Libav
cavsdsp.c
Go to the documentation of this file.
1 /*
2  * Chinese AVS video (AVS1-P2, JiZhun profile) decoder.
3  * Copyright (c) 2006 Stefan Gehrer <stefan.gehrer@gmx.de>
4  *
5  * MMX-optimized DSP functions, based on H.264 optimizations by
6  * Michael Niedermayer and Loren Merritt
7  *
8  * This file is part of Libav.
9  *
10  * Libav is free software; you can redistribute it and/or
11  * modify it under the terms of the GNU Lesser General Public
12  * License as published by the Free Software Foundation; either
13  * version 2.1 of the License, or (at your option) any later version.
14  *
15  * Libav is distributed in the hope that it will be useful,
16  * but WITHOUT ANY WARRANTY; without even the implied warranty of
17  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
18  * Lesser General Public License for more details.
19  *
20  * You should have received a copy of the GNU Lesser General Public
21  * License along with Libav; if not, write to the Free Software
22  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
23  */
24 
25 #include "libavutil/attributes.h"
26 #include "libavutil/common.h"
27 #include "libavutil/cpu.h"
28 #include "libavutil/x86/asm.h"
29 #include "libavutil/x86/cpu.h"
30 #include "libavcodec/cavsdsp.h"
31 #include "libavcodec/idctdsp.h"
32 #include "constants.h"
33 #include "fpel.h"
34 #include "idctdsp.h"
35 #include "config.h"
36 
37 #if HAVE_MMX_INLINE
38 
39 /* in/out: mma=mma+mmb, mmb=mmb-mma */
40 #define SUMSUB_BA( a, b ) \
41  "paddw "#b", "#a" \n\t"\
42  "paddw "#b", "#b" \n\t"\
43  "psubw "#a", "#b" \n\t"
44 
45 /*****************************************************************************
46  *
47  * inverse transform
48  *
49  ****************************************************************************/
50 
51 static inline void cavs_idct8_1d(int16_t *block, uint64_t bias)
52 {
53  __asm__ volatile(
54  "movq 112(%0), %%mm4 \n\t" /* mm4 = src7 */
55  "movq 16(%0), %%mm5 \n\t" /* mm5 = src1 */
56  "movq 80(%0), %%mm2 \n\t" /* mm2 = src5 */
57  "movq 48(%0), %%mm7 \n\t" /* mm7 = src3 */
58  "movq %%mm4, %%mm0 \n\t"
59  "movq %%mm5, %%mm3 \n\t"
60  "movq %%mm2, %%mm6 \n\t"
61  "movq %%mm7, %%mm1 \n\t"
62 
63  "paddw %%mm4, %%mm4 \n\t" /* mm4 = 2*src7 */
64  "paddw %%mm3, %%mm3 \n\t" /* mm3 = 2*src1 */
65  "paddw %%mm6, %%mm6 \n\t" /* mm6 = 2*src5 */
66  "paddw %%mm1, %%mm1 \n\t" /* mm1 = 2*src3 */
67  "paddw %%mm4, %%mm0 \n\t" /* mm0 = 3*src7 */
68  "paddw %%mm3, %%mm5 \n\t" /* mm5 = 3*src1 */
69  "paddw %%mm6, %%mm2 \n\t" /* mm2 = 3*src5 */
70  "paddw %%mm1, %%mm7 \n\t" /* mm7 = 3*src3 */
71  "psubw %%mm4, %%mm5 \n\t" /* mm5 = 3*src1 - 2*src7 = a0 */
72  "paddw %%mm6, %%mm7 \n\t" /* mm7 = 3*src3 + 2*src5 = a1 */
73  "psubw %%mm2, %%mm1 \n\t" /* mm1 = 2*src3 - 3*src5 = a2 */
74  "paddw %%mm0, %%mm3 \n\t" /* mm3 = 2*src1 + 3*src7 = a3 */
75 
76  "movq %%mm5, %%mm4 \n\t"
77  "movq %%mm7, %%mm6 \n\t"
78  "movq %%mm3, %%mm0 \n\t"
79  "movq %%mm1, %%mm2 \n\t"
80  SUMSUB_BA( %%mm7, %%mm5 ) /* mm7 = a0 + a1 mm5 = a0 - a1 */
81  "paddw %%mm3, %%mm7 \n\t" /* mm7 = a0 + a1 + a3 */
82  "paddw %%mm1, %%mm5 \n\t" /* mm5 = a0 - a1 + a2 */
83  "paddw %%mm7, %%mm7 \n\t"
84  "paddw %%mm5, %%mm5 \n\t"
85  "paddw %%mm6, %%mm7 \n\t" /* mm7 = b4 */
86  "paddw %%mm4, %%mm5 \n\t" /* mm5 = b5 */
87 
88  SUMSUB_BA( %%mm1, %%mm3 ) /* mm1 = a3 + a2 mm3 = a3 - a2 */
89  "psubw %%mm1, %%mm4 \n\t" /* mm4 = a0 - a2 - a3 */
90  "movq %%mm4, %%mm1 \n\t" /* mm1 = a0 - a2 - a3 */
91  "psubw %%mm6, %%mm3 \n\t" /* mm3 = a3 - a2 - a1 */
92  "paddw %%mm1, %%mm1 \n\t"
93  "paddw %%mm3, %%mm3 \n\t"
94  "psubw %%mm2, %%mm1 \n\t" /* mm1 = b7 */
95  "paddw %%mm0, %%mm3 \n\t" /* mm3 = b6 */
96 
97  "movq 32(%0), %%mm2 \n\t" /* mm2 = src2 */
98  "movq 96(%0), %%mm6 \n\t" /* mm6 = src6 */
99  "movq %%mm2, %%mm4 \n\t"
100  "movq %%mm6, %%mm0 \n\t"
101  "psllw $2, %%mm4 \n\t" /* mm4 = 4*src2 */
102  "psllw $2, %%mm6 \n\t" /* mm6 = 4*src6 */
103  "paddw %%mm4, %%mm2 \n\t" /* mm2 = 5*src2 */
104  "paddw %%mm6, %%mm0 \n\t" /* mm0 = 5*src6 */
105  "paddw %%mm2, %%mm2 \n\t"
106  "paddw %%mm0, %%mm0 \n\t"
107  "psubw %%mm0, %%mm4 \n\t" /* mm4 = 4*src2 - 10*src6 = a7 */
108  "paddw %%mm2, %%mm6 \n\t" /* mm6 = 4*src6 + 10*src2 = a6 */
109 
110  "movq (%0), %%mm2 \n\t" /* mm2 = src0 */
111  "movq 64(%0), %%mm0 \n\t" /* mm0 = src4 */
112  SUMSUB_BA( %%mm0, %%mm2 ) /* mm0 = src0+src4 mm2 = src0-src4 */
113  "psllw $3, %%mm0 \n\t"
114  "psllw $3, %%mm2 \n\t"
115  "paddw %1, %%mm0 \n\t" /* add rounding bias */
116  "paddw %1, %%mm2 \n\t" /* add rounding bias */
117 
118  SUMSUB_BA( %%mm6, %%mm0 ) /* mm6 = a4 + a6 mm0 = a4 - a6 */
119  SUMSUB_BA( %%mm4, %%mm2 ) /* mm4 = a5 + a7 mm2 = a5 - a7 */
120  SUMSUB_BA( %%mm7, %%mm6 ) /* mm7 = dst0 mm6 = dst7 */
121  SUMSUB_BA( %%mm5, %%mm4 ) /* mm5 = dst1 mm4 = dst6 */
122  SUMSUB_BA( %%mm3, %%mm2 ) /* mm3 = dst2 mm2 = dst5 */
123  SUMSUB_BA( %%mm1, %%mm0 ) /* mm1 = dst3 mm0 = dst4 */
124  :: "r"(block), "m"(bias)
125  );
126 }
127 
128 #define SBUTTERFLY(a,b,t,n,m)\
129  "mov" #m " " #a ", " #t " \n\t" /* abcd */\
130  "punpckl" #n " " #b ", " #a " \n\t" /* aebf */\
131  "punpckh" #n " " #b ", " #t " \n\t" /* cgdh */\
132 
133 #define TRANSPOSE4(a,b,c,d,t)\
134  SBUTTERFLY(a,b,t,wd,q) /* a=aebf t=cgdh */\
135  SBUTTERFLY(c,d,b,wd,q) /* c=imjn b=kolp */\
136  SBUTTERFLY(a,c,d,dq,q) /* a=aeim d=bfjn */\
137  SBUTTERFLY(t,b,c,dq,q) /* t=cgko c=dhlp */
138 
139 static void cavs_idct8_add_mmx(uint8_t *dst, int16_t *block, ptrdiff_t stride)
140 {
141  int i;
142  DECLARE_ALIGNED(8, int16_t, b2)[64];
143 
144  for(i=0; i<2; i++){
145  cavs_idct8_1d(block + 4 * i, ff_pw_4.a);
146 
147  __asm__ volatile(
148  "psraw $3, %%mm7 \n\t"
149  "psraw $3, %%mm6 \n\t"
150  "psraw $3, %%mm5 \n\t"
151  "psraw $3, %%mm4 \n\t"
152  "psraw $3, %%mm3 \n\t"
153  "psraw $3, %%mm2 \n\t"
154  "psraw $3, %%mm1 \n\t"
155  "psraw $3, %%mm0 \n\t"
156  "movq %%mm7, (%0) \n\t"
157  TRANSPOSE4( %%mm0, %%mm2, %%mm4, %%mm6, %%mm7 )
158  "movq %%mm0, 8(%0) \n\t"
159  "movq %%mm6, 24(%0) \n\t"
160  "movq %%mm7, 40(%0) \n\t"
161  "movq %%mm4, 56(%0) \n\t"
162  "movq (%0), %%mm7 \n\t"
163  TRANSPOSE4( %%mm7, %%mm5, %%mm3, %%mm1, %%mm0 )
164  "movq %%mm7, (%0) \n\t"
165  "movq %%mm1, 16(%0) \n\t"
166  "movq %%mm0, 32(%0) \n\t"
167  "movq %%mm3, 48(%0) \n\t"
168  :
169  : "r"(b2 + 32 * i)
170  : "memory"
171  );
172  }
173 
174  for(i=0; i<2; i++){
175  cavs_idct8_1d(b2+4*i, ff_pw_64.a);
176 
177  __asm__ volatile(
178  "psraw $7, %%mm7 \n\t"
179  "psraw $7, %%mm6 \n\t"
180  "psraw $7, %%mm5 \n\t"
181  "psraw $7, %%mm4 \n\t"
182  "psraw $7, %%mm3 \n\t"
183  "psraw $7, %%mm2 \n\t"
184  "psraw $7, %%mm1 \n\t"
185  "psraw $7, %%mm0 \n\t"
186  "movq %%mm7, (%0) \n\t"
187  "movq %%mm5, 16(%0) \n\t"
188  "movq %%mm3, 32(%0) \n\t"
189  "movq %%mm1, 48(%0) \n\t"
190  "movq %%mm0, 64(%0) \n\t"
191  "movq %%mm2, 80(%0) \n\t"
192  "movq %%mm4, 96(%0) \n\t"
193  "movq %%mm6, 112(%0) \n\t"
194  :: "r"(b2+4*i)
195  : "memory"
196  );
197  }
198 
199  ff_add_pixels_clamped_mmx(b2, dst, stride);
200 }
201 
202 #endif /* HAVE_MMX_INLINE */
203 
204 #if (HAVE_MMXEXT_INLINE || HAVE_AMD3DNOW_INLINE)
205 
206 /*****************************************************************************
207  *
208  * motion compensation
209  *
210  ****************************************************************************/
211 
212 /* vertical filter [-1 -2 96 42 -7 0] */
213 #define QPEL_CAVSV1(A,B,C,D,E,F,OP,MUL2) \
214  "movd (%0), "#F" \n\t"\
215  "movq "#C", %%mm6 \n\t"\
216  "pmullw %5, %%mm6 \n\t"\
217  "movq "#D", %%mm7 \n\t"\
218  "pmullw "MANGLE(MUL2)", %%mm7\n\t"\
219  "psllw $3, "#E" \n\t"\
220  "psubw "#E", %%mm6 \n\t"\
221  "psraw $3, "#E" \n\t"\
222  "paddw %%mm7, %%mm6 \n\t"\
223  "paddw "#E", %%mm6 \n\t"\
224  "paddw "#B", "#B" \n\t"\
225  "pxor %%mm7, %%mm7 \n\t"\
226  "add %2, %0 \n\t"\
227  "punpcklbw %%mm7, "#F" \n\t"\
228  "psubw "#B", %%mm6 \n\t"\
229  "psraw $1, "#B" \n\t"\
230  "psubw "#A", %%mm6 \n\t"\
231  "paddw %4, %%mm6 \n\t"\
232  "psraw $7, %%mm6 \n\t"\
233  "packuswb %%mm6, %%mm6 \n\t"\
234  OP(%%mm6, (%1), A, d) \
235  "add %3, %1 \n\t"
236 
237 /* vertical filter [ 0 -1 5 5 -1 0] */
238 #define QPEL_CAVSV2(A,B,C,D,E,F,OP,MUL2) \
239  "movd (%0), "#F" \n\t"\
240  "movq "#C", %%mm6 \n\t"\
241  "paddw "#D", %%mm6 \n\t"\
242  "pmullw %5, %%mm6 \n\t"\
243  "add %2, %0 \n\t"\
244  "punpcklbw %%mm7, "#F" \n\t"\
245  "psubw "#B", %%mm6 \n\t"\
246  "psubw "#E", %%mm6 \n\t"\
247  "paddw %4, %%mm6 \n\t"\
248  "psraw $3, %%mm6 \n\t"\
249  "packuswb %%mm6, %%mm6 \n\t"\
250  OP(%%mm6, (%1), A, d) \
251  "add %3, %1 \n\t"
252 
253 /* vertical filter [ 0 -7 42 96 -2 -1] */
254 #define QPEL_CAVSV3(A,B,C,D,E,F,OP,MUL2) \
255  "movd (%0), "#F" \n\t"\
256  "movq "#C", %%mm6 \n\t"\
257  "pmullw "MANGLE(MUL2)", %%mm6\n\t"\
258  "movq "#D", %%mm7 \n\t"\
259  "pmullw %5, %%mm7 \n\t"\
260  "psllw $3, "#B" \n\t"\
261  "psubw "#B", %%mm6 \n\t"\
262  "psraw $3, "#B" \n\t"\
263  "paddw %%mm7, %%mm6 \n\t"\
264  "paddw "#B", %%mm6 \n\t"\
265  "paddw "#E", "#E" \n\t"\
266  "pxor %%mm7, %%mm7 \n\t"\
267  "add %2, %0 \n\t"\
268  "punpcklbw %%mm7, "#F" \n\t"\
269  "psubw "#E", %%mm6 \n\t"\
270  "psraw $1, "#E" \n\t"\
271  "psubw "#F", %%mm6 \n\t"\
272  "paddw %4, %%mm6 \n\t"\
273  "psraw $7, %%mm6 \n\t"\
274  "packuswb %%mm6, %%mm6 \n\t"\
275  OP(%%mm6, (%1), A, d) \
276  "add %3, %1 \n\t"
277 
278 
279 #define QPEL_CAVSVNUM(VOP,OP,ADD,MUL1,MUL2)\
280  int w= 2;\
281  src -= 2*srcStride;\
282  \
283  while(w--){\
284  __asm__ volatile(\
285  "pxor %%mm7, %%mm7 \n\t"\
286  "movd (%0), %%mm0 \n\t"\
287  "add %2, %0 \n\t"\
288  "movd (%0), %%mm1 \n\t"\
289  "add %2, %0 \n\t"\
290  "movd (%0), %%mm2 \n\t"\
291  "add %2, %0 \n\t"\
292  "movd (%0), %%mm3 \n\t"\
293  "add %2, %0 \n\t"\
294  "movd (%0), %%mm4 \n\t"\
295  "add %2, %0 \n\t"\
296  "punpcklbw %%mm7, %%mm0 \n\t"\
297  "punpcklbw %%mm7, %%mm1 \n\t"\
298  "punpcklbw %%mm7, %%mm2 \n\t"\
299  "punpcklbw %%mm7, %%mm3 \n\t"\
300  "punpcklbw %%mm7, %%mm4 \n\t"\
301  VOP(%%mm0, %%mm1, %%mm2, %%mm3, %%mm4, %%mm5, OP, MUL2)\
302  VOP(%%mm1, %%mm2, %%mm3, %%mm4, %%mm5, %%mm0, OP, MUL2)\
303  VOP(%%mm2, %%mm3, %%mm4, %%mm5, %%mm0, %%mm1, OP, MUL2)\
304  VOP(%%mm3, %%mm4, %%mm5, %%mm0, %%mm1, %%mm2, OP, MUL2)\
305  VOP(%%mm4, %%mm5, %%mm0, %%mm1, %%mm2, %%mm3, OP, MUL2)\
306  VOP(%%mm5, %%mm0, %%mm1, %%mm2, %%mm3, %%mm4, OP, MUL2)\
307  VOP(%%mm0, %%mm1, %%mm2, %%mm3, %%mm4, %%mm5, OP, MUL2)\
308  VOP(%%mm1, %%mm2, %%mm3, %%mm4, %%mm5, %%mm0, OP, MUL2)\
309  \
310  : "+a"(src), "+c"(dst)\
311  : "S"((x86_reg)srcStride), "r"((x86_reg)dstStride), "m"(ADD), "m"(MUL1)\
312  : "memory"\
313  );\
314  if(h==16){\
315  __asm__ volatile(\
316  VOP(%%mm2, %%mm3, %%mm4, %%mm5, %%mm0, %%mm1, OP, MUL2)\
317  VOP(%%mm3, %%mm4, %%mm5, %%mm0, %%mm1, %%mm2, OP, MUL2)\
318  VOP(%%mm4, %%mm5, %%mm0, %%mm1, %%mm2, %%mm3, OP, MUL2)\
319  VOP(%%mm5, %%mm0, %%mm1, %%mm2, %%mm3, %%mm4, OP, MUL2)\
320  VOP(%%mm0, %%mm1, %%mm2, %%mm3, %%mm4, %%mm5, OP, MUL2)\
321  VOP(%%mm1, %%mm2, %%mm3, %%mm4, %%mm5, %%mm0, OP, MUL2)\
322  VOP(%%mm2, %%mm3, %%mm4, %%mm5, %%mm0, %%mm1, OP, MUL2)\
323  VOP(%%mm3, %%mm4, %%mm5, %%mm0, %%mm1, %%mm2, OP, MUL2)\
324  \
325  : "+a"(src), "+c"(dst)\
326  : "S"((x86_reg)srcStride), "r"((x86_reg)dstStride), "m"(ADD), "m"(MUL1)\
327  : "memory"\
328  );\
329  }\
330  src += 4-(h+5)*srcStride;\
331  dst += 4-h*dstStride;\
332  }
333 
334 #define QPEL_CAVS(OPNAME, OP, MMX)\
335 static void OPNAME ## cavs_qpel8_h_ ## MMX(uint8_t *dst, const uint8_t *src, ptrdiff_t dstStride, ptrdiff_t srcStride)\
336 {\
337  int h=8;\
338  __asm__ volatile(\
339  "pxor %%mm7, %%mm7 \n\t"\
340  "movq %5, %%mm6 \n\t"\
341  "1: \n\t"\
342  "movq (%0), %%mm0 \n\t"\
343  "movq 1(%0), %%mm2 \n\t"\
344  "movq %%mm0, %%mm1 \n\t"\
345  "movq %%mm2, %%mm3 \n\t"\
346  "punpcklbw %%mm7, %%mm0 \n\t"\
347  "punpckhbw %%mm7, %%mm1 \n\t"\
348  "punpcklbw %%mm7, %%mm2 \n\t"\
349  "punpckhbw %%mm7, %%mm3 \n\t"\
350  "paddw %%mm2, %%mm0 \n\t"\
351  "paddw %%mm3, %%mm1 \n\t"\
352  "pmullw %%mm6, %%mm0 \n\t"\
353  "pmullw %%mm6, %%mm1 \n\t"\
354  "movq -1(%0), %%mm2 \n\t"\
355  "movq 2(%0), %%mm4 \n\t"\
356  "movq %%mm2, %%mm3 \n\t"\
357  "movq %%mm4, %%mm5 \n\t"\
358  "punpcklbw %%mm7, %%mm2 \n\t"\
359  "punpckhbw %%mm7, %%mm3 \n\t"\
360  "punpcklbw %%mm7, %%mm4 \n\t"\
361  "punpckhbw %%mm7, %%mm5 \n\t"\
362  "paddw %%mm4, %%mm2 \n\t"\
363  "paddw %%mm3, %%mm5 \n\t"\
364  "psubw %%mm2, %%mm0 \n\t"\
365  "psubw %%mm5, %%mm1 \n\t"\
366  "movq %6, %%mm5 \n\t"\
367  "paddw %%mm5, %%mm0 \n\t"\
368  "paddw %%mm5, %%mm1 \n\t"\
369  "psraw $3, %%mm0 \n\t"\
370  "psraw $3, %%mm1 \n\t"\
371  "packuswb %%mm1, %%mm0 \n\t"\
372  OP(%%mm0, (%1),%%mm5, q) \
373  "add %3, %0 \n\t"\
374  "add %4, %1 \n\t"\
375  "decl %2 \n\t"\
376  " jnz 1b \n\t"\
377  : "+a"(src), "+c"(dst), "+m"(h)\
378  : "d"((x86_reg)srcStride), "S"((x86_reg)dstStride), "m"(ff_pw_5), "m"(ff_pw_4)\
379  : "memory"\
380  );\
381 }\
382 \
383 static inline void OPNAME ## cavs_qpel8or16_v1_ ## MMX(uint8_t *dst, const uint8_t *src, ptrdiff_t dstStride, ptrdiff_t srcStride, int h)\
384 { \
385  QPEL_CAVSVNUM(QPEL_CAVSV1,OP,ff_pw_64,ff_pw_96,ff_pw_42) \
386 }\
387 \
388 static inline void OPNAME ## cavs_qpel8or16_v2_ ## MMX(uint8_t *dst, const uint8_t *src, ptrdiff_t dstStride, ptrdiff_t srcStride, int h)\
389 { \
390  QPEL_CAVSVNUM(QPEL_CAVSV2,OP,ff_pw_4,ff_pw_5,ff_pw_5) \
391 }\
392 \
393 static inline void OPNAME ## cavs_qpel8or16_v3_ ## MMX(uint8_t *dst, const uint8_t *src, ptrdiff_t dstStride, ptrdiff_t srcStride, int h)\
394 { \
395  QPEL_CAVSVNUM(QPEL_CAVSV3,OP,ff_pw_64,ff_pw_96,ff_pw_42) \
396 }\
397 \
398 static void OPNAME ## cavs_qpel8_v1_ ## MMX(uint8_t *dst, const uint8_t *src, ptrdiff_t dstStride, ptrdiff_t srcStride)\
399 { \
400  OPNAME ## cavs_qpel8or16_v1_ ## MMX(dst , src , dstStride, srcStride, 8);\
401 }\
402 static void OPNAME ## cavs_qpel16_v1_ ## MMX(uint8_t *dst, const uint8_t *src, ptrdiff_t dstStride, ptrdiff_t srcStride)\
403 { \
404  OPNAME ## cavs_qpel8or16_v1_ ## MMX(dst , src , dstStride, srcStride, 16);\
405  OPNAME ## cavs_qpel8or16_v1_ ## MMX(dst+8, src+8, dstStride, srcStride, 16);\
406 }\
407 \
408 static void OPNAME ## cavs_qpel8_v2_ ## MMX(uint8_t *dst, const uint8_t *src, ptrdiff_t dstStride, ptrdiff_t srcStride)\
409 { \
410  OPNAME ## cavs_qpel8or16_v2_ ## MMX(dst , src , dstStride, srcStride, 8);\
411 }\
412 static void OPNAME ## cavs_qpel16_v2_ ## MMX(uint8_t *dst, const uint8_t *src, ptrdiff_t dstStride, ptrdiff_t srcStride)\
413 { \
414  OPNAME ## cavs_qpel8or16_v2_ ## MMX(dst , src , dstStride, srcStride, 16);\
415  OPNAME ## cavs_qpel8or16_v2_ ## MMX(dst+8, src+8, dstStride, srcStride, 16);\
416 }\
417 \
418 static void OPNAME ## cavs_qpel8_v3_ ## MMX(uint8_t *dst, const uint8_t *src, ptrdiff_t dstStride, ptrdiff_t srcStride)\
419 { \
420  OPNAME ## cavs_qpel8or16_v3_ ## MMX(dst , src , dstStride, srcStride, 8);\
421 }\
422 static void OPNAME ## cavs_qpel16_v3_ ## MMX(uint8_t *dst, const uint8_t *src, ptrdiff_t dstStride, ptrdiff_t srcStride)\
423 { \
424  OPNAME ## cavs_qpel8or16_v3_ ## MMX(dst , src , dstStride, srcStride, 16);\
425  OPNAME ## cavs_qpel8or16_v3_ ## MMX(dst+8, src+8, dstStride, srcStride, 16);\
426 }\
427 \
428 static void OPNAME ## cavs_qpel16_h_ ## MMX(uint8_t *dst, const uint8_t *src, ptrdiff_t dstStride, ptrdiff_t srcStride)\
429 { \
430  OPNAME ## cavs_qpel8_h_ ## MMX(dst , src , dstStride, srcStride);\
431  OPNAME ## cavs_qpel8_h_ ## MMX(dst+8, src+8, dstStride, srcStride);\
432  src += 8*srcStride;\
433  dst += 8*dstStride;\
434  OPNAME ## cavs_qpel8_h_ ## MMX(dst , src , dstStride, srcStride);\
435  OPNAME ## cavs_qpel8_h_ ## MMX(dst+8, src+8, dstStride, srcStride);\
436 }\
437 
438 #define CAVS_MC(OPNAME, SIZE, MMX) \
439 static void OPNAME ## cavs_qpel ## SIZE ## _mc20_ ## MMX(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
440 {\
441  OPNAME ## cavs_qpel ## SIZE ## _h_ ## MMX(dst, src, stride, stride);\
442 }\
443 \
444 static void OPNAME ## cavs_qpel ## SIZE ## _mc01_ ## MMX(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
445 {\
446  OPNAME ## cavs_qpel ## SIZE ## _v1_ ## MMX(dst, src, stride, stride);\
447 }\
448 \
449 static void OPNAME ## cavs_qpel ## SIZE ## _mc02_ ## MMX(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
450 {\
451  OPNAME ## cavs_qpel ## SIZE ## _v2_ ## MMX(dst, src, stride, stride);\
452 }\
453 \
454 static void OPNAME ## cavs_qpel ## SIZE ## _mc03_ ## MMX(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
455 {\
456  OPNAME ## cavs_qpel ## SIZE ## _v3_ ## MMX(dst, src, stride, stride);\
457 }\
458 
459 #define PUT_OP(a,b,temp, size) "mov" #size " " #a ", " #b " \n\t"
460 #define AVG_3DNOW_OP(a,b,temp, size) \
461 "mov" #size " " #b ", " #temp " \n\t"\
462 "pavgusb " #temp ", " #a " \n\t"\
463 "mov" #size " " #a ", " #b " \n\t"
464 #define AVG_MMXEXT_OP(a, b, temp, size) \
465 "mov" #size " " #b ", " #temp " \n\t"\
466 "pavgb " #temp ", " #a " \n\t"\
467 "mov" #size " " #a ", " #b " \n\t"
468 
469 #endif /* (HAVE_MMXEXT_INLINE || HAVE_AMD3DNOW_INLINE) */
470 
471 #if HAVE_MMX_INLINE
472 static void put_cavs_qpel8_mc00_mmx(uint8_t *dst, const uint8_t *src,
473  ptrdiff_t stride)
474 {
475  ff_put_pixels8_mmx(dst, src, stride, 8);
476 }
477 
478 static void avg_cavs_qpel8_mc00_mmx(uint8_t *dst, const uint8_t *src,
479  ptrdiff_t stride)
480 {
481  ff_avg_pixels8_mmx(dst, src, stride, 8);
482 }
483 
484 static void put_cavs_qpel16_mc00_mmx(uint8_t *dst, const uint8_t *src,
485  ptrdiff_t stride)
486 {
487  ff_put_pixels16_mmx(dst, src, stride, 16);
488 }
489 
490 static void avg_cavs_qpel16_mc00_mmx(uint8_t *dst, const uint8_t *src,
491  ptrdiff_t stride)
492 {
493  ff_avg_pixels16_mmx(dst, src, stride, 16);
494 }
495 
496 static av_cold void cavsdsp_init_mmx(CAVSDSPContext *c,
497  AVCodecContext *avctx)
498 {
499  c->put_cavs_qpel_pixels_tab[0][0] = put_cavs_qpel16_mc00_mmx;
500  c->put_cavs_qpel_pixels_tab[1][0] = put_cavs_qpel8_mc00_mmx;
501  c->avg_cavs_qpel_pixels_tab[0][0] = avg_cavs_qpel16_mc00_mmx;
502  c->avg_cavs_qpel_pixels_tab[1][0] = avg_cavs_qpel8_mc00_mmx;
503 
504  c->cavs_idct8_add = cavs_idct8_add_mmx;
506 }
507 #endif /* HAVE_MMX_INLINE */
508 
509 #define DSPFUNC(PFX, IDX, NUM, EXT) \
510  c->PFX ## _cavs_qpel_pixels_tab[IDX][ 2] = PFX ## _cavs_qpel ## NUM ## _mc20_ ## EXT; \
511  c->PFX ## _cavs_qpel_pixels_tab[IDX][ 4] = PFX ## _cavs_qpel ## NUM ## _mc01_ ## EXT; \
512  c->PFX ## _cavs_qpel_pixels_tab[IDX][ 8] = PFX ## _cavs_qpel ## NUM ## _mc02_ ## EXT; \
513  c->PFX ## _cavs_qpel_pixels_tab[IDX][12] = PFX ## _cavs_qpel ## NUM ## _mc03_ ## EXT; \
514 
515 #if HAVE_MMXEXT_INLINE
516 QPEL_CAVS(put_, PUT_OP, mmxext)
517 QPEL_CAVS(avg_, AVG_MMXEXT_OP, mmxext)
518 
519 CAVS_MC(put_, 8, mmxext)
520 CAVS_MC(put_, 16, mmxext)
521 CAVS_MC(avg_, 8, mmxext)
522 CAVS_MC(avg_, 16, mmxext)
523 
524 static av_cold void cavsdsp_init_mmxext(CAVSDSPContext *c,
525  AVCodecContext *avctx)
526 {
527  DSPFUNC(put, 0, 16, mmxext);
528  DSPFUNC(put, 1, 8, mmxext);
529  DSPFUNC(avg, 0, 16, mmxext);
530  DSPFUNC(avg, 1, 8, mmxext);
531 }
532 #endif /* HAVE_MMXEXT_INLINE */
533 
534 #if HAVE_AMD3DNOW_INLINE
535 QPEL_CAVS(put_, PUT_OP, 3dnow)
536 QPEL_CAVS(avg_, AVG_3DNOW_OP, 3dnow)
537 
538 CAVS_MC(put_, 8, 3dnow)
539 CAVS_MC(put_, 16,3dnow)
540 CAVS_MC(avg_, 8, 3dnow)
541 CAVS_MC(avg_, 16,3dnow)
542 
543 static av_cold void cavsdsp_init_3dnow(CAVSDSPContext *c,
544  AVCodecContext *avctx)
545 {
546  DSPFUNC(put, 0, 16, 3dnow);
547  DSPFUNC(put, 1, 8, 3dnow);
548  DSPFUNC(avg, 0, 16, 3dnow);
549  DSPFUNC(avg, 1, 8, 3dnow);
550 }
551 #endif /* HAVE_AMD3DNOW_INLINE */
552 
554 {
555 #if HAVE_MMX_INLINE
556  int cpu_flags = av_get_cpu_flags();
557 
558  if (INLINE_MMX(cpu_flags))
559  cavsdsp_init_mmx(c, avctx);
560 #endif /* HAVE_MMX_INLINE */
561 #if HAVE_AMD3DNOW_INLINE
562  if (INLINE_AMD3DNOW(cpu_flags))
563  cavsdsp_init_3dnow(c, avctx);
564 #endif /* HAVE_AMD3DNOW_INLINE */
565 #if HAVE_MMXEXT_INLINE
566  if (INLINE_MMXEXT(cpu_flags))
567  cavsdsp_init_mmxext(c, avctx);
568 #endif /* HAVE_MMXEXT_INLINE */
569 }
void ff_put_pixels8_mmx(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)
#define DECLARE_ALIGNED(n, t, v)
Definition: mem.h:58
int stride
Definition: mace.c:144
Macro definitions for various function/variable attributes.
static int16_t block[64]
Definition: dct.c:97
#define DSPFUNC(PFX, IDX, NUM, EXT)
Definition: cavsdsp.c:509
void(* cavs_idct8_add)(uint8_t *dst, int16_t *block, ptrdiff_t stride)
Definition: cavsdsp.h:37
const xmm_reg ff_pw_64
Definition: constants.c:42
uint8_t
#define av_cold
Definition: attributes.h:66
uint64_t a
Definition: asm.h:27
#define src
Definition: vp8dsp.c:254
#define INLINE_MMX(flags)
Definition: cpu.h:79
qpel_mc_func put_cavs_qpel_pixels_tab[2][16]
Definition: cavsdsp.h:31
#define CAVS_MC(OPNAME, SIZE)
Definition: cavsdsp.c:440
#define INLINE_AMD3DNOW(flags)
Definition: cpu.h:77
main external API structure.
Definition: avcodec.h:1409
int idct_perm
Definition: cavsdsp.h:38
qpel_mc_func avg_cavs_qpel_pixels_tab[2][16]
Definition: cavsdsp.h:32
av_cold void ff_cavsdsp_init_x86(CAVSDSPContext *c, AVCodecContext *avctx)
Definition: cavsdsp.c:553
int av_get_cpu_flags(void)
Return the flags which specify extensions supported by the CPU.
Definition: cpu.c:47
#define INLINE_MMXEXT(flags)
Definition: cpu.h:80
void ff_avg_pixels8_mmx(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)
common internal and external API header
void ff_avg_pixels16_mmx(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)
const xmm_reg ff_pw_4
Definition: constants.c:30
void ff_add_pixels_clamped_mmx(const int16_t *block, uint8_t *pixels, int line_size)
void ff_put_pixels16_mmx(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)