Lines Matching defs:__a
97 /// vector \a __a and returns each value in the corresponding byte of
104 /// \param __a
108 _mm256_abs_epi8(__m256i __a)
110 return (__m256i)__builtin_elementwise_abs((__v32qs)__a);
114 /// vector of [16 x i16] in \a __a and returns each value in the
121 /// \param __a
125 _mm256_abs_epi16(__m256i __a)
127 return (__m256i)__builtin_elementwise_abs((__v16hi)__a);
131 /// vector of [8 x i32] in \a __a and returns each value in the
138 /// \param __a
142 _mm256_abs_epi32(__m256i __a)
144 return (__m256i)__builtin_elementwise_abs((__v8si)__a);
154 /// result[7+k:k] := SATURATE8(__a[15+j:j])
156 /// result[135+k:128+k] := SATURATE8(__a[143+j:128+j])
165 /// \param __a
173 _mm256_packs_epi16(__m256i __a, __m256i __b)
175 return (__m256i)__builtin_ia32_packsswb256((__v16hi)__a, (__v16hi)__b);
186 /// result[15+k:k] := SATURATE16(__a[31+j:j])
188 /// result[143+k:128+k] := SATURATE16(__a[159+j:128+j])
197 /// \param __a
205 _mm256_packs_epi32(__m256i __a, __m256i __b)
207 return (__m256i)__builtin_ia32_packssdw256((__v8si)__a, (__v8si)__b);
217 /// result[7+k:k] := SATURATE8U(__a[15+j:j])
219 /// result[135+k:128+k] := SATURATE8U(__a[143+j:128+j])
228 /// \param __a
236 _mm256_packus_epi16(__m256i __a, __m256i __b)
238 return (__m256i)__builtin_ia32_packuswb256((__v16hi)__a, (__v16hi)__b);
281 /// \param __a
287 _mm256_add_epi8(__m256i __a, __m256i __b)
289 return (__m256i)((__v32qu)__a + (__v32qu)__b);
300 /// \param __a
306 _mm256_add_epi16(__m256i __a, __m256i __b)
308 return (__m256i)((__v16hu)__a + (__v16hu)__b);
319 /// \param __a
325 _mm256_add_epi32(__m256i __a, __m256i __b)
327 return (__m256i)((__v8su)__a + (__v8su)__b);
338 /// \param __a
344 _mm256_add_epi64(__m256i __a, __m256i __b)
346 return (__m256i)((__v4du)__a + (__v4du)__b);
357 /// \param __a
363 _mm256_adds_epi8(__m256i __a, __m256i __b)
365 return (__m256i)__builtin_elementwise_add_sat((__v32qs)__a, (__v32qs)__b);
375 /// \param __a
381 _mm256_adds_epi16(__m256i __a, __m256i __b)
383 return (__m256i)__builtin_elementwise_add_sat((__v16hi)__a, (__v16hi)__b);
394 /// \param __a
400 _mm256_adds_epu8(__m256i __a, __m256i __b)
402 return (__m256i)__builtin_elementwise_add_sat((__v32qu)__a, (__v32qu)__b);
412 /// \param __a
418 _mm256_adds_epu16(__m256i __a, __m256i __b)
420 return (__m256i)__builtin_elementwise_add_sat((__v16hu)__a, (__v16hu)__b);
451 /// Computes the bitwise AND of the 256-bit integer vectors in \a __a and
458 /// \param __a
464 _mm256_and_si256(__m256i __a, __m256i __b)
466 return (__m256i)((__v4du)__a & (__v4du)__b);
470 /// the bitwise NOT of the 256-bit integer vector in \a __a.
476 /// \param __a
482 _mm256_andnot_si256(__m256i __a, __m256i __b)
484 return (__m256i)(~(__v4du)__a & (__v4du)__b);
488 /// 256-bit integer vectors in \a __a and \a __b and returns each
494 /// result[j+7:j] := (__a[j+7:j] + __b[j+7:j] + 1) >> 1
502 /// \param __a
508 _mm256_avg_epu8(__m256i __a, __m256i __b)
510 return (__m256i)__builtin_ia32_pavgb256((__v32qi)__a, (__v32qi)__b);
514 /// the two 256-bit vectors of [16 x i16] in \a __a and \a __b and returns
520 /// result[j+15:j] := (__a[j+15:j] + __b[j+15:j] + 1) >> 1
528 /// \param __a
534 _mm256_avg_epu16(__m256i __a, __m256i __b)
536 return (__m256i)__builtin_ia32_pavgw256((__v16hi)__a, (__v16hi)__b);
616 /// Compares corresponding bytes in the 256-bit integer vectors in \a __a and
623 /// result[j+7:j] := (__a[j+7:j] == __b[j+7:j]) ? 0xFF : 0
631 /// \param __a
637 _mm256_cmpeq_epi8(__m256i __a, __m256i __b)
639 return (__m256i)((__v32qi)__a == (__v32qi)__b);
643 /// \a __a and \a __b for equality and returns the outcomes in the
649 /// result[j+15:j] := (__a[j+15:j] == __b[j+15:j]) ? 0xFFFF : 0
657 /// \param __a
663 _mm256_cmpeq_epi16(__m256i __a, __m256i __b)
665 return (__m256i)((__v16hi)__a == (__v16hi)__b);
669 /// \a __a and \a __b for equality and returns the outcomes in the
675 /// result[j+31:j] := (__a[j+31:j] == __b[j+31:j]) ? 0xFFFFFFFF : 0
683 /// \param __a
689 _mm256_cmpeq_epi32(__m256i __a, __m256i __b)
691 return (__m256i)((__v8si)__a == (__v8si)__b);
695 /// \a __a and \a __b for equality and returns the outcomes in the
701 /// result[j+63:j] := (__a[j+63:j] == __b[j+63:j]) ? 0xFFFFFFFFFFFFFFFF : 0
709 /// \param __a
715 _mm256_cmpeq_epi64(__m256i __a, __m256i __b)
717 return (__m256i)((__v4di)__a == (__v4di)__b);
721 /// \a __a and \a __b for greater-than and returns the outcomes in the
727 /// result[j+7:j] := (__a[j+7:j] > __b[j+7:j]) ? 0xFF : 0
735 /// \param __a
741 _mm256_cmpgt_epi8(__m256i __a, __m256i __b)
745 return (__m256i)((__v32qs)__a > (__v32qs)__b);
749 /// [16 x i16] in \a __a and \a __b for greater-than and returns the
755 /// result[j+15:j] := (__a[j+15:j] > __b[j+15:j]) ? 0xFFFF : 0
763 /// \param __a
769 _mm256_cmpgt_epi16(__m256i __a, __m256i __b)
771 return (__m256i)((__v16hi)__a > (__v16hi)__b);
775 /// [8 x i32] in \a __a and \a __b for greater-than and returns the
781 /// result[j+31:j] := (__a[j+31:j] > __b[j+31:j]) ? 0xFFFFFFFF : 0
789 /// \param __a
795 _mm256_cmpgt_epi32(__m256i __a, __m256i __b)
797 return (__m256i)((__v8si)__a > (__v8si)__b);
801 /// [4 x i64] in \a __a and \a __b for greater-than and returns the
807 /// result[j+63:j] := (__a[j+63:j] > __b[j+63:j]) ? 0xFFFFFFFFFFFFFFFF : 0
815 /// \param __a
821 _mm256_cmpgt_epi64(__m256i __a, __m256i __b)
823 return (__m256i)((__v4di)__a > (__v4di)__b);
829 /// \a __a are returned in the lower 64 bits of each 128-bit half of the
836 /// result[j+15:j] := __a[j+15:j] + __a[j+31:j+16]
837 /// result[j+31:j+16] := __a[j+47:j+32] + __a[j+63:j+48]
838 /// result[j+47:j+32] := __a[j+79:j+64] + __a[j+95:j+80]
839 /// result[j+63:j+48] := __a[j+111:j+96] + __a[j+127:j+112]
851 /// \param __a
857 _mm256_hadd_epi16(__m256i __a, __m256i __b)
859 return (__m256i)__builtin_ia32_phaddw256((__v16hi)__a, (__v16hi)__b);
864 /// element of the [8 x i32] result (overflow is ignored). Sums from \a __a
872 /// result[j+31:j] := __a[j+31:j] + __a[j+63:j+32]
873 /// result[j+63:j+32] := __a[j+95:j+64] + __a[j+127:j+96]
883 /// \param __a
889 _mm256_hadd_epi32(__m256i __a, __m256i __b)
891 return (__m256i)__builtin_ia32_phaddd256((__v8si)__a, (__v8si)__b);
896 /// an element of the [16 x i16] result. Sums from \a __a are returned in
903 /// result[j+15:j] := SATURATE16(__a[j+15:j] + __a[j+31:j+16])
904 /// result[j+31:j+16] := SATURATE16(__a[j+47:j+32] + __a[j+63:j+48])
905 /// result[j+47:j+32] := SATURATE16(__a[j+79:j+64] + __a[j+95:j+80])
906 /// result[j+63:j+48] := SATURATE16(__a[j+111:j+96] + __a[j+127:j+112])
918 /// \param __a
924 _mm256_hadds_epi16(__m256i __a, __m256i __b)
926 return (__m256i)__builtin_ia32_phaddsw256((__v16hi)__a, (__v16hi)__b);
932 /// Differences from \a __a are returned in the lower 64 bits of each
939 /// result[j+15:j] := __a[j+15:j] - __a[j+31:j+16]
940 /// result[j+31:j+16] := __a[j+47:j+32] - __a[j+63:j+48]
941 /// result[j+47:j+32] := __a[j+79:j+64] - __a[j+95:j+80]
942 /// result[j+63:j+48] := __a[j+111:j+96] - __a[j+127:j+112]
954 /// \param __a
960 _mm256_hsub_epi16(__m256i __a, __m256i __b)
962 return (__m256i)__builtin_ia32_phsubw256((__v16hi)__a, (__v16hi)__b);
968 /// from \a __a are returned in the lower 64 bits of each 128-bit half of
975 /// result[j+31:j] := __a[j+31:j] - __a[j+63:j+32]
976 /// result[j+63:j+32] := __a[j+95:j+64] - __a[j+127:j+96]
986 /// \param __a
992 _mm256_hsub_epi32(__m256i __a, __m256i __b)
994 return (__m256i)__builtin_ia32_phsubd256((__v8si)__a, (__v8si)__b);
999 /// an element of the [16 x i16] result. Differences from \a __a are
1007 /// result[j+15:j] := SATURATE16(__a[j+15:j] - __a[j+31:j+16])
1008 /// result[j+31:j+16] := SATURATE16(__a[j+47:j+32] - __a[j+63:j+48])
1009 /// result[j+47:j+32] := SATURATE16(__a[j+79:j+64] - __a[j+95:j+80])
1010 /// result[j+63:j+48] := SATURATE16(__a[j+111:j+96] - __a[j+127:j+112])
1022 /// \param __a
1028 _mm256_hsubs_epi16(__m256i __a, __m256i __b)
1030 return (__m256i)__builtin_ia32_phsubsw256((__v16hi)__a, (__v16hi)__b);
1033 /// Multiplies each unsigned byte from the 256-bit integer vector in \a __a
1042 /// temp1 := __a[j+7:j] * __b[j+7:j]
1043 /// temp2 := __a[j+15:j+8] * __b[j+15:j+8]
1052 /// \param __a
1058 _mm256_maddubs_epi16(__m256i __a, __m256i __b)
1060 return (__m256i)__builtin_ia32_pmaddubsw256((__v32qi)__a, (__v32qi)__b);
1074 /// temp1 := __a[j+15:j] * __b[j+15:j]
1075 /// temp2 := __a[j+31:j+16] * __b[j+31:j+16]
1084 /// \param __a
1090 _mm256_madd_epi16(__m256i __a, __m256i __b)
1092 return (__m256i)__builtin_ia32_pmaddwd256((__v16hi)__a, (__v16hi)__b);
1096 /// in \a __a and \a __b and returns the larger of each pair in the
1103 /// \param __a
1109 _mm256_max_epi8(__m256i __a, __m256i __b)
1111 return (__m256i)__builtin_elementwise_max((__v32qs)__a, (__v32qs)__b);
1115 /// vectors of [16 x i16] in \a __a and \a __b and returns the larger of
1122 /// \param __a
1128 _mm256_max_epi16(__m256i __a, __m256i __b)
1130 return (__m256i)__builtin_elementwise_max((__v16hi)__a, (__v16hi)__b);
1134 /// vectors of [8 x i32] in \a __a and \a __b and returns the larger of
1141 /// \param __a
1147 _mm256_max_epi32(__m256i __a, __m256i __b)
1149 return (__m256i)__builtin_elementwise_max((__v8si)__a, (__v8si)__b);
1153 /// vectors in \a __a and \a __b and returns the larger of each pair in
1160 /// \param __a
1166 _mm256_max_epu8(__m256i __a, __m256i __b)
1168 return (__m256i)__builtin_elementwise_max((__v32qu)__a, (__v32qu)__b);
1172 /// vectors of [16 x i16] in \a __a and \a __b and returns the larger of
1179 /// \param __a
1185 _mm256_max_epu16(__m256i __a, __m256i __b)
1187 return (__m256i)__builtin_elementwise_max((__v16hu)__a, (__v16hu)__b);
1191 /// vectors of [8 x i32] in \a __a and \a __b and returns the larger of
1198 /// \param __a
1204 _mm256_max_epu32(__m256i __a, __m256i __b)
1206 return (__m256i)__builtin_elementwise_max((__v8su)__a, (__v8su)__b);
1210 /// in \a __a and \a __b and returns the smaller of each pair in the
1217 /// \param __a
1223 _mm256_min_epi8(__m256i __a, __m256i __b)
1225 return (__m256i)__builtin_elementwise_min((__v32qs)__a, (__v32qs)__b);
1229 /// vectors of [16 x i16] in \a __a and \a __b and returns the smaller of
1236 /// \param __a
1242 _mm256_min_epi16(__m256i __a, __m256i __b)
1244 return (__m256i)__builtin_elementwise_min((__v16hi)__a, (__v16hi)__b);
1248 /// vectors of [8 x i32] in \a __a and \a __b and returns the smaller of
1255 /// \param __a
1261 _mm256_min_epi32(__m256i __a, __m256i __b)
1263 return (__m256i)__builtin_elementwise_min((__v8si)__a, (__v8si)__b);
1267 /// vectors in \a __a and \a __b and returns the smaller of each pair in
1274 /// \param __a
1280 _mm256_min_epu8(__m256i __a, __m256i __b)
1282 return (__m256i)__builtin_elementwise_min((__v32qu)__a, (__v32qu)__b);
1286 /// vectors of [16 x i16] in \a __a and \a __b and returns the smaller of
1293 /// \param __a
1299 _mm256_min_epu16(__m256i __a, __m256i __b)
1301 return (__m256i)__builtin_elementwise_min((__v16hu)__a, (__v16hu)__b);
1305 /// vectors of [8 x i32] in \a __a and \a __b and returns the smaller of
1312 /// \param __a
1318 _mm256_min_epu32(__m256i __a, __m256i __b)
1320 return (__m256i)__builtin_elementwise_min((__v8su)__a, (__v8su)__b);
1324 /// in the 256-bit integer vector in \a __a and returns the result.
1329 /// result[i] := __a[j+7]
1337 /// \param __a
1341 _mm256_movemask_epi8(__m256i __a)
1343 return __builtin_ia32_pmovmskb256((__v32qi)__a);
1663 /// result[63:0] := __a[31:0] * __b[31:0]
1664 /// result[127:64] := __a[95:64] * __b[95:64]
1665 /// result[191:128] := __a[159:128] * __b[159:128]
1666 /// result[255:192] := __a[223:192] * __b[223:192]
1673 /// \param __a
1679 _mm256_mul_epi32(__m256i __a, __m256i __b)
1681 return (__m256i)__builtin_ia32_pmuldq256((__v8si)__a, (__v8si)__b);
1692 /// temp := ((__a[j+15:j] * __b[j+15:j]) >> 14) + 1
1700 /// \param __a
1706 _mm256_mulhrs_epi16(__m256i __a, __m256i __b)
1708 return (__m256i)__builtin_ia32_pmulhrsw256((__v16hi)__a, (__v16hi)__b);
1719 /// \param __a
1725 _mm256_mulhi_epu16(__m256i __a, __m256i __b)
1727 return (__m256i)__builtin_ia32_pmulhuw256((__v16hi)__a, (__v16hi)__b);
1738 /// \param __a
1744 _mm256_mulhi_epi16(__m256i __a, __m256i __b)
1746 return (__m256i)__builtin_ia32_pmulhw256((__v16hi)__a, (__v16hi)__b);
1757 /// \param __a
1763 _mm256_mullo_epi16(__m256i __a, __m256i __b)
1765 return (__m256i)((__v16hu)__a * (__v16hu)__b);
1776 /// \param __a
1782 _mm256_mullo_epi32 (__m256i __a, __m256i __b)
1784 return (__m256i)((__v8su)__a * (__v8su)__b);
1792 /// result[63:0] := __a[31:0] * __b[31:0]
1793 /// result[127:64] := __a[95:64] * __b[95:64]
1794 /// result[191:128] := __a[159:128] * __b[159:128]
1795 /// result[255:192] := __a[223:192] * __b[223:192]
1802 /// \param __a
1808 _mm256_mul_epu32(__m256i __a, __m256i __b)
1810 return __builtin_ia32_pmuludq256((__v8si)__a, (__v8si)__b);
1813 /// Computes the bitwise OR of the 256-bit integer vectors in \a __a and
1820 /// \param __a
1826 _mm256_or_si256(__m256i __a, __m256i __b)
1828 return (__m256i)((__v4du)__a | (__v4du)__b);
1832 /// unsigned 8-bit integers from the 256-bit integer vectors \a __a and
1835 /// One SAD result is computed for each set of eight bytes from \a __a and
1840 /// bytes of \a __a and \a __b, takes the absolute value of each difference,
1847 /// temp0 := ABS(__a[j+7:j] - __b[j+7:j])
1848 /// temp1 := ABS(__a[j+15:j+8] - __b[j+15:j+8])
1849 /// temp2 := ABS(__a[j+23:j+16] - __b[j+23:j+16])
1850 /// temp3 := ABS(__a[j+31:j+24] - __b[j+31:j+24])
1851 /// temp4 := ABS(__a[j+39:j+32] - __b[j+39:j+32])
1852 /// temp5 := ABS(__a[j+47:j+40] - __b[j+47:j+40])
1853 /// temp6 := ABS(__a[j+55:j+48] - __b[j+55:j+48])
1854 /// temp7 := ABS(__a[j+63:j+56] - __b[j+63:j+56])
1865 /// \param __a
1871 _mm256_sad_epu8(__m256i __a, __m256i __b)
1873 return __builtin_ia32_psadbw256((__v32qi)__a, (__v32qi)__b);
1876 /// Shuffles 8-bit integers in the 256-bit integer vector \a __a according
1891 /// result[j+7:j] := __a[k+7:k]
1900 /// \param __a
1906 /// control byte specify the index (within the same 128-bit half) of \a __a
1910 _mm256_shuffle_epi8(__m256i __a, __m256i __b)
1912 return (__m256i)__builtin_ia32_pshufb256((__v32qi)__a, (__v32qi)__b);
2022 /// integer vector in \a __a, the negative of that byte, or zero, depending
2031 /// \param __a
2037 _mm256_sign_epi8(__m256i __a, __m256i __b)
2039 return (__m256i)__builtin_ia32_psignb256((__v32qi)__a, (__v32qi)__b);
2043 /// 256-bit vector of [16 x i16] in \a __a, the negative of that element,
2052 /// \param __a
2058 _mm256_sign_epi16(__m256i __a, __m256i __b)
2060 return (__m256i)__builtin_ia32_psignw256((__v16hi)__a, (__v16hi)__b);
2064 /// 256-bit vector of [8 x i32] in \a __a, the negative of that element, or
2073 /// \param __a
2079 _mm256_sign_epi32(__m256i __a, __m256i __b)
2081 return (__m256i)__builtin_ia32_psignd256((__v8si)__a, (__v8si)__b);
2124 /// Shifts each 16-bit element of the 256-bit vector of [16 x i16] in \a __a
2132 /// \param __a
2138 _mm256_slli_epi16(__m256i __a, int __count)
2140 return (__m256i)__builtin_ia32_psllwi256((__v16hi)__a, __count);
2143 /// Shifts each 16-bit element of the 256-bit vector of [16 x i16] in \a __a
2152 /// \param __a
2159 _mm256_sll_epi16(__m256i __a, __m128i __count)
2161 return (__m256i)__builtin_ia32_psllw256((__v16hi)__a, (__v8hi)__count);
2164 /// Shifts each 32-bit element of the 256-bit vector of [8 x i32] in \a __a
2172 /// \param __a
2178 _mm256_slli_epi32(__m256i __a, int __count)
2180 return (__m256i)__builtin_ia32_pslldi256((__v8si)__a, __count);
2183 /// Shifts each 32-bit element of the 256-bit vector of [8 x i32] in \a __a
2192 /// \param __a
2199 _mm256_sll_epi32(__m256i __a, __m128i __count)
2201 return (__m256i)__builtin_ia32_pslld256((__v8si)__a, (__v4si)__count);
2204 /// Shifts each 64-bit element of the 256-bit vector of [4 x i64] in \a __a
2212 /// \param __a
2218 _mm256_slli_epi64(__m256i __a, int __count)
2220 return __builtin_ia32_psllqi256((__v4di)__a, __count);
2223 /// Shifts each 64-bit element of the 256-bit vector of [4 x i64] in \a __a
2232 /// \param __a
2239 _mm256_sll_epi64(__m256i __a, __m128i __count)
2241 return __builtin_ia32_psllq256((__v4di)__a, __count);
2244 /// Shifts each 16-bit element of the 256-bit vector of [16 x i16] in \a __a
2253 /// \param __a
2259 _mm256_srai_epi16(__m256i __a, int __count)
2261 return (__m256i)__builtin_ia32_psrawi256((__v16hi)__a, __count);
2264 /// Shifts each 16-bit element of the 256-bit vector of [16 x i16] in \a __a
2274 /// \param __a
2281 _mm256_sra_epi16(__m256i __a, __m128i __count)
2283 return (__m256i)__builtin_ia32_psraw256((__v16hi)__a, (__v8hi)__count);
2286 /// Shifts each 32-bit element of the 256-bit vector of [8 x i32] in \a __a
2295 /// \param __a
2301 _mm256_srai_epi32(__m256i __a, int __count)
2303 return (__m256i)__builtin_ia32_psradi256((__v8si)__a, __count);
2306 /// Shifts each 32-bit element of the 256-bit vector of [8 x i32] in \a __a
2316 /// \param __a
2323 _mm256_sra_epi32(__m256i __a, __m128i __count)
2325 return (__m256i)__builtin_ia32_psrad256((__v8si)__a, (__v4si)__count);
2368 /// Shifts each 16-bit element of the 256-bit vector of [16 x i16] in \a __a
2376 /// \param __a
2382 _mm256_srli_epi16(__m256i __a, int __count)
2384 return (__m256i)__builtin_ia32_psrlwi256((__v16hi)__a, __count);
2387 /// Shifts each 16-bit element of the 256-bit vector of [16 x i16] in \a __a
2396 /// \param __a
2403 _mm256_srl_epi16(__m256i __a, __m128i __count)
2405 return (__m256i)__builtin_ia32_psrlw256((__v16hi)__a, (__v8hi)__count);
2408 /// Shifts each 32-bit element of the 256-bit vector of [8 x i32] in \a __a
2416 /// \param __a
2422 _mm256_srli_epi32(__m256i __a, int __count)
2424 return (__m256i)__builtin_ia32_psrldi256((__v8si)__a, __count);
2427 /// Shifts each 32-bit element of the 256-bit vector of [8 x i32] in \a __a
2436 /// \param __a
2443 _mm256_srl_epi32(__m256i __a, __m128i __count)
2445 return (__m256i)__builtin_ia32_psrld256((__v8si)__a, (__v4si)__count);
2448 /// Shifts each 64-bit element of the 256-bit vector of [4 x i64] in \a __a
2456 /// \param __a
2462 _mm256_srli_epi64(__m256i __a, int __count)
2464 return __builtin_ia32_psrlqi256((__v4di)__a, __count);
2467 /// Shifts each 64-bit element of the 256-bit vector of [4 x i64] in \a __a
2476 /// \param __a
2483 _mm256_srl_epi64(__m256i __a, __m128i __count)
2485 return __builtin_ia32_psrlq256((__v4di)__a, __count);
2496 /// result[j+7:j] := __a[j+7:j] - __b[j+7:j]
2504 /// \param __a
2510 _mm256_sub_epi8(__m256i __a, __m256i __b)
2512 return (__m256i)((__v32qu)__a - (__v32qu)__b);
2523 /// result[j+15:j] := __a[j+15:j] - __b[j+15:j]
2531 /// \param __a
2537 _mm256_sub_epi16(__m256i __a, __m256i __b)
2539 return (__m256i)((__v16hu)__a - (__v16hu)__b);
2549 /// result[j+31:j] := __a[j+31:j] - __b[j+31:j]
2557 /// \param __a
2563 _mm256_sub_epi32(__m256i __a, __m256i __b)
2565 return (__m256i)((__v8su)__a - (__v8su)__b);
2575 /// result[j+63:j] := __a[j+63:j] - __b[j+63:j]
2583 /// \param __a
2589 _mm256_sub_epi64(__m256i __a, __m256i __b)
2591 return (__m256i)((__v4du)__a - (__v4du)__b);
2601 /// result[j+7:j] := SATURATE8(__a[j+7:j] - __b[j+7:j])
2609 /// \param __a
2615 _mm256_subs_epi8(__m256i __a, __m256i __b)
2617 return (__m256i)__builtin_elementwise_sub_sat((__v32qs)__a, (__v32qs)__b);
2627 /// result[j+7:j] := SATURATE16(__a[j+7:j] - __b[j+7:j])
2635 /// \param __a
2641 _mm256_subs_epi16(__m256i __a, __m256i __b)
2643 return (__m256i)__builtin_elementwise_sub_sat((__v16hi)__a, (__v16hi)__b);
2649 /// computes <c> result = __a - __b </c>.
2654 /// result[j+7:j] := SATURATE8U(__a[j+7:j] - __b[j+7:j])
2662 /// \param __a
2668 _mm256_subs_epu8(__m256i __a, __m256i __b)
2670 return (__m256i)__builtin_elementwise_sub_sat((__v32qu)__a, (__v32qu)__b);
2680 /// result[j+15:j] := SATURATE16U(__a[j+15:j] - __b[j+15:j])
2688 /// \param __a
2694 _mm256_subs_epu16(__m256i __a, __m256i __b)
2696 return (__m256i)__builtin_elementwise_sub_sat((__v16hu)__a, (__v16hu)__b);
2700 /// vectors in \a __a and \a __b to form the 256-bit result. Specifically,
2701 /// uses the upper 64 bits of each 128-bit half of \a __a and \a __b as
2705 /// result[7:0] := __a[71:64]
2707 /// result[23:16] := __a[79:72]
2711 /// result[135:128] := __a[199:192]
2720 /// \param __a
2728 _mm256_unpackhi_epi8(__m256i __a, __m256i __b)
2730 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);
2734 /// of [16 x i16] in \a __a and \a __b to return the resulting 256-bit
2736 /// 128-bit half of \a __a and \a __b as input; other bits in these
2740 /// result[15:0] := __a[79:64]
2742 /// result[47:32] := __a[95:80]
2746 /// result[143:128] := __a[211:196]
2755 /// \param __a
2763 _mm256_unpackhi_epi16(__m256i __a, __m256i __b)
2765 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);
2769 /// of [8 x i32] in \a __a and \a __b to return the resulting 256-bit vector
2771 /// of \a __a and \a __b as input; other bits in these parameters are
2775 /// result[31:0] := __a[95:64]
2777 /// result[95:64] := __a[127:96]
2779 /// result[159:128] := __a[223:192]
2781 /// result[223:192] := __a[255:224]
2789 /// \param __a
2797 _mm256_unpackhi_epi32(__m256i __a, __m256i __b)
2799 return (__m256i)__builtin_shufflevector((__v8si)__a, (__v8si)__b, 2, 8+2, 3, 8+3, 6, 8+6, 7, 8+7);
2803 /// of [4 x i64] in \a __a and \a __b to return the resulting 256-bit vector
2805 /// of \a __a and \a __b as input; other bits in these parameters are
2809 /// result[63:0] := __a[127:64]
2811 /// result[191:128] := __a[255:192]
2819 /// \param __a
2827 _mm256_unpackhi_epi64(__m256i __a, __m256i __b)
2829 return (__m256i)__builtin_shufflevector((__v4di)__a, (__v4di)__b, 1, 4+1, 3, 4+3);
2833 /// vectors in \a __a and \a __b to form the 256-bit result. Specifically,
2834 /// uses the lower 64 bits of each 128-bit half of \a __a and \a __b as
2838 /// result[7:0] := __a[7:0]
2840 /// result[23:16] := __a[15:8]
2844 /// result[135:128] := __a[135:128]
2853 /// \param __a
2861 _mm256_unpacklo_epi8(__m256i __a, __m256i __b)
2863 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);
2867 /// of [16 x i16] in \a __a and \a __b to return the resulting 256-bit
2869 /// 128-bit half of \a __a and \a __b as input; other bits in these
2873 /// result[15:0] := __a[15:0]
2875 /// result[47:32] := __a[31:16]
2879 /// result[143:128] := __a[143:128]
2888 /// \param __a
2896 _mm256_unpacklo_epi16(__m256i __a, __m256i __b)
2898 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);
2902 /// of [8 x i32] in \a __a and \a __b to return the resulting 256-bit vector
2904 /// of \a __a and \a __b as input; other bits in these parameters are
2908 /// result[31:0] := __a[31:0]
2910 /// result[95:64] := __a[63:32]
2912 /// result[159:128] := __a[159:128]
2914 /// result[223:192] := __a[191:160]
2922 /// \param __a
2930 _mm256_unpacklo_epi32(__m256i __a, __m256i __b)
2932 return (__m256i)__builtin_shufflevector((__v8si)__a, (__v8si)__b, 0, 8+0, 1, 8+1, 4, 8+4, 5, 8+5);
2936 /// of [4 x i64] in \a __a and \a __b to return the resulting 256-bit vector
2938 /// of \a __a and \a __b as input; other bits in these parameters are
2942 /// result[63:0] := __a[63:0]
2944 /// result[191:128] := __a[191:128]
2952 /// \param __a
2960 _mm256_unpacklo_epi64(__m256i __a, __m256i __b)
2962 return (__m256i)__builtin_shufflevector((__v4di)__a, (__v4di)__b, 0, 4+0, 2, 4+2);
2965 /// Computes the bitwise XOR of the 256-bit integer vectors in \a __a and
2972 /// \param __a
2978 _mm256_xor_si256(__m256i __a, __m256i __b)
2980 return (__m256i)((__v4du)__a ^ (__v4du)__b);
3019 /// 128-bit vector of [2 x double] in \a __a to both elements of the
3026 /// \param __a
3030 _mm_broadcastsd_pd(__m128d __a)
3032 return __builtin_shufflevector((__v2df)__a, (__v2df)__a, 0, 0);
3290 /// 256-bit vector of [8 x i32] in \a __a as specified by indexes in the
3297 /// result[j+31:j] := __a[k+31:k]
3305 /// \param __a
3309 /// \a __a.
3312 _mm256_permutevar8x32_epi32(__m256i __a, __m256i __b)
3314 return (__m256i)__builtin_ia32_permvarsi256((__v8si)__a, (__v8si)__b);
3348 /// the 256-bit vector of [8 x float] in \a __a as specified by indexes in
3355 /// result[j+31:j] := __a[k+31:k]
3363 /// \param __a
3367 /// \a __a.
3370 _mm256_permutevar8x32_ps(__m256 __a, __m256i __b)
3372 return (__m256)__builtin_ia32_permvarsf256((__v8sf)__a, (__v8si)__b);