1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
|
/*===---- smmintrin.h - SSE intrinsics -------------------------------------===
*
* Permission is hereby granted, free of charge, to any person obtaining a copy
* of this software and associated documentation files (the "Software"), to deal
* in the Software without restriction, including without limitation the rights
* to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
* copies of the Software, and to permit persons to whom the Software is
* furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be included in
* all copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
* AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
* THE SOFTWARE.
*
*===-----------------------------------------------------------------------===
*/
#ifndef _SMMINTRIN_H
#define _SMMINTRIN_H
#ifndef __SSE4_1__
#error "SSE4.1 instruction set not enabled"
#else
#include <tmmintrin.h>
/* Type defines. */
typedef double __v2df __attribute__ ((__vector_size__ (16)));
typedef long long __v2di __attribute__ ((__vector_size__ (16)));
/* SSE4 Rounding macros. */
#define _MM_FROUND_TO_NEAREST_INT 0x00
#define _MM_FROUND_TO_NEG_INF 0x01
#define _MM_FROUND_TO_POS_INF 0x02
#define _MM_FROUND_TO_ZERO 0x03
#define _MM_FROUND_CUR_DIRECTION 0x04
#define _MM_FROUND_RAISE_EXC 0x00
#define _MM_FROUND_NO_EXC 0x08
#define _MM_FROUND_NINT (_MM_FROUND_RAISE_EXC | _MM_FROUND_TO_NEAREST_INT)
#define _MM_FROUND_FLOOR (_MM_FROUND_RAISE_EXC | _MM_FROUND_TO_NEG_INF)
#define _MM_FROUND_CEIL (_MM_FROUND_RAISE_EXC | _MM_FROUND_TO_POS_INF)
#define _MM_FROUND_TRUNC (_MM_FROUND_RAISE_EXC | _MM_FROUND_TO_ZERO)
#define _MM_FROUND_RINT (_MM_FROUND_RAISE_EXC | _MM_FROUND_CUR_DIRECTION)
#define _MM_FROUND_NEARBYINT (_MM_FROUND_NO_EXC | _MM_FROUND_CUR_DIRECTION)
#define _mm_ceil_ps(X) _mm_round_ps((X), _MM_FROUND_CEIL)
#define _mm_ceil_pd(X) _mm_round_pd((X), _MM_FROUND_CEIL)
#define _mm_ceil_ss(X, Y) _mm_round_ss((X), (Y), _MM_FROUND_CEIL)
#define _mm_ceil_sd(X, Y) _mm_round_sd((X), (Y), _MM_FROUND_CEIL)
#define _mm_floor_ps(X) _mm_round_ps((X), _MM_FROUND_FLOOR)
#define _mm_floor_pd(X) _mm_round_pd((X), _MM_FROUND_FLOOR)
#define _mm_floor_ss(X, Y) _mm_round_ss((X), (Y), _MM_FROUND_FLOOR)
#define _mm_floor_sd(X, Y) _mm_round_sd((X), (Y), _MM_FROUND_FLOOR)
#define _mm_round_ps(X, Y) __builtin_ia32_roundps((X), (Y))
#define _mm_round_ss(X, Y, M) __builtin_ia32_roundss((X), (Y), (M))
#define _mm_round_pd(X, M) __builtin_ia32_roundpd((X), (M))
#define _mm_round_sd(X, Y, M) __builtin_ia32_roundsd((X), (Y), (M))
/* SSE4 Packed Blending Intrinsics. */
static inline __m128d __attribute__((__always_inline__, __nodebug__))
_mm_blend_pd (__m128d __V1, __m128d __V2, const int __M)
{
return (__m128d) __builtin_ia32_blendpd ((__v2df)__V1, (__v2df)__V2, __M);
}
static inline __m128 __attribute__((__always_inline__, __nodebug__))
_mm_blend_ps (__m128 __V1, __m128 __V2, const int __M)
{
return (__m128) __builtin_ia32_blendps ((__v4sf)__V1, (__v4sf)__V2, __M);
}
static inline __m128d __attribute__((__always_inline__, __nodebug__))
_mm_blendv_pd (__m128d __V1, __m128d __V2, __m128d __M)
{
return (__m128d) __builtin_ia32_blendvpd ((__v2df)__V1, (__v2df)__V2,
(__v2df)__M);
}
static inline __m128 __attribute__((__always_inline__, __nodebug__))
_mm_blendv_ps (__m128 __V1, __m128 __V2, __m128 __M)
{
return (__m128) __builtin_ia32_blendvps ((__v4sf)__V1, (__v4sf)__V2,
(__v4sf)__M);
}
static inline __m128i __attribute__((__always_inline__, __nodebug__))
_mm_blendv_epi8 (__m128i __V1, __m128i __V2, __m128i __M)
{
return (__m128i) __builtin_ia32_pblendvb128 ((__v16qi)__V1, (__v16qi)__V2,
(__v16qi)__M);
}
static inline __m128i __attribute__((__always_inline__, __nodebug__))
_mm_blend_epi16 (__m128i __V1, __m128i __V2, const int __M)
{
return (__m128i) __builtin_ia32_pblendw128 ((__v8hi)__V1, (__v8hi)__V2, __M);
}
/* SSE4 Dword Multiply Instructions. */
static inline __m128i __attribute__((__always_inline__, __nodebug__))
_mm_mullo_epi32 (__m128i __V1, __m128i __V2)
{
return (__m128i) __builtin_ia32_pmulld128((__v4si)__V1, (__v4si)__V2);
}
static inline __m128i __attribute__((__always_inline__, __nodebug__))
_mm_mul_epi32 (__m128i __V1, __m128i __V2)
{
return (__m128i) __builtin_ia32_pmuldq128 ((__v4si)__V1, (__v4si)__V2);
}
/* SSE4 Floating Point Dot Product Instructions. */
#define _mm_dp_ps(X, Y, M) __builtin_ia32_dpps ((X), (Y), (M))
#define _mm_dp_pd(X, Y, M) __builtin_ia32_dppd ((X), (Y), (M))
/* SSE4 Streaming Load Hint Instruction. */
static inline __m128i __attribute__((__always_inline__, __nodebug__))
_mm_stream_load_si128 (__m128i *__V)
{
return (__m128i) __builtin_ia32_movntdqa ((__v2di *) __V);
}
/* SSE4 Packed Integer Min/Max Instructions. */
static inline __m128i __attribute__((__always_inline__, __nodebug__))
_mm_min_epi8 (__m128i __V1, __m128i __V2)
{
return (__m128i) __builtin_ia32_pminsb128 ((__v16qi) __V1, (__v16qi) __V2);
}
static inline __m128i __attribute__((__always_inline__, __nodebug__))
_mm_max_epi8 (__m128i __V1, __m128i __V2)
{
return (__m128i) __builtin_ia32_pmaxsb128 ((__v16qi) __V1, (__v16qi) __V2);
}
static inline __m128i __attribute__((__always_inline__, __nodebug__))
_mm_min_epu16 (__m128i __V1, __m128i __V2)
{
return (__m128i) __builtin_ia32_pminuw128 ((__v8hi) __V1, (__v8hi) __V2);
}
static inline __m128i __attribute__((__always_inline__, __nodebug__))
_mm_max_epu16 (__m128i __V1, __m128i __V2)
{
return (__m128i) __builtin_ia32_pmaxuw128 ((__v8hi) __V1, (__v8hi) __V2);
}
static inline __m128i __attribute__((__always_inline__, __nodebug__))
_mm_min_epi32 (__m128i __V1, __m128i __V2)
{
return (__m128i) __builtin_ia32_pminsd128 ((__v4si) __V1, (__v4si) __V2);
}
static inline __m128i __attribute__((__always_inline__, __nodebug__))
_mm_max_epi32 (__m128i __V1, __m128i __V2)
{
return (__m128i) __builtin_ia32_pmaxsd128 ((__v4si) __V1, (__v4si) __V2);
}
static inline __m128i __attribute__((__always_inline__, __nodebug__))
_mm_min_epu32 (__m128i __V1, __m128i __V2)
{
return (__m128i) __builtin_ia32_pminud128((__v4si) __V1, (__v4si) __V2);
}
static inline __m128i __attribute__((__always_inline__, __nodebug__))
_mm_max_epu32 (__m128i __V1, __m128i __V2)
{
return (__m128i) __builtin_ia32_pmaxud128((__v4si) __V1, (__v4si) __V2);
}
/* SSE4 Insertion and Extraction from XMM Register Instructions. */
#define _mm_insert_ps(X, Y, N) __builtin_ia32_insertps128((X), (Y), (N))
#define _mm_extract_ps(X, N) (__extension__ \
({ union { int i; float f; } __t; \
__v4sf __a = (__v4sf)X; \
__t.f = __a[N]; \
__t.i;}))
/* Miscellaneous insert and extract macros. */
/* Extract a single-precision float from X at index N into D. */
#define _MM_EXTRACT_FLOAT(D, X, N) (__extension__ ({ __v4sf __a = (__v4sf)X; \
(D) = __a[N]; }))
/* Or together 2 sets of indexes (X and Y) with the zeroing bits (Z) to create
an index suitable for _mm_insert_ps. */
#define _MM_MK_INSERTPS_NDX(X, Y, Z) (((X) << 6) | ((Y) << 4) | (Z))
/* Extract a float from X at index N into the first index of the return. */
#define _MM_PICK_OUT_PS(X, N) _mm_insert_ps (_mm_setzero_ps(), (X), \
_MM_MK_INSERTPS_NDX((N), 0, 0x0e))
/* Insert int into packed integer array at index. */
#define _mm_insert_epi8(X, I, N) (__extension__ ({ __v16qi __a = (__v16qi)X; \
__a[N] = I; \
__a;}))
#define _mm_insert_epi32(X, I, N) (__extension__ ({ __v4si __a = (__v4si)X; \
__a[N] = I; \
__a;}))
#ifdef __x86_64__
#define _mm_insert_epi64(X, I, N) (__extension__ ({ __v2di __a = (__v2di)X; \
__a[N] = I; \
__a;}))
#endif /* __x86_64__ */
/* Extract int from packed integer array at index. */
#define _mm_extract_epi8(X, N) (__extension__ ({ __v16qi __a = (__v16qi)X; \
__a[N];}))
#define _mm_extract_epi32(X, N) (__extension__ ({ __v4si __a = (__v4si)X; \
__a[N];}))
#ifdef __x86_64__
#define _mm_extract_epi64(X, N) (__extension__ ({ __v2di __a = (__v2di)X; \
__a[N];}))
#endif /* __x86_64 */
/* SSE4 128-bit Packed Integer Comparisons. */
static inline int __attribute__((__always_inline__, __nodebug__))
_mm_testz_si128(__m128i __M, __m128i __V)
{
return __builtin_ia32_ptestz128((__v2di)__M, (__v2di)__V);
}
static inline int __attribute__((__always_inline__, __nodebug__))
_mm_testc_si128(__m128i __M, __m128i __V)
{
return __builtin_ia32_ptestc128((__v2di)__M, (__v2di)__V);
}
static inline int __attribute__((__always_inline__, __nodebug__))
_mm_testnzc_si128(__m128i __M, __m128i __V)
{
return __builtin_ia32_ptestnzc128((__v2di)__M, (__v2di)__V);
}
#define _mm_test_all_ones(V) _mm_testc_si128((V), _mm_cmpeq_epi32((V), (V)))
#define _mm_test_mix_ones_zeros(M, V) _mm_testnzc_si128((M), (V))
#define _mm_test_all_zeros(M, V) _mm_testz_si128 ((V), (V))
/* SSE4 64-bit Packed Integer Comparisons. */
static inline __m128i __attribute__((__always_inline__, __nodebug__))
_mm_cmpeq_epi64(__m128i __V1, __m128i __V2)
{
return (__m128i) __builtin_ia32_pcmpeqq((__v2di)__V1, (__v2di)__V2);
}
/* SSE4 Packed Integer Sign-Extension. */
static inline __m128i __attribute__((__always_inline__, __nodebug__))
_mm_cvtepi8_epi16(__m128i __V)
{
return (__m128i) __builtin_ia32_pmovsxbw128((__v16qi) __V);
}
static inline __m128i __attribute__((__always_inline__, __nodebug__))
_mm_cvtepi8_epi32(__m128i __V)
{
return (__m128i) __builtin_ia32_pmovsxbd128((__v16qi) __V);
}
static inline __m128i __attribute__((__always_inline__, __nodebug__))
_mm_cvtepi8_epi64(__m128i __V)
{
return (__m128i) __builtin_ia32_pmovsxbq128((__v16qi) __V);
}
static inline __m128i __attribute__((__always_inline__, __nodebug__))
_mm_cvtepi16_epi32(__m128i __V)
{
return (__m128i) __builtin_ia32_pmovsxwd128((__v8hi) __V);
}
static inline __m128i __attribute__((__always_inline__, __nodebug__))
_mm_cvtepi16_epi64(__m128i __V)
{
return (__m128i) __builtin_ia32_pmovsxwq128((__v8hi)__V);
}
static inline __m128i __attribute__((__always_inline__, __nodebug__))
_mm_cvtepi32_epi64(__m128i __V)
{
return (__m128i) __builtin_ia32_pmovsxdq128((__v4si)__V);
}
/* SSE4 Packed Integer Zero-Extension. */
static inline __m128i __attribute__((__always_inline__, __nodebug__))
_mm_cvtepu8_epi16(__m128i __V)
{
return (__m128i) __builtin_ia32_pmovzxbw128((__v16qi) __V);
}
static inline __m128i __attribute__((__always_inline__, __nodebug__))
_mm_cvtepu8_epi32(__m128i __V)
{
return (__m128i) __builtin_ia32_pmovzxbd128((__v16qi)__V);
}
static inline __m128i __attribute__((__always_inline__, __nodebug__))
_mm_cvtepu8_epi64(__m128i __V)
{
return (__m128i) __builtin_ia32_pmovzxbq128((__v16qi)__V);
}
static inline __m128i __attribute__((__always_inline__, __nodebug__))
_mm_cvtepu16_epi32(__m128i __V)
{
return (__m128i) __builtin_ia32_pmovzxwd128((__v8hi)__V);
}
static inline __m128i __attribute__((__always_inline__, __nodebug__))
_mm_cvtepu16_epi64(__m128i __V)
{
return (__m128i) __builtin_ia32_pmovzxwq128((__v8hi)__V);
}
static inline __m128i __attribute__((__always_inline__, __nodebug__))
_mm_cvtepu32_epi64(__m128i __V)
{
return (__m128i) __builtin_ia32_pmovzxdq128((__v4si)__V);
}
/* SSE4 Pack with Unsigned Saturation. */
static inline __m128i __attribute__((__always_inline__, __nodebug__))
_mm_packus_epi32(__m128i __V1, __m128i __V2)
{
return (__m128i) __builtin_ia32_packusdw128((__v4si)__V1, (__v4si)__V2);
}
/* SSE4 Multiple Packed Sums of Absolute Difference. */
#define _mm_mpsadbw_epu8(X, Y, M) __builtin_ia32_mpsadbw128((X), (Y), (M))
#endif /* __SSE4_1__ */
#endif /* _SMMINTRIN_H */
|
