + (float const *)(m), \ + (__v4si)(__m128i)(i), \ + (__v4sf)(__m128)(mask), (s)); }) #define _mm256_mask_i32gather_ps(a, m, i, mask, s) __extension__ ({ \ - __m256 __a = (a); \ - float const *__m = (m); \ - __m256i __i = (i); \ - __m256 __mask = (mask); \ - (__m256)__builtin_ia32_gatherd_ps256((__v8sf)__a, (const __v8sf *)__m, \ - (__v8si)__i, (__v8sf)__mask, (s)); }) + (__m256)__builtin_ia32_gatherd_ps256((__v8sf)(__m256)(a), \ + (float const *)(m), \ + (__v8si)(__m256i)(i), \ + (__v8sf)(__m256)(mask), (s)); }) #define _mm_mask_i64gather_ps(a, m, i, mask, s) __extension__ ({ \ - __m128 __a = (a); \ - float const *__m = (m); \ - __m128i __i = (i); \ - __m128 __mask = (mask); \ - (__m128)__builtin_ia32_gatherq_ps((__v4sf)__a, (const __v4sf *)__m, \ - (__v2di)__i, (__v4sf)__mask, (s)); }) + (__m128)__builtin_ia32_gatherq_ps((__v4sf)(__m128)(a), \ + (float const *)(m), \ + (__v2di)(__m128i)(i), \ + (__v4sf)(__m128)(mask), (s)); }) #define _mm256_mask_i64gather_ps(a, m, i, mask, s) __extension__ ({ \ - __m128 __a = (a); \ - float const *__m = (m); \ - __m256i __i = (i); \ - __m128 __mask = (mask); \ - (__m128)__builtin_ia32_gatherq_ps256((__v4sf)__a, (const __v4sf *)__m, \ - (__v4di)__i, (__v4sf)__mask, (s)); }) + (__m128)__builtin_ia32_gatherq_ps256((__v4sf)(__m128)(a), \ + (float const *)(m), \ + (__v4di)(__m256i)(i), \ + (__v4sf)(__m128)(mask), (s)); }) #define _mm_mask_i32gather_epi32(a, m, i, mask, s) __extension__ ({ \ - __m128i __a = (a); \ - int const *__m = (m); \ - __m128i __i = (i); \ - __m128i __mask = (mask); \ - (__m128i)__builtin_ia32_gatherd_d((__v4si)__a, (const __v4si *)__m, \ - (__v4si)__i, (__v4si)__mask, (s)); }) + (__m128i)__builtin_ia32_gatherd_d((__v4si)(__m128i)(a), \ + (int const *)(m), \ + (__v4si)(__m128i)(i), \ + (__v4si)(__m128i)(mask), (s)); }) #define _mm256_mask_i32gather_epi32(a, m, i, mask, s) __extension__ ({ \ - __m256i __a = (a); \ - int const *__m = (m); \ - __m256i __i = (i); \ - __m256i __mask = (mask); \ - (__m256i)__builtin_ia32_gatherd_d256((__v8si)__a, (const __v8si *)__m, \ - (__v8si)__i, (__v8si)__mask, (s)); }) + (__m256i)__builtin_ia32_gatherd_d256((__v8si)(__m256i)(a), \ + (int const *)(m), \ + (__v8si)(__m256i)(i), \ + (__v8si)(__m256i)(mask), (s)); }) #define _mm_mask_i64gather_epi32(a, m, i, mask, s) __extension__ ({ \ - __m128i __a = (a); \ - int const *__m = (m); \ - __m128i __i = (i); \ - __m128i __mask = (mask); \ - (__m128i)__builtin_ia32_gatherq_d((__v4si)__a, (const __v4si *)__m, \ - (__v2di)__i, (__v4si)__mask, (s)); }) + (__m128i)__builtin_ia32_gatherq_d((__v4si)(__m128i)(a), \ + (int const *)(m), \ + (__v2di)(__m128i)(i), \ + (__v4si)(__m128i)(mask), (s)); }) #define _mm256_mask_i64gather_epi32(a, m, i, mask, s) __extension__ ({ \ - __m128i __a = (a); \ - int const *__m = (m); \ - __m256i __i = (i); \ - __m128i __mask = (mask); \ - (__m128i)__builtin_ia32_gatherq_d256((__v4si)__a, (const __v4si *)__m, \ - (__v4di)__i, (__v4si)__mask, (s)); }) + (__m128i)__builtin_ia32_gatherq_d256((__v4si)(__m128i)(a), \ + (int const *)(m), \ + (__v4di)(__m256i)(i), \ + (__v4si)(__m128i)(mask), (s)); }) #define _mm_mask_i32gather_epi64(a, m, i, mask, s) __extension__ ({ \ - __m128i __a = (a); \ - long long const *__m = (m); \ - __m128i __i = (i); \ - __m128i __mask = (mask); \ - (__m128i)__builtin_ia32_gatherd_q((__v2di)__a, (const __v2di *)__m, \ - (__v4si)__i, (__v2di)__mask, (s)); }) + (__m128i)__builtin_ia32_gatherd_q((__v2di)(__m128i)(a), \ + (long long const *)(m), \ + (__v4si)(__m128i)(i), \ + (__v2di)(__m128i)(mask), (s)); }) #define _mm256_mask_i32gather_epi64(a, m, i, mask, s) __extension__ ({ \ - __m256i __a = (a); \ - long long const *__m = (m); \ - __m128i __i = (i); \ - __m256i __mask = (mask); \ - (__m256i)__builtin_ia32_gatherd_q256((__v4di)__a, (const __v4di *)__m, \ - (__v4si)__i, (__v4di)__mask, (s)); }) + (__m256i)__builtin_ia32_gatherd_q256((__v4di)(__m256i)(a), \ + (long long const *)(m), \ + (__v4si)(__m128i)(i), \ + (__v4di)(__m256i)(mask), (s)); }) #define _mm_mask_i64gather_epi64(a, m, i, mask, s) __extension__ ({ \ - __m128i __a = (a); \ - long long const *__m = (m); \ - __m128i __i = (i); \ - __m128i __mask = (mask); \ - (__m128i)__builtin_ia32_gatherq_q((__v2di)__a, (const __v2di *)__m, \ - (__v2di)__i, (__v2di)__mask, (s)); }) + (__m128i)__builtin_ia32_gatherq_q((__v2di)(__m128i)(a), \ + (long long const *)(m), \ + (__v2di)(__m128i)(i), \ + (__v2di)(__m128i)(mask), (s)); }) #define _mm256_mask_i64gather_epi64(a, m, i, mask, s) __extension__ ({ \ - __m256i __a = (a); \ - long long const *__m = (m); \ - __m256i __i = (i); \ - __m256i __mask = (mask); \ - (__m256i)__builtin_ia32_gatherq_q256((__v4di)__a, (const __v4di *)__m, \ - (__v4di)__i, (__v4di)__mask, (s)); }) + (__m256i)__builtin_ia32_gatherq_q256((__v4di)(__m256i)(a), \ + (long long const *)(m), \ + (__v4di)(__m256i)(i), \ + (__v4di)(__m256i)(mask), (s)); }) #define _mm_i32gather_pd(m, i, s) __extension__ ({ \ - double const *__m = (m); \ - __m128i __i = (i); \ - (__m128d)__builtin_ia32_gatherd_pd((__v2df)_mm_setzero_pd(), \ - (const __v2df *)__m, (__v4si)__i, \ - (__v2df)_mm_set1_pd((double)(long long int)-1), (s)); }) + (__m128d)__builtin_ia32_gatherd_pd((__v2df)_mm_undefined_pd(), \ + (double const *)(m), \ + (__v4si)(__m128i)(i), \ + (__v2df)_mm_cmpeq_pd(_mm_setzero_pd(), \ + _mm_setzero_pd()), \ + (s)); }) #define _mm256_i32gather_pd(m, i, s) __extension__ ({ \ - double const *__m = (m); \ - __m128i __i = (i); \ - (__m256d)__builtin_ia32_gatherd_pd256((__v4df)_mm256_setzero_pd(), \ - (const __v4df *)__m, (__v4si)__i, \ - (__v4df)_mm256_set1_pd((double)(long long int)-1), (s)); }) + (__m256d)__builtin_ia32_gatherd_pd256((__v4df)_mm256_undefined_pd(), \ + (double const *)(m), \ + (__v4si)(__m128i)(i), \ + (__v4df)_mm256_cmp_pd(_mm256_setzero_pd(), \ + _mm256_setzero_pd(), \ + _CMP_EQ_OQ), \ + (s)); }) #define _mm_i64gather_pd(m, i, s) __extension__ ({ \ - double const *__m = (m); \ - __m128i __i = (i); \ - (__m128d)__builtin_ia32_gatherq_pd((__v2df)_mm_setzero_pd(), \ - (const __v2df *)__m, (__v2di)__i, \ - (__v2df)_mm_set1_pd((double)(long long int)-1), (s)); }) + (__m128d)__builtin_ia32_gatherq_pd((__v2df)_mm_undefined_pd(), \ + (double const *)(m), \ + (__v2di)(__m128i)(i), \ + (__v2df)_mm_cmpeq_pd(_mm_setzero_pd(), \ + _mm_setzero_pd()), \ + (s)); }) #define _mm256_i64gather_pd(m, i, s) __extension__ ({ \ - double const *__m = (m); \ - __m256i __i = (i); \ - (__m256d)__builtin_ia32_gatherq_pd256((__v4df)_mm256_setzero_pd(), \ - (const __v4df *)__m, (__v4di)__i, \ - (__v4df)_mm256_set1_pd((double)(long long int)-1), (s)); }) + (__m256d)__builtin_ia32_gatherq_pd256((__v4df)_mm256_undefined_pd(), \ + (double const *)(m), \ + (__v4di)(__m256i)(i), \ + (__v4df)_mm256_cmp_pd(_mm256_setzero_pd(), \ + _mm256_setzero_pd(), \ + _CMP_EQ_OQ), \ + (s)); }) #define _mm_i32gather_ps(m, i, s) __extension__ ({ \ - float const *__m = (m); \ - __m128i __i = (i); \ - (__m128)__builtin_ia32_gatherd_ps((__v4sf)_mm_setzero_ps(), \ - (const __v4sf *)__m, (__v4si)__i, \ - (__v4sf)_mm_set1_ps((float)(int)-1), (s)); }) + (__m128)__builtin_ia32_gatherd_ps((__v4sf)_mm_undefined_ps(), \ + (float const *)(m), \ + (__v4si)(__m128i)(i), \ + (__v4sf)_mm_cmpeq_ps(_mm_setzero_ps(), \ + _mm_setzero_ps()), \ + (s)); }) #define _mm256_i32gather_ps(m, i, s) __extension__ ({ \ - float const *__m = (m); \ - __m256i __i = (i); \ - (__m256)__builtin_ia32_gatherd_ps256((__v8sf)_mm256_setzero_ps(), \ - (const __v8sf *)__m, (__v8si)__i, \ - (__v8sf)_mm256_set1_ps((float)(int)-1), (s)); }) + (__m256)__builtin_ia32_gatherd_ps256((__v8sf)_mm256_undefined_ps(), \ + (float const *)(m), \ + (__v8si)(__m256i)(i), \ + (__v8sf)_mm256_cmp_ps(_mm256_setzero_ps(), \ + _mm256_setzero_ps(), \ + _CMP_EQ_OQ), \ + (s)); }) #define _mm_i64gather_ps(m, i, s) __extension__ ({ \ - float const *__m = (m); \ - __m128i __i = (i); \ - (__m128)__builtin_ia32_gatherq_ps((__v4sf)_mm_setzero_ps(), \ - (const __v4sf *)__m, (__v2di)__i, \ - (__v4sf)_mm_set1_ps((float)(int)-1), (s)); }) + (__m128)__builtin_ia32_gatherq_ps((__v4sf)_mm_undefined_ps(), \ + (float const *)(m), \ + (__v2di)(__m128i)(i), \ + (__v4sf)_mm_cmpeq_ps(_mm_setzero_ps(), \ + _mm_setzero_ps()), \ + (s)); }) #define _mm256_i64gather_ps(m, i, s) __extension__ ({ \ - float const *__m = (m); \ - __m256i __i = (i); \ - (__m128)__builtin_ia32_gatherq_ps256((__v4sf)_mm_setzero_ps(), \ - (const __v4sf *)__m, (__v4di)__i, \ - (__v4sf)_mm_set1_ps((float)(int)-1), (s)); }) + (__m128)__builtin_ia32_gatherq_ps256((__v4sf)_mm_undefined_ps(), \ + (float const *)(m), \ + (__v4di)(__m256i)(i), \ + (__v4sf)_mm_cmpeq_ps(_mm_setzero_ps(), \ + _mm_setzero_ps()), \ + (s)); }) #define _mm_i32gather_epi32(m, i, s) __extension__ ({ \ - int const *__m = (m); \ - __m128i __i = (i); \ - (__m128i)__builtin_ia32_gatherd_d((__v4si)_mm_setzero_si128(), \ - (const __v4si *)__m, (__v4si)__i, \ - (__v4si)_mm_set1_epi32(-1), (s)); }) + (__m128i)__builtin_ia32_gatherd_d((__v4si)_mm_undefined_si128(), \ + (int const *)(m), (__v4si)(__m128i)(i), \ + (__v4si)_mm_set1_epi32(-1), (s)); }) #define _mm256_i32gather_epi32(m, i, s) __extension__ ({ \ - int const *__m = (m); \ - __m256i __i = (i); \ - (__m256i)__builtin_ia32_gatherd_d256((__v8si)_mm256_setzero_si256(), \ - (const __v8si *)__m, (__v8si)__i, \ - (__v8si)_mm256_set1_epi32(-1), (s)); }) + (__m256i)__builtin_ia32_gatherd_d256((__v8si)_mm256_undefined_si256(), \ + (int const *)(m), (__v8si)(__m256i)(i), \ + (__v8si)_mm256_set1_epi32(-1), (s)); }) #define _mm_i64gather_epi32(m, i, s) __extension__ ({ \ - int const *__m = (m); \ - __m128i __i = (i); \ - (__m128i)__builtin_ia32_gatherq_d((__v4si)_mm_setzero_si128(), \ - (const __v4si *)__m, (__v2di)__i, \ - (__v4si)_mm_set1_epi32(-1), (s)); }) + (__m128i)__builtin_ia32_gatherq_d((__v4si)_mm_undefined_si128(), \ + (int const *)(m), (__v2di)(__m128i)(i), \ + (__v4si)_mm_set1_epi32(-1), (s)); }) #define _mm256_i64gather_epi32(m, i, s) __extension__ ({ \ - int const *__m = (m); \ - __m256i __i = (i); \ - (__m128i)__builtin_ia32_gatherq_d256((__v4si)_mm_setzero_si128(), \ - (const __v4si *)__m, (__v4di)__i, \ - (__v4si)_mm_set1_epi32(-1), (s)); }) + (__m128i)__builtin_ia32_gatherq_d256((__v4si)_mm_undefined_si128(), \ + (int const *)(m), (__v4di)(__m256i)(i), \ + (__v4si)_mm_set1_epi32(-1), (s)); }) #define _mm_i32gather_epi64(m, i, s) __extension__ ({ \ - long long const *__m = (m); \ - __m128i __i = (i); \ - (__m128i)__builtin_ia32_gatherd_q((__v2di)_mm_setzero_si128(), \ - (const __v2di *)__m, (__v4si)__i, \ - (__v2di)_mm_set1_epi64x(-1), (s)); }) + (__m128i)__builtin_ia32_gatherd_q((__v2di)_mm_undefined_si128(), \ + (long long const *)(m), \ + (__v4si)(__m128i)(i), \ + (__v2di)_mm_set1_epi64x(-1), (s)); }) #define _mm256_i32gather_epi64(m, i, s) __extension__ ({ \ - long long const *__m = (m); \ - __m128i __i = (i); \ - (__m256i)__builtin_ia32_gatherd_q256((__v4di)_mm256_setzero_si256(), \ - (const __v4di *)__m, (__v4si)__i, \ - (__v4di)_mm256_set1_epi64x(-1), (s)); }) + (__m256i)__builtin_ia32_gatherd_q256((__v4di)_mm256_undefined_si256(), \ + (long long const *)(m), \ + (__v4si)(__m128i)(i), \ + (__v4di)_mm256_set1_epi64x(-1), (s)); }) #define _mm_i64gather_epi64(m, i, s) __extension__ ({ \ - long long const *__m = (m); \ - __m128i __i = (i); \ - (__m128i)__builtin_ia32_gatherq_q((__v2di)_mm_setzero_si128(), \ - (const __v2di *)__m, (__v2di)__i, \ - (__v2di)_mm_set1_epi64x(-1), (s)); }) + (__m128i)__builtin_ia32_gatherq_q((__v2di)_mm_undefined_si128(), \ + (long long const *)(m), \ + (__v2di)(__m128i)(i), \ + (__v2di)_mm_set1_epi64x(-1), (s)); }) #define _mm256_i64gather_epi64(m, i, s) __extension__ ({ \ - long long const *__m = (m); \ - __m256i __i = (i); \ - (__m256i)__builtin_ia32_gatherq_q256((__v4di)_mm256_setzero_si256(), \ - (const __v4di *)__m, (__v4di)__i, \ - (__v4di)_mm256_set1_epi64x(-1), (s)); }) + (__m256i)__builtin_ia32_gatherq_q256((__v4di)_mm256_undefined_si256(), \ + (long long const *)(m), \ + (__v4di)(__m256i)(i), \ + (__v4di)_mm256_set1_epi64x(-1), (s)); }) #undef __DEFAULT_FN_ATTRS |
