1 /* Copyright (C) 2007-2021 Free Software Foundation, Inc.
2
3 This file is part of GCC.
4
5 GCC is free software; you can redistribute it and/or modify
6 it under the terms of the GNU General Public License as published by
7 the Free Software Foundation; either version 3, or (at your option)
8 any later version.
9
10 GCC is distributed in the hope that it will be useful,
11 but WITHOUT ANY WARRANTY; without even the implied warranty of
12 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 GNU General Public License for more details.
14
15 Under Section 7 of GPL version 3, you are granted additional
16 permissions described in the GCC Runtime Library Exception, version
17 3.1, as published by the Free Software Foundation.
18
19 You should have received a copy of the GNU General Public License and
20 a copy of the GCC Runtime Library Exception along with this program;
21 see the files COPYING3 and COPYING.RUNTIME respectively. If not, see
22 <http://www.gnu.org/licenses/>. */
23
24 #ifndef _X86INTRIN_H_INCLUDED
25 # error "Never use <fma4intrin.h> directly; include <x86intrin.h> instead."
26 #endif
27
28 #ifndef _FMA4INTRIN_H_INCLUDED
29 #define _FMA4INTRIN_H_INCLUDED
30
31 /* We need definitions from the SSE4A, SSE3, SSE2 and SSE header files. */
32 #include <ammintrin.h>
33
34 #ifndef __FMA4__
35 #pragma GCC push_options
36 #pragma GCC target("fma4")
37 #define __DISABLE_FMA4__
38 #endif /* __FMA4__ */
39
40 /* 128b Floating point multiply/add type instructions. */
41 extern __inline __m128 __attribute__((__gnu_inline__, __always_inline__, __artificial__))
_mm_macc_ps(__m128 __A,__m128 __B,__m128 __C)42 _mm_macc_ps (__m128 __A, __m128 __B, __m128 __C)
43 {
44 return (__m128) __builtin_ia32_vfmaddps ((__v4sf)__A, (__v4sf)__B, (__v4sf)__C);
45 }
46
47 extern __inline __m128d __attribute__((__gnu_inline__, __always_inline__, __artificial__))
_mm_macc_pd(__m128d __A,__m128d __B,__m128d __C)48 _mm_macc_pd (__m128d __A, __m128d __B, __m128d __C)
49 {
50 return (__m128d) __builtin_ia32_vfmaddpd ((__v2df)__A, (__v2df)__B, (__v2df)__C);
51 }
52
53 extern __inline __m128 __attribute__((__gnu_inline__, __always_inline__, __artificial__))
_mm_macc_ss(__m128 __A,__m128 __B,__m128 __C)54 _mm_macc_ss (__m128 __A, __m128 __B, __m128 __C)
55 {
56 return (__m128) __builtin_ia32_vfmaddss ((__v4sf)__A, (__v4sf)__B, (__v4sf)__C);
57 }
58
59 extern __inline __m128d __attribute__((__gnu_inline__, __always_inline__, __artificial__))
_mm_macc_sd(__m128d __A,__m128d __B,__m128d __C)60 _mm_macc_sd (__m128d __A, __m128d __B, __m128d __C)
61 {
62 return (__m128d) __builtin_ia32_vfmaddsd ((__v2df)__A, (__v2df)__B, (__v2df)__C);
63 }
64
65 extern __inline __m128 __attribute__((__gnu_inline__, __always_inline__, __artificial__))
_mm_msub_ps(__m128 __A,__m128 __B,__m128 __C)66 _mm_msub_ps (__m128 __A, __m128 __B, __m128 __C)
67
68 {
69 return (__m128) __builtin_ia32_vfmaddps ((__v4sf)__A, (__v4sf)__B, -(__v4sf)__C);
70 }
71
72 extern __inline __m128d __attribute__((__gnu_inline__, __always_inline__, __artificial__))
_mm_msub_pd(__m128d __A,__m128d __B,__m128d __C)73 _mm_msub_pd (__m128d __A, __m128d __B, __m128d __C)
74 {
75 return (__m128d) __builtin_ia32_vfmaddpd ((__v2df)__A, (__v2df)__B, -(__v2df)__C);
76 }
77
78 extern __inline __m128 __attribute__((__gnu_inline__, __always_inline__, __artificial__))
_mm_msub_ss(__m128 __A,__m128 __B,__m128 __C)79 _mm_msub_ss (__m128 __A, __m128 __B, __m128 __C)
80 {
81 return (__m128) __builtin_ia32_vfmaddss ((__v4sf)__A, (__v4sf)__B, -(__v4sf)__C);
82 }
83
84 extern __inline __m128d __attribute__((__gnu_inline__, __always_inline__, __artificial__))
_mm_msub_sd(__m128d __A,__m128d __B,__m128d __C)85 _mm_msub_sd (__m128d __A, __m128d __B, __m128d __C)
86 {
87 return (__m128d) __builtin_ia32_vfmaddsd ((__v2df)__A, (__v2df)__B, -(__v2df)__C);
88 }
89
90 extern __inline __m128 __attribute__((__gnu_inline__, __always_inline__, __artificial__))
_mm_nmacc_ps(__m128 __A,__m128 __B,__m128 __C)91 _mm_nmacc_ps (__m128 __A, __m128 __B, __m128 __C)
92 {
93 return (__m128) __builtin_ia32_vfmaddps (-(__v4sf)__A, (__v4sf)__B, (__v4sf)__C);
94 }
95
96 extern __inline __m128d __attribute__((__gnu_inline__, __always_inline__, __artificial__))
_mm_nmacc_pd(__m128d __A,__m128d __B,__m128d __C)97 _mm_nmacc_pd (__m128d __A, __m128d __B, __m128d __C)
98 {
99 return (__m128d) __builtin_ia32_vfmaddpd (-(__v2df)__A, (__v2df)__B, (__v2df)__C);
100 }
101
102 extern __inline __m128 __attribute__((__gnu_inline__, __always_inline__, __artificial__))
_mm_nmacc_ss(__m128 __A,__m128 __B,__m128 __C)103 _mm_nmacc_ss (__m128 __A, __m128 __B, __m128 __C)
104 {
105 return (__m128) __builtin_ia32_vfmaddss (-(__v4sf)__A, (__v4sf)__B, (__v4sf)__C);
106 }
107
108 extern __inline __m128d __attribute__((__gnu_inline__, __always_inline__, __artificial__))
_mm_nmacc_sd(__m128d __A,__m128d __B,__m128d __C)109 _mm_nmacc_sd (__m128d __A, __m128d __B, __m128d __C)
110 {
111 return (__m128d) __builtin_ia32_vfmaddsd (-(__v2df)__A, (__v2df)__B, (__v2df)__C);
112 }
113
114 extern __inline __m128 __attribute__((__gnu_inline__, __always_inline__, __artificial__))
_mm_nmsub_ps(__m128 __A,__m128 __B,__m128 __C)115 _mm_nmsub_ps (__m128 __A, __m128 __B, __m128 __C)
116 {
117 return (__m128) __builtin_ia32_vfmaddps (-(__v4sf)__A, (__v4sf)__B, -(__v4sf)__C);
118 }
119
120 extern __inline __m128d __attribute__((__gnu_inline__, __always_inline__, __artificial__))
_mm_nmsub_pd(__m128d __A,__m128d __B,__m128d __C)121 _mm_nmsub_pd (__m128d __A, __m128d __B, __m128d __C)
122 {
123 return (__m128d) __builtin_ia32_vfmaddpd (-(__v2df)__A, (__v2df)__B, -(__v2df)__C);
124 }
125
126 extern __inline __m128 __attribute__((__gnu_inline__, __always_inline__, __artificial__))
_mm_nmsub_ss(__m128 __A,__m128 __B,__m128 __C)127 _mm_nmsub_ss (__m128 __A, __m128 __B, __m128 __C)
128 {
129 return (__m128) __builtin_ia32_vfmaddss (-(__v4sf)__A, (__v4sf)__B, -(__v4sf)__C);
130 }
131
132 extern __inline __m128d __attribute__((__gnu_inline__, __always_inline__, __artificial__))
_mm_nmsub_sd(__m128d __A,__m128d __B,__m128d __C)133 _mm_nmsub_sd (__m128d __A, __m128d __B, __m128d __C)
134 {
135 return (__m128d) __builtin_ia32_vfmaddsd (-(__v2df)__A, (__v2df)__B, -(__v2df)__C);
136 }
137
138 extern __inline __m128 __attribute__((__gnu_inline__, __always_inline__, __artificial__))
_mm_maddsub_ps(__m128 __A,__m128 __B,__m128 __C)139 _mm_maddsub_ps (__m128 __A, __m128 __B, __m128 __C)
140 {
141 return (__m128) __builtin_ia32_vfmaddsubps ((__v4sf)__A, (__v4sf)__B, (__v4sf)__C);
142 }
143
144 extern __inline __m128d __attribute__((__gnu_inline__, __always_inline__, __artificial__))
_mm_maddsub_pd(__m128d __A,__m128d __B,__m128d __C)145 _mm_maddsub_pd (__m128d __A, __m128d __B, __m128d __C)
146 {
147 return (__m128d) __builtin_ia32_vfmaddsubpd ((__v2df)__A, (__v2df)__B, (__v2df)__C);
148 }
149
150 extern __inline __m128 __attribute__((__gnu_inline__, __always_inline__, __artificial__))
_mm_msubadd_ps(__m128 __A,__m128 __B,__m128 __C)151 _mm_msubadd_ps (__m128 __A, __m128 __B, __m128 __C)
152 {
153 return (__m128) __builtin_ia32_vfmaddsubps ((__v4sf)__A, (__v4sf)__B, -(__v4sf)__C);
154 }
155
156 extern __inline __m128d __attribute__((__gnu_inline__, __always_inline__, __artificial__))
_mm_msubadd_pd(__m128d __A,__m128d __B,__m128d __C)157 _mm_msubadd_pd (__m128d __A, __m128d __B, __m128d __C)
158 {
159 return (__m128d) __builtin_ia32_vfmaddsubpd ((__v2df)__A, (__v2df)__B, -(__v2df)__C);
160 }
161
162 /* 256b Floating point multiply/add type instructions. */
163 extern __inline __m256 __attribute__((__gnu_inline__, __always_inline__, __artificial__))
_mm256_macc_ps(__m256 __A,__m256 __B,__m256 __C)164 _mm256_macc_ps (__m256 __A, __m256 __B, __m256 __C)
165 {
166 return (__m256) __builtin_ia32_vfmaddps256 ((__v8sf)__A, (__v8sf)__B, (__v8sf)__C);
167 }
168
169 extern __inline __m256d __attribute__((__gnu_inline__, __always_inline__, __artificial__))
_mm256_macc_pd(__m256d __A,__m256d __B,__m256d __C)170 _mm256_macc_pd (__m256d __A, __m256d __B, __m256d __C)
171 {
172 return (__m256d) __builtin_ia32_vfmaddpd256 ((__v4df)__A, (__v4df)__B, (__v4df)__C);
173 }
174
175 extern __inline __m256 __attribute__((__gnu_inline__, __always_inline__, __artificial__))
_mm256_msub_ps(__m256 __A,__m256 __B,__m256 __C)176 _mm256_msub_ps (__m256 __A, __m256 __B, __m256 __C)
177
178 {
179 return (__m256) __builtin_ia32_vfmaddps256 ((__v8sf)__A, (__v8sf)__B, -(__v8sf)__C);
180 }
181
182 extern __inline __m256d __attribute__((__gnu_inline__, __always_inline__, __artificial__))
_mm256_msub_pd(__m256d __A,__m256d __B,__m256d __C)183 _mm256_msub_pd (__m256d __A, __m256d __B, __m256d __C)
184 {
185 return (__m256d) __builtin_ia32_vfmaddpd256 ((__v4df)__A, (__v4df)__B, -(__v4df)__C);
186 }
187
188 extern __inline __m256 __attribute__((__gnu_inline__, __always_inline__, __artificial__))
_mm256_nmacc_ps(__m256 __A,__m256 __B,__m256 __C)189 _mm256_nmacc_ps (__m256 __A, __m256 __B, __m256 __C)
190 {
191 return (__m256) __builtin_ia32_vfmaddps256 (-(__v8sf)__A, (__v8sf)__B, (__v8sf)__C);
192 }
193
194 extern __inline __m256d __attribute__((__gnu_inline__, __always_inline__, __artificial__))
_mm256_nmacc_pd(__m256d __A,__m256d __B,__m256d __C)195 _mm256_nmacc_pd (__m256d __A, __m256d __B, __m256d __C)
196 {
197 return (__m256d) __builtin_ia32_vfmaddpd256 (-(__v4df)__A, (__v4df)__B, (__v4df)__C);
198 }
199
200 extern __inline __m256 __attribute__((__gnu_inline__, __always_inline__, __artificial__))
_mm256_nmsub_ps(__m256 __A,__m256 __B,__m256 __C)201 _mm256_nmsub_ps (__m256 __A, __m256 __B, __m256 __C)
202 {
203 return (__m256) __builtin_ia32_vfmaddps256 (-(__v8sf)__A, (__v8sf)__B, -(__v8sf)__C);
204 }
205
206 extern __inline __m256d __attribute__((__gnu_inline__, __always_inline__, __artificial__))
_mm256_nmsub_pd(__m256d __A,__m256d __B,__m256d __C)207 _mm256_nmsub_pd (__m256d __A, __m256d __B, __m256d __C)
208 {
209 return (__m256d) __builtin_ia32_vfmaddpd256 (-(__v4df)__A, (__v4df)__B, -(__v4df)__C);
210 }
211
212 extern __inline __m256 __attribute__((__gnu_inline__, __always_inline__, __artificial__))
_mm256_maddsub_ps(__m256 __A,__m256 __B,__m256 __C)213 _mm256_maddsub_ps (__m256 __A, __m256 __B, __m256 __C)
214 {
215 return (__m256) __builtin_ia32_vfmaddsubps256 ((__v8sf)__A, (__v8sf)__B, (__v8sf)__C);
216 }
217
218 extern __inline __m256d __attribute__((__gnu_inline__, __always_inline__, __artificial__))
_mm256_maddsub_pd(__m256d __A,__m256d __B,__m256d __C)219 _mm256_maddsub_pd (__m256d __A, __m256d __B, __m256d __C)
220 {
221 return (__m256d) __builtin_ia32_vfmaddsubpd256 ((__v4df)__A, (__v4df)__B, (__v4df)__C);
222 }
223
224 extern __inline __m256 __attribute__((__gnu_inline__, __always_inline__, __artificial__))
_mm256_msubadd_ps(__m256 __A,__m256 __B,__m256 __C)225 _mm256_msubadd_ps (__m256 __A, __m256 __B, __m256 __C)
226 {
227 return (__m256) __builtin_ia32_vfmaddsubps256 ((__v8sf)__A, (__v8sf)__B, -(__v8sf)__C);
228 }
229
230 extern __inline __m256d __attribute__((__gnu_inline__, __always_inline__, __artificial__))
_mm256_msubadd_pd(__m256d __A,__m256d __B,__m256d __C)231 _mm256_msubadd_pd (__m256d __A, __m256d __B, __m256d __C)
232 {
233 return (__m256d) __builtin_ia32_vfmaddsubpd256 ((__v4df)__A, (__v4df)__B, -(__v4df)__C);
234 }
235
236 #ifdef __DISABLE_FMA4__
237 #undef __DISABLE_FMA4__
238 #pragma GCC pop_options
239 #endif /* __DISABLE_FMA4__ */
240
241 #endif
242