avx512vpopcntdqvlintrin.h revision 327952 
1/*===------------- avx512vpopcntdqintrin.h - AVX512VPOPCNTDQ intrinsics
2 *------------------===
3 *
4 *
5 * Permission is hereby granted, free of charge, to any person obtaining a copy
6 * of this software and associated documentation files (the "Software"), to deal
7 * in the Software without restriction, including without limitation the rights
8 * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
9 * copies of the Software, and to permit persons to whom the Software is
10 * furnished to do so, subject to the following conditions:
11 *
12 * The above copyright notice and this permission notice shall be included in
13 * all copies or substantial portions of the Software.
14 *
15 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
16 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
17 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
18 * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
19 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
20 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
21 * THE SOFTWARE.
22 *
23 *===-----------------------------------------------------------------------===
24 */
25#ifndef __IMMINTRIN_H
26#error                                                                         \
27    "Never use <avx512vpopcntdqvlintrin.h> directly; include <immintrin.h> instead."
28#endif
29
30#ifndef __AVX512VPOPCNTDQVLINTRIN_H
31#define __AVX512VPOPCNTDQVLINTRIN_H
32
33/* Define the default attributes for the functions in this file. */
34#define __DEFAULT_FN_ATTRS                                                     \
35  __attribute__((__always_inline__, __nodebug__, __target__("avx512vpopcntdq,avx512vl")))
36
37static __inline__ __m128i __DEFAULT_FN_ATTRS _mm_popcnt_epi64(__m128i __A) {
38  return (__m128i)__builtin_ia32_vpopcntq_128((__v2di)__A);
39}
40
41static __inline__ __m128i __DEFAULT_FN_ATTRS
42_mm_mask_popcnt_epi64(__m128i __W, __mmask8 __U, __m128i __A) {
43  return (__m128i)__builtin_ia32_selectq_128(
44      (__mmask8)__U, (__v2di)_mm_popcnt_epi64(__A), (__v2di)__W);
45}
46
47static __inline__ __m128i __DEFAULT_FN_ATTRS
48_mm_maskz_popcnt_epi64(__mmask8 __U, __m128i __A) {
49  return _mm_mask_popcnt_epi64((__m128i)_mm_setzero_si128(), __U, __A);
50}
51
52static __inline__ __m128i __DEFAULT_FN_ATTRS _mm_popcnt_epi32(__m128i __A) {
53  return (__m128i)__builtin_ia32_vpopcntd_128((__v4si)__A);
54}
55
56static __inline__ __m128i __DEFAULT_FN_ATTRS
57_mm_mask_popcnt_epi32(__m128i __W, __mmask8 __U, __m128i __A) {
58  return (__m128i)__builtin_ia32_selectd_128(
59      (__mmask8)__U, (__v4si)_mm_popcnt_epi32(__A), (__v4si)__W);
60}
61
62static __inline__ __m128i __DEFAULT_FN_ATTRS
63_mm_maskz_popcnt_epi32(__mmask8 __U, __m128i __A) {
64  return _mm_mask_popcnt_epi32((__m128i)_mm_setzero_si128(), __U, __A);
65}
66
67static __inline__ __m256i __DEFAULT_FN_ATTRS _mm256_popcnt_epi64(__m256i __A) {
68  return (__m256i)__builtin_ia32_vpopcntq_256((__v4di)__A);
69}
70
71static __inline__ __m256i __DEFAULT_FN_ATTRS
72_mm256_mask_popcnt_epi64(__m256i __W, __mmask8 __U, __m256i __A) {
73  return (__m256i)__builtin_ia32_selectq_256(
74      (__mmask8)__U, (__v4di)_mm256_popcnt_epi64(__A), (__v4di)__W);
75}
76
77static __inline__ __m256i __DEFAULT_FN_ATTRS
78_mm256_maskz_popcnt_epi64(__mmask8 __U, __m256i __A) {
79  return _mm256_mask_popcnt_epi64((__m256i)_mm256_setzero_si256(), __U, __A);
80}
81
82static __inline__ __m256i __DEFAULT_FN_ATTRS _mm256_popcnt_epi32(__m256i __A) {
83  return (__m256i)__builtin_ia32_vpopcntd_256((__v8si)__A);
84}
85
86static __inline__ __m256i __DEFAULT_FN_ATTRS
87_mm256_mask_popcnt_epi32(__m256i __W, __mmask8 __U, __m256i __A) {
88  return (__m256i)__builtin_ia32_selectd_256(
89      (__mmask8)__U, (__v8si)_mm256_popcnt_epi32(__A), (__v8si)__W);
90}
91
92static __inline__ __m256i __DEFAULT_FN_ATTRS
93_mm256_maskz_popcnt_epi32(__mmask8 __U, __m256i __A) {
94  return _mm256_mask_popcnt_epi32((__m256i)_mm256_setzero_si256(), __U, __A);
95}
96
97#undef __DEFAULT_FN_ATTRS
98
99#endif
100