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
28 #ifndef __VBMIINTRIN_H
29 #define __VBMIINTRIN_H
30
31 /* Define the default attributes for the functions in this file. */
32 #define __DEFAULT_FN_ATTRS __attribute__((__always_inline__, __nodebug__, __target__("avx512vbmi"), __min_vector_width__(512)))
33
34
35 static __inline__ __m512i __DEFAULT_FN_ATTRS
_mm512_permutex2var_epi8(__m512i __A,__m512i __I,__m512i __B)36 _mm512_permutex2var_epi8(__m512i __A, __m512i __I, __m512i __B)
37 {
38 return (__m512i)__builtin_ia32_vpermi2varqi512((__v64qi)__A, (__v64qi)__I,
39 (__v64qi) __B);
40 }
41
42 static __inline__ __m512i __DEFAULT_FN_ATTRS
_mm512_mask_permutex2var_epi8(__m512i __A,__mmask64 __U,__m512i __I,__m512i __B)43 _mm512_mask_permutex2var_epi8(__m512i __A, __mmask64 __U, __m512i __I,
44 __m512i __B)
45 {
46 return (__m512i)__builtin_ia32_selectb_512(__U,
47 (__v64qi)_mm512_permutex2var_epi8(__A, __I, __B),
48 (__v64qi)__A);
49 }
50
51 static __inline__ __m512i __DEFAULT_FN_ATTRS
_mm512_mask2_permutex2var_epi8(__m512i __A,__m512i __I,__mmask64 __U,__m512i __B)52 _mm512_mask2_permutex2var_epi8(__m512i __A, __m512i __I, __mmask64 __U,
53 __m512i __B)
54 {
55 return (__m512i)__builtin_ia32_selectb_512(__U,
56 (__v64qi)_mm512_permutex2var_epi8(__A, __I, __B),
57 (__v64qi)__I);
58 }
59
60 static __inline__ __m512i __DEFAULT_FN_ATTRS
_mm512_maskz_permutex2var_epi8(__mmask64 __U,__m512i __A,__m512i __I,__m512i __B)61 _mm512_maskz_permutex2var_epi8(__mmask64 __U, __m512i __A, __m512i __I,
62 __m512i __B)
63 {
64 return (__m512i)__builtin_ia32_selectb_512(__U,
65 (__v64qi)_mm512_permutex2var_epi8(__A, __I, __B),
66 (__v64qi)_mm512_setzero_si512());
67 }
68
69 static __inline__ __m512i __DEFAULT_FN_ATTRS
_mm512_permutexvar_epi8(__m512i __A,__m512i __B)70 _mm512_permutexvar_epi8 (__m512i __A, __m512i __B)
71 {
72 return (__m512i)__builtin_ia32_permvarqi512((__v64qi) __B, (__v64qi) __A);
73 }
74
75 static __inline__ __m512i __DEFAULT_FN_ATTRS
_mm512_maskz_permutexvar_epi8(__mmask64 __M,__m512i __A,__m512i __B)76 _mm512_maskz_permutexvar_epi8 (__mmask64 __M, __m512i __A,
77 __m512i __B)
78 {
79 return (__m512i)__builtin_ia32_selectb_512((__mmask64)__M,
80 (__v64qi)_mm512_permutexvar_epi8(__A, __B),
81 (__v64qi)_mm512_setzero_si512());
82 }
83
84 static __inline__ __m512i __DEFAULT_FN_ATTRS
_mm512_mask_permutexvar_epi8(__m512i __W,__mmask64 __M,__m512i __A,__m512i __B)85 _mm512_mask_permutexvar_epi8 (__m512i __W, __mmask64 __M, __m512i __A,
86 __m512i __B)
87 {
88 return (__m512i)__builtin_ia32_selectb_512((__mmask64)__M,
89 (__v64qi)_mm512_permutexvar_epi8(__A, __B),
90 (__v64qi)__W);
91 }
92
93 static __inline__ __m512i __DEFAULT_FN_ATTRS
_mm512_mask_multishift_epi64_epi8(__m512i __W,__mmask64 __M,__m512i __X,__m512i __Y)94 _mm512_mask_multishift_epi64_epi8 (__m512i __W, __mmask64 __M, __m512i __X, __m512i __Y)
95 {
96 return (__m512i) __builtin_ia32_vpmultishiftqb512_mask ((__v64qi) __X,
97 (__v64qi) __Y,
98 (__v64qi) __W,
99 (__mmask64) __M);
100 }
101
102 static __inline__ __m512i __DEFAULT_FN_ATTRS
_mm512_maskz_multishift_epi64_epi8(__mmask64 __M,__m512i __X,__m512i __Y)103 _mm512_maskz_multishift_epi64_epi8 (__mmask64 __M, __m512i __X, __m512i __Y)
104 {
105 return (__m512i) __builtin_ia32_vpmultishiftqb512_mask ((__v64qi) __X,
106 (__v64qi) __Y,
107 (__v64qi) _mm512_setzero_si512 (),
108 (__mmask64) __M);
109 }
110
111 static __inline__ __m512i __DEFAULT_FN_ATTRS
_mm512_multishift_epi64_epi8(__m512i __X,__m512i __Y)112 _mm512_multishift_epi64_epi8 (__m512i __X, __m512i __Y)
113 {
114 return (__m512i) __builtin_ia32_vpmultishiftqb512_mask ((__v64qi) __X,
115 (__v64qi) __Y,
116 (__v64qi) _mm512_undefined_epi32 (),
117 (__mmask64) -1);
118 }
119
120
121 #undef __DEFAULT_FN_ATTRS
122
123 #endif
124