1 /*===------------- avx512vbmivlintrin.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 <avx512vbmivlintrin.h> directly; include <immintrin.h> instead."
26 #endif
27
28 #ifndef __VBMIVLINTRIN_H
29 #define __VBMIVLINTRIN_H
30
31 /* Define the default attributes for the functions in this file. */
32 #define __DEFAULT_FN_ATTRS128 __attribute__((__always_inline__, __nodebug__, __target__("avx512vbmi,avx512vl"), __min_vector_width__(128)))
33 #define __DEFAULT_FN_ATTRS256 __attribute__((__always_inline__, __nodebug__, __target__("avx512vbmi,avx512vl"), __min_vector_width__(256)))
34
35
36 static __inline__ __m128i __DEFAULT_FN_ATTRS128
_mm_permutex2var_epi8(__m128i __A,__m128i __I,__m128i __B)37 _mm_permutex2var_epi8(__m128i __A, __m128i __I, __m128i __B)
38 {
39 return (__m128i)__builtin_ia32_vpermi2varqi128((__v16qi)__A,
40 (__v16qi)__I,
41 (__v16qi)__B);
42 }
43
44 static __inline__ __m128i __DEFAULT_FN_ATTRS128
_mm_mask_permutex2var_epi8(__m128i __A,__mmask16 __U,__m128i __I,__m128i __B)45 _mm_mask_permutex2var_epi8(__m128i __A, __mmask16 __U, __m128i __I,
46 __m128i __B)
47 {
48 return (__m128i)__builtin_ia32_selectb_128(__U,
49 (__v16qi)_mm_permutex2var_epi8(__A, __I, __B),
50 (__v16qi)__A);
51 }
52
53 static __inline__ __m128i __DEFAULT_FN_ATTRS128
_mm_mask2_permutex2var_epi8(__m128i __A,__m128i __I,__mmask16 __U,__m128i __B)54 _mm_mask2_permutex2var_epi8(__m128i __A, __m128i __I, __mmask16 __U,
55 __m128i __B)
56 {
57 return (__m128i)__builtin_ia32_selectb_128(__U,
58 (__v16qi)_mm_permutex2var_epi8(__A, __I, __B),
59 (__v16qi)__I);
60 }
61
62 static __inline__ __m128i __DEFAULT_FN_ATTRS128
_mm_maskz_permutex2var_epi8(__mmask16 __U,__m128i __A,__m128i __I,__m128i __B)63 _mm_maskz_permutex2var_epi8(__mmask16 __U, __m128i __A, __m128i __I,
64 __m128i __B)
65 {
66 return (__m128i)__builtin_ia32_selectb_128(__U,
67 (__v16qi)_mm_permutex2var_epi8(__A, __I, __B),
68 (__v16qi)_mm_setzero_si128());
69 }
70
71 static __inline__ __m256i __DEFAULT_FN_ATTRS256
_mm256_permutex2var_epi8(__m256i __A,__m256i __I,__m256i __B)72 _mm256_permutex2var_epi8(__m256i __A, __m256i __I, __m256i __B)
73 {
74 return (__m256i)__builtin_ia32_vpermi2varqi256((__v32qi)__A, (__v32qi)__I,
75 (__v32qi)__B);
76 }
77
78 static __inline__ __m256i __DEFAULT_FN_ATTRS256
_mm256_mask_permutex2var_epi8(__m256i __A,__mmask32 __U,__m256i __I,__m256i __B)79 _mm256_mask_permutex2var_epi8(__m256i __A, __mmask32 __U, __m256i __I,
80 __m256i __B)
81 {
82 return (__m256i)__builtin_ia32_selectb_256(__U,
83 (__v32qi)_mm256_permutex2var_epi8(__A, __I, __B),
84 (__v32qi)__A);
85 }
86
87 static __inline__ __m256i __DEFAULT_FN_ATTRS256
_mm256_mask2_permutex2var_epi8(__m256i __A,__m256i __I,__mmask32 __U,__m256i __B)88 _mm256_mask2_permutex2var_epi8(__m256i __A, __m256i __I, __mmask32 __U,
89 __m256i __B)
90 {
91 return (__m256i)__builtin_ia32_selectb_256(__U,
92 (__v32qi)_mm256_permutex2var_epi8(__A, __I, __B),
93 (__v32qi)__I);
94 }
95
96 static __inline__ __m256i __DEFAULT_FN_ATTRS256
_mm256_maskz_permutex2var_epi8(__mmask32 __U,__m256i __A,__m256i __I,__m256i __B)97 _mm256_maskz_permutex2var_epi8(__mmask32 __U, __m256i __A, __m256i __I,
98 __m256i __B)
99 {
100 return (__m256i)__builtin_ia32_selectb_256(__U,
101 (__v32qi)_mm256_permutex2var_epi8(__A, __I, __B),
102 (__v32qi)_mm256_setzero_si256());
103 }
104
105 static __inline__ __m128i __DEFAULT_FN_ATTRS128
_mm_permutexvar_epi8(__m128i __A,__m128i __B)106 _mm_permutexvar_epi8 (__m128i __A, __m128i __B)
107 {
108 return (__m128i)__builtin_ia32_permvarqi128((__v16qi)__B, (__v16qi)__A);
109 }
110
111 static __inline__ __m128i __DEFAULT_FN_ATTRS128
_mm_maskz_permutexvar_epi8(__mmask16 __M,__m128i __A,__m128i __B)112 _mm_maskz_permutexvar_epi8 (__mmask16 __M, __m128i __A, __m128i __B)
113 {
114 return (__m128i)__builtin_ia32_selectb_128((__mmask16)__M,
115 (__v16qi)_mm_permutexvar_epi8(__A, __B),
116 (__v16qi)_mm_setzero_si128());
117 }
118
119 static __inline__ __m128i __DEFAULT_FN_ATTRS128
_mm_mask_permutexvar_epi8(__m128i __W,__mmask16 __M,__m128i __A,__m128i __B)120 _mm_mask_permutexvar_epi8 (__m128i __W, __mmask16 __M, __m128i __A,
121 __m128i __B)
122 {
123 return (__m128i)__builtin_ia32_selectb_128((__mmask16)__M,
124 (__v16qi)_mm_permutexvar_epi8(__A, __B),
125 (__v16qi)__W);
126 }
127
128 static __inline__ __m256i __DEFAULT_FN_ATTRS256
_mm256_permutexvar_epi8(__m256i __A,__m256i __B)129 _mm256_permutexvar_epi8 (__m256i __A, __m256i __B)
130 {
131 return (__m256i)__builtin_ia32_permvarqi256((__v32qi) __B, (__v32qi) __A);
132 }
133
134 static __inline__ __m256i __DEFAULT_FN_ATTRS256
_mm256_maskz_permutexvar_epi8(__mmask32 __M,__m256i __A,__m256i __B)135 _mm256_maskz_permutexvar_epi8 (__mmask32 __M, __m256i __A,
136 __m256i __B)
137 {
138 return (__m256i)__builtin_ia32_selectb_256((__mmask32)__M,
139 (__v32qi)_mm256_permutexvar_epi8(__A, __B),
140 (__v32qi)_mm256_setzero_si256());
141 }
142
143 static __inline__ __m256i __DEFAULT_FN_ATTRS256
_mm256_mask_permutexvar_epi8(__m256i __W,__mmask32 __M,__m256i __A,__m256i __B)144 _mm256_mask_permutexvar_epi8 (__m256i __W, __mmask32 __M, __m256i __A,
145 __m256i __B)
146 {
147 return (__m256i)__builtin_ia32_selectb_256((__mmask32)__M,
148 (__v32qi)_mm256_permutexvar_epi8(__A, __B),
149 (__v32qi)__W);
150 }
151
152 static __inline__ __m128i __DEFAULT_FN_ATTRS128
_mm_mask_multishift_epi64_epi8(__m128i __W,__mmask16 __M,__m128i __X,__m128i __Y)153 _mm_mask_multishift_epi64_epi8 (__m128i __W, __mmask16 __M, __m128i __X, __m128i __Y)
154 {
155 return (__m128i) __builtin_ia32_vpmultishiftqb128_mask ((__v16qi) __X,
156 (__v16qi) __Y,
157 (__v16qi) __W,
158 (__mmask16) __M);
159 }
160
161 static __inline__ __m128i __DEFAULT_FN_ATTRS128
_mm_maskz_multishift_epi64_epi8(__mmask16 __M,__m128i __X,__m128i __Y)162 _mm_maskz_multishift_epi64_epi8 (__mmask16 __M, __m128i __X, __m128i __Y)
163 {
164 return (__m128i) __builtin_ia32_vpmultishiftqb128_mask ((__v16qi) __X,
165 (__v16qi) __Y,
166 (__v16qi)
167 _mm_setzero_si128 (),
168 (__mmask16) __M);
169 }
170
171 static __inline__ __m128i __DEFAULT_FN_ATTRS128
_mm_multishift_epi64_epi8(__m128i __X,__m128i __Y)172 _mm_multishift_epi64_epi8 (__m128i __X, __m128i __Y)
173 {
174 return (__m128i) __builtin_ia32_vpmultishiftqb128_mask ((__v16qi) __X,
175 (__v16qi) __Y,
176 (__v16qi)
177 _mm_undefined_si128 (),
178 (__mmask16) -1);
179 }
180
181 static __inline__ __m256i __DEFAULT_FN_ATTRS256
_mm256_mask_multishift_epi64_epi8(__m256i __W,__mmask32 __M,__m256i __X,__m256i __Y)182 _mm256_mask_multishift_epi64_epi8 (__m256i __W, __mmask32 __M, __m256i __X, __m256i __Y)
183 {
184 return (__m256i) __builtin_ia32_vpmultishiftqb256_mask ((__v32qi) __X,
185 (__v32qi) __Y,
186 (__v32qi) __W,
187 (__mmask32) __M);
188 }
189
190 static __inline__ __m256i __DEFAULT_FN_ATTRS256
_mm256_maskz_multishift_epi64_epi8(__mmask32 __M,__m256i __X,__m256i __Y)191 _mm256_maskz_multishift_epi64_epi8 (__mmask32 __M, __m256i __X, __m256i __Y)
192 {
193 return (__m256i) __builtin_ia32_vpmultishiftqb256_mask ((__v32qi) __X,
194 (__v32qi) __Y,
195 (__v32qi)
196 _mm256_setzero_si256 (),
197 (__mmask32) __M);
198 }
199
200 static __inline__ __m256i __DEFAULT_FN_ATTRS256
_mm256_multishift_epi64_epi8(__m256i __X,__m256i __Y)201 _mm256_multishift_epi64_epi8 (__m256i __X, __m256i __Y)
202 {
203 return (__m256i) __builtin_ia32_vpmultishiftqb256_mask ((__v32qi) __X,
204 (__v32qi) __Y,
205 (__v32qi)
206 _mm256_undefined_si256 (),
207 (__mmask32) -1);
208 }
209
210
211 #undef __DEFAULT_FN_ATTRS128
212 #undef __DEFAULT_FN_ATTRS256
213
214 #endif
215