avx512vbmiintrin.h 4.6 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123
  1. /*===------------- avx512vbmiintrin.h - VBMI intrinsics ------------------===
  2. *
  3. *
  4. * Permission is hereby granted, free of charge, to any person obtaining a copy
  5. * of this software and associated documentation files (the "Software"), to deal
  6. * in the Software without restriction, including without limitation the rights
  7. * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
  8. * copies of the Software, and to permit persons to whom the Software is
  9. * furnished to do so, subject to the following conditions:
  10. *
  11. * The above copyright notice and this permission notice shall be included in
  12. * all copies or substantial portions of the Software.
  13. *
  14. * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
  15. * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
  16. * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
  17. * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
  18. * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
  19. * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
  20. * THE SOFTWARE.
  21. *
  22. *===-----------------------------------------------------------------------===
  23. */
  24. #ifndef __IMMINTRIN_H
  25. #error "Never use <avx512vbmiintrin.h> directly; include <immintrin.h> instead."
  26. #endif
  27. #ifndef __VBMIINTRIN_H
  28. #define __VBMIINTRIN_H
  29. /* Define the default attributes for the functions in this file. */
  30. #define __DEFAULT_FN_ATTRS __attribute__((__always_inline__, __nodebug__, __target__("avx512vbmi"), __min_vector_width__(512)))
  31. static __inline__ __m512i __DEFAULT_FN_ATTRS
  32. _mm512_permutex2var_epi8(__m512i __A, __m512i __I, __m512i __B)
  33. {
  34. return (__m512i)__builtin_ia32_vpermi2varqi512((__v64qi)__A, (__v64qi)__I,
  35. (__v64qi) __B);
  36. }
  37. static __inline__ __m512i __DEFAULT_FN_ATTRS
  38. _mm512_mask_permutex2var_epi8(__m512i __A, __mmask64 __U, __m512i __I,
  39. __m512i __B)
  40. {
  41. return (__m512i)__builtin_ia32_selectb_512(__U,
  42. (__v64qi)_mm512_permutex2var_epi8(__A, __I, __B),
  43. (__v64qi)__A);
  44. }
  45. static __inline__ __m512i __DEFAULT_FN_ATTRS
  46. _mm512_mask2_permutex2var_epi8(__m512i __A, __m512i __I, __mmask64 __U,
  47. __m512i __B)
  48. {
  49. return (__m512i)__builtin_ia32_selectb_512(__U,
  50. (__v64qi)_mm512_permutex2var_epi8(__A, __I, __B),
  51. (__v64qi)__I);
  52. }
  53. static __inline__ __m512i __DEFAULT_FN_ATTRS
  54. _mm512_maskz_permutex2var_epi8(__mmask64 __U, __m512i __A, __m512i __I,
  55. __m512i __B)
  56. {
  57. return (__m512i)__builtin_ia32_selectb_512(__U,
  58. (__v64qi)_mm512_permutex2var_epi8(__A, __I, __B),
  59. (__v64qi)_mm512_setzero_si512());
  60. }
  61. static __inline__ __m512i __DEFAULT_FN_ATTRS
  62. _mm512_permutexvar_epi8 (__m512i __A, __m512i __B)
  63. {
  64. return (__m512i)__builtin_ia32_permvarqi512((__v64qi) __B, (__v64qi) __A);
  65. }
  66. static __inline__ __m512i __DEFAULT_FN_ATTRS
  67. _mm512_maskz_permutexvar_epi8 (__mmask64 __M, __m512i __A,
  68. __m512i __B)
  69. {
  70. return (__m512i)__builtin_ia32_selectb_512((__mmask64)__M,
  71. (__v64qi)_mm512_permutexvar_epi8(__A, __B),
  72. (__v64qi)_mm512_setzero_si512());
  73. }
  74. static __inline__ __m512i __DEFAULT_FN_ATTRS
  75. _mm512_mask_permutexvar_epi8 (__m512i __W, __mmask64 __M, __m512i __A,
  76. __m512i __B)
  77. {
  78. return (__m512i)__builtin_ia32_selectb_512((__mmask64)__M,
  79. (__v64qi)_mm512_permutexvar_epi8(__A, __B),
  80. (__v64qi)__W);
  81. }
  82. static __inline__ __m512i __DEFAULT_FN_ATTRS
  83. _mm512_mask_multishift_epi64_epi8 (__m512i __W, __mmask64 __M, __m512i __X, __m512i __Y)
  84. {
  85. return (__m512i) __builtin_ia32_vpmultishiftqb512_mask ((__v64qi) __X,
  86. (__v64qi) __Y,
  87. (__v64qi) __W,
  88. (__mmask64) __M);
  89. }
  90. static __inline__ __m512i __DEFAULT_FN_ATTRS
  91. _mm512_maskz_multishift_epi64_epi8 (__mmask64 __M, __m512i __X, __m512i __Y)
  92. {
  93. return (__m512i) __builtin_ia32_vpmultishiftqb512_mask ((__v64qi) __X,
  94. (__v64qi) __Y,
  95. (__v64qi) _mm512_setzero_si512 (),
  96. (__mmask64) __M);
  97. }
  98. static __inline__ __m512i __DEFAULT_FN_ATTRS
  99. _mm512_multishift_epi64_epi8 (__m512i __X, __m512i __Y)
  100. {
  101. return (__m512i) __builtin_ia32_vpmultishiftqb512_mask ((__v64qi) __X,
  102. (__v64qi) __Y,
  103. (__v64qi) _mm512_undefined_epi32 (),
  104. (__mmask64) -1);
  105. }
  106. #undef __DEFAULT_FN_ATTRS
  107. #endif