1 /*===---- fma4intrin.h - FMA4 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
24 #ifndef __IMMINTRIN_H
25 #error "Never use <fmaintrin.h> directly; include <immintrin.h> instead."
26 #endif
27
28 #ifndef __FMAINTRIN_H
29 #define __FMAINTRIN_H
30
31 /* Define the default attributes for the functions in this file. */
32 #define __DEFAULT_FN_ATTRS __attribute__((__always_inline__, __nodebug__, __target__("fma")))
33
34 static __inline__ __m128 __DEFAULT_FN_ATTRS
_mm_fmadd_ps(__m128 __A,__m128 __B,__m128 __C)35 _mm_fmadd_ps(__m128 __A, __m128 __B, __m128 __C)
36 {
37 return (__m128)__builtin_ia32_vfmaddps((__v4sf)__A, (__v4sf)__B, (__v4sf)__C);
38 }
39
40 static __inline__ __m128d __DEFAULT_FN_ATTRS
_mm_fmadd_pd(__m128d __A,__m128d __B,__m128d __C)41 _mm_fmadd_pd(__m128d __A, __m128d __B, __m128d __C)
42 {
43 return (__m128d)__builtin_ia32_vfmaddpd((__v2df)__A, (__v2df)__B, (__v2df)__C);
44 }
45
46 static __inline__ __m128 __DEFAULT_FN_ATTRS
_mm_fmadd_ss(__m128 __A,__m128 __B,__m128 __C)47 _mm_fmadd_ss(__m128 __A, __m128 __B, __m128 __C)
48 {
49 return (__m128)__builtin_ia32_vfmaddss((__v4sf)__A, (__v4sf)__B, (__v4sf)__C);
50 }
51
52 static __inline__ __m128d __DEFAULT_FN_ATTRS
_mm_fmadd_sd(__m128d __A,__m128d __B,__m128d __C)53 _mm_fmadd_sd(__m128d __A, __m128d __B, __m128d __C)
54 {
55 return (__m128d)__builtin_ia32_vfmaddsd((__v2df)__A, (__v2df)__B, (__v2df)__C);
56 }
57
58 static __inline__ __m128 __DEFAULT_FN_ATTRS
_mm_fmsub_ps(__m128 __A,__m128 __B,__m128 __C)59 _mm_fmsub_ps(__m128 __A, __m128 __B, __m128 __C)
60 {
61 return (__m128)__builtin_ia32_vfmsubps((__v4sf)__A, (__v4sf)__B, (__v4sf)__C);
62 }
63
64 static __inline__ __m128d __DEFAULT_FN_ATTRS
_mm_fmsub_pd(__m128d __A,__m128d __B,__m128d __C)65 _mm_fmsub_pd(__m128d __A, __m128d __B, __m128d __C)
66 {
67 return (__m128d)__builtin_ia32_vfmsubpd((__v2df)__A, (__v2df)__B, (__v2df)__C);
68 }
69
70 static __inline__ __m128 __DEFAULT_FN_ATTRS
_mm_fmsub_ss(__m128 __A,__m128 __B,__m128 __C)71 _mm_fmsub_ss(__m128 __A, __m128 __B, __m128 __C)
72 {
73 return (__m128)__builtin_ia32_vfmsubss((__v4sf)__A, (__v4sf)__B, (__v4sf)__C);
74 }
75
76 static __inline__ __m128d __DEFAULT_FN_ATTRS
_mm_fmsub_sd(__m128d __A,__m128d __B,__m128d __C)77 _mm_fmsub_sd(__m128d __A, __m128d __B, __m128d __C)
78 {
79 return (__m128d)__builtin_ia32_vfmsubsd((__v2df)__A, (__v2df)__B, (__v2df)__C);
80 }
81
82 static __inline__ __m128 __DEFAULT_FN_ATTRS
_mm_fnmadd_ps(__m128 __A,__m128 __B,__m128 __C)83 _mm_fnmadd_ps(__m128 __A, __m128 __B, __m128 __C)
84 {
85 return (__m128)__builtin_ia32_vfnmaddps((__v4sf)__A, (__v4sf)__B, (__v4sf)__C);
86 }
87
88 static __inline__ __m128d __DEFAULT_FN_ATTRS
_mm_fnmadd_pd(__m128d __A,__m128d __B,__m128d __C)89 _mm_fnmadd_pd(__m128d __A, __m128d __B, __m128d __C)
90 {
91 return (__m128d)__builtin_ia32_vfnmaddpd((__v2df)__A, (__v2df)__B, (__v2df)__C);
92 }
93
94 static __inline__ __m128 __DEFAULT_FN_ATTRS
_mm_fnmadd_ss(__m128 __A,__m128 __B,__m128 __C)95 _mm_fnmadd_ss(__m128 __A, __m128 __B, __m128 __C)
96 {
97 return (__m128)__builtin_ia32_vfnmaddss((__v4sf)__A, (__v4sf)__B, (__v4sf)__C);
98 }
99
100 static __inline__ __m128d __DEFAULT_FN_ATTRS
_mm_fnmadd_sd(__m128d __A,__m128d __B,__m128d __C)101 _mm_fnmadd_sd(__m128d __A, __m128d __B, __m128d __C)
102 {
103 return (__m128d)__builtin_ia32_vfnmaddsd((__v2df)__A, (__v2df)__B, (__v2df)__C);
104 }
105
106 static __inline__ __m128 __DEFAULT_FN_ATTRS
_mm_fnmsub_ps(__m128 __A,__m128 __B,__m128 __C)107 _mm_fnmsub_ps(__m128 __A, __m128 __B, __m128 __C)
108 {
109 return (__m128)__builtin_ia32_vfnmsubps((__v4sf)__A, (__v4sf)__B, (__v4sf)__C);
110 }
111
112 static __inline__ __m128d __DEFAULT_FN_ATTRS
_mm_fnmsub_pd(__m128d __A,__m128d __B,__m128d __C)113 _mm_fnmsub_pd(__m128d __A, __m128d __B, __m128d __C)
114 {
115 return (__m128d)__builtin_ia32_vfnmsubpd((__v2df)__A, (__v2df)__B, (__v2df)__C);
116 }
117
118 static __inline__ __m128 __DEFAULT_FN_ATTRS
_mm_fnmsub_ss(__m128 __A,__m128 __B,__m128 __C)119 _mm_fnmsub_ss(__m128 __A, __m128 __B, __m128 __C)
120 {
121 return (__m128)__builtin_ia32_vfnmsubss((__v4sf)__A, (__v4sf)__B, (__v4sf)__C);
122 }
123
124 static __inline__ __m128d __DEFAULT_FN_ATTRS
_mm_fnmsub_sd(__m128d __A,__m128d __B,__m128d __C)125 _mm_fnmsub_sd(__m128d __A, __m128d __B, __m128d __C)
126 {
127 return (__m128d)__builtin_ia32_vfnmsubsd((__v2df)__A, (__v2df)__B, (__v2df)__C);
128 }
129
130 static __inline__ __m128 __DEFAULT_FN_ATTRS
_mm_fmaddsub_ps(__m128 __A,__m128 __B,__m128 __C)131 _mm_fmaddsub_ps(__m128 __A, __m128 __B, __m128 __C)
132 {
133 return (__m128)__builtin_ia32_vfmaddsubps((__v4sf)__A, (__v4sf)__B, (__v4sf)__C);
134 }
135
136 static __inline__ __m128d __DEFAULT_FN_ATTRS
_mm_fmaddsub_pd(__m128d __A,__m128d __B,__m128d __C)137 _mm_fmaddsub_pd(__m128d __A, __m128d __B, __m128d __C)
138 {
139 return (__m128d)__builtin_ia32_vfmaddsubpd((__v2df)__A, (__v2df)__B, (__v2df)__C);
140 }
141
142 static __inline__ __m128 __DEFAULT_FN_ATTRS
_mm_fmsubadd_ps(__m128 __A,__m128 __B,__m128 __C)143 _mm_fmsubadd_ps(__m128 __A, __m128 __B, __m128 __C)
144 {
145 return (__m128)__builtin_ia32_vfmsubaddps((__v4sf)__A, (__v4sf)__B, (__v4sf)__C);
146 }
147
148 static __inline__ __m128d __DEFAULT_FN_ATTRS
_mm_fmsubadd_pd(__m128d __A,__m128d __B,__m128d __C)149 _mm_fmsubadd_pd(__m128d __A, __m128d __B, __m128d __C)
150 {
151 return (__m128d)__builtin_ia32_vfmsubaddpd((__v2df)__A, (__v2df)__B, (__v2df)__C);
152 }
153
154 static __inline__ __m256 __DEFAULT_FN_ATTRS
_mm256_fmadd_ps(__m256 __A,__m256 __B,__m256 __C)155 _mm256_fmadd_ps(__m256 __A, __m256 __B, __m256 __C)
156 {
157 return (__m256)__builtin_ia32_vfmaddps256((__v8sf)__A, (__v8sf)__B, (__v8sf)__C);
158 }
159
160 static __inline__ __m256d __DEFAULT_FN_ATTRS
_mm256_fmadd_pd(__m256d __A,__m256d __B,__m256d __C)161 _mm256_fmadd_pd(__m256d __A, __m256d __B, __m256d __C)
162 {
163 return (__m256d)__builtin_ia32_vfmaddpd256((__v4df)__A, (__v4df)__B, (__v4df)__C);
164 }
165
166 static __inline__ __m256 __DEFAULT_FN_ATTRS
_mm256_fmsub_ps(__m256 __A,__m256 __B,__m256 __C)167 _mm256_fmsub_ps(__m256 __A, __m256 __B, __m256 __C)
168 {
169 return (__m256)__builtin_ia32_vfmsubps256((__v8sf)__A, (__v8sf)__B, (__v8sf)__C);
170 }
171
172 static __inline__ __m256d __DEFAULT_FN_ATTRS
_mm256_fmsub_pd(__m256d __A,__m256d __B,__m256d __C)173 _mm256_fmsub_pd(__m256d __A, __m256d __B, __m256d __C)
174 {
175 return (__m256d)__builtin_ia32_vfmsubpd256((__v4df)__A, (__v4df)__B, (__v4df)__C);
176 }
177
178 static __inline__ __m256 __DEFAULT_FN_ATTRS
_mm256_fnmadd_ps(__m256 __A,__m256 __B,__m256 __C)179 _mm256_fnmadd_ps(__m256 __A, __m256 __B, __m256 __C)
180 {
181 return (__m256)__builtin_ia32_vfnmaddps256((__v8sf)__A, (__v8sf)__B, (__v8sf)__C);
182 }
183
184 static __inline__ __m256d __DEFAULT_FN_ATTRS
_mm256_fnmadd_pd(__m256d __A,__m256d __B,__m256d __C)185 _mm256_fnmadd_pd(__m256d __A, __m256d __B, __m256d __C)
186 {
187 return (__m256d)__builtin_ia32_vfnmaddpd256((__v4df)__A, (__v4df)__B, (__v4df)__C);
188 }
189
190 static __inline__ __m256 __DEFAULT_FN_ATTRS
_mm256_fnmsub_ps(__m256 __A,__m256 __B,__m256 __C)191 _mm256_fnmsub_ps(__m256 __A, __m256 __B, __m256 __C)
192 {
193 return (__m256)__builtin_ia32_vfnmsubps256((__v8sf)__A, (__v8sf)__B, (__v8sf)__C);
194 }
195
196 static __inline__ __m256d __DEFAULT_FN_ATTRS
_mm256_fnmsub_pd(__m256d __A,__m256d __B,__m256d __C)197 _mm256_fnmsub_pd(__m256d __A, __m256d __B, __m256d __C)
198 {
199 return (__m256d)__builtin_ia32_vfnmsubpd256((__v4df)__A, (__v4df)__B, (__v4df)__C);
200 }
201
202 static __inline__ __m256 __DEFAULT_FN_ATTRS
_mm256_fmaddsub_ps(__m256 __A,__m256 __B,__m256 __C)203 _mm256_fmaddsub_ps(__m256 __A, __m256 __B, __m256 __C)
204 {
205 return (__m256)__builtin_ia32_vfmaddsubps256((__v8sf)__A, (__v8sf)__B, (__v8sf)__C);
206 }
207
208 static __inline__ __m256d __DEFAULT_FN_ATTRS
_mm256_fmaddsub_pd(__m256d __A,__m256d __B,__m256d __C)209 _mm256_fmaddsub_pd(__m256d __A, __m256d __B, __m256d __C)
210 {
211 return (__m256d)__builtin_ia32_vfmaddsubpd256((__v4df)__A, (__v4df)__B, (__v4df)__C);
212 }
213
214 static __inline__ __m256 __DEFAULT_FN_ATTRS
_mm256_fmsubadd_ps(__m256 __A,__m256 __B,__m256 __C)215 _mm256_fmsubadd_ps(__m256 __A, __m256 __B, __m256 __C)
216 {
217 return (__m256)__builtin_ia32_vfmsubaddps256((__v8sf)__A, (__v8sf)__B, (__v8sf)__C);
218 }
219
220 static __inline__ __m256d __DEFAULT_FN_ATTRS
_mm256_fmsubadd_pd(__m256d __A,__m256d __B,__m256d __C)221 _mm256_fmsubadd_pd(__m256d __A, __m256d __B, __m256d __C)
222 {
223 return (__m256d)__builtin_ia32_vfmsubaddpd256((__v4df)__A, (__v4df)__B, (__v4df)__C);
224 }
225
226 #undef __DEFAULT_FN_ATTRS
227
228 #endif /* __FMAINTRIN_H */
229