xopintrin.h revision 1.1
1 1.1 mrg /* Copyright (C) 2007, 2008, 2009, 2010 Free Software Foundation, Inc. 2 1.1 mrg 3 1.1 mrg This file is part of GCC. 4 1.1 mrg 5 1.1 mrg GCC is free software; you can redistribute it and/or modify 6 1.1 mrg it under the terms of the GNU General Public License as published by 7 1.1 mrg the Free Software Foundation; either version 3, or (at your option) 8 1.1 mrg any later version. 9 1.1 mrg 10 1.1 mrg GCC is distributed in the hope that it will be useful, 11 1.1 mrg but WITHOUT ANY WARRANTY; without even the implied warranty of 12 1.1 mrg MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the 13 1.1 mrg GNU General Public License for more details. 14 1.1 mrg 15 1.1 mrg Under Section 7 of GPL version 3, you are granted additional 16 1.1 mrg permissions described in the GCC Runtime Library Exception, version 17 1.1 mrg 3.1, as published by the Free Software Foundation. 18 1.1 mrg 19 1.1 mrg You should have received a copy of the GNU General Public License and 20 1.1 mrg a copy of the GCC Runtime Library Exception along with this program; 21 1.1 mrg see the files COPYING3 and COPYING.RUNTIME respectively. If not, see 22 1.1 mrg <http://www.gnu.org/licenses/>. */ 23 1.1 mrg 24 1.1 mrg #ifndef _X86INTRIN_H_INCLUDED 25 1.1 mrg # error "Never use <xopintrin.h> directly; include <x86intrin.h> instead." 26 1.1 mrg #endif 27 1.1 mrg 28 1.1 mrg #ifndef _XOPMMINTRIN_H_INCLUDED 29 1.1 mrg #define _XOPMMINTRIN_H_INCLUDED 30 1.1 mrg 31 1.1 mrg #ifndef __XOP__ 32 1.1 mrg # error "XOP instruction set not enabled" 33 1.1 mrg #else 34 1.1 mrg 35 1.1 mrg #include <fma4intrin.h> 36 1.1 mrg 37 1.1 mrg /* Integer multiply/add intructions. */ 38 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 39 1.1 mrg _mm_maccs_epi16(__m128i __A, __m128i __B, __m128i __C) 40 1.1 mrg { 41 1.1 mrg return (__m128i) __builtin_ia32_vpmacssww ((__v8hi)__A,(__v8hi)__B, (__v8hi)__C); 42 1.1 mrg } 43 1.1 mrg 44 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 45 1.1 mrg _mm_macc_epi16(__m128i __A, __m128i __B, __m128i __C) 46 1.1 mrg { 47 1.1 mrg return (__m128i) __builtin_ia32_vpmacsww ((__v8hi)__A, (__v8hi)__B, (__v8hi)__C); 48 1.1 mrg } 49 1.1 mrg 50 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 51 1.1 mrg _mm_maccsd_epi16(__m128i __A, __m128i __B, __m128i __C) 52 1.1 mrg { 53 1.1 mrg return (__m128i) __builtin_ia32_vpmacsswd ((__v8hi)__A, (__v8hi)__B, (__v4si)__C); 54 1.1 mrg } 55 1.1 mrg 56 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 57 1.1 mrg _mm_maccd_epi16(__m128i __A, __m128i __B, __m128i __C) 58 1.1 mrg { 59 1.1 mrg return (__m128i) __builtin_ia32_vpmacswd ((__v8hi)__A, (__v8hi)__B, (__v4si)__C); 60 1.1 mrg } 61 1.1 mrg 62 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 63 1.1 mrg _mm_maccs_epi32(__m128i __A, __m128i __B, __m128i __C) 64 1.1 mrg { 65 1.1 mrg return (__m128i) __builtin_ia32_vpmacssdd ((__v4si)__A, (__v4si)__B, (__v4si)__C); 66 1.1 mrg } 67 1.1 mrg 68 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 69 1.1 mrg _mm_macc_epi32(__m128i __A, __m128i __B, __m128i __C) 70 1.1 mrg { 71 1.1 mrg return (__m128i) __builtin_ia32_vpmacsdd ((__v4si)__A, (__v4si)__B, (__v4si)__C); 72 1.1 mrg } 73 1.1 mrg 74 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 75 1.1 mrg _mm_maccslo_epi32(__m128i __A, __m128i __B, __m128i __C) 76 1.1 mrg { 77 1.1 mrg return (__m128i) __builtin_ia32_vpmacssdql ((__v4si)__A, (__v4si)__B, (__v2di)__C); 78 1.1 mrg } 79 1.1 mrg 80 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 81 1.1 mrg _mm_macclo_epi32(__m128i __A, __m128i __B, __m128i __C) 82 1.1 mrg { 83 1.1 mrg return (__m128i) __builtin_ia32_vpmacsdql ((__v4si)__A, (__v4si)__B, (__v2di)__C); 84 1.1 mrg } 85 1.1 mrg 86 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 87 1.1 mrg _mm_maccshi_epi32(__m128i __A, __m128i __B, __m128i __C) 88 1.1 mrg { 89 1.1 mrg return (__m128i) __builtin_ia32_vpmacssdqh ((__v4si)__A, (__v4si)__B, (__v2di)__C); 90 1.1 mrg } 91 1.1 mrg 92 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 93 1.1 mrg _mm_macchi_epi32(__m128i __A, __m128i __B, __m128i __C) 94 1.1 mrg { 95 1.1 mrg return (__m128i) __builtin_ia32_vpmacsdqh ((__v4si)__A, (__v4si)__B, (__v2di)__C); 96 1.1 mrg } 97 1.1 mrg 98 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 99 1.1 mrg _mm_maddsd_epi16(__m128i __A, __m128i __B, __m128i __C) 100 1.1 mrg { 101 1.1 mrg return (__m128i) __builtin_ia32_vpmadcsswd ((__v8hi)__A,(__v8hi)__B,(__v4si)__C); 102 1.1 mrg } 103 1.1 mrg 104 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 105 1.1 mrg _mm_maddd_epi16(__m128i __A, __m128i __B, __m128i __C) 106 1.1 mrg { 107 1.1 mrg return (__m128i) __builtin_ia32_vpmadcswd ((__v8hi)__A,(__v8hi)__B,(__v4si)__C); 108 1.1 mrg } 109 1.1 mrg 110 1.1 mrg /* Packed Integer Horizontal Add and Subtract */ 111 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 112 1.1 mrg _mm_haddw_epi8(__m128i __A) 113 1.1 mrg { 114 1.1 mrg return (__m128i) __builtin_ia32_vphaddbw ((__v16qi)__A); 115 1.1 mrg } 116 1.1 mrg 117 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 118 1.1 mrg _mm_haddd_epi8(__m128i __A) 119 1.1 mrg { 120 1.1 mrg return (__m128i) __builtin_ia32_vphaddbd ((__v16qi)__A); 121 1.1 mrg } 122 1.1 mrg 123 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 124 1.1 mrg _mm_haddq_epi8(__m128i __A) 125 1.1 mrg { 126 1.1 mrg return (__m128i) __builtin_ia32_vphaddbq ((__v16qi)__A); 127 1.1 mrg } 128 1.1 mrg 129 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 130 1.1 mrg _mm_haddd_epi16(__m128i __A) 131 1.1 mrg { 132 1.1 mrg return (__m128i) __builtin_ia32_vphaddwd ((__v8hi)__A); 133 1.1 mrg } 134 1.1 mrg 135 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 136 1.1 mrg _mm_haddq_epi16(__m128i __A) 137 1.1 mrg { 138 1.1 mrg return (__m128i) __builtin_ia32_vphaddwq ((__v8hi)__A); 139 1.1 mrg } 140 1.1 mrg 141 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 142 1.1 mrg _mm_haddq_epi32(__m128i __A) 143 1.1 mrg { 144 1.1 mrg return (__m128i) __builtin_ia32_vphadddq ((__v4si)__A); 145 1.1 mrg } 146 1.1 mrg 147 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 148 1.1 mrg _mm_haddw_epu8(__m128i __A) 149 1.1 mrg { 150 1.1 mrg return (__m128i) __builtin_ia32_vphaddubw ((__v16qi)__A); 151 1.1 mrg } 152 1.1 mrg 153 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 154 1.1 mrg _mm_haddd_epu8(__m128i __A) 155 1.1 mrg { 156 1.1 mrg return (__m128i) __builtin_ia32_vphaddubd ((__v16qi)__A); 157 1.1 mrg } 158 1.1 mrg 159 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 160 1.1 mrg _mm_haddq_epu8(__m128i __A) 161 1.1 mrg { 162 1.1 mrg return (__m128i) __builtin_ia32_vphaddubq ((__v16qi)__A); 163 1.1 mrg } 164 1.1 mrg 165 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 166 1.1 mrg _mm_haddd_epu16(__m128i __A) 167 1.1 mrg { 168 1.1 mrg return (__m128i) __builtin_ia32_vphadduwd ((__v8hi)__A); 169 1.1 mrg } 170 1.1 mrg 171 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 172 1.1 mrg _mm_haddq_epu16(__m128i __A) 173 1.1 mrg { 174 1.1 mrg return (__m128i) __builtin_ia32_vphadduwq ((__v8hi)__A); 175 1.1 mrg } 176 1.1 mrg 177 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 178 1.1 mrg _mm_haddq_epu32(__m128i __A) 179 1.1 mrg { 180 1.1 mrg return (__m128i) __builtin_ia32_vphaddudq ((__v4si)__A); 181 1.1 mrg } 182 1.1 mrg 183 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 184 1.1 mrg _mm_hsubw_epi8(__m128i __A) 185 1.1 mrg { 186 1.1 mrg return (__m128i) __builtin_ia32_vphsubbw ((__v16qi)__A); 187 1.1 mrg } 188 1.1 mrg 189 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 190 1.1 mrg _mm_hsubd_epi16(__m128i __A) 191 1.1 mrg { 192 1.1 mrg return (__m128i) __builtin_ia32_vphsubwd ((__v8hi)__A); 193 1.1 mrg } 194 1.1 mrg 195 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 196 1.1 mrg _mm_hsubq_epi32(__m128i __A) 197 1.1 mrg { 198 1.1 mrg return (__m128i) __builtin_ia32_vphsubdq ((__v4si)__A); 199 1.1 mrg } 200 1.1 mrg 201 1.1 mrg /* Vector conditional move and permute */ 202 1.1 mrg 203 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 204 1.1 mrg _mm_cmov_si128(__m128i __A, __m128i __B, __m128i __C) 205 1.1 mrg { 206 1.1 mrg return (__m128i) __builtin_ia32_vpcmov (__A, __B, __C); 207 1.1 mrg } 208 1.1 mrg 209 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 210 1.1 mrg _mm_perm_epi8(__m128i __A, __m128i __B, __m128i __C) 211 1.1 mrg { 212 1.1 mrg return (__m128i) __builtin_ia32_vpperm ((__v16qi)__A, (__v16qi)__B, (__v16qi)__C); 213 1.1 mrg } 214 1.1 mrg 215 1.1 mrg /* Packed Integer Rotates and Shifts 216 1.1 mrg Rotates - Non-Immediate form */ 217 1.1 mrg 218 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 219 1.1 mrg _mm_rot_epi8(__m128i __A, __m128i __B) 220 1.1 mrg { 221 1.1 mrg return (__m128i) __builtin_ia32_vprotb ((__v16qi)__A, (__v16qi)__B); 222 1.1 mrg } 223 1.1 mrg 224 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 225 1.1 mrg _mm_rot_epi16(__m128i __A, __m128i __B) 226 1.1 mrg { 227 1.1 mrg return (__m128i) __builtin_ia32_vprotw ((__v8hi)__A, (__v8hi)__B); 228 1.1 mrg } 229 1.1 mrg 230 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 231 1.1 mrg _mm_rot_epi32(__m128i __A, __m128i __B) 232 1.1 mrg { 233 1.1 mrg return (__m128i) __builtin_ia32_vprotd ((__v4si)__A, (__v4si)__B); 234 1.1 mrg } 235 1.1 mrg 236 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 237 1.1 mrg _mm_rot_epi64(__m128i __A, __m128i __B) 238 1.1 mrg { 239 1.1 mrg return (__m128i) __builtin_ia32_vprotq ((__v2di)__A, (__v2di)__B); 240 1.1 mrg } 241 1.1 mrg 242 1.1 mrg /* Rotates - Immediate form */ 243 1.1 mrg 244 1.1 mrg #ifdef __OPTIMIZE__ 245 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 246 1.1 mrg _mm_roti_epi8(__m128i __A, const int __B) 247 1.1 mrg { 248 1.1 mrg return (__m128i) __builtin_ia32_vprotbi ((__v16qi)__A, __B); 249 1.1 mrg } 250 1.1 mrg 251 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 252 1.1 mrg _mm_roti_epi16(__m128i __A, const int __B) 253 1.1 mrg { 254 1.1 mrg return (__m128i) __builtin_ia32_vprotwi ((__v8hi)__A, __B); 255 1.1 mrg } 256 1.1 mrg 257 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 258 1.1 mrg _mm_roti_epi32(__m128i __A, const int __B) 259 1.1 mrg { 260 1.1 mrg return (__m128i) __builtin_ia32_vprotdi ((__v4si)__A, __B); 261 1.1 mrg } 262 1.1 mrg 263 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 264 1.1 mrg _mm_roti_epi64(__m128i __A, const int __B) 265 1.1 mrg { 266 1.1 mrg return (__m128i) __builtin_ia32_vprotqi ((__v2di)__A, __B); 267 1.1 mrg } 268 1.1 mrg #else 269 1.1 mrg #define _mm_roti_epi8(A, N) \ 270 1.1 mrg ((__m128i) __builtin_ia32_vprotbi ((__v16qi)(__m128i)(A), (int)(N))) 271 1.1 mrg #define _mm_roti_epi16(A, N) \ 272 1.1 mrg ((__m128i) __builtin_ia32_vprotwi ((__v8hi)(__m128i)(A), (int)(N))) 273 1.1 mrg #define _mm_roti_epi32(A, N) \ 274 1.1 mrg ((__m128i) __builtin_ia32_vprotdi ((__v4si)(__m128i)(A), (int)(N))) 275 1.1 mrg #define _mm_roti_epi64(A, N) \ 276 1.1 mrg ((__m128i) __builtin_ia32_vprotqi ((__v2di)(__m128i)(A), (int)(N))) 277 1.1 mrg #endif 278 1.1 mrg 279 1.1 mrg /* Shifts */ 280 1.1 mrg 281 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 282 1.1 mrg _mm_shl_epi8(__m128i __A, __m128i __B) 283 1.1 mrg { 284 1.1 mrg return (__m128i) __builtin_ia32_vpshlb ((__v16qi)__A, (__v16qi)__B); 285 1.1 mrg } 286 1.1 mrg 287 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 288 1.1 mrg _mm_shl_epi16(__m128i __A, __m128i __B) 289 1.1 mrg { 290 1.1 mrg return (__m128i) __builtin_ia32_vpshlw ((__v8hi)__A, (__v8hi)__B); 291 1.1 mrg } 292 1.1 mrg 293 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 294 1.1 mrg _mm_shl_epi32(__m128i __A, __m128i __B) 295 1.1 mrg { 296 1.1 mrg return (__m128i) __builtin_ia32_vpshld ((__v4si)__A, (__v4si)__B); 297 1.1 mrg } 298 1.1 mrg 299 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 300 1.1 mrg _mm_shl_epi64(__m128i __A, __m128i __B) 301 1.1 mrg { 302 1.1 mrg return (__m128i) __builtin_ia32_vpshlq ((__v2di)__A, (__v2di)__B); 303 1.1 mrg } 304 1.1 mrg 305 1.1 mrg 306 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 307 1.1 mrg _mm_sha_epi8(__m128i __A, __m128i __B) 308 1.1 mrg { 309 1.1 mrg return (__m128i) __builtin_ia32_vpshab ((__v16qi)__A, (__v16qi)__B); 310 1.1 mrg } 311 1.1 mrg 312 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 313 1.1 mrg _mm_sha_epi16(__m128i __A, __m128i __B) 314 1.1 mrg { 315 1.1 mrg return (__m128i) __builtin_ia32_vpshaw ((__v8hi)__A, (__v8hi)__B); 316 1.1 mrg } 317 1.1 mrg 318 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 319 1.1 mrg _mm_sha_epi32(__m128i __A, __m128i __B) 320 1.1 mrg { 321 1.1 mrg return (__m128i) __builtin_ia32_vpshad ((__v4si)__A, (__v4si)__B); 322 1.1 mrg } 323 1.1 mrg 324 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 325 1.1 mrg _mm_sha_epi64(__m128i __A, __m128i __B) 326 1.1 mrg { 327 1.1 mrg return (__m128i) __builtin_ia32_vpshaq ((__v2di)__A, (__v2di)__B); 328 1.1 mrg } 329 1.1 mrg 330 1.1 mrg /* Compare and Predicate Generation 331 1.1 mrg pcom (integer, unsinged bytes) */ 332 1.1 mrg 333 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 334 1.1 mrg _mm_comlt_epu8(__m128i __A, __m128i __B) 335 1.1 mrg { 336 1.1 mrg return (__m128i) __builtin_ia32_vpcomltub ((__v16qi)__A, (__v16qi)__B); 337 1.1 mrg } 338 1.1 mrg 339 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 340 1.1 mrg _mm_comle_epu8(__m128i __A, __m128i __B) 341 1.1 mrg { 342 1.1 mrg return (__m128i) __builtin_ia32_vpcomleub ((__v16qi)__A, (__v16qi)__B); 343 1.1 mrg } 344 1.1 mrg 345 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 346 1.1 mrg _mm_comgt_epu8(__m128i __A, __m128i __B) 347 1.1 mrg { 348 1.1 mrg return (__m128i) __builtin_ia32_vpcomgtub ((__v16qi)__A, (__v16qi)__B); 349 1.1 mrg } 350 1.1 mrg 351 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 352 1.1 mrg _mm_comge_epu8(__m128i __A, __m128i __B) 353 1.1 mrg { 354 1.1 mrg return (__m128i) __builtin_ia32_vpcomgeub ((__v16qi)__A, (__v16qi)__B); 355 1.1 mrg } 356 1.1 mrg 357 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 358 1.1 mrg _mm_comeq_epu8(__m128i __A, __m128i __B) 359 1.1 mrg { 360 1.1 mrg return (__m128i) __builtin_ia32_vpcomequb ((__v16qi)__A, (__v16qi)__B); 361 1.1 mrg } 362 1.1 mrg 363 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 364 1.1 mrg _mm_comneq_epu8(__m128i __A, __m128i __B) 365 1.1 mrg { 366 1.1 mrg return (__m128i) __builtin_ia32_vpcomnequb ((__v16qi)__A, (__v16qi)__B); 367 1.1 mrg } 368 1.1 mrg 369 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 370 1.1 mrg _mm_comfalse_epu8(__m128i __A, __m128i __B) 371 1.1 mrg { 372 1.1 mrg return (__m128i) __builtin_ia32_vpcomfalseub ((__v16qi)__A, (__v16qi)__B); 373 1.1 mrg } 374 1.1 mrg 375 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 376 1.1 mrg _mm_comtrue_epu8(__m128i __A, __m128i __B) 377 1.1 mrg { 378 1.1 mrg return (__m128i) __builtin_ia32_vpcomtrueub ((__v16qi)__A, (__v16qi)__B); 379 1.1 mrg } 380 1.1 mrg 381 1.1 mrg /*pcom (integer, unsinged words) */ 382 1.1 mrg 383 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 384 1.1 mrg _mm_comlt_epu16(__m128i __A, __m128i __B) 385 1.1 mrg { 386 1.1 mrg return (__m128i) __builtin_ia32_vpcomltuw ((__v8hi)__A, (__v8hi)__B); 387 1.1 mrg } 388 1.1 mrg 389 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 390 1.1 mrg _mm_comle_epu16(__m128i __A, __m128i __B) 391 1.1 mrg { 392 1.1 mrg return (__m128i) __builtin_ia32_vpcomleuw ((__v8hi)__A, (__v8hi)__B); 393 1.1 mrg } 394 1.1 mrg 395 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 396 1.1 mrg _mm_comgt_epu16(__m128i __A, __m128i __B) 397 1.1 mrg { 398 1.1 mrg return (__m128i) __builtin_ia32_vpcomgtuw ((__v8hi)__A, (__v8hi)__B); 399 1.1 mrg } 400 1.1 mrg 401 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 402 1.1 mrg _mm_comge_epu16(__m128i __A, __m128i __B) 403 1.1 mrg { 404 1.1 mrg return (__m128i) __builtin_ia32_vpcomgeuw ((__v8hi)__A, (__v8hi)__B); 405 1.1 mrg } 406 1.1 mrg 407 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 408 1.1 mrg _mm_comeq_epu16(__m128i __A, __m128i __B) 409 1.1 mrg { 410 1.1 mrg return (__m128i) __builtin_ia32_vpcomequw ((__v8hi)__A, (__v8hi)__B); 411 1.1 mrg } 412 1.1 mrg 413 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 414 1.1 mrg _mm_comneq_epu16(__m128i __A, __m128i __B) 415 1.1 mrg { 416 1.1 mrg return (__m128i) __builtin_ia32_vpcomnequw ((__v8hi)__A, (__v8hi)__B); 417 1.1 mrg } 418 1.1 mrg 419 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 420 1.1 mrg _mm_comfalse_epu16(__m128i __A, __m128i __B) 421 1.1 mrg { 422 1.1 mrg return (__m128i) __builtin_ia32_vpcomfalseuw ((__v8hi)__A, (__v8hi)__B); 423 1.1 mrg } 424 1.1 mrg 425 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 426 1.1 mrg _mm_comtrue_epu16(__m128i __A, __m128i __B) 427 1.1 mrg { 428 1.1 mrg return (__m128i) __builtin_ia32_vpcomtrueuw ((__v8hi)__A, (__v8hi)__B); 429 1.1 mrg } 430 1.1 mrg 431 1.1 mrg /*pcom (integer, unsinged double words) */ 432 1.1 mrg 433 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 434 1.1 mrg _mm_comlt_epu32(__m128i __A, __m128i __B) 435 1.1 mrg { 436 1.1 mrg return (__m128i) __builtin_ia32_vpcomltud ((__v4si)__A, (__v4si)__B); 437 1.1 mrg } 438 1.1 mrg 439 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 440 1.1 mrg _mm_comle_epu32(__m128i __A, __m128i __B) 441 1.1 mrg { 442 1.1 mrg return (__m128i) __builtin_ia32_vpcomleud ((__v4si)__A, (__v4si)__B); 443 1.1 mrg } 444 1.1 mrg 445 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 446 1.1 mrg _mm_comgt_epu32(__m128i __A, __m128i __B) 447 1.1 mrg { 448 1.1 mrg return (__m128i) __builtin_ia32_vpcomgtud ((__v4si)__A, (__v4si)__B); 449 1.1 mrg } 450 1.1 mrg 451 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 452 1.1 mrg _mm_comge_epu32(__m128i __A, __m128i __B) 453 1.1 mrg { 454 1.1 mrg return (__m128i) __builtin_ia32_vpcomgeud ((__v4si)__A, (__v4si)__B); 455 1.1 mrg } 456 1.1 mrg 457 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 458 1.1 mrg _mm_comeq_epu32(__m128i __A, __m128i __B) 459 1.1 mrg { 460 1.1 mrg return (__m128i) __builtin_ia32_vpcomequd ((__v4si)__A, (__v4si)__B); 461 1.1 mrg } 462 1.1 mrg 463 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 464 1.1 mrg _mm_comneq_epu32(__m128i __A, __m128i __B) 465 1.1 mrg { 466 1.1 mrg return (__m128i) __builtin_ia32_vpcomnequd ((__v4si)__A, (__v4si)__B); 467 1.1 mrg } 468 1.1 mrg 469 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 470 1.1 mrg _mm_comfalse_epu32(__m128i __A, __m128i __B) 471 1.1 mrg { 472 1.1 mrg return (__m128i) __builtin_ia32_vpcomfalseud ((__v4si)__A, (__v4si)__B); 473 1.1 mrg } 474 1.1 mrg 475 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 476 1.1 mrg _mm_comtrue_epu32(__m128i __A, __m128i __B) 477 1.1 mrg { 478 1.1 mrg return (__m128i) __builtin_ia32_vpcomtrueud ((__v4si)__A, (__v4si)__B); 479 1.1 mrg } 480 1.1 mrg 481 1.1 mrg /*pcom (integer, unsinged quad words) */ 482 1.1 mrg 483 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 484 1.1 mrg _mm_comlt_epu64(__m128i __A, __m128i __B) 485 1.1 mrg { 486 1.1 mrg return (__m128i) __builtin_ia32_vpcomltuq ((__v2di)__A, (__v2di)__B); 487 1.1 mrg } 488 1.1 mrg 489 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 490 1.1 mrg _mm_comle_epu64(__m128i __A, __m128i __B) 491 1.1 mrg { 492 1.1 mrg return (__m128i) __builtin_ia32_vpcomleuq ((__v2di)__A, (__v2di)__B); 493 1.1 mrg } 494 1.1 mrg 495 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 496 1.1 mrg _mm_comgt_epu64(__m128i __A, __m128i __B) 497 1.1 mrg { 498 1.1 mrg return (__m128i) __builtin_ia32_vpcomgtuq ((__v2di)__A, (__v2di)__B); 499 1.1 mrg } 500 1.1 mrg 501 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 502 1.1 mrg _mm_comge_epu64(__m128i __A, __m128i __B) 503 1.1 mrg { 504 1.1 mrg return (__m128i) __builtin_ia32_vpcomgeuq ((__v2di)__A, (__v2di)__B); 505 1.1 mrg } 506 1.1 mrg 507 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 508 1.1 mrg _mm_comeq_epu64(__m128i __A, __m128i __B) 509 1.1 mrg { 510 1.1 mrg return (__m128i) __builtin_ia32_vpcomequq ((__v2di)__A, (__v2di)__B); 511 1.1 mrg } 512 1.1 mrg 513 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 514 1.1 mrg _mm_comneq_epu64(__m128i __A, __m128i __B) 515 1.1 mrg { 516 1.1 mrg return (__m128i) __builtin_ia32_vpcomnequq ((__v2di)__A, (__v2di)__B); 517 1.1 mrg } 518 1.1 mrg 519 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 520 1.1 mrg _mm_comfalse_epu64(__m128i __A, __m128i __B) 521 1.1 mrg { 522 1.1 mrg return (__m128i) __builtin_ia32_vpcomfalseuq ((__v2di)__A, (__v2di)__B); 523 1.1 mrg } 524 1.1 mrg 525 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 526 1.1 mrg _mm_comtrue_epu64(__m128i __A, __m128i __B) 527 1.1 mrg { 528 1.1 mrg return (__m128i) __builtin_ia32_vpcomtrueuq ((__v2di)__A, (__v2di)__B); 529 1.1 mrg } 530 1.1 mrg 531 1.1 mrg /*pcom (integer, signed bytes) */ 532 1.1 mrg 533 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 534 1.1 mrg _mm_comlt_epi8(__m128i __A, __m128i __B) 535 1.1 mrg { 536 1.1 mrg return (__m128i) __builtin_ia32_vpcomltb ((__v16qi)__A, (__v16qi)__B); 537 1.1 mrg } 538 1.1 mrg 539 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 540 1.1 mrg _mm_comle_epi8(__m128i __A, __m128i __B) 541 1.1 mrg { 542 1.1 mrg return (__m128i) __builtin_ia32_vpcomleb ((__v16qi)__A, (__v16qi)__B); 543 1.1 mrg } 544 1.1 mrg 545 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 546 1.1 mrg _mm_comgt_epi8(__m128i __A, __m128i __B) 547 1.1 mrg { 548 1.1 mrg return (__m128i) __builtin_ia32_vpcomgtb ((__v16qi)__A, (__v16qi)__B); 549 1.1 mrg } 550 1.1 mrg 551 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 552 1.1 mrg _mm_comge_epi8(__m128i __A, __m128i __B) 553 1.1 mrg { 554 1.1 mrg return (__m128i) __builtin_ia32_vpcomgeb ((__v16qi)__A, (__v16qi)__B); 555 1.1 mrg } 556 1.1 mrg 557 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 558 1.1 mrg _mm_comeq_epi8(__m128i __A, __m128i __B) 559 1.1 mrg { 560 1.1 mrg return (__m128i) __builtin_ia32_vpcomeqb ((__v16qi)__A, (__v16qi)__B); 561 1.1 mrg } 562 1.1 mrg 563 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 564 1.1 mrg _mm_comneq_epi8(__m128i __A, __m128i __B) 565 1.1 mrg { 566 1.1 mrg return (__m128i) __builtin_ia32_vpcomneqb ((__v16qi)__A, (__v16qi)__B); 567 1.1 mrg } 568 1.1 mrg 569 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 570 1.1 mrg _mm_comfalse_epi8(__m128i __A, __m128i __B) 571 1.1 mrg { 572 1.1 mrg return (__m128i) __builtin_ia32_vpcomfalseb ((__v16qi)__A, (__v16qi)__B); 573 1.1 mrg } 574 1.1 mrg 575 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 576 1.1 mrg _mm_comtrue_epi8(__m128i __A, __m128i __B) 577 1.1 mrg { 578 1.1 mrg return (__m128i) __builtin_ia32_vpcomtrueb ((__v16qi)__A, (__v16qi)__B); 579 1.1 mrg } 580 1.1 mrg 581 1.1 mrg /*pcom (integer, signed words) */ 582 1.1 mrg 583 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 584 1.1 mrg _mm_comlt_epi16(__m128i __A, __m128i __B) 585 1.1 mrg { 586 1.1 mrg return (__m128i) __builtin_ia32_vpcomltw ((__v8hi)__A, (__v8hi)__B); 587 1.1 mrg } 588 1.1 mrg 589 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 590 1.1 mrg _mm_comle_epi16(__m128i __A, __m128i __B) 591 1.1 mrg { 592 1.1 mrg return (__m128i) __builtin_ia32_vpcomlew ((__v8hi)__A, (__v8hi)__B); 593 1.1 mrg } 594 1.1 mrg 595 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 596 1.1 mrg _mm_comgt_epi16(__m128i __A, __m128i __B) 597 1.1 mrg { 598 1.1 mrg return (__m128i) __builtin_ia32_vpcomgtw ((__v8hi)__A, (__v8hi)__B); 599 1.1 mrg } 600 1.1 mrg 601 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 602 1.1 mrg _mm_comge_epi16(__m128i __A, __m128i __B) 603 1.1 mrg { 604 1.1 mrg return (__m128i) __builtin_ia32_vpcomgew ((__v8hi)__A, (__v8hi)__B); 605 1.1 mrg } 606 1.1 mrg 607 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 608 1.1 mrg _mm_comeq_epi16(__m128i __A, __m128i __B) 609 1.1 mrg { 610 1.1 mrg return (__m128i) __builtin_ia32_vpcomeqw ((__v8hi)__A, (__v8hi)__B); 611 1.1 mrg } 612 1.1 mrg 613 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 614 1.1 mrg _mm_comneq_epi16(__m128i __A, __m128i __B) 615 1.1 mrg { 616 1.1 mrg return (__m128i) __builtin_ia32_vpcomneqw ((__v8hi)__A, (__v8hi)__B); 617 1.1 mrg } 618 1.1 mrg 619 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 620 1.1 mrg _mm_comfalse_epi16(__m128i __A, __m128i __B) 621 1.1 mrg { 622 1.1 mrg return (__m128i) __builtin_ia32_vpcomfalsew ((__v8hi)__A, (__v8hi)__B); 623 1.1 mrg } 624 1.1 mrg 625 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 626 1.1 mrg _mm_comtrue_epi16(__m128i __A, __m128i __B) 627 1.1 mrg { 628 1.1 mrg return (__m128i) __builtin_ia32_vpcomtruew ((__v8hi)__A, (__v8hi)__B); 629 1.1 mrg } 630 1.1 mrg 631 1.1 mrg /*pcom (integer, signed double words) */ 632 1.1 mrg 633 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 634 1.1 mrg _mm_comlt_epi32(__m128i __A, __m128i __B) 635 1.1 mrg { 636 1.1 mrg return (__m128i) __builtin_ia32_vpcomltd ((__v4si)__A, (__v4si)__B); 637 1.1 mrg } 638 1.1 mrg 639 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 640 1.1 mrg _mm_comle_epi32(__m128i __A, __m128i __B) 641 1.1 mrg { 642 1.1 mrg return (__m128i) __builtin_ia32_vpcomled ((__v4si)__A, (__v4si)__B); 643 1.1 mrg } 644 1.1 mrg 645 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 646 1.1 mrg _mm_comgt_epi32(__m128i __A, __m128i __B) 647 1.1 mrg { 648 1.1 mrg return (__m128i) __builtin_ia32_vpcomgtd ((__v4si)__A, (__v4si)__B); 649 1.1 mrg } 650 1.1 mrg 651 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 652 1.1 mrg _mm_comge_epi32(__m128i __A, __m128i __B) 653 1.1 mrg { 654 1.1 mrg return (__m128i) __builtin_ia32_vpcomged ((__v4si)__A, (__v4si)__B); 655 1.1 mrg } 656 1.1 mrg 657 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 658 1.1 mrg _mm_comeq_epi32(__m128i __A, __m128i __B) 659 1.1 mrg { 660 1.1 mrg return (__m128i) __builtin_ia32_vpcomeqd ((__v4si)__A, (__v4si)__B); 661 1.1 mrg } 662 1.1 mrg 663 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 664 1.1 mrg _mm_comneq_epi32(__m128i __A, __m128i __B) 665 1.1 mrg { 666 1.1 mrg return (__m128i) __builtin_ia32_vpcomneqd ((__v4si)__A, (__v4si)__B); 667 1.1 mrg } 668 1.1 mrg 669 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 670 1.1 mrg _mm_comfalse_epi32(__m128i __A, __m128i __B) 671 1.1 mrg { 672 1.1 mrg return (__m128i) __builtin_ia32_vpcomfalsed ((__v4si)__A, (__v4si)__B); 673 1.1 mrg } 674 1.1 mrg 675 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 676 1.1 mrg _mm_comtrue_epi32(__m128i __A, __m128i __B) 677 1.1 mrg { 678 1.1 mrg return (__m128i) __builtin_ia32_vpcomtrued ((__v4si)__A, (__v4si)__B); 679 1.1 mrg } 680 1.1 mrg 681 1.1 mrg /*pcom (integer, signed quad words) */ 682 1.1 mrg 683 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 684 1.1 mrg _mm_comlt_epi64(__m128i __A, __m128i __B) 685 1.1 mrg { 686 1.1 mrg return (__m128i) __builtin_ia32_vpcomltq ((__v2di)__A, (__v2di)__B); 687 1.1 mrg } 688 1.1 mrg 689 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 690 1.1 mrg _mm_comle_epi64(__m128i __A, __m128i __B) 691 1.1 mrg { 692 1.1 mrg return (__m128i) __builtin_ia32_vpcomleq ((__v2di)__A, (__v2di)__B); 693 1.1 mrg } 694 1.1 mrg 695 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 696 1.1 mrg _mm_comgt_epi64(__m128i __A, __m128i __B) 697 1.1 mrg { 698 1.1 mrg return (__m128i) __builtin_ia32_vpcomgtq ((__v2di)__A, (__v2di)__B); 699 1.1 mrg } 700 1.1 mrg 701 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 702 1.1 mrg _mm_comge_epi64(__m128i __A, __m128i __B) 703 1.1 mrg { 704 1.1 mrg return (__m128i) __builtin_ia32_vpcomgeq ((__v2di)__A, (__v2di)__B); 705 1.1 mrg } 706 1.1 mrg 707 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 708 1.1 mrg _mm_comeq_epi64(__m128i __A, __m128i __B) 709 1.1 mrg { 710 1.1 mrg return (__m128i) __builtin_ia32_vpcomeqq ((__v2di)__A, (__v2di)__B); 711 1.1 mrg } 712 1.1 mrg 713 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 714 1.1 mrg _mm_comneq_epi64(__m128i __A, __m128i __B) 715 1.1 mrg { 716 1.1 mrg return (__m128i) __builtin_ia32_vpcomneqq ((__v2di)__A, (__v2di)__B); 717 1.1 mrg } 718 1.1 mrg 719 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 720 1.1 mrg _mm_comfalse_epi64(__m128i __A, __m128i __B) 721 1.1 mrg { 722 1.1 mrg return (__m128i) __builtin_ia32_vpcomfalseq ((__v2di)__A, (__v2di)__B); 723 1.1 mrg } 724 1.1 mrg 725 1.1 mrg extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 726 1.1 mrg _mm_comtrue_epi64(__m128i __A, __m128i __B) 727 1.1 mrg { 728 1.1 mrg return (__m128i) __builtin_ia32_vpcomtrueq ((__v2di)__A, (__v2di)__B); 729 1.1 mrg } 730 1.1 mrg 731 1.1 mrg /* FRCZ */ 732 1.1 mrg 733 1.1 mrg extern __inline __m128 __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 734 1.1 mrg _mm_frcz_ps (__m128 __A) 735 1.1 mrg { 736 1.1 mrg return (__m128) __builtin_ia32_vfrczps ((__v4sf)__A); 737 1.1 mrg } 738 1.1 mrg 739 1.1 mrg extern __inline __m128d __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 740 1.1 mrg _mm_frcz_pd (__m128d __A) 741 1.1 mrg { 742 1.1 mrg return (__m128d) __builtin_ia32_vfrczpd ((__v2df)__A); 743 1.1 mrg } 744 1.1 mrg 745 1.1 mrg extern __inline __m128 __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 746 1.1 mrg _mm_frcz_ss (__m128 __A, __m128 __B) 747 1.1 mrg { 748 1.1 mrg return (__m128) __builtin_ia32_vfrczss ((__v4sf)__A, (__v4sf)__B); 749 1.1 mrg } 750 1.1 mrg 751 1.1 mrg extern __inline __m128d __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 752 1.1 mrg _mm_frcz_sd (__m128d __A, __m128d __B) 753 1.1 mrg { 754 1.1 mrg return (__m128d) __builtin_ia32_vfrczsd ((__v2df)__A, (__v2df)__B); 755 1.1 mrg } 756 1.1 mrg 757 1.1 mrg extern __inline __m256 __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 758 1.1 mrg _mm256_frcz_ps (__m256 __A) 759 1.1 mrg { 760 1.1 mrg return (__m256) __builtin_ia32_vfrczps256 ((__v8sf)__A); 761 1.1 mrg } 762 1.1 mrg 763 1.1 mrg extern __inline __m256d __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 764 1.1 mrg _mm256_frcz_pd (__m256d __A) 765 1.1 mrg { 766 1.1 mrg return (__m256d) __builtin_ia32_vfrczpd256 ((__v4df)__A); 767 1.1 mrg } 768 1.1 mrg 769 1.1 mrg /* PERMIL2 */ 770 1.1 mrg 771 1.1 mrg #ifdef __OPTIMIZE__ 772 1.1 mrg extern __inline __m128d __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 773 1.1 mrg _mm_permute2_pd (__m128d __X, __m128d __Y, __m128i __C, const int __I) 774 1.1 mrg { 775 1.1 mrg return (__m128d) __builtin_ia32_vpermil2pd ((__v2df)__X, 776 1.1 mrg (__v2df)__Y, 777 1.1 mrg (__v2di)__C, 778 1.1 mrg __I); 779 1.1 mrg } 780 1.1 mrg 781 1.1 mrg extern __inline __m256d __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 782 1.1 mrg _mm256_permute2_pd (__m256d __X, __m256d __Y, __m256i __C, const int __I) 783 1.1 mrg { 784 1.1 mrg return (__m256d) __builtin_ia32_vpermil2pd256 ((__v4df)__X, 785 1.1 mrg (__v4df)__Y, 786 1.1 mrg (__v4di)__C, 787 1.1 mrg __I); 788 1.1 mrg } 789 1.1 mrg 790 1.1 mrg extern __inline __m128 __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 791 1.1 mrg _mm_permute2_ps (__m128 __X, __m128 __Y, __m128i __C, const int __I) 792 1.1 mrg { 793 1.1 mrg return (__m128) __builtin_ia32_vpermil2ps ((__v4sf)__X, 794 1.1 mrg (__v4sf)__Y, 795 1.1 mrg (__v4si)__C, 796 1.1 mrg __I); 797 1.1 mrg } 798 1.1 mrg 799 1.1 mrg extern __inline __m256 __attribute__((__gnu_inline__, __always_inline__, __artificial__)) 800 1.1 mrg _mm256_permute2_ps (__m256 __X, __m256 __Y, __m256i __C, const int __I) 801 1.1 mrg { 802 1.1 mrg return (__m256) __builtin_ia32_vpermil2ps256 ((__v8sf)__X, 803 1.1 mrg (__v8sf)__Y, 804 1.1 mrg (__v8si)__C, 805 1.1 mrg __I); 806 1.1 mrg } 807 1.1 mrg #else 808 1.1 mrg #define _mm_permute2_pd(X, Y, C, I) \ 809 1.1 mrg ((__m128d) __builtin_ia32_vpermil2pd ((__v2df)(__m128d)(X), \ 810 1.1 mrg (__v2df)(__m128d)(Y), \ 811 1.1 mrg (__v2di)(__m128d)(C), \ 812 1.1 mrg (int)(I))) 813 1.1 mrg 814 1.1 mrg #define _mm256_permute2_pd(X, Y, C, I) \ 815 1.1 mrg ((__m256d) __builtin_ia32_vpermil2pd256 ((__v4df)(__m256d)(X), \ 816 1.1 mrg (__v4df)(__m256d)(Y), \ 817 1.1 mrg (__v4di)(__m256d)(C), \ 818 1.1 mrg (int)(I))) 819 1.1 mrg 820 1.1 mrg #define _mm_permute2_ps(X, Y, C, I) \ 821 1.1 mrg ((__m128) __builtin_ia32_vpermil2ps ((__v4sf)(__m128)(X), \ 822 1.1 mrg (__v4sf)(__m128)(Y), \ 823 1.1 mrg (__v4si)(__m128)(C), \ 824 1.1 mrg (int)(I))) 825 1.1 mrg 826 1.1 mrg #define _mm256_permute2_ps(X, Y, C, I) \ 827 1.1 mrg ((__m256) __builtin_ia32_vpermil2ps256 ((__v8sf)(__m256)(X), \ 828 1.1 mrg (__v8sf)(__m256)(Y), \ 829 1.1 mrg (__v8si)(__m256)(C), \ 830 1.1 mrg (int)(I))) 831 1.1 mrg #endif /* __OPTIMIZE__ */ 832 1.1 mrg 833 1.1 mrg #endif /* __XOP__ */ 834 1.1 mrg 835 1.1 mrg #endif /* _XOPMMINTRIN_H_INCLUDED */ 836
Indexes created Thu Apr 09 00:23:28 UTC 2026