1 /*===---- avx512fintrin.h - AVX2 intrinsics -----------------------------------===
2  *
3  * Permission is hereby granted, free of charge, to any person obtaining a copy
4  * of this software and associated documentation files (the "Software"), to deal
5  * in the Software without restriction, including without limitation the rights
6  * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
7  * copies of the Software, and to permit persons to whom the Software is
8  * furnished to do so, subject to the following conditions:
9  *
10  * The above copyright notice and this permission notice shall be included in
11  * all copies or substantial portions of the Software.
12  *
13  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
14  * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
15  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
16  * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
17  * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
18  * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
19  * THE SOFTWARE.
20  *
21  *===-----------------------------------------------------------------------===
22  */
23 #ifndef __IMMINTRIN_H
24 #error "Never use <avx512erintrin.h> directly; include <immintrin.h> instead."
25 #endif
26 
27 #ifndef __AVX512ERINTRIN_H
28 #define __AVX512ERINTRIN_H
29 
30 
31 // rsqrt28
32 static  __inline__ __m512d __attribute__((__always_inline__, __nodebug__))
_mm512_rsqrt28_round_pd(__m512d __A,int __R)33 _mm512_rsqrt28_round_pd (__m512d __A, int __R)
34 {
35   return (__m512d)__builtin_ia32_rsqrt28pd_mask ((__v8df)__A,
36                                                  (__v8df)_mm512_setzero_pd(),
37                                                  (__mmask8)-1,
38                                                  __R);
39 }
40 static  __inline__ __m512 __attribute__((__always_inline__, __nodebug__))
_mm512_rsqrt28_round_ps(__m512 __A,int __R)41 _mm512_rsqrt28_round_ps(__m512 __A, int __R)
42 {
43   return (__m512)__builtin_ia32_rsqrt28ps_mask ((__v16sf)__A,
44                                                 (__v16sf)_mm512_setzero_ps(),
45                                                 (__mmask16)-1,
46                                                 __R);
47 }
48 
49 static  __inline__ __m128 __attribute__((__always_inline__, __nodebug__))
_mm_rsqrt28_round_ss(__m128 __A,__m128 __B,int __R)50 _mm_rsqrt28_round_ss(__m128 __A, __m128 __B, int __R)
51 {
52   return (__m128) __builtin_ia32_rsqrt28ss_mask ((__v4sf) __A,
53              (__v4sf) __B,
54              (__v4sf)
55              _mm_setzero_ps (),
56              (__mmask8) -1,
57              __R);
58 }
59 
60 static  __inline__ __m128d __attribute__((__always_inline__, __nodebug__))
_mm_rsqrt28_round_sd(__m128d __A,__m128d __B,int __R)61 _mm_rsqrt28_round_sd (__m128d __A, __m128d __B, int __R)
62 {
63   return (__m128d) __builtin_ia32_rsqrt28sd_mask ((__v2df) __A,
64               (__v2df) __B,
65               (__v2df)
66               _mm_setzero_pd (),
67               (__mmask8) -1,
68              __R);
69 }
70 
71 
72 // rcp28
73 static  __inline__ __m512d __attribute__((__always_inline__, __nodebug__))
_mm512_rcp28_round_pd(__m512d __A,int __R)74 _mm512_rcp28_round_pd (__m512d __A, int __R)
75 {
76   return (__m512d)__builtin_ia32_rcp28pd_mask ((__v8df)__A,
77                                                (__v8df)_mm512_setzero_pd(),
78                                                (__mmask8)-1,
79                                                __R);
80 }
81 
82 static  __inline__ __m512 __attribute__((__always_inline__, __nodebug__))
_mm512_rcp28_round_ps(__m512 __A,int __R)83 _mm512_rcp28_round_ps (__m512 __A, int __R)
84 {
85   return (__m512)__builtin_ia32_rcp28ps_mask ((__v16sf)__A,
86                                               (__v16sf)_mm512_setzero_ps (),
87                                               (__mmask16)-1,
88                                               __R);
89 }
90 
91 static  __inline__ __m128 __attribute__((__always_inline__, __nodebug__))
_mm_rcp28_round_ss(__m128 __A,__m128 __B,int __R)92 _mm_rcp28_round_ss (__m128 __A, __m128 __B, int __R)
93 {
94   return (__m128) __builtin_ia32_rcp28ss_mask ((__v4sf) __A,
95              (__v4sf) __B,
96              (__v4sf)
97              _mm_setzero_ps (),
98              (__mmask8) -1,
99              __R);
100 }
101 static  __inline__ __m128d __attribute__((__always_inline__, __nodebug__))
_mm_rcp28_round_sd(__m128d __A,__m128d __B,int __R)102 _mm_rcp28_round_sd (__m128d __A, __m128d __B, int __R)
103 {
104   return (__m128d) __builtin_ia32_rcp28sd_mask ((__v2df) __A,
105               (__v2df) __B,
106               (__v2df)
107               _mm_setzero_pd (),
108               (__mmask8) -1,
109              __R);
110 }
111 
112 #endif // __AVX512ERINTRIN_H
113