Lines Matching defs:__b
155 /// result[71+k:64+k] := SATURATE8(__b[15+j:j])
157 /// result[199+k:192+k] := SATURATE8(__b[143+j:128+j])
168 /// \param __b
173 _mm256_packs_epi16(__m256i __a, __m256i __b)
175 return (__m256i)__builtin_ia32_packsswb256((__v16hi)__a, (__v16hi)__b);
187 /// result[79+k:64+k] := SATURATE16(__b[31+j:j])
189 /// result[207+k:192+k] := SATURATE16(__b[159+j:128+j])
200 /// \param __b
205 _mm256_packs_epi32(__m256i __a, __m256i __b)
207 return (__m256i)__builtin_ia32_packssdw256((__v8si)__a, (__v8si)__b);
218 /// result[71+k:64+k] := SATURATE8U(__b[15+j:j])
220 /// result[199+k:192+k] := SATURATE8U(__b[143+j:128+j])
231 /// \param __b
236 _mm256_packus_epi16(__m256i __a, __m256i __b)
238 return (__m256i)__builtin_ia32_packuswb256((__v16hi)__a, (__v16hi)__b);
283 /// \param __b
287 _mm256_add_epi8(__m256i __a, __m256i __b)
289 return (__m256i)((__v32qu)__a + (__v32qu)__b);
302 /// \param __b
306 _mm256_add_epi16(__m256i __a, __m256i __b)
308 return (__m256i)((__v16hu)__a + (__v16hu)__b);
321 /// \param __b
325 _mm256_add_epi32(__m256i __a, __m256i __b)
327 return (__m256i)((__v8su)__a + (__v8su)__b);
340 /// \param __b
344 _mm256_add_epi64(__m256i __a, __m256i __b)
346 return (__m256i)((__v4du)__a + (__v4du)__b);
359 /// \param __b
363 _mm256_adds_epi8(__m256i __a, __m256i __b)
365 return (__m256i)__builtin_elementwise_add_sat((__v32qs)__a, (__v32qs)__b);
377 /// \param __b
381 _mm256_adds_epi16(__m256i __a, __m256i __b)
383 return (__m256i)__builtin_elementwise_add_sat((__v16hi)__a, (__v16hi)__b);
396 /// \param __b
400 _mm256_adds_epu8(__m256i __a, __m256i __b)
402 return (__m256i)__builtin_elementwise_add_sat((__v32qu)__a, (__v32qu)__b);
414 /// \param __b
418 _mm256_adds_epu16(__m256i __a, __m256i __b)
420 return (__m256i)__builtin_elementwise_add_sat((__v16hu)__a, (__v16hu)__b);
452 /// \a __b.
460 /// \param __b
464 _mm256_and_si256(__m256i __a, __m256i __b)
466 return (__m256i)((__v4du)__a & (__v4du)__b);
469 /// Computes the bitwise AND of the 256-bit integer vector in \a __b with
478 /// \param __b
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
504 /// \param __b
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
530 /// \param __b
534 _mm256_avg_epu16(__m256i __a, __m256i __b)
536 return (__m256i)__builtin_ia32_pavgw256((__v16hi)__a, (__v16hi)__b);
617 /// \a __b for equality and returns the outcomes in the corresponding
623 /// result[j+7:j] := (__a[j+7:j] == __b[j+7:j]) ? 0xFF : 0
633 /// \param __b
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
659 /// \param __b
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
685 /// \param __b
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
711 /// \param __b
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
737 /// \param __b
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
765 /// \param __b
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
791 /// \param __b
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
817 /// \param __b
821 _mm256_cmpgt_epi64(__m256i __a, __m256i __b)
823 return (__m256i)((__v4di)__a > (__v4di)__b);
830 /// result; sums from \a __b are returned in the upper 64 bits of each
840 /// result[j+79:j+64] := __b[j+15:j] + __b[j+31:j+16]
841 /// result[j+95:j+80] := __b[j+47:j+32] + __b[j+63:j+48]
842 /// result[j+111:j+96] := __b[j+79:j+64] + __b[j+95:j+80]
843 /// result[j+127:j+112] := __b[j+111:j+96] + __b[j+127:j+112]
853 /// \param __b
857 _mm256_hadd_epi16(__m256i __a, __m256i __b)
859 return (__m256i)__builtin_ia32_phaddw256((__v16hi)__a, (__v16hi)__b);
866 /// sums from \a __b are returned in the upper 64 bits of each 128-bit half
874 /// result[j+95:j+64] := __b[j+31:j] + __b[j+63:j+32]
875 /// result[j+127:j+96] := __b[j+95:j+64] + __b[j+127:j+96]
885 /// \param __b
889 _mm256_hadd_epi32(__m256i __a, __m256i __b)
891 return (__m256i)__builtin_ia32_phaddd256((__v8si)__a, (__v8si)__b);
897 /// the lower 64 bits of each 128-bit half of the result; sums from \a __b
907 /// result[j+79:j+64] := SATURATE16(__b[j+15:j] + __b[j+31:j+16])
908 /// result[j+95:j+80] := SATURATE16(__b[j+47:j+32] + __b[j+63:j+48])
909 /// result[j+111:j+96] := SATURATE16(__b[j+79:j+64] + __b[j+95:j+80])
910 /// result[j+127:j+112] := SATURATE16(__b[j+111:j+96] + __b[j+127:j+112])
920 /// \param __b
924 _mm256_hadds_epi16(__m256i __a, __m256i __b)
926 return (__m256i)__builtin_ia32_phaddsw256((__v16hi)__a, (__v16hi)__b);
933 /// 128-bit half of the result; differences from \a __b are returned in the
943 /// result[j+79:j+64] := __b[j+15:j] - __b[j+31:j+16]
944 /// result[j+95:j+80] := __b[j+47:j+32] - __b[j+63:j+48]
945 /// result[j+111:j+96] := __b[j+79:j+64] - __b[j+95:j+80]
946 /// result[j+127:j+112] := __b[j+111:j+96] - __b[j+127:j+112]
956 /// \param __b
960 _mm256_hsub_epi16(__m256i __a, __m256i __b)
962 return (__m256i)__builtin_ia32_phsubw256((__v16hi)__a, (__v16hi)__b);
969 /// the result; differences from \a __b are returned in the upper 64 bits
977 /// result[j+95:j+64] := __b[j+31:j] - __b[j+63:j+32]
978 /// result[j+127:j+96] := __b[j+95:j+64] - __b[j+127:j+96]
988 /// \param __b
992 _mm256_hsub_epi32(__m256i __a, __m256i __b)
994 return (__m256i)__builtin_ia32_phsubd256((__v8si)__a, (__v8si)__b);
1001 /// differences from \a __b are returned in the upper 64 bits of each
1011 /// result[j+79:j+64] := SATURATE16(__b[j+15:j] - __b[j+31:j+16])
1012 /// result[j+95:j+80] := SATURATE16(__b[j+47:j+32] - __b[j+63:j+48])
1013 /// result[j+111:j+96] := SATURATE16(__b[j+79:j+64] - __b[j+95:j+80])
1014 /// result[j+127:j+112] := SATURATE16(__b[j+111:j+96] - __b[j+127:j+112])
1024 /// \param __b
1028 _mm256_hsubs_epi16(__m256i __a, __m256i __b)
1030 return (__m256i)__builtin_ia32_phsubsw256((__v16hi)__a, (__v16hi)__b);
1035 /// \a __b, forming signed 16-bit intermediate products. Adds adjacent
1042 /// temp1 := __a[j+7:j] * __b[j+7:j]
1043 /// temp2 := __a[j+15:j+8] * __b[j+15:j+8]
1054 /// \param __b
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]
1086 /// \param __b
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
1105 /// \param __b
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
1124 /// \param __b
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
1143 /// \param __b
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
1162 /// \param __b
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
1181 /// \param __b
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
1200 /// \param __b
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
1219 /// \param __b
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
1238 /// \param __b
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
1257 /// \param __b
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
1276 /// \param __b
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
1295 /// \param __b
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
1314 /// \param __b
1318 _mm256_min_epu32(__m256i __a, __m256i __b)
1320 return (__m256i)__builtin_elementwise_min((__v8su)__a, (__v8su)__b);
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]
1675 /// \param __b
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
1702 /// \param __b
1706 _mm256_mulhrs_epi16(__m256i __a, __m256i __b)
1708 return (__m256i)__builtin_ia32_pmulhrsw256((__v16hi)__a, (__v16hi)__b);
1721 /// \param __b
1725 _mm256_mulhi_epu16(__m256i __a, __m256i __b)
1727 return (__m256i)__builtin_ia32_pmulhuw256((__v16hi)__a, (__v16hi)__b);
1740 /// \param __b
1744 _mm256_mulhi_epi16(__m256i __a, __m256i __b)
1746 return (__m256i)__builtin_ia32_pmulhw256((__v16hi)__a, (__v16hi)__b);
1759 /// \param __b
1763 _mm256_mullo_epi16(__m256i __a, __m256i __b)
1765 return (__m256i)((__v16hu)__a * (__v16hu)__b);
1778 /// \param __b
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]
1804 /// \param __b
1808 _mm256_mul_epu32(__m256i __a, __m256i __b)
1810 return __builtin_ia32_pmuludq256((__v8si)__a, (__v8si)__b);
1814 /// \a __b.
1822 /// \param __b
1826 _mm256_or_si256(__m256i __a, __m256i __b)
1828 return (__m256i)((__v4du)__a | (__v4du)__b);
1833 /// \a __b.
1836 /// eight bytes from \a __b. The zero-extended SAD value is returned in the
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])
1867 /// \param __b
1871 _mm256_sad_epu8(__m256i __a, __m256i __b)
1873 return __builtin_ia32_psadbw256((__v32qi)__a, (__v32qi)__b);
1877 /// to control information in the 256-bit integer vector \a __b, and
1884 /// IF __b[j+7] == 1
1887 /// k := __b[j+3:j] * 8
1902 /// \param __b
1910 _mm256_shuffle_epi8(__m256i __a, __m256i __b)
1912 return (__m256i)__builtin_ia32_pshufb256((__v32qi)__a, (__v32qi)__b);
2024 /// \a __b is greater than zero, less than zero, or equal to zero,
2033 /// \param __b
2037 _mm256_sign_epi8(__m256i __a, __m256i __b)
2039 return (__m256i)__builtin_ia32_psignb256((__v32qi)__a, (__v32qi)__b);
2045 /// vector of [16 x i16] in \a __b is greater than zero, less than zero, or
2054 /// \param __b
2058 _mm256_sign_epi16(__m256i __a, __m256i __b)
2060 return (__m256i)__builtin_ia32_psignw256((__v16hi)__a, (__v16hi)__b);
2066 /// vector of [8 x i32] in \a __b is greater than zero, less than zero, or
2075 /// \param __b
2079 _mm256_sign_epi32(__m256i __a, __m256i __b)
2081 return (__m256i)__builtin_ia32_psignd256((__v8si)__a, (__v8si)__b);
2496 /// result[j+7:j] := __a[j+7:j] - __b[j+7:j]
2506 /// \param __b
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]
2533 /// \param __b
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]
2559 /// \param __b
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]
2585 /// \param __b
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])
2611 /// \param __b
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])
2637 /// \param __b
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])
2664 /// \param __b
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])
2690 /// \param __b
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
2706 /// result[15:8] := __b[71:64]
2708 /// result[31:24] := __b[79:72]
2710 /// result[127:120] := __b[127:120]
2713 /// result[255:248] := __b[255:248]
2723 /// \param __b
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
2741 /// result[31:16] := __b[79:64]
2743 /// result[63:48] := __b[95:80]
2745 /// result[127:112] := __b[127:112]
2748 /// result[255:240] := __b[255:240]
2758 /// \param __b
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
2776 /// result[63:32] := __b[95:64]
2778 /// result[127:96] := __b[127:96]
2780 /// result[191:160] := __b[223:192]
2782 /// result[255:224] := __b[255:224]
2792 /// \param __b
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
2810 /// result[127:64] := __b[127:64]
2812 /// result[255:192] := __b[255:192]
2822 /// \param __b
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
2839 /// result[15:8] := __b[7:0]
2841 /// result[31:24] := __b[15:8]
2843 /// result[127:120] := __b[63:56]
2846 /// result[255:248] := __b[191:184]
2856 /// \param __b
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
2874 /// result[31:16] := __b[15:0]
2876 /// result[63:48] := __b[31:16]
2878 /// result[127:112] := __b[63:48]
2881 /// result[255:239] := __b[191:176]
2891 /// \param __b
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
2909 /// result[63:32] := __b[31:0]
2911 /// result[127:96] := __b[63:32]
2913 /// result[191:160] := __b[159:128]
2915 /// result[255:224] := __b[191:190]
2925 /// \param __b
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
2943 /// result[127:64] := __b[63:0]
2945 /// result[255:192] := __b[191:128]
2955 /// \param __b
2960 _mm256_unpacklo_epi64(__m256i __a, __m256i __b)
2962 return (__m256i)__builtin_shufflevector((__v4di)__a, (__v4di)__b, 0, 4+0, 2, 4+2);
2966 /// \a __b.
2974 /// \param __b
2978 _mm256_xor_si256(__m256i __a, __m256i __b)
2980 return (__m256i)((__v4du)__a ^ (__v4du)__b);
3291 /// elements of the 256-bit vector of [8 x i32] in \a __b.
3296 /// k := __b[j+2:j] * 32
3307 /// \param __b
3312 _mm256_permutevar8x32_epi32(__m256i __a, __m256i __b)
3314 return (__m256i)__builtin_ia32_permvarsi256((__v8si)__a, (__v8si)__b);
3349 /// the elements of the 256-bit vector of [8 x i32] in \a __b.
3354 /// k := __b[j+2:j] * 32
3365 /// \param __b
3370 _mm256_permutevar8x32_ps(__m256 __a, __m256i __b)
3372 return (__m256)__builtin_ia32_permvarsf256((__v8sf)__a, (__v8si)__b);