| 1 |
/* Copyright (C) 2013-2021 Free Software Foundation, Inc. |
| 2 |
|
| 3 |
This file is part of GCC. |
| 4 |
|
| 5 |
GCC is free software; you can redistribute it and/or modify |
| 6 |
it under the terms of the GNU General Public License as published by |
| 7 |
the Free Software Foundation; either version 3, or (at your option) |
| 8 |
any later version. |
| 9 |
|
| 10 |
GCC is distributed in the hope that it will be useful, |
| 11 |
but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 12 |
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 13 |
GNU General Public License for more details. |
| 14 |
|
| 15 |
Under Section 7 of GPL version 3, you are granted additional |
| 16 |
permissions described in the GCC Runtime Library Exception, version |
| 17 |
3.1, as published by the Free Software Foundation. |
| 18 |
|
| 19 |
You should have received a copy of the GNU General Public License and |
| 20 |
a copy of the GCC Runtime Library Exception along with this program; |
| 21 |
see the files COPYING3 and COPYING.RUNTIME respectively. If not, see |
| 22 |
<http://www.gnu.org/licenses/>. */ |
| 23 |
|
| 24 |
#ifndef _IMMINTRIN_H_INCLUDED |
| 25 |
#error "Never use <avx512vbmi2vlintrin.h> directly; include <immintrin.h> instead." |
| 26 |
#endif |
| 27 |
|
| 28 |
#ifndef _AVX512VBMI2VLINTRIN_H_INCLUDED |
| 29 |
#define _AVX512VBMI2VLINTRIN_H_INCLUDED |
| 30 |
|
| 31 |
#if !defined(__AVX512VL__) || !defined(__AVX512VBMI2__) |
| 32 |
#pragma GCC push_options |
| 33 |
#pragma GCC target("avx512vbmi2,avx512vl") |
| 34 |
#define __DISABLE_AVX512VBMI2VL__ |
| 35 |
#endif /* __AVX512VBMIVL__ */ |
| 36 |
|
| 37 |
extern __inline __m128i |
| 38 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 39 |
_mm_mask_compress_epi8 (__m128i __A, __mmask16 __B, __m128i __C) |
| 40 |
{ |
| 41 |
return (__m128i) __builtin_ia32_compressqi128_mask ((__v16qi)__C, |
| 42 |
(__v16qi)__A, (__mmask16)__B); |
| 43 |
} |
| 44 |
|
| 45 |
extern __inline __m128i |
| 46 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 47 |
_mm_maskz_compress_epi8 (__mmask16 __A, __m128i __B) |
| 48 |
{ |
| 49 |
return (__m128i) __builtin_ia32_compressqi128_mask ((__v16qi) __B, |
| 50 |
(__v16qi) _mm_setzero_si128 (), (__mmask16) __A); |
| 51 |
} |
| 52 |
|
| 53 |
|
| 54 |
extern __inline void |
| 55 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 56 |
_mm256_mask_compressstoreu_epi16 (void * __A, __mmask16 __B, __m256i __C) |
| 57 |
{ |
| 58 |
__builtin_ia32_compressstoreuhi256_mask ((__v16hi *) __A, (__v16hi) __C, |
| 59 |
(__mmask16) __B); |
| 60 |
} |
| 61 |
|
| 62 |
extern __inline __m128i |
| 63 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 64 |
_mm_mask_compress_epi16 (__m128i __A, __mmask8 __B, __m128i __C) |
| 65 |
{ |
| 66 |
return (__m128i) __builtin_ia32_compresshi128_mask ((__v8hi)__C, (__v8hi)__A, |
| 67 |
(__mmask8)__B); |
| 68 |
} |
| 69 |
|
| 70 |
extern __inline __m128i |
| 71 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 72 |
_mm_maskz_compress_epi16 (__mmask8 __A, __m128i __B) |
| 73 |
{ |
| 74 |
return (__m128i) __builtin_ia32_compresshi128_mask ((__v8hi) __B, |
| 75 |
(__v8hi) _mm_setzero_si128 (), (__mmask8) __A); |
| 76 |
} |
| 77 |
|
| 78 |
extern __inline __m256i |
| 79 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 80 |
_mm256_mask_compress_epi16 (__m256i __A, __mmask16 __B, __m256i __C) |
| 81 |
{ |
| 82 |
return (__m256i) __builtin_ia32_compresshi256_mask ((__v16hi)__C, |
| 83 |
(__v16hi)__A, (__mmask16)__B); |
| 84 |
} |
| 85 |
|
| 86 |
extern __inline __m256i |
| 87 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 88 |
_mm256_maskz_compress_epi16 (__mmask16 __A, __m256i __B) |
| 89 |
{ |
| 90 |
return (__m256i) __builtin_ia32_compresshi256_mask ((__v16hi) __B, |
| 91 |
(__v16hi) _mm256_setzero_si256 (), (__mmask16) __A); |
| 92 |
} |
| 93 |
|
| 94 |
extern __inline void |
| 95 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 96 |
_mm_mask_compressstoreu_epi8 (void * __A, __mmask16 __B, __m128i __C) |
| 97 |
{ |
| 98 |
__builtin_ia32_compressstoreuqi128_mask ((__v16qi *) __A, (__v16qi) __C, |
| 99 |
(__mmask16) __B); |
| 100 |
} |
| 101 |
|
| 102 |
extern __inline void |
| 103 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 104 |
_mm_mask_compressstoreu_epi16 (void * __A, __mmask8 __B, __m128i __C) |
| 105 |
{ |
| 106 |
__builtin_ia32_compressstoreuhi128_mask ((__v8hi *) __A, (__v8hi) __C, |
| 107 |
(__mmask8) __B); |
| 108 |
} |
| 109 |
|
| 110 |
extern __inline __m128i |
| 111 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 112 |
_mm_mask_expand_epi8 (__m128i __A, __mmask16 __B, __m128i __C) |
| 113 |
{ |
| 114 |
return (__m128i) __builtin_ia32_expandqi128_mask ((__v16qi) __C, |
| 115 |
(__v16qi) __A, |
| 116 |
(__mmask16) __B); |
| 117 |
} |
| 118 |
|
| 119 |
extern __inline __m128i |
| 120 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 121 |
_mm_maskz_expand_epi8 (__mmask16 __A, __m128i __B) |
| 122 |
{ |
| 123 |
return (__m128i) __builtin_ia32_expandqi128_maskz ((__v16qi) __B, |
| 124 |
(__v16qi) _mm_setzero_si128 (), (__mmask16) __A); |
| 125 |
} |
| 126 |
|
| 127 |
extern __inline __m128i |
| 128 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 129 |
_mm_mask_expandloadu_epi8 (__m128i __A, __mmask16 __B, const void * __C) |
| 130 |
{ |
| 131 |
return (__m128i) __builtin_ia32_expandloadqi128_mask ((const __v16qi *) __C, |
| 132 |
(__v16qi) __A, (__mmask16) __B); |
| 133 |
} |
| 134 |
|
| 135 |
extern __inline __m128i |
| 136 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 137 |
_mm_maskz_expandloadu_epi8 (__mmask16 __A, const void * __B) |
| 138 |
{ |
| 139 |
return (__m128i) __builtin_ia32_expandloadqi128_maskz ((const __v16qi *) __B, |
| 140 |
(__v16qi) _mm_setzero_si128 (), (__mmask16) __A); |
| 141 |
} |
| 142 |
|
| 143 |
extern __inline __m128i |
| 144 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 145 |
_mm_mask_expand_epi16 (__m128i __A, __mmask8 __B, __m128i __C) |
| 146 |
{ |
| 147 |
return (__m128i) __builtin_ia32_expandhi128_mask ((__v8hi) __C, |
| 148 |
(__v8hi) __A, |
| 149 |
(__mmask8) __B); |
| 150 |
} |
| 151 |
|
| 152 |
extern __inline __m128i |
| 153 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 154 |
_mm_maskz_expand_epi16 (__mmask8 __A, __m128i __B) |
| 155 |
{ |
| 156 |
return (__m128i) __builtin_ia32_expandhi128_maskz ((__v8hi) __B, |
| 157 |
(__v8hi) _mm_setzero_si128 (), (__mmask8) __A); |
| 158 |
} |
| 159 |
|
| 160 |
extern __inline __m128i |
| 161 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 162 |
_mm_mask_expandloadu_epi16 (__m128i __A, __mmask8 __B, const void * __C) |
| 163 |
{ |
| 164 |
return (__m128i) __builtin_ia32_expandloadhi128_mask ((const __v8hi *) __C, |
| 165 |
(__v8hi) __A, (__mmask8) __B); |
| 166 |
} |
| 167 |
|
| 168 |
extern __inline __m128i |
| 169 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 170 |
_mm_maskz_expandloadu_epi16 (__mmask8 __A, const void * __B) |
| 171 |
{ |
| 172 |
return (__m128i) __builtin_ia32_expandloadhi128_maskz ((const __v8hi *) __B, |
| 173 |
(__v8hi) _mm_setzero_si128 (), (__mmask8) __A); |
| 174 |
} |
| 175 |
extern __inline __m256i |
| 176 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 177 |
_mm256_mask_expand_epi16 (__m256i __A, __mmask16 __B, __m256i __C) |
| 178 |
{ |
| 179 |
return (__m256i) __builtin_ia32_expandhi256_mask ((__v16hi) __C, |
| 180 |
(__v16hi) __A, |
| 181 |
(__mmask16) __B); |
| 182 |
} |
| 183 |
|
| 184 |
extern __inline __m256i |
| 185 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 186 |
_mm256_maskz_expand_epi16 (__mmask16 __A, __m256i __B) |
| 187 |
{ |
| 188 |
return (__m256i) __builtin_ia32_expandhi256_maskz ((__v16hi) __B, |
| 189 |
(__v16hi) _mm256_setzero_si256 (), (__mmask16) __A); |
| 190 |
} |
| 191 |
|
| 192 |
extern __inline __m256i |
| 193 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 194 |
_mm256_mask_expandloadu_epi16 (__m256i __A, __mmask16 __B, const void * __C) |
| 195 |
{ |
| 196 |
return (__m256i) __builtin_ia32_expandloadhi256_mask ((const __v16hi *) __C, |
| 197 |
(__v16hi) __A, (__mmask16) __B); |
| 198 |
} |
| 199 |
|
| 200 |
extern __inline __m256i |
| 201 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 202 |
_mm256_maskz_expandloadu_epi16 (__mmask16 __A, const void * __B) |
| 203 |
{ |
| 204 |
return (__m256i) __builtin_ia32_expandloadhi256_maskz ((const __v16hi *) __B, |
| 205 |
(__v16hi) _mm256_setzero_si256 (), (__mmask16) __A); |
| 206 |
} |
| 207 |
|
| 208 |
#ifdef __OPTIMIZE__ |
| 209 |
extern __inline __m256i |
| 210 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 211 |
_mm256_shrdi_epi16 (__m256i __A, __m256i __B, int __C) |
| 212 |
{ |
| 213 |
return (__m256i) __builtin_ia32_vpshrd_v16hi ((__v16hi)__A, (__v16hi) __B, |
| 214 |
__C); |
| 215 |
} |
| 216 |
|
| 217 |
extern __inline __m256i |
| 218 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 219 |
_mm256_mask_shrdi_epi16 (__m256i __A, __mmask16 __B, __m256i __C, __m256i __D, |
| 220 |
int __E) |
| 221 |
{ |
| 222 |
return (__m256i)__builtin_ia32_vpshrd_v16hi_mask ((__v16hi)__C, |
| 223 |
(__v16hi) __D, __E, (__v16hi) __A, (__mmask16)__B); |
| 224 |
} |
| 225 |
|
| 226 |
extern __inline __m256i |
| 227 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 228 |
_mm256_maskz_shrdi_epi16 (__mmask16 __A, __m256i __B, __m256i __C, int __D) |
| 229 |
{ |
| 230 |
return (__m256i)__builtin_ia32_vpshrd_v16hi_mask ((__v16hi)__B, |
| 231 |
(__v16hi) __C, __D, (__v16hi) _mm256_setzero_si256 (), (__mmask16)__A); |
| 232 |
} |
| 233 |
|
| 234 |
extern __inline __m256i |
| 235 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 236 |
_mm256_mask_shrdi_epi32 (__m256i __A, __mmask8 __B, __m256i __C, __m256i __D, |
| 237 |
int __E) |
| 238 |
{ |
| 239 |
return (__m256i)__builtin_ia32_vpshrd_v8si_mask ((__v8si)__C, (__v8si) __D, |
| 240 |
__E, (__v8si) __A, (__mmask8)__B); |
| 241 |
} |
| 242 |
|
| 243 |
extern __inline __m256i |
| 244 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 245 |
_mm256_maskz_shrdi_epi32 (__mmask8 __A, __m256i __B, __m256i __C, int __D) |
| 246 |
{ |
| 247 |
return (__m256i)__builtin_ia32_vpshrd_v8si_mask ((__v8si)__B, (__v8si) __C, |
| 248 |
__D, (__v8si) _mm256_setzero_si256 (), (__mmask8)__A); |
| 249 |
} |
| 250 |
|
| 251 |
extern __inline __m256i |
| 252 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 253 |
_mm256_shrdi_epi32 (__m256i __A, __m256i __B, int __C) |
| 254 |
{ |
| 255 |
return (__m256i) __builtin_ia32_vpshrd_v8si ((__v8si)__A, (__v8si) __B, __C); |
| 256 |
} |
| 257 |
|
| 258 |
extern __inline __m256i |
| 259 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 260 |
_mm256_mask_shrdi_epi64 (__m256i __A, __mmask8 __B, __m256i __C, __m256i __D, |
| 261 |
int __E) |
| 262 |
{ |
| 263 |
return (__m256i)__builtin_ia32_vpshrd_v4di_mask ((__v4di)__C, (__v4di) __D, |
| 264 |
__E, (__v4di) __A, (__mmask8)__B); |
| 265 |
} |
| 266 |
|
| 267 |
extern __inline __m256i |
| 268 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 269 |
_mm256_maskz_shrdi_epi64 (__mmask8 __A, __m256i __B, __m256i __C, int __D) |
| 270 |
{ |
| 271 |
return (__m256i)__builtin_ia32_vpshrd_v4di_mask ((__v4di)__B, (__v4di) __C, |
| 272 |
__D, (__v4di) _mm256_setzero_si256 (), (__mmask8)__A); |
| 273 |
} |
| 274 |
|
| 275 |
extern __inline __m256i |
| 276 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 277 |
_mm256_shrdi_epi64 (__m256i __A, __m256i __B, int __C) |
| 278 |
{ |
| 279 |
return (__m256i) __builtin_ia32_vpshrd_v4di ((__v4di)__A, (__v4di) __B, __C); |
| 280 |
} |
| 281 |
|
| 282 |
extern __inline __m128i |
| 283 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 284 |
_mm_mask_shrdi_epi16 (__m128i __A, __mmask8 __B, __m128i __C, __m128i __D, |
| 285 |
int __E) |
| 286 |
{ |
| 287 |
return (__m128i)__builtin_ia32_vpshrd_v8hi_mask ((__v8hi)__C, (__v8hi) __D, |
| 288 |
__E, (__v8hi) __A, (__mmask8)__B); |
| 289 |
} |
| 290 |
|
| 291 |
extern __inline __m128i |
| 292 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 293 |
_mm_maskz_shrdi_epi16 (__mmask8 __A, __m128i __B, __m128i __C, int __D) |
| 294 |
{ |
| 295 |
return (__m128i)__builtin_ia32_vpshrd_v8hi_mask ((__v8hi)__B, (__v8hi) __C, |
| 296 |
__D, (__v8hi) _mm_setzero_si128 (), (__mmask8)__A); |
| 297 |
} |
| 298 |
|
| 299 |
extern __inline __m128i |
| 300 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 301 |
_mm_shrdi_epi16 (__m128i __A, __m128i __B, int __C) |
| 302 |
{ |
| 303 |
return (__m128i) __builtin_ia32_vpshrd_v8hi ((__v8hi)__A, (__v8hi) __B, __C); |
| 304 |
} |
| 305 |
|
| 306 |
extern __inline __m128i |
| 307 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 308 |
_mm_mask_shrdi_epi32 (__m128i __A, __mmask8 __B, __m128i __C, __m128i __D, |
| 309 |
int __E) |
| 310 |
{ |
| 311 |
return (__m128i)__builtin_ia32_vpshrd_v4si_mask ((__v4si)__C, (__v4si) __D, |
| 312 |
__E, (__v4si) __A, (__mmask8)__B); |
| 313 |
} |
| 314 |
|
| 315 |
extern __inline __m128i |
| 316 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 317 |
_mm_maskz_shrdi_epi32 (__mmask8 __A, __m128i __B, __m128i __C, int __D) |
| 318 |
{ |
| 319 |
return (__m128i)__builtin_ia32_vpshrd_v4si_mask ((__v4si)__B, (__v4si) __C, |
| 320 |
__D, (__v4si) _mm_setzero_si128 (), (__mmask8)__A); |
| 321 |
} |
| 322 |
|
| 323 |
extern __inline __m128i |
| 324 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 325 |
_mm_shrdi_epi32 (__m128i __A, __m128i __B, int __C) |
| 326 |
{ |
| 327 |
return (__m128i) __builtin_ia32_vpshrd_v4si ((__v4si)__A, (__v4si) __B, __C); |
| 328 |
} |
| 329 |
|
| 330 |
extern __inline __m128i |
| 331 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 332 |
_mm_mask_shrdi_epi64 (__m128i __A, __mmask8 __B, __m128i __C, __m128i __D, |
| 333 |
int __E) |
| 334 |
{ |
| 335 |
return (__m128i)__builtin_ia32_vpshrd_v2di_mask ((__v2di)__C, (__v2di) __D, |
| 336 |
__E, (__v2di) __A, (__mmask8)__B); |
| 337 |
} |
| 338 |
|
| 339 |
extern __inline __m128i |
| 340 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 341 |
_mm_maskz_shrdi_epi64 (__mmask8 __A, __m128i __B, __m128i __C, int __D) |
| 342 |
{ |
| 343 |
return (__m128i)__builtin_ia32_vpshrd_v2di_mask ((__v2di)__B, (__v2di) __C, |
| 344 |
__D, (__v2di) _mm_setzero_si128 (), (__mmask8)__A); |
| 345 |
} |
| 346 |
|
| 347 |
extern __inline __m128i |
| 348 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 349 |
_mm_shrdi_epi64 (__m128i __A, __m128i __B, int __C) |
| 350 |
{ |
| 351 |
return (__m128i) __builtin_ia32_vpshrd_v2di ((__v2di)__A, (__v2di) __B, __C); |
| 352 |
} |
| 353 |
|
| 354 |
extern __inline __m256i |
| 355 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 356 |
_mm256_shldi_epi16 (__m256i __A, __m256i __B, int __C) |
| 357 |
{ |
| 358 |
return (__m256i) __builtin_ia32_vpshld_v16hi ((__v16hi)__A, (__v16hi) __B, |
| 359 |
__C); |
| 360 |
} |
| 361 |
|
| 362 |
extern __inline __m256i |
| 363 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 364 |
_mm256_mask_shldi_epi16 (__m256i __A, __mmask16 __B, __m256i __C, __m256i __D, |
| 365 |
int __E) |
| 366 |
{ |
| 367 |
return (__m256i)__builtin_ia32_vpshld_v16hi_mask ((__v16hi)__C, |
| 368 |
(__v16hi) __D, __E, (__v16hi) __A, (__mmask16)__B); |
| 369 |
} |
| 370 |
|
| 371 |
extern __inline __m256i |
| 372 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 373 |
_mm256_maskz_shldi_epi16 (__mmask16 __A, __m256i __B, __m256i __C, int __D) |
| 374 |
{ |
| 375 |
return (__m256i)__builtin_ia32_vpshld_v16hi_mask ((__v16hi)__B, |
| 376 |
(__v16hi) __C, __D, (__v16hi) _mm256_setzero_si256 (), (__mmask16)__A); |
| 377 |
} |
| 378 |
|
| 379 |
extern __inline __m256i |
| 380 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 381 |
_mm256_mask_shldi_epi32 (__m256i __A, __mmask8 __B, __m256i __C, __m256i __D, |
| 382 |
int __E) |
| 383 |
{ |
| 384 |
return (__m256i)__builtin_ia32_vpshld_v8si_mask ((__v8si)__C, (__v8si) __D, |
| 385 |
__E, (__v8si) __A, (__mmask8)__B); |
| 386 |
} |
| 387 |
|
| 388 |
extern __inline __m256i |
| 389 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 390 |
_mm256_maskz_shldi_epi32 (__mmask8 __A, __m256i __B, __m256i __C, int __D) |
| 391 |
{ |
| 392 |
return (__m256i)__builtin_ia32_vpshld_v8si_mask ((__v8si)__B, (__v8si) __C, |
| 393 |
__D, (__v8si) _mm256_setzero_si256 (), (__mmask8)__A); |
| 394 |
} |
| 395 |
|
| 396 |
extern __inline __m256i |
| 397 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 398 |
_mm256_shldi_epi32 (__m256i __A, __m256i __B, int __C) |
| 399 |
{ |
| 400 |
return (__m256i) __builtin_ia32_vpshld_v8si ((__v8si)__A, (__v8si) __B, __C); |
| 401 |
} |
| 402 |
|
| 403 |
extern __inline __m256i |
| 404 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 405 |
_mm256_mask_shldi_epi64 (__m256i __A, __mmask8 __B, __m256i __C, __m256i __D, |
| 406 |
int __E) |
| 407 |
{ |
| 408 |
return (__m256i)__builtin_ia32_vpshld_v4di_mask ((__v4di)__C, (__v4di) __D, |
| 409 |
__E, (__v4di) __A, (__mmask8)__B); |
| 410 |
} |
| 411 |
|
| 412 |
extern __inline __m256i |
| 413 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 414 |
_mm256_maskz_shldi_epi64 (__mmask8 __A, __m256i __B, __m256i __C, int __D) |
| 415 |
{ |
| 416 |
return (__m256i)__builtin_ia32_vpshld_v4di_mask ((__v4di)__B, (__v4di) __C, |
| 417 |
__D, (__v4di) _mm256_setzero_si256 (), (__mmask8)__A); |
| 418 |
} |
| 419 |
|
| 420 |
extern __inline __m256i |
| 421 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 422 |
_mm256_shldi_epi64 (__m256i __A, __m256i __B, int __C) |
| 423 |
{ |
| 424 |
return (__m256i) __builtin_ia32_vpshld_v4di ((__v4di)__A, (__v4di) __B, __C); |
| 425 |
} |
| 426 |
|
| 427 |
extern __inline __m128i |
| 428 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 429 |
_mm_mask_shldi_epi16 (__m128i __A, __mmask8 __B, __m128i __C, __m128i __D, |
| 430 |
int __E) |
| 431 |
{ |
| 432 |
return (__m128i)__builtin_ia32_vpshld_v8hi_mask ((__v8hi)__C, (__v8hi) __D, |
| 433 |
__E, (__v8hi) __A, (__mmask8)__B); |
| 434 |
} |
| 435 |
|
| 436 |
extern __inline __m128i |
| 437 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 438 |
_mm_maskz_shldi_epi16 (__mmask8 __A, __m128i __B, __m128i __C, int __D) |
| 439 |
{ |
| 440 |
return (__m128i)__builtin_ia32_vpshld_v8hi_mask ((__v8hi)__B, (__v8hi) __C, |
| 441 |
__D, (__v8hi) _mm_setzero_si128 (), (__mmask8)__A); |
| 442 |
} |
| 443 |
|
| 444 |
extern __inline __m128i |
| 445 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 446 |
_mm_shldi_epi16 (__m128i __A, __m128i __B, int __C) |
| 447 |
{ |
| 448 |
return (__m128i) __builtin_ia32_vpshld_v8hi ((__v8hi)__A, (__v8hi) __B, __C); |
| 449 |
} |
| 450 |
|
| 451 |
extern __inline __m128i |
| 452 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 453 |
_mm_mask_shldi_epi32 (__m128i __A, __mmask8 __B, __m128i __C, __m128i __D, |
| 454 |
int __E) |
| 455 |
{ |
| 456 |
return (__m128i)__builtin_ia32_vpshld_v4si_mask ((__v4si)__C, (__v4si) __D, |
| 457 |
__E, (__v4si) __A, (__mmask8)__B); |
| 458 |
} |
| 459 |
|
| 460 |
extern __inline __m128i |
| 461 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 462 |
_mm_maskz_shldi_epi32 (__mmask8 __A, __m128i __B, __m128i __C, int __D) |
| 463 |
{ |
| 464 |
return (__m128i)__builtin_ia32_vpshld_v4si_mask ((__v4si)__B, (__v4si) __C, |
| 465 |
__D, (__v4si) _mm_setzero_si128 (), (__mmask8)__A); |
| 466 |
} |
| 467 |
|
| 468 |
extern __inline __m128i |
| 469 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 470 |
_mm_shldi_epi32 (__m128i __A, __m128i __B, int __C) |
| 471 |
{ |
| 472 |
return (__m128i) __builtin_ia32_vpshld_v4si ((__v4si)__A, (__v4si) __B, __C); |
| 473 |
} |
| 474 |
|
| 475 |
extern __inline __m128i |
| 476 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 477 |
_mm_mask_shldi_epi64 (__m128i __A, __mmask8 __B, __m128i __C, __m128i __D, |
| 478 |
int __E) |
| 479 |
{ |
| 480 |
return (__m128i)__builtin_ia32_vpshld_v2di_mask ((__v2di)__C, (__v2di) __D, |
| 481 |
__E, (__v2di) __A, (__mmask8)__B); |
| 482 |
} |
| 483 |
|
| 484 |
extern __inline __m128i |
| 485 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 486 |
_mm_maskz_shldi_epi64 (__mmask8 __A, __m128i __B, __m128i __C, int __D) |
| 487 |
{ |
| 488 |
return (__m128i)__builtin_ia32_vpshld_v2di_mask ((__v2di)__B, (__v2di) __C, |
| 489 |
__D, (__v2di) _mm_setzero_si128 (), (__mmask8)__A); |
| 490 |
} |
| 491 |
|
| 492 |
extern __inline __m128i |
| 493 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 494 |
_mm_shldi_epi64 (__m128i __A, __m128i __B, int __C) |
| 495 |
{ |
| 496 |
return (__m128i) __builtin_ia32_vpshld_v2di ((__v2di)__A, (__v2di) __B, __C); |
| 497 |
} |
| 498 |
#else |
| 499 |
#define _mm256_shrdi_epi16(A, B, C) \ |
| 500 |
((__m256i) __builtin_ia32_vpshrd_v16hi ((__v16hi)(__m256i)(A), \ |
| 501 |
(__v16hi)(__m256i)(B),(int)(C))) |
| 502 |
#define _mm256_mask_shrdi_epi16(A, B, C, D, E) \ |
| 503 |
((__m256i) __builtin_ia32_vpshrd_v16hi_mask ((__v16hi)(__m256i)(C), \ |
| 504 |
(__v16hi)(__m256i)(D), \ |
| 505 |
(int)(E), \ |
| 506 |
(__v16hi)(__m256i)(A), \ |
| 507 |
(__mmask16)(B))) |
| 508 |
#define _mm256_maskz_shrdi_epi16(A, B, C, D) \ |
| 509 |
((__m256i) \ |
| 510 |
__builtin_ia32_vpshrd_v16hi_mask ((__v16hi)(__m256i)(B), \ |
| 511 |
(__v16hi)(__m256i)(C),(int)(D), \ |
| 512 |
(__v16hi)(__m256i)_mm256_setzero_si256 (), \ |
| 513 |
(__mmask16)(A))) |
| 514 |
#define _mm256_shrdi_epi32(A, B, C) \ |
| 515 |
((__m256i) __builtin_ia32_vpshrd_v8si ((__v8si)(__m256i)(A), \ |
| 516 |
(__v8si)(__m256i)(B),(int)(C))) |
| 517 |
#define _mm256_mask_shrdi_epi32(A, B, C, D, E) \ |
| 518 |
((__m256i) __builtin_ia32_vpshrd_v8si_mask ((__v8si)(__m256i)(C), \ |
| 519 |
(__v8si)(__m256i)(D), \ |
| 520 |
(int)(E), \ |
| 521 |
(__v8si)(__m256i)(A), \ |
| 522 |
(__mmask8)(B))) |
| 523 |
#define _mm256_maskz_shrdi_epi32(A, B, C, D) \ |
| 524 |
((__m256i) \ |
| 525 |
__builtin_ia32_vpshrd_v8si_mask ((__v8si)(__m256i)(B), \ |
| 526 |
(__v8si)(__m256i)(C),(int)(D), \ |
| 527 |
(__v8si)(__m256i)_mm256_setzero_si256 (), \ |
| 528 |
(__mmask8)(A))) |
| 529 |
#define _mm256_shrdi_epi64(A, B, C) \ |
| 530 |
((__m256i) __builtin_ia32_vpshrd_v4di ((__v4di)(__m256i)(A), \ |
| 531 |
(__v4di)(__m256i)(B),(int)(C))) |
| 532 |
#define _mm256_mask_shrdi_epi64(A, B, C, D, E) \ |
| 533 |
((__m256i) __builtin_ia32_vpshrd_v4di_mask ((__v4di)(__m256i)(C), \ |
| 534 |
(__v4di)(__m256i)(D), (int)(E), \ |
| 535 |
(__v4di)(__m256i)(A), \ |
| 536 |
(__mmask8)(B))) |
| 537 |
#define _mm256_maskz_shrdi_epi64(A, B, C, D) \ |
| 538 |
((__m256i) \ |
| 539 |
__builtin_ia32_vpshrd_v4di_mask ((__v4di)(__m256i)(B), \ |
| 540 |
(__v4di)(__m256i)(C),(int)(D), \ |
| 541 |
(__v4di)(__m256i)_mm256_setzero_si256 (), \ |
| 542 |
(__mmask8)(A))) |
| 543 |
#define _mm_shrdi_epi16(A, B, C) \ |
| 544 |
((__m128i) __builtin_ia32_vpshrd_v8hi ((__v8hi)(__m128i)(A), \ |
| 545 |
(__v8hi)(__m128i)(B),(int)(C))) |
| 546 |
#define _mm_mask_shrdi_epi16(A, B, C, D, E) \ |
| 547 |
((__m128i) __builtin_ia32_vpshrd_v8hi_mask ((__v8hi)(__m128i)(C), \ |
| 548 |
(__v8hi)(__m128i)(D), (int)(E), \ |
| 549 |
(__v8hi)(__m128i)(A), \ |
| 550 |
(__mmask8)(B))) |
| 551 |
#define _mm_maskz_shrdi_epi16(A, B, C, D) \ |
| 552 |
((__m128i) \ |
| 553 |
__builtin_ia32_vpshrd_v8hi_mask ((__v8hi)(__m128i)(B), \ |
| 554 |
(__v8hi)(__m128i)(C),(int)(D), \ |
| 555 |
(__v8hi)(__m128i)_mm_setzero_si128 (), \ |
| 556 |
(__mmask8)(A))) |
| 557 |
#define _mm_shrdi_epi32(A, B, C) \ |
| 558 |
((__m128i) __builtin_ia32_vpshrd_v4si ((__v4si)(__m128i)(A), \ |
| 559 |
(__v4si)(__m128i)(B),(int)(C))) |
| 560 |
#define _mm_mask_shrdi_epi32(A, B, C, D, E) \ |
| 561 |
((__m128i) __builtin_ia32_vpshrd_v4si_mask ((__v4si)(__m128i)(C), \ |
| 562 |
(__v4si)(__m128i)(D), (int)(E), \ |
| 563 |
(__v4si)(__m128i)(A), \ |
| 564 |
(__mmask8)(B))) |
| 565 |
#define _mm_maskz_shrdi_epi32(A, B, C, D) \ |
| 566 |
((__m128i) \ |
| 567 |
__builtin_ia32_vpshrd_v4si_mask ((__v4si)(__m128i)(B), \ |
| 568 |
(__v4si)(__m128i)(C),(int)(D), \ |
| 569 |
(__v4si)(__m128i)_mm_setzero_si128 (), \ |
| 570 |
(__mmask8)(A))) |
| 571 |
#define _mm_shrdi_epi64(A, B, C) \ |
| 572 |
((__m128i) __builtin_ia32_vpshrd_v2di ((__v2di)(__m128i)(A), \ |
| 573 |
(__v2di)(__m128i)(B),(int)(C))) |
| 574 |
#define _mm_mask_shrdi_epi64(A, B, C, D, E) \ |
| 575 |
((__m128i) __builtin_ia32_vpshrd_v2di_mask ((__v2di)(__m128i)(C), \ |
| 576 |
(__v2di)(__m128i)(D), (int)(E), \ |
| 577 |
(__v2di)(__m128i)(A), \ |
| 578 |
(__mmask8)(B))) |
| 579 |
#define _mm_maskz_shrdi_epi64(A, B, C, D) \ |
| 580 |
((__m128i) \ |
| 581 |
__builtin_ia32_vpshrd_v2di_mask ((__v2di)(__m128i)(B), \ |
| 582 |
(__v2di)(__m128i)(C),(int)(D), \ |
| 583 |
(__v2di)(__m128i)_mm_setzero_si128 (), \ |
| 584 |
(__mmask8)(A))) |
| 585 |
#define _mm256_shldi_epi16(A, B, C) \ |
| 586 |
((__m256i) __builtin_ia32_vpshld_v16hi ((__v16hi)(__m256i)(A), \ |
| 587 |
(__v16hi)(__m256i)(B),(int)(C))) |
| 588 |
#define _mm256_mask_shldi_epi16(A, B, C, D, E) \ |
| 589 |
((__m256i) __builtin_ia32_vpshld_v16hi_mask ((__v16hi)(__m256i)(C), \ |
| 590 |
(__v16hi)(__m256i)(D), \ |
| 591 |
(int)(E), \ |
| 592 |
(__v16hi)(__m256i)(A), \ |
| 593 |
(__mmask16)(B))) |
| 594 |
#define _mm256_maskz_shldi_epi16(A, B, C, D) \ |
| 595 |
((__m256i) \ |
| 596 |
__builtin_ia32_vpshld_v16hi_mask ((__v16hi)(__m256i)(B), \ |
| 597 |
(__v16hi)(__m256i)(C),(int)(D), \ |
| 598 |
(__v16hi)(__m256i)_mm256_setzero_si256 (), \ |
| 599 |
(__mmask16)(A))) |
| 600 |
#define _mm256_shldi_epi32(A, B, C) \ |
| 601 |
((__m256i) __builtin_ia32_vpshld_v8si ((__v8si)(__m256i)(A), \ |
| 602 |
(__v8si)(__m256i)(B),(int)(C))) |
| 603 |
#define _mm256_mask_shldi_epi32(A, B, C, D, E) \ |
| 604 |
((__m256i) __builtin_ia32_vpshld_v8si_mask ((__v8si)(__m256i)(C), \ |
| 605 |
(__v8si)(__m256i)(D), (int)(E), \ |
| 606 |
(__v8si)(__m256i)(A), \ |
| 607 |
(__mmask8)(B))) |
| 608 |
#define _mm256_maskz_shldi_epi32(A, B, C, D) \ |
| 609 |
((__m256i) \ |
| 610 |
__builtin_ia32_vpshld_v8si_mask ((__v8si)(__m256i)(B), \ |
| 611 |
(__v8si)(__m256i)(C),(int)(D), \ |
| 612 |
(__v8si)(__m256i)_mm256_setzero_si256 (), \ |
| 613 |
(__mmask8)(A))) |
| 614 |
#define _mm256_shldi_epi64(A, B, C) \ |
| 615 |
((__m256i) __builtin_ia32_vpshld_v4di ((__v4di)(__m256i)(A), \ |
| 616 |
(__v4di)(__m256i)(B),(int)(C))) |
| 617 |
#define _mm256_mask_shldi_epi64(A, B, C, D, E) \ |
| 618 |
((__m256i) __builtin_ia32_vpshld_v4di_mask ((__v4di)(__m256i)(C), \ |
| 619 |
(__v4di)(__m256i)(D), (int)(E), \ |
| 620 |
(__v4di)(__m256i)(A), \ |
| 621 |
(__mmask8)(B))) |
| 622 |
#define _mm256_maskz_shldi_epi64(A, B, C, D) \ |
| 623 |
((__m256i) \ |
| 624 |
__builtin_ia32_vpshld_v4di_mask ((__v4di)(__m256i)(B), \ |
| 625 |
(__v4di)(__m256i)(C),(int)(D), \ |
| 626 |
(__v4di)(__m256i)_mm256_setzero_si256 (), \ |
| 627 |
(__mmask8)(A))) |
| 628 |
#define _mm_shldi_epi16(A, B, C) \ |
| 629 |
((__m128i) __builtin_ia32_vpshld_v8hi ((__v8hi)(__m128i)(A), \ |
| 630 |
(__v8hi)(__m128i)(B),(int)(C))) |
| 631 |
#define _mm_mask_shldi_epi16(A, B, C, D, E) \ |
| 632 |
((__m128i) __builtin_ia32_vpshld_v8hi_mask ((__v8hi)(__m128i)(C), \ |
| 633 |
(__v8hi)(__m128i)(D), (int)(E), \ |
| 634 |
(__v8hi)(__m128i)(A), \ |
| 635 |
(__mmask8)(B))) |
| 636 |
#define _mm_maskz_shldi_epi16(A, B, C, D) \ |
| 637 |
((__m128i) \ |
| 638 |
__builtin_ia32_vpshld_v8hi_mask ((__v8hi)(__m128i)(B), \ |
| 639 |
(__v8hi)(__m128i)(C),(int)(D), \ |
| 640 |
(__v8hi)(__m128i)_mm_setzero_si128 (), \ |
| 641 |
(__mmask8)(A))) |
| 642 |
#define _mm_shldi_epi32(A, B, C) \ |
| 643 |
((__m128i) __builtin_ia32_vpshld_v4si ((__v4si)(__m128i)(A), \ |
| 644 |
(__v4si)(__m128i)(B),(int)(C))) |
| 645 |
#define _mm_mask_shldi_epi32(A, B, C, D, E) \ |
| 646 |
((__m128i) __builtin_ia32_vpshld_v4si_mask ((__v4si)(__m128i)(C), \ |
| 647 |
(__v4si)(__m128i)(D), (int)(E), \ |
| 648 |
(__v4si)(__m128i)(A), \ |
| 649 |
(__mmask8)(B))) |
| 650 |
#define _mm_maskz_shldi_epi32(A, B, C, D) \ |
| 651 |
((__m128i) \ |
| 652 |
__builtin_ia32_vpshld_v4si_mask ((__v4si)(__m128i)(B), \ |
| 653 |
(__v4si)(__m128i)(C),(int)(D), \ |
| 654 |
(__v4si)(__m128i)_mm_setzero_si128 (), \ |
| 655 |
(__mmask8)(A))) |
| 656 |
#define _mm_shldi_epi64(A, B, C) \ |
| 657 |
((__m128i) __builtin_ia32_vpshld_v2di ((__v2di)(__m128i)(A), \ |
| 658 |
(__v2di)(__m128i)(B),(int)(C))) |
| 659 |
#define _mm_mask_shldi_epi64(A, B, C, D, E) \ |
| 660 |
((__m128i) __builtin_ia32_vpshld_v2di_mask ((__v2di)(__m128i)(C), \ |
| 661 |
(__v2di)(__m128i)(D), (int)(E), \ |
| 662 |
(__v2di)(__m128i)(A), \ |
| 663 |
(__mmask8)(B))) |
| 664 |
#define _mm_maskz_shldi_epi64(A, B, C, D) \ |
| 665 |
((__m128i) \ |
| 666 |
__builtin_ia32_vpshld_v2di_mask ((__v2di)(__m128i)(B), \ |
| 667 |
(__v2di)(__m128i)(C),(int)(D), \ |
| 668 |
(__v2di)(__m128i)_mm_setzero_si128 (), \ |
| 669 |
(__mmask8)(A))) |
| 670 |
#endif |
| 671 |
|
| 672 |
extern __inline __m256i |
| 673 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 674 |
_mm256_shrdv_epi16 (__m256i __A, __m256i __B, __m256i __C) |
| 675 |
{ |
| 676 |
return (__m256i) __builtin_ia32_vpshrdv_v16hi ((__v16hi)__A, (__v16hi) __B, |
| 677 |
(__v16hi) __C); |
| 678 |
} |
| 679 |
|
| 680 |
extern __inline __m256i |
| 681 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 682 |
_mm256_mask_shrdv_epi16 (__m256i __A, __mmask16 __B, __m256i __C, __m256i __D) |
| 683 |
{ |
| 684 |
return (__m256i)__builtin_ia32_vpshrdv_v16hi_mask ((__v16hi)__A, |
| 685 |
(__v16hi) __C, (__v16hi) __D, (__mmask16)__B); |
| 686 |
} |
| 687 |
|
| 688 |
extern __inline __m256i |
| 689 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 690 |
_mm256_maskz_shrdv_epi16 (__mmask16 __A, __m256i __B, __m256i __C, __m256i __D) |
| 691 |
{ |
| 692 |
return (__m256i)__builtin_ia32_vpshrdv_v16hi_maskz ((__v16hi)__B, |
| 693 |
(__v16hi) __C, (__v16hi) __D, (__mmask16)__A); |
| 694 |
} |
| 695 |
|
| 696 |
extern __inline __m256i |
| 697 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 698 |
_mm256_shrdv_epi32 (__m256i __A, __m256i __B, __m256i __C) |
| 699 |
{ |
| 700 |
return (__m256i) __builtin_ia32_vpshrdv_v8si ((__v8si)__A, (__v8si) __B, |
| 701 |
(__v8si) __C); |
| 702 |
} |
| 703 |
|
| 704 |
extern __inline __m256i |
| 705 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 706 |
_mm256_mask_shrdv_epi32 (__m256i __A, __mmask8 __B, __m256i __C, __m256i __D) |
| 707 |
{ |
| 708 |
return (__m256i)__builtin_ia32_vpshrdv_v8si_mask ((__v8si)__A, (__v8si) __C, |
| 709 |
(__v8si) __D, (__mmask8)__B); |
| 710 |
} |
| 711 |
|
| 712 |
extern __inline __m256i |
| 713 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 714 |
_mm256_maskz_shrdv_epi32 (__mmask8 __A, __m256i __B, __m256i __C, __m256i __D) |
| 715 |
{ |
| 716 |
return (__m256i)__builtin_ia32_vpshrdv_v8si_maskz ((__v8si)__B, (__v8si) __C, |
| 717 |
(__v8si) __D, (__mmask8)__A); |
| 718 |
} |
| 719 |
|
| 720 |
extern __inline __m256i |
| 721 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 722 |
_mm256_shrdv_epi64 (__m256i __A, __m256i __B, __m256i __C) |
| 723 |
{ |
| 724 |
return (__m256i) __builtin_ia32_vpshrdv_v4di ((__v4di)__A, (__v4di) __B, |
| 725 |
(__v4di) __C); |
| 726 |
} |
| 727 |
|
| 728 |
extern __inline __m256i |
| 729 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 730 |
_mm256_mask_shrdv_epi64 (__m256i __A, __mmask8 __B, __m256i __C, __m256i __D) |
| 731 |
{ |
| 732 |
return (__m256i)__builtin_ia32_vpshrdv_v4di_mask ((__v4di)__A, (__v4di) __C, |
| 733 |
(__v4di) __D, (__mmask8)__B); |
| 734 |
} |
| 735 |
|
| 736 |
extern __inline __m256i |
| 737 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 738 |
_mm256_maskz_shrdv_epi64 (__mmask8 __A, __m256i __B, __m256i __C, __m256i __D) |
| 739 |
{ |
| 740 |
return (__m256i)__builtin_ia32_vpshrdv_v4di_maskz ((__v4di)__B, (__v4di) __C, |
| 741 |
(__v4di) __D, (__mmask8)__A); |
| 742 |
} |
| 743 |
|
| 744 |
extern __inline __m128i |
| 745 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 746 |
_mm_shrdv_epi16 (__m128i __A, __m128i __B, __m128i __C) |
| 747 |
{ |
| 748 |
return (__m128i) __builtin_ia32_vpshrdv_v8hi ((__v8hi)__A, (__v8hi) __B, |
| 749 |
(__v8hi) __C); |
| 750 |
} |
| 751 |
|
| 752 |
extern __inline __m128i |
| 753 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 754 |
_mm_mask_shrdv_epi16 (__m128i __A, __mmask8 __B, __m128i __C, __m128i __D) |
| 755 |
{ |
| 756 |
return (__m128i)__builtin_ia32_vpshrdv_v8hi_mask ((__v8hi)__A, (__v8hi) __C, |
| 757 |
(__v8hi) __D, (__mmask8)__B); |
| 758 |
} |
| 759 |
|
| 760 |
extern __inline __m128i |
| 761 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 762 |
_mm_maskz_shrdv_epi16 (__mmask8 __A, __m128i __B, __m128i __C, __m128i __D) |
| 763 |
{ |
| 764 |
return (__m128i)__builtin_ia32_vpshrdv_v8hi_maskz ((__v8hi)__B, (__v8hi) __C, |
| 765 |
(__v8hi) __D, (__mmask8)__A); |
| 766 |
} |
| 767 |
|
| 768 |
extern __inline __m128i |
| 769 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 770 |
_mm_shrdv_epi32 (__m128i __A, __m128i __B, __m128i __C) |
| 771 |
{ |
| 772 |
return (__m128i) __builtin_ia32_vpshrdv_v4si ((__v4si)__A, (__v4si) __B, |
| 773 |
(__v4si) __C); |
| 774 |
} |
| 775 |
|
| 776 |
extern __inline __m128i |
| 777 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 778 |
_mm_mask_shrdv_epi32 (__m128i __A, __mmask8 __B, __m128i __C, __m128i __D) |
| 779 |
{ |
| 780 |
return (__m128i)__builtin_ia32_vpshrdv_v4si_mask ((__v4si)__A, (__v4si) __C, |
| 781 |
(__v4si) __D, (__mmask8)__B); |
| 782 |
} |
| 783 |
|
| 784 |
extern __inline __m128i |
| 785 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 786 |
_mm_maskz_shrdv_epi32 (__mmask8 __A, __m128i __B, __m128i __C, __m128i __D) |
| 787 |
{ |
| 788 |
return (__m128i)__builtin_ia32_vpshrdv_v4si_maskz ((__v4si)__B, (__v4si) __C, |
| 789 |
(__v4si) __D, (__mmask8)__A); |
| 790 |
} |
| 791 |
|
| 792 |
extern __inline __m128i |
| 793 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 794 |
_mm_shrdv_epi64 (__m128i __A, __m128i __B, __m128i __C) |
| 795 |
{ |
| 796 |
return (__m128i) __builtin_ia32_vpshrdv_v2di ((__v2di)__A, (__v2di) __B, |
| 797 |
(__v2di) __C); |
| 798 |
} |
| 799 |
|
| 800 |
extern __inline __m128i |
| 801 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 802 |
_mm_mask_shrdv_epi64 (__m128i __A, __mmask8 __B, __m128i __C, __m128i __D) |
| 803 |
{ |
| 804 |
return (__m128i)__builtin_ia32_vpshrdv_v2di_mask ((__v2di)__A, (__v2di) __C, |
| 805 |
(__v2di) __D, (__mmask8)__B); |
| 806 |
} |
| 807 |
|
| 808 |
extern __inline __m128i |
| 809 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 810 |
_mm_maskz_shrdv_epi64 (__mmask8 __A, __m128i __B, __m128i __C, __m128i __D) |
| 811 |
{ |
| 812 |
return (__m128i)__builtin_ia32_vpshrdv_v2di_maskz ((__v2di)__B, (__v2di) __C, |
| 813 |
(__v2di) __D, (__mmask8)__A); |
| 814 |
} |
| 815 |
|
| 816 |
extern __inline __m256i |
| 817 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 818 |
_mm256_shldv_epi16 (__m256i __A, __m256i __B, __m256i __C) |
| 819 |
{ |
| 820 |
return (__m256i) __builtin_ia32_vpshldv_v16hi ((__v16hi)__A, (__v16hi) __B, |
| 821 |
(__v16hi) __C); |
| 822 |
} |
| 823 |
|
| 824 |
extern __inline __m256i |
| 825 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 826 |
_mm256_mask_shldv_epi16 (__m256i __A, __mmask16 __B, __m256i __C, __m256i __D) |
| 827 |
{ |
| 828 |
return (__m256i)__builtin_ia32_vpshldv_v16hi_mask ((__v16hi)__A, |
| 829 |
(__v16hi) __C, (__v16hi) __D, (__mmask16)__B); |
| 830 |
} |
| 831 |
|
| 832 |
extern __inline __m256i |
| 833 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 834 |
_mm256_maskz_shldv_epi16 (__mmask16 __A, __m256i __B, __m256i __C, __m256i __D) |
| 835 |
{ |
| 836 |
return (__m256i)__builtin_ia32_vpshldv_v16hi_maskz ((__v16hi)__B, |
| 837 |
(__v16hi) __C, (__v16hi) __D, (__mmask16)__A); |
| 838 |
} |
| 839 |
|
| 840 |
extern __inline __m256i |
| 841 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 842 |
_mm256_shldv_epi32 (__m256i __A, __m256i __B, __m256i __C) |
| 843 |
{ |
| 844 |
return (__m256i) __builtin_ia32_vpshldv_v8si ((__v8si)__A, (__v8si) __B, |
| 845 |
(__v8si) __C); |
| 846 |
} |
| 847 |
|
| 848 |
extern __inline __m256i |
| 849 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 850 |
_mm256_mask_shldv_epi32 (__m256i __A, __mmask8 __B, __m256i __C, __m256i __D) |
| 851 |
{ |
| 852 |
return (__m256i)__builtin_ia32_vpshldv_v8si_mask ((__v8si)__A, (__v8si) __C, |
| 853 |
(__v8si) __D, (__mmask8)__B) ; |
| 854 |
} |
| 855 |
|
| 856 |
extern __inline __m256i |
| 857 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 858 |
_mm256_maskz_shldv_epi32 (__mmask8 __A, __m256i __B, __m256i __C, __m256i __D) |
| 859 |
{ |
| 860 |
return (__m256i)__builtin_ia32_vpshldv_v8si_maskz ((__v8si)__B, (__v8si) __C, |
| 861 |
(__v8si) __D, (__mmask8)__A); |
| 862 |
} |
| 863 |
|
| 864 |
extern __inline __m256i |
| 865 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 866 |
_mm256_shldv_epi64 (__m256i __A, __m256i __B, __m256i __C) |
| 867 |
{ |
| 868 |
return (__m256i) __builtin_ia32_vpshldv_v4di ((__v4di)__A, (__v4di) __B, |
| 869 |
(__v4di) __C); |
| 870 |
} |
| 871 |
|
| 872 |
extern __inline __m256i |
| 873 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 874 |
_mm256_mask_shldv_epi64 (__m256i __A, __mmask8 __B, __m256i __C, __m256i __D) |
| 875 |
{ |
| 876 |
return (__m256i)__builtin_ia32_vpshldv_v4di_mask ((__v4di)__A, (__v4di) __C, |
| 877 |
(__v4di) __D, (__mmask8)__B); |
| 878 |
} |
| 879 |
|
| 880 |
extern __inline __m256i |
| 881 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 882 |
_mm256_maskz_shldv_epi64 (__mmask8 __A, __m256i __B, __m256i __C, __m256i __D) |
| 883 |
{ |
| 884 |
return (__m256i)__builtin_ia32_vpshldv_v4di_maskz ((__v4di)__B, (__v4di) __C, |
| 885 |
(__v4di) __D, (__mmask8)__A); |
| 886 |
} |
| 887 |
|
| 888 |
extern __inline __m128i |
| 889 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 890 |
_mm_shldv_epi16 (__m128i __A, __m128i __B, __m128i __C) |
| 891 |
{ |
| 892 |
return (__m128i) __builtin_ia32_vpshldv_v8hi ((__v8hi)__A, (__v8hi) __B, |
| 893 |
(__v8hi) __C); |
| 894 |
} |
| 895 |
|
| 896 |
extern __inline __m128i |
| 897 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 898 |
_mm_mask_shldv_epi16 (__m128i __A, __mmask8 __B, __m128i __C, __m128i __D) |
| 899 |
{ |
| 900 |
return (__m128i)__builtin_ia32_vpshldv_v8hi_mask ((__v8hi)__A, (__v8hi) __C, |
| 901 |
(__v8hi) __D, (__mmask8)__B); |
| 902 |
} |
| 903 |
|
| 904 |
extern __inline __m128i |
| 905 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 906 |
_mm_maskz_shldv_epi16 (__mmask8 __A, __m128i __B, __m128i __C, __m128i __D) |
| 907 |
{ |
| 908 |
return (__m128i)__builtin_ia32_vpshldv_v8hi_maskz ((__v8hi)__B, (__v8hi) __C, |
| 909 |
(__v8hi) __D, (__mmask8)__A); |
| 910 |
} |
| 911 |
|
| 912 |
extern __inline __m128i |
| 913 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 914 |
_mm_shldv_epi32 (__m128i __A, __m128i __B, __m128i __C) |
| 915 |
{ |
| 916 |
return (__m128i) __builtin_ia32_vpshldv_v4si ((__v4si)__A, (__v4si) __B, |
| 917 |
(__v4si) __C); |
| 918 |
} |
| 919 |
|
| 920 |
extern __inline __m128i |
| 921 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 922 |
_mm_mask_shldv_epi32 (__m128i __A, __mmask8 __B, __m128i __C, __m128i __D) |
| 923 |
{ |
| 924 |
return (__m128i)__builtin_ia32_vpshldv_v4si_mask ((__v4si)__A, (__v4si) __C, |
| 925 |
(__v4si) __D, (__mmask8)__B); |
| 926 |
} |
| 927 |
|
| 928 |
extern __inline __m128i |
| 929 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 930 |
_mm_maskz_shldv_epi32 (__mmask8 __A, __m128i __B, __m128i __C, __m128i __D) |
| 931 |
{ |
| 932 |
return (__m128i)__builtin_ia32_vpshldv_v4si_maskz ((__v4si)__B, (__v4si) __C, |
| 933 |
(__v4si) __D, (__mmask8)__A); |
| 934 |
} |
| 935 |
|
| 936 |
extern __inline __m128i |
| 937 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 938 |
_mm_shldv_epi64 (__m128i __A, __m128i __B, __m128i __C) |
| 939 |
{ |
| 940 |
return (__m128i) __builtin_ia32_vpshldv_v2di ((__v2di)__A, (__v2di) __B, |
| 941 |
(__v2di) __C); |
| 942 |
} |
| 943 |
|
| 944 |
extern __inline __m128i |
| 945 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 946 |
_mm_mask_shldv_epi64 (__m128i __A, __mmask8 __B, __m128i __C, __m128i __D) |
| 947 |
{ |
| 948 |
return (__m128i)__builtin_ia32_vpshldv_v2di_mask ((__v2di)__A, (__v2di) __C, |
| 949 |
(__v2di) __D, (__mmask8)__B); |
| 950 |
} |
| 951 |
|
| 952 |
extern __inline __m128i |
| 953 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 954 |
_mm_maskz_shldv_epi64 (__mmask8 __A, __m128i __B, __m128i __C, __m128i __D) |
| 955 |
{ |
| 956 |
return (__m128i)__builtin_ia32_vpshldv_v2di_maskz ((__v2di)__B, (__v2di) __C, |
| 957 |
(__v2di) __D, (__mmask8)__A); |
| 958 |
} |
| 959 |
|
| 960 |
|
| 961 |
|
| 962 |
|
| 963 |
#ifdef __DISABLE_AVX512VBMI2VL__ |
| 964 |
#undef __DISABLE_AVX512VBMI2VL__ |
| 965 |
#pragma GCC pop_options |
| 966 |
#endif /* __DISABLE_AVX512VBMIVL__ */ |
| 967 |
|
| 968 |
#if !defined(__AVX512VL__) || !defined(__AVX512VBMI2__) || \ |
| 969 |
!defined(__AVX512BW__) |
| 970 |
#pragma GCC push_options |
| 971 |
#pragma GCC target("avx512vbmi2,avx512vl,avx512bw") |
| 972 |
#define __DISABLE_AVX512VBMI2VLBW__ |
| 973 |
#endif /* __AVX512VBMIVLBW__ */ |
| 974 |
|
| 975 |
extern __inline __m256i |
| 976 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 977 |
_mm256_mask_compress_epi8 (__m256i __A, __mmask32 __B, __m256i __C) |
| 978 |
{ |
| 979 |
return (__m256i) __builtin_ia32_compressqi256_mask ((__v32qi)__C, |
| 980 |
(__v32qi)__A, (__mmask32)__B); |
| 981 |
} |
| 982 |
|
| 983 |
extern __inline __m256i |
| 984 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 985 |
_mm256_maskz_compress_epi8 (__mmask32 __A, __m256i __B) |
| 986 |
{ |
| 987 |
return (__m256i) __builtin_ia32_compressqi256_mask ((__v32qi) __B, |
| 988 |
(__v32qi) _mm256_setzero_si256 (), (__mmask32) __A); |
| 989 |
} |
| 990 |
|
| 991 |
extern __inline void |
| 992 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 993 |
_mm256_mask_compressstoreu_epi8 (void * __A, __mmask32 __B, __m256i __C) |
| 994 |
{ |
| 995 |
__builtin_ia32_compressstoreuqi256_mask ((__v32qi *) __A, (__v32qi) __C, |
| 996 |
(__mmask32) __B); |
| 997 |
} |
| 998 |
|
| 999 |
extern __inline __m256i |
| 1000 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 1001 |
_mm256_mask_expand_epi8 (__m256i __A, __mmask32 __B, __m256i __C) |
| 1002 |
{ |
| 1003 |
return (__m256i) __builtin_ia32_expandqi256_mask ((__v32qi) __C, |
| 1004 |
(__v32qi) __A, |
| 1005 |
(__mmask32) __B); |
| 1006 |
} |
| 1007 |
|
| 1008 |
extern __inline __m256i |
| 1009 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 1010 |
_mm256_maskz_expand_epi8 (__mmask32 __A, __m256i __B) |
| 1011 |
{ |
| 1012 |
return (__m256i) __builtin_ia32_expandqi256_maskz ((__v32qi) __B, |
| 1013 |
(__v32qi) _mm256_setzero_si256 (), (__mmask32) __A); |
| 1014 |
} |
| 1015 |
|
| 1016 |
extern __inline __m256i |
| 1017 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 1018 |
_mm256_mask_expandloadu_epi8 (__m256i __A, __mmask32 __B, const void * __C) |
| 1019 |
{ |
| 1020 |
return (__m256i) __builtin_ia32_expandloadqi256_mask ((const __v32qi *) __C, |
| 1021 |
(__v32qi) __A, (__mmask32) __B); |
| 1022 |
} |
| 1023 |
|
| 1024 |
extern __inline __m256i |
| 1025 |
__attribute__((__gnu_inline__, __always_inline__, __artificial__)) |
| 1026 |
_mm256_maskz_expandloadu_epi8 (__mmask32 __A, const void * __B) |
| 1027 |
{ |
| 1028 |
return (__m256i) __builtin_ia32_expandloadqi256_maskz ((const __v32qi *) __B, |
| 1029 |
(__v32qi) _mm256_setzero_si256 (), (__mmask32) __A); |
| 1030 |
} |
| 1031 |
|
| 1032 |
#ifdef __DISABLE_AVX512VBMI2VLBW__ |
| 1033 |
#undef __DISABLE_AVX512VBMI2VLBW__ |
| 1034 |
#pragma GCC pop_options |
| 1035 |
#endif /* __DISABLE_AVX512VBMIVLBW__ */ |
| 1036 |
|
| 1037 |
#endif /* _AVX512VBMIVLINTRIN_H_INCLUDED */ |
