1 /*===---- avx512vlbitalgintrin.h - BITALG intrinsics -----------------------===
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:
11 * The above copyright notice and this permission notice shall be included in
12 * all copies or substantial portions of the Software.
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
22 *===-----------------------------------------------------------------------===
25 #error "Never use <avx512vlbitalgintrin.h> directly; include <immintrin.h> instead."
28 #ifndef __AVX512VLBITALGINTRIN_H
29 #define __AVX512VLBITALGINTRIN_H
31 /* Define the default attributes for the functions in this file. */
32 #define __DEFAULT_FN_ATTRS128 __attribute__((__always_inline__, __nodebug__, __target__("avx512vl,avx512bitalg"), __min_vector_width__(128)))
33 #define __DEFAULT_FN_ATTRS256 __attribute__((__always_inline__, __nodebug__, __target__("avx512vl,avx512bitalg"), __min_vector_width__(256)))
35 static __inline__ __m256i __DEFAULT_FN_ATTRS256
36 _mm256_popcnt_epi16(__m256i __A)
38 return (__m256i) __builtin_ia32_vpopcntw_256((__v16hi) __A);
41 static __inline__ __m256i __DEFAULT_FN_ATTRS256
42 _mm256_mask_popcnt_epi16(__m256i __A, __mmask16 __U, __m256i __B)
44 return (__m256i) __builtin_ia32_selectw_256((__mmask16) __U,
45 (__v16hi) _mm256_popcnt_epi16(__B),
49 static __inline__ __m256i __DEFAULT_FN_ATTRS256
50 _mm256_maskz_popcnt_epi16(__mmask16 __U, __m256i __B)
52 return _mm256_mask_popcnt_epi16((__m256i) _mm256_setzero_si256(),
57 static __inline__ __m128i __DEFAULT_FN_ATTRS128
58 _mm_popcnt_epi16(__m128i __A)
60 return (__m128i) __builtin_ia32_vpopcntw_128((__v8hi) __A);
63 static __inline__ __m128i __DEFAULT_FN_ATTRS128
64 _mm_mask_popcnt_epi16(__m128i __A, __mmask8 __U, __m128i __B)
66 return (__m128i) __builtin_ia32_selectw_128((__mmask8) __U,
67 (__v8hi) _mm_popcnt_epi16(__B),
71 static __inline__ __m128i __DEFAULT_FN_ATTRS128
72 _mm_maskz_popcnt_epi16(__mmask8 __U, __m128i __B)
74 return _mm_mask_popcnt_epi16((__m128i) _mm_setzero_si128(),
79 static __inline__ __m256i __DEFAULT_FN_ATTRS256
80 _mm256_popcnt_epi8(__m256i __A)
82 return (__m256i) __builtin_ia32_vpopcntb_256((__v32qi) __A);
85 static __inline__ __m256i __DEFAULT_FN_ATTRS256
86 _mm256_mask_popcnt_epi8(__m256i __A, __mmask32 __U, __m256i __B)
88 return (__m256i) __builtin_ia32_selectb_256((__mmask32) __U,
89 (__v32qi) _mm256_popcnt_epi8(__B),
93 static __inline__ __m256i __DEFAULT_FN_ATTRS256
94 _mm256_maskz_popcnt_epi8(__mmask32 __U, __m256i __B)
96 return _mm256_mask_popcnt_epi8((__m256i) _mm256_setzero_si256(),
101 static __inline__ __m128i __DEFAULT_FN_ATTRS128
102 _mm_popcnt_epi8(__m128i __A)
104 return (__m128i) __builtin_ia32_vpopcntb_128((__v16qi) __A);
107 static __inline__ __m128i __DEFAULT_FN_ATTRS128
108 _mm_mask_popcnt_epi8(__m128i __A, __mmask16 __U, __m128i __B)
110 return (__m128i) __builtin_ia32_selectb_128((__mmask16) __U,
111 (__v16qi) _mm_popcnt_epi8(__B),
115 static __inline__ __m128i __DEFAULT_FN_ATTRS128
116 _mm_maskz_popcnt_epi8(__mmask16 __U, __m128i __B)
118 return _mm_mask_popcnt_epi8((__m128i) _mm_setzero_si128(),
123 static __inline__ __mmask32 __DEFAULT_FN_ATTRS256
124 _mm256_mask_bitshuffle_epi64_mask(__mmask32 __U, __m256i __A, __m256i __B)
126 return (__mmask32) __builtin_ia32_vpshufbitqmb256_mask((__v32qi) __A,
131 static __inline__ __mmask32 __DEFAULT_FN_ATTRS256
132 _mm256_bitshuffle_epi64_mask(__m256i __A, __m256i __B)
134 return _mm256_mask_bitshuffle_epi64_mask((__mmask32) -1,
139 static __inline__ __mmask16 __DEFAULT_FN_ATTRS128
140 _mm_mask_bitshuffle_epi64_mask(__mmask16 __U, __m128i __A, __m128i __B)
142 return (__mmask16) __builtin_ia32_vpshufbitqmb128_mask((__v16qi) __A,
147 static __inline__ __mmask16 __DEFAULT_FN_ATTRS128
148 _mm_bitshuffle_epi64_mask(__m128i __A, __m128i __B)
150 return _mm_mask_bitshuffle_epi64_mask((__mmask16) -1,
156 #undef __DEFAULT_FN_ATTRS128
157 #undef __DEFAULT_FN_ATTRS256