11#error "Never use <avx2intrin.h> directly; include <immintrin.h> instead."
18#define __DEFAULT_FN_ATTRS256 \
19 __attribute__((__always_inline__, __nodebug__, __target__("avx2"), \
20 __min_vector_width__(256)))
21#define __DEFAULT_FN_ATTRS128 \
22 __attribute__((__always_inline__, __nodebug__, __target__("avx2"), \
23 __min_vector_width__(128)))
25#if defined(__cplusplus) && (__cplusplus >= 201103L)
26#define __DEFAULT_FN_ATTRS256_CONSTEXPR __DEFAULT_FN_ATTRS256 constexpr
27#define __DEFAULT_FN_ATTRS128_CONSTEXPR __DEFAULT_FN_ATTRS128 constexpr
29#define __DEFAULT_FN_ATTRS256_CONSTEXPR __DEFAULT_FN_ATTRS256
30#define __DEFAULT_FN_ATTRS128_CONSTEXPR __DEFAULT_FN_ATTRS128
91#define _mm256_mpsadbw_epu8(X, Y, M) \
92 ((__m256i)__builtin_ia32_mpsadbw256((__v32qi)(__m256i)(X), \
93 (__v32qi)(__m256i)(Y), (int)(M)))
108 return (__m256i)__builtin_elementwise_abs((__v32qs)
__a);
124 return (__m256i)__builtin_elementwise_abs((__v16hi)
__a);
140 return (__m256i)__builtin_elementwise_abs((__v8si)
__a);
170 return (__m256i)__builtin_ia32_packsswb256((__v16hi)
__a, (__v16hi)
__b);
201 return (__m256i)__builtin_ia32_packssdw256((__v8si)
__a, (__v8si)
__b);
231 return (__m256i)__builtin_ia32_packuswb256((__v16hi)
__a, (__v16hi)
__b);
262 return (__m256i) __builtin_ia32_packusdw256((__v8si)__V1, (__v8si)__V2);
280 return (__m256i)((__v32qu)
__a + (__v32qu)
__b);
298 return (__m256i)((__v16hu)
__a + (__v16hu)
__b);
316 return (__m256i)((__v8su)
__a + (__v8su)
__b);
334 return (__m256i)((__v4du)
__a + (__v4du)
__b);
352 return (__m256i)__builtin_elementwise_add_sat((__v32qs)
__a, (__v32qs)
__b);
369 return (__m256i)__builtin_elementwise_add_sat((__v16hi)
__a, (__v16hi)
__b);
387 return (__m256i)__builtin_elementwise_add_sat((__v32qu)
__a, (__v32qu)
__b);
404 return (__m256i)__builtin_elementwise_add_sat((__v16hu)
__a, (__v16hu)
__b);
431#define _mm256_alignr_epi8(a, b, n) \
432 ((__m256i)__builtin_ia32_palignr256((__v32qi)(__m256i)(a), \
433 (__v32qi)(__m256i)(b), (n)))
450 return (__m256i)((__v4du)
__a & (__v4du)
__b);
468 return (__m256i)(~(__v4du)
__a & (__v4du)
__b);
493 return (__m256i)__builtin_ia32_pavgb256((__v32qu)
__a, (__v32qu)
__b);
518 return (__m256i)__builtin_ia32_pavgw256((__v16hu)
__a, (__v16hu)
__b);
552 return (__m256i)__builtin_ia32_pblendvb256((__v32qi)__V1, (__v32qi)__V2,
593#define _mm256_blend_epi16(V1, V2, M) \
594 ((__m256i)__builtin_ia32_pblendw256((__v16hi)(__m256i)(V1), \
595 (__v16hi)(__m256i)(V2), (int)(M)))
620 return (__m256i)((__v32qi)
__a == (__v32qi)
__b);
646 return (__m256i)((__v16hi)
__a == (__v16hi)
__b);
672 return (__m256i)((__v8si)
__a == (__v8si)
__b);
698 return (__m256i)((__v4di)
__a == (__v4di)
__b);
726 return (__m256i)((__v32qs)
__a > (__v32qs)
__b);
752 return (__m256i)((__v16hi)
__a > (__v16hi)
__b);
778 return (__m256i)((__v8si)
__a > (__v8si)
__b);
804 return (__m256i)((__v4di)
__a > (__v4di)
__b);
839 return (__m256i)__builtin_ia32_phaddw256((__v16hi)
__a, (__v16hi)
__b);
870 return (__m256i)__builtin_ia32_phaddd256((__v8si)
__a, (__v8si)
__b);
904 return (__m256i)__builtin_ia32_phaddsw256((__v16hi)
__a, (__v16hi)
__b);
939 return (__m256i)__builtin_ia32_phsubw256((__v16hi)
__a, (__v16hi)
__b);
970 return (__m256i)__builtin_ia32_phsubd256((__v8si)
__a, (__v8si)
__b);
1005 return (__m256i)__builtin_ia32_phsubsw256((__v16hi)
__a, (__v16hi)
__b);
1034 return (__m256i)__builtin_ia32_pmaddubsw256((__v32qi)
__a, (__v32qi)
__b);
1065 return (__m256i)__builtin_ia32_pmaddwd256((__v16hi)
__a, (__v16hi)
__b);
1083 return (__m256i)__builtin_elementwise_max((__v32qs)
__a, (__v32qs)
__b);
1101 return (__m256i)__builtin_elementwise_max((__v16hi)
__a, (__v16hi)
__b);
1119 return (__m256i)__builtin_elementwise_max((__v8si)
__a, (__v8si)
__b);
1137 return (__m256i)__builtin_elementwise_max((__v32qu)
__a, (__v32qu)
__b);
1155 return (__m256i)__builtin_elementwise_max((__v16hu)
__a, (__v16hu)
__b);
1173 return (__m256i)__builtin_elementwise_max((__v8su)
__a, (__v8su)
__b);
1191 return (__m256i)__builtin_elementwise_min((__v32qs)
__a, (__v32qs)
__b);
1209 return (__m256i)__builtin_elementwise_min((__v16hi)
__a, (__v16hi)
__b);
1227 return (__m256i)__builtin_elementwise_min((__v8si)
__a, (__v8si)
__b);
1245 return (__m256i)__builtin_elementwise_min((__v32qu)
__a, (__v32qu)
__b);
1263 return (__m256i)__builtin_elementwise_min((__v16hu)
__a, (__v16hu)
__b);
1281 return (__m256i)__builtin_elementwise_min((__v8su)
__a, (__v8su)
__b);
1304 return __builtin_ia32_pmovmskb256((__v32qi)
__a);
1331 return (__m256i)__builtin_convertvector((__v16qs)__V, __v16hi);
1358 return (__m256i)__builtin_convertvector(__builtin_shufflevector((__v16qs)__V, (__v16qs)__V, 0, 1, 2, 3, 4, 5, 6, 7), __v8si);
1384 return (__m256i)__builtin_convertvector(__builtin_shufflevector((__v16qs)__V, (__v16qs)__V, 0, 1, 2, 3), __v4di);
1409 return (__m256i)__builtin_convertvector((__v8hi)__V, __v8si);
1433 return (__m256i)__builtin_convertvector(__builtin_shufflevector((__v8hi)__V, (__v8hi)__V, 0, 1, 2, 3), __v4di);
1457 return (__m256i)__builtin_convertvector((__v4si)__V, __v4di);
1482 return (__m256i)__builtin_convertvector((__v16qu)__V, __v16hi);
1507 return (__m256i)__builtin_convertvector(__builtin_shufflevector((__v16qu)__V, (__v16qu)__V, 0, 1, 2, 3, 4, 5, 6, 7), __v8si);
1531 return (__m256i)__builtin_convertvector(__builtin_shufflevector((__v16qu)__V, (__v16qu)__V, 0, 1, 2, 3), __v4di);
1556 return (__m256i)__builtin_convertvector((__v8hu)__V, __v8si);
1580 return (__m256i)__builtin_convertvector(__builtin_shufflevector((__v8hu)__V, (__v8hu)__V, 0, 1, 2, 3), __v4di);
1604 return (__m256i)__builtin_convertvector((__v4su)__V, __v4di);
1629 return (__m256i)__builtin_ia32_pmuldq256((__v8si)
__a, (__v8si)
__b);
1656 return (__m256i)__builtin_ia32_pmulhrsw256((__v16hi)
__a, (__v16hi)
__b);
1675 return (__m256i)__builtin_ia32_pmulhuw256((__v16hu)
__a, (__v16hu)
__b);
1694 return (__m256i)__builtin_ia32_pmulhw256((__v16hi)
__a, (__v16hi)
__b);
1713 return (__m256i)((__v16hu)
__a * (__v16hu)
__b);
1731 return (__m256i)((__v8su)
__a * (__v8su)
__b);
1756 return __builtin_ia32_pmuludq256((__v8si)
__a, (__v8si)
__b);
1774 return (__m256i)((__v4du)
__a | (__v4du)
__b);
1819 return __builtin_ia32_psadbw256((__v32qi)
__a, (__v32qi)
__b);
1858 return (__m256i)__builtin_ia32_pshufb256((__v32qi)
__a, (__v32qi)
__b);
1891#define _mm256_shuffle_epi32(a, imm) \
1892 ((__m256i)__builtin_ia32_pshufd256((__v8si)(__m256i)(a), (int)(imm)))
1927#define _mm256_shufflehi_epi16(a, imm) \
1928 ((__m256i)__builtin_ia32_pshufhw256((__v16hi)(__m256i)(a), (int)(imm)))
1964#define _mm256_shufflelo_epi16(a, imm) \
1965 ((__m256i)__builtin_ia32_pshuflw256((__v16hi)(__m256i)(a), (int)(imm)))
1985 return (__m256i)__builtin_ia32_psignb256((__v32qi)
__a, (__v32qi)
__b);
2006 return (__m256i)__builtin_ia32_psignw256((__v16hi)
__a, (__v16hi)
__b);
2027 return (__m256i)__builtin_ia32_psignd256((__v8si)
__a, (__v8si)
__b);
2047#define _mm256_slli_si256(a, imm) \
2048 ((__m256i)__builtin_ia32_pslldqi256_byteshift((__v32qi)(__m256i)(a), \
2068#define _mm256_bslli_epi128(a, imm) \
2069 ((__m256i)__builtin_ia32_pslldqi256_byteshift((__v32qi)(__m256i)(a), \
2087 return (__m256i)__builtin_ia32_psllwi256((__v16hi)
__a, __count);
2108 return (__m256i)__builtin_ia32_psllw256((__v16hi)
__a, (__v8hi)__count);
2126 return (__m256i)__builtin_ia32_pslldi256((__v8si)
__a, __count);
2147 return (__m256i)__builtin_ia32_pslld256((__v8si)
__a, (__v4si)__count);
2165 return __builtin_ia32_psllqi256((__v4di)
__a, __count);
2186 return __builtin_ia32_psllq256((__v4di)
__a, __count);
2205 return (__m256i)__builtin_ia32_psrawi256((__v16hi)
__a, __count);
2227 return (__m256i)__builtin_ia32_psraw256((__v16hi)
__a, (__v8hi)__count);
2246 return (__m256i)__builtin_ia32_psradi256((__v8si)
__a, __count);
2268 return (__m256i)__builtin_ia32_psrad256((__v8si)
__a, (__v4si)__count);
2288#define _mm256_srli_si256(a, imm) \
2289 ((__m256i)__builtin_ia32_psrldqi256_byteshift((__v32qi)(__m256i)(a), \
2309#define _mm256_bsrli_epi128(a, imm) \
2310 ((__m256i)__builtin_ia32_psrldqi256_byteshift((__v32qi)(__m256i)(a), \
2328 return (__m256i)__builtin_ia32_psrlwi256((__v16hi)
__a, __count);
2349 return (__m256i)__builtin_ia32_psrlw256((__v16hi)
__a, (__v8hi)__count);
2367 return (__m256i)__builtin_ia32_psrldi256((__v8si)
__a, __count);
2388 return (__m256i)__builtin_ia32_psrld256((__v8si)
__a, (__v4si)__count);
2406 return __builtin_ia32_psrlqi256((__v4di)
__a, __count);
2427 return __builtin_ia32_psrlq256((__v4di)
__a, __count);
2453 return (__m256i)((__v32qu)
__a - (__v32qu)
__b);
2479 return (__m256i)((__v16hu)
__a - (__v16hu)
__b);
2504 return (__m256i)((__v8su)
__a - (__v8su)
__b);
2529 return (__m256i)((__v4du)
__a - (__v4du)
__b);
2554 return (__m256i)__builtin_elementwise_sub_sat((__v32qs)
__a, (__v32qs)
__b);
2579 return (__m256i)__builtin_elementwise_sub_sat((__v16hi)
__a, (__v16hi)
__b);
2605 return (__m256i)__builtin_elementwise_sub_sat((__v32qu)
__a, (__v32qu)
__b);
2630 return (__m256i)__builtin_elementwise_sub_sat((__v16hu)
__a, (__v16hu)
__b);
2663 return (__m256i)__builtin_shufflevector((__v32qi)
__a, (__v32qi)
__b, 8, 32+8, 9, 32+9, 10, 32+10, 11, 32+11, 12, 32+12, 13, 32+13, 14, 32+14, 15, 32+15, 24, 32+24, 25, 32+25, 26, 32+26, 27, 32+27, 28, 32+28, 29, 32+29, 30, 32+30, 31, 32+31);
2697 return (__m256i)__builtin_shufflevector((__v16hi)
__a, (__v16hi)
__b, 4, 16+4, 5, 16+5, 6, 16+6, 7, 16+7, 12, 16+12, 13, 16+13, 14, 16+14, 15, 16+15);
2730 return (__m256i)__builtin_shufflevector((__v8si)
__a, (__v8si)
__b, 2, 8+2, 3, 8+3, 6, 8+6, 7, 8+7);
2759 return (__m256i)__builtin_shufflevector((__v4di)
__a, (__v4di)
__b, 1, 4+1, 3, 4+3);
2792 return (__m256i)__builtin_shufflevector((__v32qi)
__a, (__v32qi)
__b, 0, 32+0, 1, 32+1, 2, 32+2, 3, 32+3, 4, 32+4, 5, 32+5, 6, 32+6, 7, 32+7, 16, 32+16, 17, 32+17, 18, 32+18, 19, 32+19, 20, 32+20, 21, 32+21, 22, 32+22, 23, 32+23);
2826 return (__m256i)__builtin_shufflevector((__v16hi)
__a, (__v16hi)
__b, 0, 16+0, 1, 16+1, 2, 16+2, 3, 16+3, 8, 16+8, 9, 16+9, 10, 16+10, 11, 16+11);
2859 return (__m256i)__builtin_shufflevector((__v8si)
__a, (__v8si)
__b, 0, 8+0, 1, 8+1, 4, 8+4, 5, 8+5);
2888 return (__m256i)__builtin_shufflevector((__v4di)
__a, (__v4di)
__b, 0, 4+0, 2, 4+2);
2906 return (__m256i)((__v4du)
__a ^ (__v4du)
__b);
2924 return (__m256i)__builtin_nontemporal_load((
const __v4di_aligned *)__V);
2940 return (__m128)__builtin_shufflevector((__v4sf)__X, (__v4sf)__X, 0, 0, 0, 0);
2956 return __builtin_shufflevector((__v2df)
__a, (__v2df)
__a, 0, 0);
2972 return (__m256)__builtin_shufflevector((__v4sf)__X, (__v4sf)__X, 0, 0, 0, 0, 0, 0, 0, 0);
2988 return (__m256d)__builtin_shufflevector((__v2df)__X, (__v2df)__X, 0, 0, 0, 0);
3003 return (__m256i)__builtin_shufflevector((__v2di)__X, (__v2di)__X, 0, 1, 0, 1);
3006#define _mm_broadcastsi128_si256(X) _mm256_broadcastsi128_si256(X)
3041#define _mm_blend_epi32(V1, V2, M) \
3042 ((__m128i)__builtin_ia32_pblendd128((__v4si)(__m128i)(V1), \
3043 (__v4si)(__m128i)(V2), (int)(M)))
3078#define _mm256_blend_epi32(V1, V2, M) \
3079 ((__m256i)__builtin_ia32_pblendd256((__v8si)(__m256i)(V1), \
3080 (__v8si)(__m256i)(V2), (int)(M)))
3094 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);
3109 return (__m256i)__builtin_shufflevector((__v8hi)__X, (__v8hi)__X, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0);
3124 return (__m256i)__builtin_shufflevector((__v4si)__X, (__v4si)__X, 0, 0, 0, 0, 0, 0, 0, 0);
3139 return (__m256i)__builtin_shufflevector((__v2di)__X, (__v2di)__X, 0, 0, 0, 0);
3154 return (__m128i)__builtin_shufflevector((__v16qi)__X, (__v16qi)__X, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0);
3169 return (__m128i)__builtin_shufflevector((__v8hi)__X, (__v8hi)__X, 0, 0, 0, 0, 0, 0, 0, 0);
3184 return (__m128i)__builtin_shufflevector((__v4si)__X, (__v4si)__X, 0, 0, 0, 0);
3199 return (__m128i)__builtin_shufflevector((__v2di)__X, (__v2di)__X, 0, 0);
3227 return (__m256i)__builtin_ia32_permvarsi256((__v8si)
__a, (__v8si)
__b);
3257#define _mm256_permute4x64_pd(V, M) \
3258 ((__m256d)__builtin_ia32_permdf256((__v4df)(__m256d)(V), (int)(M)))
3285 return (__m256)__builtin_ia32_permvarsf256((__v8sf)
__a, (__v8si)
__b);
3315#define _mm256_permute4x64_epi64(V, M) \
3316 ((__m256i)__builtin_ia32_permdi256((__v4di)(__m256i)(V), (int)(M)))
3361#define _mm256_permute2x128_si256(V1, V2, M) \
3362 ((__m256i)__builtin_ia32_permti256((__m256i)(V1), (__m256i)(V2), (int)(M)))
3381#define _mm256_extracti128_si256(V, M) \
3382 ((__m128i)__builtin_ia32_extract128i256((__v4di)(__m256i)(V), (int)(M)))
3404#define _mm256_inserti128_si256(V1, V2, M) \
3405 ((__m256i)__builtin_ia32_insert128i256((__v4di)(__m256i)(V1), \
3406 (__v2di)(__m128i)(V2), (int)(M)))
3437 return (__m256i)__builtin_ia32_maskloadd256((
const __v8si *)__X, (__v8si)__M);
3469 return (__m256i)__builtin_ia32_maskloadq256((
const __v4di *)__X, (__v4di)__M);
3501 return (__m128i)__builtin_ia32_maskloadd((
const __v4si *)__X, (__v4si)__M);
3533 return (__m128i)__builtin_ia32_maskloadq((
const __v2di *)__X, (__v2di)__M);
3563 __builtin_ia32_maskstored256((__v8si *)__X, (__v8si)__M, (__v8si)
__Y);
3593 __builtin_ia32_maskstoreq256((__v4di *)__X, (__v4di)__M, (__v4di)
__Y);
3623 __builtin_ia32_maskstored((__v4si *)__X, (__v4si)__M, (__v4si)
__Y);
3653 __builtin_ia32_maskstoreq(( __v2di *)__X, (__v2di)__M, (__v2di)
__Y);
3675 return (__m256i)__builtin_ia32_psllv8si((__v8si)__X, (__v8si)
__Y);
3697 return (__m128i)__builtin_ia32_psllv4si((__v4si)__X, (__v4si)
__Y);
3719 return (__m256i)__builtin_ia32_psllv4di((__v4di)__X, (__v4di)
__Y);
3741 return (__m128i)__builtin_ia32_psllv2di((__v2di)__X, (__v2di)
__Y);
3764 return (__m256i)__builtin_ia32_psrav8si((__v8si)__X, (__v8si)
__Y);
3787 return (__m128i)__builtin_ia32_psrav4si((__v4si)__X, (__v4si)
__Y);
3809 return (__m256i)__builtin_ia32_psrlv8si((__v8si)__X, (__v8si)
__Y);
3831 return (__m128i)__builtin_ia32_psrlv4si((__v4si)__X, (__v4si)
__Y);
3853 return (__m256i)__builtin_ia32_psrlv4di((__v4di)__X, (__v4di)
__Y);
3875 return (__m128i)__builtin_ia32_psrlv2di((__v2di)__X, (__v2di)
__Y);
3921#define _mm_mask_i32gather_pd(a, m, i, mask, s) \
3922 ((__m128d)__builtin_ia32_gatherd_pd((__v2df)(__m128i)(a), \
3923 (double const *)(m), \
3924 (__v4si)(__m128i)(i), \
3925 (__v2df)(__m128d)(mask), (s)))
3969#define _mm256_mask_i32gather_pd(a, m, i, mask, s) \
3970 ((__m256d)__builtin_ia32_gatherd_pd256((__v4df)(__m256d)(a), \
3971 (double const *)(m), \
3972 (__v4si)(__m128i)(i), \
3973 (__v4df)(__m256d)(mask), (s)))
4017#define _mm_mask_i64gather_pd(a, m, i, mask, s) \
4018 ((__m128d)__builtin_ia32_gatherq_pd((__v2df)(__m128d)(a), \
4019 (double const *)(m), \
4020 (__v2di)(__m128i)(i), \
4021 (__v2df)(__m128d)(mask), (s)))
4065#define _mm256_mask_i64gather_pd(a, m, i, mask, s) \
4066 ((__m256d)__builtin_ia32_gatherq_pd256((__v4df)(__m256d)(a), \
4067 (double const *)(m), \
4068 (__v4di)(__m256i)(i), \
4069 (__v4df)(__m256d)(mask), (s)))
4113#define _mm_mask_i32gather_ps(a, m, i, mask, s) \
4114 ((__m128)__builtin_ia32_gatherd_ps((__v4sf)(__m128)(a), \
4115 (float const *)(m), \
4116 (__v4si)(__m128i)(i), \
4117 (__v4sf)(__m128)(mask), (s)))
4161#define _mm256_mask_i32gather_ps(a, m, i, mask, s) \
4162 ((__m256)__builtin_ia32_gatherd_ps256((__v8sf)(__m256)(a), \
4163 (float const *)(m), \
4164 (__v8si)(__m256i)(i), \
4165 (__v8sf)(__m256)(mask), (s)))
4212#define _mm_mask_i64gather_ps(a, m, i, mask, s) \
4213 ((__m128)__builtin_ia32_gatherq_ps((__v4sf)(__m128)(a), \
4214 (float const *)(m), \
4215 (__v2di)(__m128i)(i), \
4216 (__v4sf)(__m128)(mask), (s)))
4260#define _mm256_mask_i64gather_ps(a, m, i, mask, s) \
4261 ((__m128)__builtin_ia32_gatherq_ps256((__v4sf)(__m128)(a), \
4262 (float const *)(m), \
4263 (__v4di)(__m256i)(i), \
4264 (__v4sf)(__m128)(mask), (s)))
4308#define _mm_mask_i32gather_epi32(a, m, i, mask, s) \
4309 ((__m128i)__builtin_ia32_gatherd_d((__v4si)(__m128i)(a), \
4311 (__v4si)(__m128i)(i), \
4312 (__v4si)(__m128i)(mask), (s)))
4356#define _mm256_mask_i32gather_epi32(a, m, i, mask, s) \
4357 ((__m256i)__builtin_ia32_gatherd_d256((__v8si)(__m256i)(a), \
4359 (__v8si)(__m256i)(i), \
4360 (__v8si)(__m256i)(mask), (s)))
4407#define _mm_mask_i64gather_epi32(a, m, i, mask, s) \
4408 ((__m128i)__builtin_ia32_gatherq_d((__v4si)(__m128i)(a), \
4410 (__v2di)(__m128i)(i), \
4411 (__v4si)(__m128i)(mask), (s)))
4455#define _mm256_mask_i64gather_epi32(a, m, i, mask, s) \
4456 ((__m128i)__builtin_ia32_gatherq_d256((__v4si)(__m128i)(a), \
4458 (__v4di)(__m256i)(i), \
4459 (__v4si)(__m128i)(mask), (s)))
4504#define _mm_mask_i32gather_epi64(a, m, i, mask, s) \
4505 ((__m128i)__builtin_ia32_gatherd_q((__v2di)(__m128i)(a), \
4506 (long long const *)(m), \
4507 (__v4si)(__m128i)(i), \
4508 (__v2di)(__m128i)(mask), (s)))
4552#define _mm256_mask_i32gather_epi64(a, m, i, mask, s) \
4553 ((__m256i)__builtin_ia32_gatherd_q256((__v4di)(__m256i)(a), \
4554 (long long const *)(m), \
4555 (__v4si)(__m128i)(i), \
4556 (__v4di)(__m256i)(mask), (s)))
4600#define _mm_mask_i64gather_epi64(a, m, i, mask, s) \
4601 ((__m128i)__builtin_ia32_gatherq_q((__v2di)(__m128i)(a), \
4602 (long long const *)(m), \
4603 (__v2di)(__m128i)(i), \
4604 (__v2di)(__m128i)(mask), (s)))
4648#define _mm256_mask_i64gather_epi64(a, m, i, mask, s) \
4649 ((__m256i)__builtin_ia32_gatherq_q256((__v4di)(__m256i)(a), \
4650 (long long const *)(m), \
4651 (__v4di)(__m256i)(i), \
4652 (__v4di)(__m256i)(mask), (s)))
4682#define _mm_i32gather_pd(m, i, s) \
4683 ((__m128d)__builtin_ia32_gatherd_pd((__v2df)_mm_undefined_pd(), \
4684 (double const *)(m), \
4685 (__v4si)(__m128i)(i), \
4686 (__v2df)_mm_cmpeq_pd(_mm_setzero_pd(), \
4687 _mm_setzero_pd()), \
4717#define _mm256_i32gather_pd(m, i, s) \
4718 ((__m256d)__builtin_ia32_gatherd_pd256((__v4df)_mm256_undefined_pd(), \
4719 (double const *)(m), \
4720 (__v4si)(__m128i)(i), \
4721 (__v4df)_mm256_cmp_pd(_mm256_setzero_pd(), \
4722 _mm256_setzero_pd(), \
4753#define _mm_i64gather_pd(m, i, s) \
4754 ((__m128d)__builtin_ia32_gatherq_pd((__v2df)_mm_undefined_pd(), \
4755 (double const *)(m), \
4756 (__v2di)(__m128i)(i), \
4757 (__v2df)_mm_cmpeq_pd(_mm_setzero_pd(), \
4758 _mm_setzero_pd()), \
4788#define _mm256_i64gather_pd(m, i, s) \
4789 ((__m256d)__builtin_ia32_gatherq_pd256((__v4df)_mm256_undefined_pd(), \
4790 (double const *)(m), \
4791 (__v4di)(__m256i)(i), \
4792 (__v4df)_mm256_cmp_pd(_mm256_setzero_pd(), \
4793 _mm256_setzero_pd(), \
4824#define _mm_i32gather_ps(m, i, s) \
4825 ((__m128)__builtin_ia32_gatherd_ps((__v4sf)_mm_undefined_ps(), \
4826 (float const *)(m), \
4827 (__v4si)(__m128i)(i), \
4828 (__v4sf)_mm_cmpeq_ps(_mm_setzero_ps(), \
4829 _mm_setzero_ps()), \
4859#define _mm256_i32gather_ps(m, i, s) \
4860 ((__m256)__builtin_ia32_gatherd_ps256((__v8sf)_mm256_undefined_ps(), \
4861 (float const *)(m), \
4862 (__v8si)(__m256i)(i), \
4863 (__v8sf)_mm256_cmp_ps(_mm256_setzero_ps(), \
4864 _mm256_setzero_ps(), \
4897#define _mm_i64gather_ps(m, i, s) \
4898 ((__m128)__builtin_ia32_gatherq_ps((__v4sf)_mm_undefined_ps(), \
4899 (float const *)(m), \
4900 (__v2di)(__m128i)(i), \
4901 (__v4sf)_mm_cmpeq_ps(_mm_setzero_ps(), \
4902 _mm_setzero_ps()), \
4932#define _mm256_i64gather_ps(m, i, s) \
4933 ((__m128)__builtin_ia32_gatherq_ps256((__v4sf)_mm_undefined_ps(), \
4934 (float const *)(m), \
4935 (__v4di)(__m256i)(i), \
4936 (__v4sf)_mm_cmpeq_ps(_mm_setzero_ps(), \
4937 _mm_setzero_ps()), \
4967#define _mm_i32gather_epi32(m, i, s) \
4968 ((__m128i)__builtin_ia32_gatherd_d((__v4si)_mm_undefined_si128(), \
4969 (int const *)(m), (__v4si)(__m128i)(i), \
4970 (__v4si)_mm_set1_epi32(-1), (s)))
4999#define _mm256_i32gather_epi32(m, i, s) \
5000 ((__m256i)__builtin_ia32_gatherd_d256((__v8si)_mm256_undefined_si256(), \
5001 (int const *)(m), (__v8si)(__m256i)(i), \
5002 (__v8si)_mm256_set1_epi32(-1), (s)))
5033#define _mm_i64gather_epi32(m, i, s) \
5034 ((__m128i)__builtin_ia32_gatherq_d((__v4si)_mm_undefined_si128(), \
5035 (int const *)(m), (__v2di)(__m128i)(i), \
5036 (__v4si)_mm_set1_epi32(-1), (s)))
5065#define _mm256_i64gather_epi32(m, i, s) \
5066 ((__m128i)__builtin_ia32_gatherq_d256((__v4si)_mm_undefined_si128(), \
5067 (int const *)(m), (__v4di)(__m256i)(i), \
5068 (__v4si)_mm_set1_epi32(-1), (s)))
5098#define _mm_i32gather_epi64(m, i, s) \
5099 ((__m128i)__builtin_ia32_gatherd_q((__v2di)_mm_undefined_si128(), \
5100 (long long const *)(m), \
5101 (__v4si)(__m128i)(i), \
5102 (__v2di)_mm_set1_epi64x(-1), (s)))
5131#define _mm256_i32gather_epi64(m, i, s) \
5132 ((__m256i)__builtin_ia32_gatherd_q256((__v4di)_mm256_undefined_si256(), \
5133 (long long const *)(m), \
5134 (__v4si)(__m128i)(i), \
5135 (__v4di)_mm256_set1_epi64x(-1), (s)))
5164#define _mm_i64gather_epi64(m, i, s) \
5165 ((__m128i)__builtin_ia32_gatherq_q((__v2di)_mm_undefined_si128(), \
5166 (long long const *)(m), \
5167 (__v2di)(__m128i)(i), \
5168 (__v2di)_mm_set1_epi64x(-1), (s)))
5197#define _mm256_i64gather_epi64(m, i, s) \
5198 ((__m256i)__builtin_ia32_gatherq_q256((__v4di)_mm256_undefined_si256(), \
5199 (long long const *)(m), \
5200 (__v4di)(__m256i)(i), \
5201 (__v4di)_mm256_set1_epi64x(-1), (s)))
5203#undef __DEFAULT_FN_ATTRS256
5204#undef __DEFAULT_FN_ATTRS128
5205#undef __DEFAULT_FN_ATTRS256_CONSTEXPR
5206#undef __DEFAULT_FN_ATTRS128_CONSTEXPR
_Float16 __2f16 __attribute__((ext_vector_type(2)))
Zeroes the upper 128 bits (bits 255:128) of all YMM registers.
static __inline__ vector float vector float __b
#define __DEFAULT_FN_ATTRS128
#define __DEFAULT_FN_ATTRS256
static __inline__ __m256i __DEFAULT_FN_ATTRS256 _mm256_srl_epi32(__m256i __a, __m128i __count)
Shifts each 32-bit element of the 256-bit vector of [8 x i32] in __a right by the number of bits give...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_sllv_epi32(__m256i __X, __m256i __Y)
Shifts each 32-bit element of the 256-bit vector of [8 x i32] in __X left by the number of bits given...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_cvtepi32_epi64(__m128i __V)
Sign-extends 32-bit elements from the 128-bit vector of [4 x i32] in __V and returns the 64-bit value...
static __inline__ __m256i __DEFAULT_FN_ATTRS256 _mm256_srl_epi16(__m256i __a, __m128i __count)
Shifts each 16-bit element of the 256-bit vector of [16 x i16] in __a right by the number of bits giv...
static __inline__ void __DEFAULT_FN_ATTRS256 _mm256_maskstore_epi64(long long *__X, __m256i __M, __m256i __Y)
Conditionally stores four 64-bit integer elements from the 256-bit vector of [4 x i64] in __Y to memo...
static __inline__ __m256i __DEFAULT_FN_ATTRS256 _mm256_stream_load_si256(const void *__V)
Loads the 256-bit integer vector from memory __V using a non-temporal memory hint and returns the vec...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_hsubs_epi16(__m256i __a, __m256i __b)
Horizontally subtracts adjacent pairs of 16-bit integers from two 256-bit vectors of [16 x i16] using...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_cmpgt_epi8(__m256i __a, __m256i __b)
Compares corresponding signed bytes in the 256-bit integer vectors in __a and __b for greater-than an...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_max_epi32(__m256i __a, __m256i __b)
Compares the corresponding signed 32-bit integers in the two 256-bit vectors of [8 x i32] in __a and ...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_mul_epu32(__m256i __a, __m256i __b)
Multiplies unsigned 32-bit integers from even-numered elements of two 256-bit vectors of [8 x i32] an...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_max_epu16(__m256i __a, __m256i __b)
Compares the corresponding unsigned 16-bit integers in the two 256-bit vectors of [16 x i16] in __a a...
static __inline__ __m256i __DEFAULT_FN_ATTRS256 _mm256_sll_epi32(__m256i __a, __m128i __count)
Shifts each 32-bit element of the 256-bit vector of [8 x i32] in __a left by the number of bits given...
static __inline__ __m128i __DEFAULT_FN_ATTRS128_CONSTEXPR _mm_srav_epi32(__m128i __X, __m128i __Y)
Shifts each 32-bit element of the 128-bit vector of [4 x i32] in __X right by the number of bits give...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_unpackhi_epi32(__m256i __a, __m256i __b)
Unpacks and interleaves 32-bit integers from parts of the 256-bit vectors of [8 x i32] in __a and __b...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_andnot_si256(__m256i __a, __m256i __b)
Computes the bitwise AND of the 256-bit integer vector in __b with the bitwise NOT of the 256-bit int...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_mulhi_epi16(__m256i __a, __m256i __b)
Multiplies signed 16-bit integer elements of two 256-bit vectors of [16 x i16], and returns the upper...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_adds_epu8(__m256i __a, __m256i __b)
Adds 8-bit integers from corresponding bytes of two 256-bit integer vectors using unsigned saturation...
static __inline__ __m256i __DEFAULT_FN_ATTRS256 _mm256_sll_epi64(__m256i __a, __m128i __count)
Shifts each 64-bit element of the 256-bit vector of [4 x i64] in __a left by the number of bits given...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_maddubs_epi16(__m256i __a, __m256i __b)
Multiplies each unsigned byte from the 256-bit integer vector in __a with the corresponding signed by...
static __inline__ __m256i __DEFAULT_FN_ATTRS256 _mm256_maskload_epi64(long long const *__X, __m256i __M)
Conditionally loads four 64-bit integer elements from memory __X, if the most significant bit of the ...
static __inline__ void __DEFAULT_FN_ATTRS256 _mm256_maskstore_epi32(int *__X, __m256i __M, __m256i __Y)
Conditionally stores eight 32-bit integer elements from the 256-bit vector of [8 x i32] in __Y to mem...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_mulhi_epu16(__m256i __a, __m256i __b)
Multiplies unsigned 16-bit integer elements of two 256-bit vectors of [16 x i16], and returns the upp...
static __inline__ __m256i __DEFAULT_FN_ATTRS256 _mm256_shuffle_epi8(__m256i __a, __m256i __b)
Shuffles 8-bit integers in the 256-bit integer vector __a according to control information in the 256...
#define __DEFAULT_FN_ATTRS128_CONSTEXPR
static __inline__ __m128i __DEFAULT_FN_ATTRS128_CONSTEXPR _mm_broadcastb_epi8(__m128i __X)
Broadcasts the low byte from the 128-bit integer vector in __X to all bytes of the 128-bit result.
static __inline__ __m128d __DEFAULT_FN_ATTRS128_CONSTEXPR _mm_broadcastsd_pd(__m128d __a)
Broadcasts the 64-bit floating-point value from the low element of the 128-bit vector of [2 x double]...
static __inline__ __m256i __DEFAULT_FN_ATTRS256 _mm256_maskload_epi32(int const *__X, __m256i __M)
Conditionally loads eight 32-bit integer elements from memory __X, if the most significant bit of the...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_cvtepu8_epi16(__m128i __V)
Zero-extends bytes from the 128-bit integer vector in __V and returns the 16-bit values in the corres...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_subs_epu16(__m256i __a, __m256i __b)
Subtracts 16-bit integers from corresponding elements of two 256-bit vectors of [16 x i16] using unsi...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_packus_epi32(__m256i __V1, __m256i __V2)
Converts elements from two 256-bit vectors of [8 x i32] to 16-bit integers using unsigned saturation,...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_broadcastb_epi8(__m128i __X)
Broadcasts the low byte from the 128-bit integer vector in __X to all bytes of the 256-bit result.
static __inline__ __m256i __DEFAULT_FN_ATTRS256 _mm256_permutevar8x32_epi32(__m256i __a, __m256i __b)
Sets the result's 256-bit vector of [8 x i32] to copies of elements of the 256-bit vector of [8 x i32...
static __inline__ __m128i __DEFAULT_FN_ATTRS128_CONSTEXPR _mm_broadcastq_epi64(__m128i __X)
Broadcasts the low element from the 128-bit vector of [2 x i64] in __X to both elements of the result...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_slli_epi64(__m256i __a, int __count)
Shifts each 64-bit element of the 256-bit vector of [4 x i64] in __a left by __count bits,...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_unpacklo_epi16(__m256i __a, __m256i __b)
Unpacks and interleaves 16-bit integers from parts of the 256-bit vectors of [16 x i16] in __a and __...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_abs_epi32(__m256i __a)
Computes the absolute value of each signed 32-bit element in the 256-bit vector of [8 x i32] in __a a...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_srav_epi32(__m256i __X, __m256i __Y)
Shifts each 32-bit element of the 256-bit vector of [8 x i32] in __X right by the number of bits give...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_mullo_epi16(__m256i __a, __m256i __b)
Multiplies signed 16-bit integer elements of two 256-bit vectors of [16 x i16], and returns the lower...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_cvtepu8_epi64(__m128i __V)
Zero-extends the first four bytes from the 128-bit integer vector in __V and returns the 64-bit value...
static __inline__ __m128i __DEFAULT_FN_ATTRS128_CONSTEXPR _mm_broadcastw_epi16(__m128i __X)
Broadcasts the low element from the 128-bit vector of [8 x i16] in __X to all elements of the result'...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_cmpeq_epi32(__m256i __a, __m256i __b)
Compares corresponding elements in the 256-bit vectors of [8 x i32] in __a and __b for equality and r...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_adds_epi16(__m256i __a, __m256i __b)
Adds 16-bit integers from corresponding elements of two 256-bit vectors of [16 x i16] using signed sa...
static __inline__ __m256i __DEFAULT_FN_ATTRS256 _mm256_mulhrs_epi16(__m256i __a, __m256i __b)
Multiplies signed 16-bit integer elements of two 256-bit vectors of [16 x i16], truncates the 32-bit ...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_min_epi8(__m256i __a, __m256i __b)
Compares the corresponding signed bytes in the two 256-bit integer vectors in __a and __b and returns...
static __inline__ __m256i __DEFAULT_FN_ATTRS256 _mm256_sll_epi16(__m256i __a, __m128i __count)
Shifts each 16-bit element of the 256-bit vector of [16 x i16] in __a left by the number of bits spec...
static __inline__ void __DEFAULT_FN_ATTRS128 _mm_maskstore_epi64(long long *__X, __m128i __M, __m128i __Y)
Conditionally stores two 64-bit integer elements from the 128-bit vector of [2 x i64] in __Y to memor...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_cvtepu32_epi64(__m128i __V)
Zero-extends 32-bit elements from the 128-bit vector of [4 x i32] in __V and returns the 64-bit value...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_cvtepu8_epi32(__m128i __V)
Zero-extends bytes from the lower half of the 128-bit integer vector in __V and returns the 32-bit va...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_srli_epi16(__m256i __a, int __count)
Shifts each 16-bit element of the 256-bit vector of [16 x i16] in __a right by __count bits,...
static __inline__ __m256 __DEFAULT_FN_ATTRS256 _mm256_permutevar8x32_ps(__m256 __a, __m256i __b)
Sets the result's 256-bit vector of [8 x float] to copies of elements of the 256-bit vector of [8 x f...
static __inline__ int __DEFAULT_FN_ATTRS256 _mm256_movemask_epi8(__m256i __a)
Creates a 32-bit integer mask from the most significant bit of each byte in the 256-bit integer vecto...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_max_epi16(__m256i __a, __m256i __b)
Compares the corresponding signed 16-bit integers in the two 256-bit vectors of [16 x i16] in __a and...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_cvtepi8_epi16(__m128i __V)
Sign-extends bytes from the 128-bit integer vector in __V and returns the 16-bit values in the corres...
static __inline__ __m128i __DEFAULT_FN_ATTRS128_CONSTEXPR _mm_srlv_epi32(__m128i __X, __m128i __Y)
Shifts each 32-bit element of the 128-bit vector of [4 x i32] in __X right by the number of bits give...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_blendv_epi8(__m256i __V1, __m256i __V2, __m256i __M)
Merges 8-bit integer values from either of the two 256-bit vectors __V1 or __V2, as specified by the ...
static __inline__ __m256 __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_broadcastss_ps(__m128 __X)
Broadcasts the 32-bit floating-point value from the low element of the 128-bit vector of [4 x float] ...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_cmpeq_epi64(__m256i __a, __m256i __b)
Compares corresponding elements in the 256-bit vectors of [4 x i64] in __a and __b for equality and r...
static __inline__ __m256i __DEFAULT_FN_ATTRS256 _mm256_sign_epi32(__m256i __a, __m256i __b)
Sets each element of the result to the corresponding element of the 256-bit vector of [8 x i32] in __...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_unpacklo_epi8(__m256i __a, __m256i __b)
Unpacks and interleaves 8-bit integers from parts of the 256-bit integer vectors in __a and __b to fo...
static __inline__ __m128i __DEFAULT_FN_ATTRS128 _mm_maskload_epi32(int const *__X, __m128i __M)
Conditionally loads four 32-bit integer elements from memory __X, if the most significant bit of the ...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_srli_epi64(__m256i __a, int __count)
Shifts each 64-bit element of the 256-bit vector of [4 x i64] in __a right by __count bits,...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_hsub_epi16(__m256i __a, __m256i __b)
Horizontally subtracts adjacent pairs of 16-bit integers from two 256-bit vectors of [16 x i16] and r...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_broadcastw_epi16(__m128i __X)
Broadcasts the low element from the 128-bit vector of [8 x i16] in __X to all elements of the result'...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_unpacklo_epi64(__m256i __a, __m256i __b)
Unpacks and interleaves 64-bit integers from parts of the 256-bit vectors of [4 x i64] in __a and __b...
static __inline__ __m256i __DEFAULT_FN_ATTRS256 _mm256_sign_epi8(__m256i __a, __m256i __b)
Sets each byte of the result to the corresponding byte of the 256-bit integer vector in __a,...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_cmpgt_epi16(__m256i __a, __m256i __b)
Compares corresponding signed elements in the 256-bit vectors of [16 x i16] in __a and __b for greate...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_hsub_epi32(__m256i __a, __m256i __b)
Horizontally subtracts adjacent pairs of 32-bit integers from two 256-bit vectors of [8 x i32] and re...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_srai_epi16(__m256i __a, int __count)
Shifts each 16-bit element of the 256-bit vector of [16 x i16] in __a right by __count bits,...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_add_epi8(__m256i __a, __m256i __b)
Adds 8-bit integers from corresponding bytes of two 256-bit integer vectors and returns the lower 8 b...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_cvtepi8_epi32(__m128i __V)
Sign-extends bytes from the lower half of the 128-bit integer vector in __V and returns the 32-bit va...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_broadcastq_epi64(__m128i __X)
Broadcasts the low element from the 128-bit vector of [2 x i64] in __X to all elements of the result'...
#define __DEFAULT_FN_ATTRS256_CONSTEXPR
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_mul_epi32(__m256i __a, __m256i __b)
Multiplies signed 32-bit integers from even-numbered elements of two 256-bit vectors of [8 x i32] and...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_min_epu16(__m256i __a, __m256i __b)
Compares the corresponding unsigned 16-bit integers in the two 256-bit vectors of [16 x i16] in __a a...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_hadd_epi32(__m256i __a, __m256i __b)
Horizontally adds the adjacent pairs of 32-bit integers from two 256-bit vectors of [8 x i32] and ret...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_mullo_epi32(__m256i __a, __m256i __b)
Multiplies signed 32-bit integer elements of two 256-bit vectors of [8 x i32], and returns the lower ...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_min_epi32(__m256i __a, __m256i __b)
Compares the corresponding signed 32-bit integers in the two 256-bit vectors of [8 x i32] in __a and ...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_cmpgt_epi32(__m256i __a, __m256i __b)
Compares corresponding signed elements in the 256-bit vectors of [8 x i32] in __a and __b for greater...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_sub_epi32(__m256i __a, __m256i __b)
Subtracts 32-bit integers from corresponding elements of two 256-bit vectors of [8 x i32].
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_unpackhi_epi64(__m256i __a, __m256i __b)
Unpacks and interleaves 64-bit integers from parts of the 256-bit vectors of [4 x i64] in __a and __b...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_subs_epu8(__m256i __a, __m256i __b)
Subtracts 8-bit integers from corresponding bytes of two 256-bit integer vectors using unsigned satur...
static __inline__ __m128 __DEFAULT_FN_ATTRS128_CONSTEXPR _mm_broadcastss_ps(__m128 __X)
Broadcasts the 32-bit floating-point value from the low element of the 128-bit vector of [4 x float] ...
static __inline__ void __DEFAULT_FN_ATTRS128 _mm_maskstore_epi32(int *__X, __m128i __M, __m128i __Y)
Conditionally stores four 32-bit integer elements from the 128-bit vector of [4 x i32] in __Y to memo...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_subs_epi16(__m256i __a, __m256i __b)
Subtracts 16-bit integers from corresponding elements of two 256-bit vectors of [16 x i16] using sign...
static __inline__ __m256i __DEFAULT_FN_ATTRS256 _mm256_sra_epi32(__m256i __a, __m128i __count)
Shifts each 32-bit element of the 256-bit vector of [8 x i32] in __a right by the number of bits give...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_cmpeq_epi8(__m256i __a, __m256i __b)
Compares corresponding bytes in the 256-bit integer vectors in __a and __b for equality and returns t...
static __inline__ __m128i __DEFAULT_FN_ATTRS128_CONSTEXPR _mm_srlv_epi64(__m128i __X, __m128i __Y)
Shifts each 64-bit element of the 128-bit vector of [2 x i64] in __X right by the number of bits give...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_srlv_epi32(__m256i __X, __m256i __Y)
Shifts each 32-bit element of the 256-bit vector of [8 x i32] in __X right by the number of bits give...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_abs_epi8(__m256i __a)
Computes the absolute value of each signed byte in the 256-bit integer vector __a and returns each va...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_cvtepu16_epi64(__m128i __V)
Zero-extends 16-bit elements from the lower half of the 128-bit vector of [8 x i16] in __V and return...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_unpacklo_epi32(__m256i __a, __m256i __b)
Unpacks and interleaves 32-bit integers from parts of the 256-bit vectors of [8 x i32] in __a and __b...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_cvtepi8_epi64(__m128i __V)
Sign-extends the first four bytes from the 128-bit integer vector in __V and returns the 64-bit value...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_hadds_epi16(__m256i __a, __m256i __b)
Horizontally adds the adjacent pairs of 16-bit integers from two 256-bit vectors of [16 x i16] using ...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_srli_epi32(__m256i __a, int __count)
Shifts each 32-bit element of the 256-bit vector of [8 x i32] in __a right by __count bits,...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_add_epi64(__m256i __a, __m256i __b)
Adds 64-bit integers from corresponding elements of two 256-bit vectors of [4 x i64] and returns the ...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_cvtepi16_epi64(__m128i __V)
Sign-extends 16-bit elements from the lower half of the 128-bit vector of [8 x i16] in __V and return...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_unpackhi_epi8(__m256i __a, __m256i __b)
Unpacks and interleaves 8-bit integers from parts of the 256-bit integer vectors in __a and __b to fo...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_sub_epi16(__m256i __a, __m256i __b)
Subtracts 16-bit integers from corresponding elements of two 256-bit vectors of [16 x i16].
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_and_si256(__m256i __a, __m256i __b)
Computes the bitwise AND of the 256-bit integer vectors in __a and __b.
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_slli_epi32(__m256i __a, int __count)
Shifts each 32-bit element of the 256-bit vector of [8 x i32] in __a left by __count bits,...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_broadcastsi128_si256(__m128i __X)
Broadcasts the 128-bit integer data from __X to both the lower and upper halves of the 256-bit result...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_srai_epi32(__m256i __a, int __count)
Shifts each 32-bit element of the 256-bit vector of [8 x i32] in __a right by __count bits,...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_add_epi32(__m256i __a, __m256i __b)
Adds 32-bit integers from corresponding elements of two 256-bit vectors of [8 x i32] and returns the ...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_packus_epi16(__m256i __a, __m256i __b)
Converts elements from two 256-bit vectors of [16 x i16] to 8-bit integers using unsigned saturation,...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_abs_epi16(__m256i __a)
Computes the absolute value of each signed 16-bit element in the 256-bit vector of [16 x i16] in __a ...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_max_epu8(__m256i __a, __m256i __b)
Compares the corresponding unsigned bytes in the two 256-bit integer vectors in __a and __b and retur...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_add_epi16(__m256i __a, __m256i __b)
Adds 16-bit integers from corresponding elements of two 256-bit vectors of [16 x i16] and returns the...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_or_si256(__m256i __a, __m256i __b)
Computes the bitwise OR of the 256-bit integer vectors in __a and __b.
static __inline__ __m256d __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_broadcastsd_pd(__m128d __X)
Broadcasts the 64-bit floating-point value from the low element of the 128-bit vector of [2 x double]...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_min_epu8(__m256i __a, __m256i __b)
Compares the corresponding unsigned bytes in the two 256-bit integer vectors in __a and __b and retur...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_cvtepu16_epi32(__m128i __V)
Zero-extends 16-bit elements from the 128-bit vector of [8 x i16] in __V and returns the 32-bit value...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_max_epu32(__m256i __a, __m256i __b)
Compares the corresponding unsigned 32-bit integers in the two 256-bit vectors of [8 x i32] in __a an...
static __inline__ __m128i __DEFAULT_FN_ATTRS128_CONSTEXPR _mm_sllv_epi64(__m128i __X, __m128i __Y)
Shifts each 64-bit element of the 128-bit vector of [2 x i64] in __X left by the number of bits given...
static __inline__ __m128i __DEFAULT_FN_ATTRS128_CONSTEXPR _mm_sllv_epi32(__m128i __X, __m128i __Y)
Shifts each 32-bit element of the 128-bit vector of [4 x i32] in __X left by the number of bits given...
static __inline__ __m256i __DEFAULT_FN_ATTRS256 _mm256_srl_epi64(__m256i __a, __m128i __count)
Shifts each 64-bit element of the 256-bit vector of [4 x i64] in __a right by the number of bits give...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_sub_epi8(__m256i __a, __m256i __b)
Subtracts 8-bit integers from corresponding bytes of two 256-bit integer vectors.
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_hadd_epi16(__m256i __a, __m256i __b)
Horizontally adds the adjacent pairs of 16-bit integers from two 256-bit vectors of [16 x i16] and re...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_min_epi16(__m256i __a, __m256i __b)
Compares the corresponding signed 16-bit integers in the two 256-bit vectors of [16 x i16] in __a and...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_avg_epu16(__m256i __a, __m256i __b)
Computes the averages of the corresponding unsigned 16-bit integers in the two 256-bit vectors of [16...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_xor_si256(__m256i __a, __m256i __b)
Computes the bitwise XOR of the 256-bit integer vectors in __a and __b.
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_min_epu32(__m256i __a, __m256i __b)
Compares the corresponding unsigned 32-bit integers in the two 256-bit vectors of [8 x i32] in __a an...
static __inline__ __m256i __DEFAULT_FN_ATTRS256 _mm256_sra_epi16(__m256i __a, __m128i __count)
Shifts each 16-bit element of the 256-bit vector of [16 x i16] in __a right by the number of bits giv...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_cvtepi16_epi32(__m128i __V)
Sign-extends 16-bit elements from the 128-bit vector of [8 x i16] in __V and returns the 32-bit value...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_adds_epu16(__m256i __a, __m256i __b)
Adds 16-bit integers from corresponding elements of two 256-bit vectors of [16 x i16] using unsigned ...
static __inline__ __m128i __DEFAULT_FN_ATTRS128 _mm_maskload_epi64(long long const *__X, __m128i __M)
Conditionally loads two 64-bit integer elements from memory __X, if the most significant bit of the c...
static __inline__ __m256i __DEFAULT_FN_ATTRS256 _mm256_sign_epi16(__m256i __a, __m256i __b)
Sets each element of the result to the corresponding element of the 256-bit vector of [16 x i16] in _...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_sub_epi64(__m256i __a, __m256i __b)
Subtracts 64-bit integers from corresponding elements of two 256-bit vectors of [4 x i64].
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_madd_epi16(__m256i __a, __m256i __b)
Multiplies corresponding 16-bit elements of two 256-bit vectors of [16 x i16], forming 32-bit interme...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_max_epi8(__m256i __a, __m256i __b)
Compares the corresponding signed bytes in the two 256-bit integer vectors in __a and __b and returns...
static __inline__ __m128i __DEFAULT_FN_ATTRS128_CONSTEXPR _mm_broadcastd_epi32(__m128i __X)
Broadcasts the low element from the 128-bit vector of [4 x i32] in __X to all elements of the result'...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_broadcastd_epi32(__m128i __X)
Broadcasts the low element from the 128-bit vector of [4 x i32] in __X to all elements of the result'...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_unpackhi_epi16(__m256i __a, __m256i __b)
Unpacks and interleaves 16-bit integers from parts of the 256-bit vectors of [16 x i16] in __a and __...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_srlv_epi64(__m256i __X, __m256i __Y)
Shifts each 64-bit element of the 256-bit vector of [4 x i64] in __X right by the number of bits give...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_subs_epi8(__m256i __a, __m256i __b)
Subtracts 8-bit integers from corresponding bytes of two 256-bit integer vectors using signed saturat...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_packs_epi32(__m256i __a, __m256i __b)
Converts the elements of two 256-bit vectors of [8 x i32] to 16-bit integers using signed saturation,...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_cmpgt_epi64(__m256i __a, __m256i __b)
Compares corresponding signed elements in the 256-bit vectors of [4 x i64] in __a and __b for greater...
static __inline__ __m256i __DEFAULT_FN_ATTRS256 _mm256_sad_epu8(__m256i __a, __m256i __b)
Computes four sum of absolute difference (SAD) operations on sets of eight unsigned 8-bit integers fr...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_adds_epi8(__m256i __a, __m256i __b)
Adds 8-bit integers from corresponding bytes of two 256-bit integer vectors using signed saturation,...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_sllv_epi64(__m256i __X, __m256i __Y)
Shifts each 64-bit element of the 256-bit vector of [4 x i64] in __X left by the number of bits given...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_slli_epi16(__m256i __a, int __count)
Shifts each 16-bit element of the 256-bit vector of [16 x i16] in __a left by __count bits,...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_cmpeq_epi16(__m256i __a, __m256i __b)
Compares corresponding elements in the 256-bit vectors of [16 x i16] in __a and __b for equality and ...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_packs_epi16(__m256i __a, __m256i __b)
Converts the elements of two 256-bit vectors of [16 x i16] to 8-bit integers using signed saturation,...
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR _mm256_avg_epu8(__m256i __a, __m256i __b)
Computes the averages of the corresponding unsigned bytes in the two 256-bit integer vectors in __a a...
static __inline__ void int __a
__inline unsigned int unsigned int __Y