rv40dsp.c
Go to the documentation of this file.
1 /*
2  * RV40 decoder motion compensation functions
3  * Copyright (c) 2008 Konstantin Shishkov
4  *
5  * This file is part of Libav.
6  *
7  * Libav is free software; you can redistribute it and/or
8  * modify it under the terms of the GNU Lesser General Public
9  * License as published by the Free Software Foundation; either
10  * version 2.1 of the License, or (at your option) any later version.
11  *
12  * Libav is distributed in the hope that it will be useful,
13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15  * Lesser General Public License for more details.
16  *
17  * You should have received a copy of the GNU Lesser General Public
18  * License along with Libav; if not, write to the Free Software
19  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
20  */
21 
27 #include "avcodec.h"
28 #include "dsputil.h"
29 #include "rv34dsp.h"
30 
31 #define RV40_LOWPASS(OPNAME, OP) \
32 static av_unused void OPNAME ## rv40_qpel8_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride,\
33  const int h, const int C1, const int C2, const int SHIFT){\
34  uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
35  int i;\
36  for(i = 0; i < h; i++)\
37  {\
38  OP(dst[0], (src[-2] + src[ 3] - 5*(src[-1]+src[2]) + src[0]*C1 + src[1]*C2 + (1<<(SHIFT-1))) >> SHIFT);\
39  OP(dst[1], (src[-1] + src[ 4] - 5*(src[ 0]+src[3]) + src[1]*C1 + src[2]*C2 + (1<<(SHIFT-1))) >> SHIFT);\
40  OP(dst[2], (src[ 0] + src[ 5] - 5*(src[ 1]+src[4]) + src[2]*C1 + src[3]*C2 + (1<<(SHIFT-1))) >> SHIFT);\
41  OP(dst[3], (src[ 1] + src[ 6] - 5*(src[ 2]+src[5]) + src[3]*C1 + src[4]*C2 + (1<<(SHIFT-1))) >> SHIFT);\
42  OP(dst[4], (src[ 2] + src[ 7] - 5*(src[ 3]+src[6]) + src[4]*C1 + src[5]*C2 + (1<<(SHIFT-1))) >> SHIFT);\
43  OP(dst[5], (src[ 3] + src[ 8] - 5*(src[ 4]+src[7]) + src[5]*C1 + src[6]*C2 + (1<<(SHIFT-1))) >> SHIFT);\
44  OP(dst[6], (src[ 4] + src[ 9] - 5*(src[ 5]+src[8]) + src[6]*C1 + src[7]*C2 + (1<<(SHIFT-1))) >> SHIFT);\
45  OP(dst[7], (src[ 5] + src[10] - 5*(src[ 6]+src[9]) + src[7]*C1 + src[8]*C2 + (1<<(SHIFT-1))) >> SHIFT);\
46  dst += dstStride;\
47  src += srcStride;\
48  }\
49 }\
50 \
51 static void OPNAME ## rv40_qpel8_v_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride,\
52  const int w, const int C1, const int C2, const int SHIFT){\
53  uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
54  int i;\
55  for(i = 0; i < w; i++)\
56  {\
57  const int srcB = src[-2*srcStride];\
58  const int srcA = src[-1*srcStride];\
59  const int src0 = src[0 *srcStride];\
60  const int src1 = src[1 *srcStride];\
61  const int src2 = src[2 *srcStride];\
62  const int src3 = src[3 *srcStride];\
63  const int src4 = src[4 *srcStride];\
64  const int src5 = src[5 *srcStride];\
65  const int src6 = src[6 *srcStride];\
66  const int src7 = src[7 *srcStride];\
67  const int src8 = src[8 *srcStride];\
68  const int src9 = src[9 *srcStride];\
69  const int src10 = src[10*srcStride];\
70  OP(dst[0*dstStride], (srcB + src3 - 5*(srcA+src2) + src0*C1 + src1*C2 + (1<<(SHIFT-1))) >> SHIFT);\
71  OP(dst[1*dstStride], (srcA + src4 - 5*(src0+src3) + src1*C1 + src2*C2 + (1<<(SHIFT-1))) >> SHIFT);\
72  OP(dst[2*dstStride], (src0 + src5 - 5*(src1+src4) + src2*C1 + src3*C2 + (1<<(SHIFT-1))) >> SHIFT);\
73  OP(dst[3*dstStride], (src1 + src6 - 5*(src2+src5) + src3*C1 + src4*C2 + (1<<(SHIFT-1))) >> SHIFT);\
74  OP(dst[4*dstStride], (src2 + src7 - 5*(src3+src6) + src4*C1 + src5*C2 + (1<<(SHIFT-1))) >> SHIFT);\
75  OP(dst[5*dstStride], (src3 + src8 - 5*(src4+src7) + src5*C1 + src6*C2 + (1<<(SHIFT-1))) >> SHIFT);\
76  OP(dst[6*dstStride], (src4 + src9 - 5*(src5+src8) + src6*C1 + src7*C2 + (1<<(SHIFT-1))) >> SHIFT);\
77  OP(dst[7*dstStride], (src5 + src10 - 5*(src6+src9) + src7*C1 + src8*C2 + (1<<(SHIFT-1))) >> SHIFT);\
78  dst++;\
79  src++;\
80  }\
81 }\
82 \
83 static void OPNAME ## rv40_qpel16_v_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride,\
84  const int w, const int C1, const int C2, const int SHIFT){\
85  OPNAME ## rv40_qpel8_v_lowpass(dst , src , dstStride, srcStride, 8, C1, C2, SHIFT);\
86  OPNAME ## rv40_qpel8_v_lowpass(dst+8, src+8, dstStride, srcStride, 8, C1, C2, SHIFT);\
87  src += 8*srcStride;\
88  dst += 8*dstStride;\
89  OPNAME ## rv40_qpel8_v_lowpass(dst , src , dstStride, srcStride, w-8, C1, C2, SHIFT);\
90  OPNAME ## rv40_qpel8_v_lowpass(dst+8, src+8, dstStride, srcStride, w-8, C1, C2, SHIFT);\
91 }\
92 \
93 static void OPNAME ## rv40_qpel16_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride,\
94  const int h, const int C1, const int C2, const int SHIFT){\
95  OPNAME ## rv40_qpel8_h_lowpass(dst , src , dstStride, srcStride, 8, C1, C2, SHIFT);\
96  OPNAME ## rv40_qpel8_h_lowpass(dst+8, src+8, dstStride, srcStride, 8, C1, C2, SHIFT);\
97  src += 8*srcStride;\
98  dst += 8*dstStride;\
99  OPNAME ## rv40_qpel8_h_lowpass(dst , src , dstStride, srcStride, h-8, C1, C2, SHIFT);\
100  OPNAME ## rv40_qpel8_h_lowpass(dst+8, src+8, dstStride, srcStride, h-8, C1, C2, SHIFT);\
101 }\
102 \
103 
104 #define RV40_MC(OPNAME, SIZE) \
105 static void OPNAME ## rv40_qpel ## SIZE ## _mc10_c(uint8_t *dst, uint8_t *src, int stride){\
106  OPNAME ## rv40_qpel ## SIZE ## _h_lowpass(dst, src, stride, stride, SIZE, 52, 20, 6);\
107 }\
108 \
109 static void OPNAME ## rv40_qpel ## SIZE ## _mc30_c(uint8_t *dst, uint8_t *src, int stride){\
110  OPNAME ## rv40_qpel ## SIZE ## _h_lowpass(dst, src, stride, stride, SIZE, 20, 52, 6);\
111 }\
112 \
113 static void OPNAME ## rv40_qpel ## SIZE ## _mc01_c(uint8_t *dst, uint8_t *src, int stride){\
114  OPNAME ## rv40_qpel ## SIZE ## _v_lowpass(dst, src, stride, stride, SIZE, 52, 20, 6);\
115 }\
116 \
117 static void OPNAME ## rv40_qpel ## SIZE ## _mc11_c(uint8_t *dst, uint8_t *src, int stride){\
118  uint8_t full[SIZE*(SIZE+5)];\
119  uint8_t * const full_mid = full + SIZE*2;\
120  put_rv40_qpel ## SIZE ## _h_lowpass(full, src - 2*stride, SIZE, stride, SIZE+5, 52, 20, 6);\
121  OPNAME ## rv40_qpel ## SIZE ## _v_lowpass(dst, full_mid, stride, SIZE, SIZE, 52, 20, 6);\
122 }\
123 \
124 static void OPNAME ## rv40_qpel ## SIZE ## _mc21_c(uint8_t *dst, uint8_t *src, int stride){\
125  uint8_t full[SIZE*(SIZE+5)];\
126  uint8_t * const full_mid = full + SIZE*2;\
127  put_rv40_qpel ## SIZE ## _h_lowpass(full, src - 2*stride, SIZE, stride, SIZE+5, 20, 20, 5);\
128  OPNAME ## rv40_qpel ## SIZE ## _v_lowpass(dst, full_mid, stride, SIZE, SIZE, 52, 20, 6);\
129 }\
130 \
131 static void OPNAME ## rv40_qpel ## SIZE ## _mc31_c(uint8_t *dst, uint8_t *src, int stride){\
132  uint8_t full[SIZE*(SIZE+5)];\
133  uint8_t * const full_mid = full + SIZE*2;\
134  put_rv40_qpel ## SIZE ## _h_lowpass(full, src - 2*stride, SIZE, stride, SIZE+5, 20, 52, 6);\
135  OPNAME ## rv40_qpel ## SIZE ## _v_lowpass(dst, full_mid, stride, SIZE, SIZE, 52, 20, 6);\
136 }\
137 \
138 static void OPNAME ## rv40_qpel ## SIZE ## _mc12_c(uint8_t *dst, uint8_t *src, int stride){\
139  uint8_t full[SIZE*(SIZE+5)];\
140  uint8_t * const full_mid = full + SIZE*2;\
141  put_rv40_qpel ## SIZE ## _h_lowpass(full, src - 2*stride, SIZE, stride, SIZE+5, 52, 20, 6);\
142  OPNAME ## rv40_qpel ## SIZE ## _v_lowpass(dst, full_mid, stride, SIZE, SIZE, 20, 20, 5);\
143 }\
144 \
145 static void OPNAME ## rv40_qpel ## SIZE ## _mc22_c(uint8_t *dst, uint8_t *src, int stride){\
146  uint8_t full[SIZE*(SIZE+5)];\
147  uint8_t * const full_mid = full + SIZE*2;\
148  put_rv40_qpel ## SIZE ## _h_lowpass(full, src - 2*stride, SIZE, stride, SIZE+5, 20, 20, 5);\
149  OPNAME ## rv40_qpel ## SIZE ## _v_lowpass(dst, full_mid, stride, SIZE, SIZE, 20, 20, 5);\
150 }\
151 \
152 static void OPNAME ## rv40_qpel ## SIZE ## _mc32_c(uint8_t *dst, uint8_t *src, int stride){\
153  uint8_t full[SIZE*(SIZE+5)];\
154  uint8_t * const full_mid = full + SIZE*2;\
155  put_rv40_qpel ## SIZE ## _h_lowpass(full, src - 2*stride, SIZE, stride, SIZE+5, 20, 52, 6);\
156  OPNAME ## rv40_qpel ## SIZE ## _v_lowpass(dst, full_mid, stride, SIZE, SIZE, 20, 20, 5);\
157 }\
158 \
159 static void OPNAME ## rv40_qpel ## SIZE ## _mc03_c(uint8_t *dst, uint8_t *src, int stride){\
160  OPNAME ## rv40_qpel ## SIZE ## _v_lowpass(dst, src, stride, stride, SIZE, 20, 52, 6);\
161 }\
162 \
163 static void OPNAME ## rv40_qpel ## SIZE ## _mc13_c(uint8_t *dst, uint8_t *src, int stride){\
164  uint8_t full[SIZE*(SIZE+5)];\
165  uint8_t * const full_mid = full + SIZE*2;\
166  put_rv40_qpel ## SIZE ## _h_lowpass(full, src - 2*stride, SIZE, stride, SIZE+5, 52, 20, 6);\
167  OPNAME ## rv40_qpel ## SIZE ## _v_lowpass(dst, full_mid, stride, SIZE, SIZE, 20, 52, 6);\
168 }\
169 \
170 static void OPNAME ## rv40_qpel ## SIZE ## _mc23_c(uint8_t *dst, uint8_t *src, int stride){\
171  uint8_t full[SIZE*(SIZE+5)];\
172  uint8_t * const full_mid = full + SIZE*2;\
173  put_rv40_qpel ## SIZE ## _h_lowpass(full, src - 2*stride, SIZE, stride, SIZE+5, 20, 20, 5);\
174  OPNAME ## rv40_qpel ## SIZE ## _v_lowpass(dst, full_mid, stride, SIZE, SIZE, 20, 52, 6);\
175 }\
176 \
177 
178 #define op_avg(a, b) a = (((a)+cm[b]+1)>>1)
179 #define op_put(a, b) a = cm[b]
180 
181 RV40_LOWPASS(put_ , op_put)
182 RV40_LOWPASS(avg_ , op_avg)
183 
184 #undef op_avg
185 #undef op_put
186 
187 RV40_MC(put_, 8)
188 RV40_MC(put_, 16)
189 RV40_MC(avg_, 8)
190 RV40_MC(avg_, 16)
191 
192 static const int rv40_bias[4][4] = {
193  { 0, 16, 32, 16 },
194  { 32, 28, 32, 28 },
195  { 0, 32, 16, 32 },
196  { 32, 28, 32, 28 }
197 };
198 
199 #define RV40_CHROMA_MC(OPNAME, OP)\
200 static void OPNAME ## rv40_chroma_mc4_c(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int stride, int h, int x, int y){\
201  const int A = (8-x) * (8-y);\
202  const int B = ( x) * (8-y);\
203  const int C = (8-x) * ( y);\
204  const int D = ( x) * ( y);\
205  int i;\
206  int bias = rv40_bias[y>>1][x>>1];\
207  \
208  assert(x<8 && y<8 && x>=0 && y>=0);\
209 \
210  if(D){\
211  for(i = 0; i < h; i++){\
212  OP(dst[0], (A*src[0] + B*src[1] + C*src[stride+0] + D*src[stride+1] + bias));\
213  OP(dst[1], (A*src[1] + B*src[2] + C*src[stride+1] + D*src[stride+2] + bias));\
214  OP(dst[2], (A*src[2] + B*src[3] + C*src[stride+2] + D*src[stride+3] + bias));\
215  OP(dst[3], (A*src[3] + B*src[4] + C*src[stride+3] + D*src[stride+4] + bias));\
216  dst += stride;\
217  src += stride;\
218  }\
219  }else{\
220  const int E = B + C;\
221  const int step = C ? stride : 1;\
222  for(i = 0; i < h; i++){\
223  OP(dst[0], (A*src[0] + E*src[step+0] + bias));\
224  OP(dst[1], (A*src[1] + E*src[step+1] + bias));\
225  OP(dst[2], (A*src[2] + E*src[step+2] + bias));\
226  OP(dst[3], (A*src[3] + E*src[step+3] + bias));\
227  dst += stride;\
228  src += stride;\
229  }\
230  }\
231 }\
232 \
233 static void OPNAME ## rv40_chroma_mc8_c(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int stride, int h, int x, int y){\
234  const int A = (8-x) * (8-y);\
235  const int B = ( x) * (8-y);\
236  const int C = (8-x) * ( y);\
237  const int D = ( x) * ( y);\
238  int i;\
239  int bias = rv40_bias[y>>1][x>>1];\
240  \
241  assert(x<8 && y<8 && x>=0 && y>=0);\
242 \
243  if(D){\
244  for(i = 0; i < h; i++){\
245  OP(dst[0], (A*src[0] + B*src[1] + C*src[stride+0] + D*src[stride+1] + bias));\
246  OP(dst[1], (A*src[1] + B*src[2] + C*src[stride+1] + D*src[stride+2] + bias));\
247  OP(dst[2], (A*src[2] + B*src[3] + C*src[stride+2] + D*src[stride+3] + bias));\
248  OP(dst[3], (A*src[3] + B*src[4] + C*src[stride+3] + D*src[stride+4] + bias));\
249  OP(dst[4], (A*src[4] + B*src[5] + C*src[stride+4] + D*src[stride+5] + bias));\
250  OP(dst[5], (A*src[5] + B*src[6] + C*src[stride+5] + D*src[stride+6] + bias));\
251  OP(dst[6], (A*src[6] + B*src[7] + C*src[stride+6] + D*src[stride+7] + bias));\
252  OP(dst[7], (A*src[7] + B*src[8] + C*src[stride+7] + D*src[stride+8] + bias));\
253  dst += stride;\
254  src += stride;\
255  }\
256  }else{\
257  const int E = B + C;\
258  const int step = C ? stride : 1;\
259  for(i = 0; i < h; i++){\
260  OP(dst[0], (A*src[0] + E*src[step+0] + bias));\
261  OP(dst[1], (A*src[1] + E*src[step+1] + bias));\
262  OP(dst[2], (A*src[2] + E*src[step+2] + bias));\
263  OP(dst[3], (A*src[3] + E*src[step+3] + bias));\
264  OP(dst[4], (A*src[4] + E*src[step+4] + bias));\
265  OP(dst[5], (A*src[5] + E*src[step+5] + bias));\
266  OP(dst[6], (A*src[6] + E*src[step+6] + bias));\
267  OP(dst[7], (A*src[7] + E*src[step+7] + bias));\
268  dst += stride;\
269  src += stride;\
270  }\
271  }\
272 }
273 
274 #define op_avg(a, b) a = (((a)+((b)>>6)+1)>>1)
275 #define op_put(a, b) a = ((b)>>6)
276 
277 RV40_CHROMA_MC(put_, op_put)
278 RV40_CHROMA_MC(avg_, op_avg)
279 
280 #define RV40_WEIGHT_FUNC(size) \
281 static void rv40_weight_func_ ## size (uint8_t *dst, uint8_t *src1, uint8_t *src2, int w1, int w2, int stride)\
282 {\
283  int i, j;\
284 \
285  for (j = 0; j < size; j++) {\
286  for (i = 0; i < size; i++)\
287  dst[i] = (((w2 * src1[i]) >> 9) + ((w1 * src2[i]) >> 9) + 0x10) >> 5;\
288  src1 += stride;\
289  src2 += stride;\
290  dst += stride;\
291  }\
292 }
293 
296 
300 static const uint8_t rv40_dither_l[16] = {
301  0x40, 0x50, 0x20, 0x60, 0x30, 0x50, 0x40, 0x30,
302  0x50, 0x40, 0x50, 0x30, 0x60, 0x20, 0x50, 0x40
303 };
304 
308 static const uint8_t rv40_dither_r[16] = {
309  0x40, 0x30, 0x60, 0x20, 0x50, 0x30, 0x30, 0x40,
310  0x40, 0x40, 0x50, 0x30, 0x20, 0x60, 0x30, 0x40
311 };
312 
313 #define CLIP_SYMM(a, b) av_clip(a, -(b), b)
314 
317 static av_always_inline void rv40_weak_loop_filter(uint8_t *src,
318  const int step,
319  const int stride,
320  const int filter_p1,
321  const int filter_q1,
322  const int alpha,
323  const int beta,
324  const int lim_p0q0,
325  const int lim_q1,
326  const int lim_p1)
327 {
328  uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
329  int i, t, u, diff;
330 
331  for (i = 0; i < 4; i++, src += stride) {
332  int diff_p1p0 = src[-2*step] - src[-1*step];
333  int diff_q1q0 = src[ 1*step] - src[ 0*step];
334  int diff_p1p2 = src[-2*step] - src[-3*step];
335  int diff_q1q2 = src[ 1*step] - src[ 2*step];
336 
337  t = src[0*step] - src[-1*step];
338  if (!t)
339  continue;
340 
341  u = (alpha * FFABS(t)) >> 7;
342  if (u > 3 - (filter_p1 && filter_q1))
343  continue;
344 
345  t <<= 2;
346  if (filter_p1 && filter_q1)
347  t += src[-2*step] - src[1*step];
348 
349  diff = CLIP_SYMM((t + 4) >> 3, lim_p0q0);
350  src[-1*step] = cm[src[-1*step] + diff];
351  src[ 0*step] = cm[src[ 0*step] - diff];
352 
353  if (filter_p1 && FFABS(diff_p1p2) <= beta) {
354  t = (diff_p1p0 + diff_p1p2 - diff) >> 1;
355  src[-2*step] = cm[src[-2*step] - CLIP_SYMM(t, lim_p1)];
356  }
357 
358  if (filter_q1 && FFABS(diff_q1q2) <= beta) {
359  t = (diff_q1q0 + diff_q1q2 + diff) >> 1;
360  src[ 1*step] = cm[src[ 1*step] - CLIP_SYMM(t, lim_q1)];
361  }
362  }
363 }
364 
365 static void rv40_h_weak_loop_filter(uint8_t *src, const int stride,
366  const int filter_p1, const int filter_q1,
367  const int alpha, const int beta,
368  const int lim_p0q0, const int lim_q1,
369  const int lim_p1)
370 {
371  rv40_weak_loop_filter(src, stride, 1, filter_p1, filter_q1,
372  alpha, beta, lim_p0q0, lim_q1, lim_p1);
373 }
374 
375 static void rv40_v_weak_loop_filter(uint8_t *src, const int stride,
376  const int filter_p1, const int filter_q1,
377  const int alpha, const int beta,
378  const int lim_p0q0, const int lim_q1,
379  const int lim_p1)
380 {
381  rv40_weak_loop_filter(src, 1, stride, filter_p1, filter_q1,
382  alpha, beta, lim_p0q0, lim_q1, lim_p1);
383 }
384 
385 static av_always_inline void rv40_strong_loop_filter(uint8_t *src,
386  const int step,
387  const int stride,
388  const int alpha,
389  const int lims,
390  const int dmode,
391  const int chroma)
392 {
393  int i;
394 
395  for(i = 0; i < 4; i++, src += stride){
396  int sflag, p0, q0, p1, q1;
397  int t = src[0*step] - src[-1*step];
398 
399  if (!t)
400  continue;
401 
402  sflag = (alpha * FFABS(t)) >> 7;
403  if (sflag > 1)
404  continue;
405 
406  p0 = (25*src[-3*step] + 26*src[-2*step] + 26*src[-1*step] +
407  26*src[ 0*step] + 25*src[ 1*step] +
408  rv40_dither_l[dmode + i]) >> 7;
409 
410  q0 = (25*src[-2*step] + 26*src[-1*step] + 26*src[ 0*step] +
411  26*src[ 1*step] + 25*src[ 2*step] +
412  rv40_dither_r[dmode + i]) >> 7;
413 
414  if (sflag) {
415  p0 = av_clip(p0, src[-1*step] - lims, src[-1*step] + lims);
416  q0 = av_clip(q0, src[ 0*step] - lims, src[ 0*step] + lims);
417  }
418 
419  p1 = (25*src[-4*step] + 26*src[-3*step] + 26*src[-2*step] + 26*p0 +
420  25*src[ 0*step] + rv40_dither_l[dmode + i]) >> 7;
421  q1 = (25*src[-1*step] + 26*q0 + 26*src[ 1*step] + 26*src[ 2*step] +
422  25*src[ 3*step] + rv40_dither_r[dmode + i]) >> 7;
423 
424  if (sflag) {
425  p1 = av_clip(p1, src[-2*step] - lims, src[-2*step] + lims);
426  q1 = av_clip(q1, src[ 1*step] - lims, src[ 1*step] + lims);
427  }
428 
429  src[-2*step] = p1;
430  src[-1*step] = p0;
431  src[ 0*step] = q0;
432  src[ 1*step] = q1;
433 
434  if(!chroma){
435  src[-3*step] = (25*src[-1*step] + 26*src[-2*step] +
436  51*src[-3*step] + 26*src[-4*step] + 64) >> 7;
437  src[ 2*step] = (25*src[ 0*step] + 26*src[ 1*step] +
438  51*src[ 2*step] + 26*src[ 3*step] + 64) >> 7;
439  }
440  }
441 }
442 
443 static void rv40_h_strong_loop_filter(uint8_t *src, const int stride,
444  const int alpha, const int lims,
445  const int dmode, const int chroma)
446 {
447  rv40_strong_loop_filter(src, stride, 1, alpha, lims, dmode, chroma);
448 }
449 
450 static void rv40_v_strong_loop_filter(uint8_t *src, const int stride,
451  const int alpha, const int lims,
452  const int dmode, const int chroma)
453 {
454  rv40_strong_loop_filter(src, 1, stride, alpha, lims, dmode, chroma);
455 }
456 
458  int step, int stride,
459  int beta, int beta2,
460  int edge,
461  int *p1, int *q1)
462 {
463  int sum_p1p0 = 0, sum_q1q0 = 0, sum_p1p2 = 0, sum_q1q2 = 0;
464  int strong0 = 0, strong1 = 0;
465  uint8_t *ptr;
466  int i;
467 
468  for (i = 0, ptr = src; i < 4; i++, ptr += stride) {
469  sum_p1p0 += ptr[-2*step] - ptr[-1*step];
470  sum_q1q0 += ptr[ 1*step] - ptr[ 0*step];
471  }
472 
473  *p1 = FFABS(sum_p1p0) < (beta << 2);
474  *q1 = FFABS(sum_q1q0) < (beta << 2);
475 
476  if(!*p1 && !*q1)
477  return 0;
478 
479  if (!edge)
480  return 0;
481 
482  for (i = 0, ptr = src; i < 4; i++, ptr += stride) {
483  sum_p1p2 += ptr[-2*step] - ptr[-3*step];
484  sum_q1q2 += ptr[ 1*step] - ptr[ 2*step];
485  }
486 
487  strong0 = *p1 && (FFABS(sum_p1p2) < beta2);
488  strong1 = *q1 && (FFABS(sum_q1q2) < beta2);
489 
490  return strong0 && strong1;
491 }
492 
493 static int rv40_h_loop_filter_strength(uint8_t *src, int stride,
494  int beta, int beta2, int edge,
495  int *p1, int *q1)
496 {
497  return rv40_loop_filter_strength(src, stride, 1, beta, beta2, edge, p1, q1);
498 }
499 
500 static int rv40_v_loop_filter_strength(uint8_t *src, int stride,
501  int beta, int beta2, int edge,
502  int *p1, int *q1)
503 {
504  return rv40_loop_filter_strength(src, 1, stride, beta, beta2, edge, p1, q1);
505 }
506 
508 
509  ff_rv34dsp_init(c, dsp);
510 
511  c->put_pixels_tab[0][ 0] = dsp->put_h264_qpel_pixels_tab[0][0];
512  c->put_pixels_tab[0][ 1] = put_rv40_qpel16_mc10_c;
513  c->put_pixels_tab[0][ 2] = dsp->put_h264_qpel_pixels_tab[0][2];
514  c->put_pixels_tab[0][ 3] = put_rv40_qpel16_mc30_c;
515  c->put_pixels_tab[0][ 4] = put_rv40_qpel16_mc01_c;
516  c->put_pixels_tab[0][ 5] = put_rv40_qpel16_mc11_c;
517  c->put_pixels_tab[0][ 6] = put_rv40_qpel16_mc21_c;
518  c->put_pixels_tab[0][ 7] = put_rv40_qpel16_mc31_c;
519  c->put_pixels_tab[0][ 8] = dsp->put_h264_qpel_pixels_tab[0][8];
520  c->put_pixels_tab[0][ 9] = put_rv40_qpel16_mc12_c;
521  c->put_pixels_tab[0][10] = put_rv40_qpel16_mc22_c;
522  c->put_pixels_tab[0][11] = put_rv40_qpel16_mc32_c;
523  c->put_pixels_tab[0][12] = put_rv40_qpel16_mc03_c;
524  c->put_pixels_tab[0][13] = put_rv40_qpel16_mc13_c;
525  c->put_pixels_tab[0][14] = put_rv40_qpel16_mc23_c;
527  c->avg_pixels_tab[0][ 0] = dsp->avg_h264_qpel_pixels_tab[0][0];
528  c->avg_pixels_tab[0][ 1] = avg_rv40_qpel16_mc10_c;
529  c->avg_pixels_tab[0][ 2] = dsp->avg_h264_qpel_pixels_tab[0][2];
530  c->avg_pixels_tab[0][ 3] = avg_rv40_qpel16_mc30_c;
531  c->avg_pixels_tab[0][ 4] = avg_rv40_qpel16_mc01_c;
532  c->avg_pixels_tab[0][ 5] = avg_rv40_qpel16_mc11_c;
533  c->avg_pixels_tab[0][ 6] = avg_rv40_qpel16_mc21_c;
534  c->avg_pixels_tab[0][ 7] = avg_rv40_qpel16_mc31_c;
535  c->avg_pixels_tab[0][ 8] = dsp->avg_h264_qpel_pixels_tab[0][8];
536  c->avg_pixels_tab[0][ 9] = avg_rv40_qpel16_mc12_c;
537  c->avg_pixels_tab[0][10] = avg_rv40_qpel16_mc22_c;
538  c->avg_pixels_tab[0][11] = avg_rv40_qpel16_mc32_c;
539  c->avg_pixels_tab[0][12] = avg_rv40_qpel16_mc03_c;
540  c->avg_pixels_tab[0][13] = avg_rv40_qpel16_mc13_c;
541  c->avg_pixels_tab[0][14] = avg_rv40_qpel16_mc23_c;
543  c->put_pixels_tab[1][ 0] = dsp->put_h264_qpel_pixels_tab[1][0];
544  c->put_pixels_tab[1][ 1] = put_rv40_qpel8_mc10_c;
545  c->put_pixels_tab[1][ 2] = dsp->put_h264_qpel_pixels_tab[1][2];
546  c->put_pixels_tab[1][ 3] = put_rv40_qpel8_mc30_c;
547  c->put_pixels_tab[1][ 4] = put_rv40_qpel8_mc01_c;
548  c->put_pixels_tab[1][ 5] = put_rv40_qpel8_mc11_c;
549  c->put_pixels_tab[1][ 6] = put_rv40_qpel8_mc21_c;
550  c->put_pixels_tab[1][ 7] = put_rv40_qpel8_mc31_c;
551  c->put_pixels_tab[1][ 8] = dsp->put_h264_qpel_pixels_tab[1][8];
552  c->put_pixels_tab[1][ 9] = put_rv40_qpel8_mc12_c;
553  c->put_pixels_tab[1][10] = put_rv40_qpel8_mc22_c;
554  c->put_pixels_tab[1][11] = put_rv40_qpel8_mc32_c;
555  c->put_pixels_tab[1][12] = put_rv40_qpel8_mc03_c;
556  c->put_pixels_tab[1][13] = put_rv40_qpel8_mc13_c;
557  c->put_pixels_tab[1][14] = put_rv40_qpel8_mc23_c;
559  c->avg_pixels_tab[1][ 0] = dsp->avg_h264_qpel_pixels_tab[1][0];
560  c->avg_pixels_tab[1][ 1] = avg_rv40_qpel8_mc10_c;
561  c->avg_pixels_tab[1][ 2] = dsp->avg_h264_qpel_pixels_tab[1][2];
562  c->avg_pixels_tab[1][ 3] = avg_rv40_qpel8_mc30_c;
563  c->avg_pixels_tab[1][ 4] = avg_rv40_qpel8_mc01_c;
564  c->avg_pixels_tab[1][ 5] = avg_rv40_qpel8_mc11_c;
565  c->avg_pixels_tab[1][ 6] = avg_rv40_qpel8_mc21_c;
566  c->avg_pixels_tab[1][ 7] = avg_rv40_qpel8_mc31_c;
567  c->avg_pixels_tab[1][ 8] = dsp->avg_h264_qpel_pixels_tab[1][8];
568  c->avg_pixels_tab[1][ 9] = avg_rv40_qpel8_mc12_c;
569  c->avg_pixels_tab[1][10] = avg_rv40_qpel8_mc22_c;
570  c->avg_pixels_tab[1][11] = avg_rv40_qpel8_mc32_c;
571  c->avg_pixels_tab[1][12] = avg_rv40_qpel8_mc03_c;
572  c->avg_pixels_tab[1][13] = avg_rv40_qpel8_mc13_c;
573  c->avg_pixels_tab[1][14] = avg_rv40_qpel8_mc23_c;
575 
576  c->put_chroma_pixels_tab[0] = put_rv40_chroma_mc8_c;
577  c->put_chroma_pixels_tab[1] = put_rv40_chroma_mc4_c;
578  c->avg_chroma_pixels_tab[0] = avg_rv40_chroma_mc8_c;
579  c->avg_chroma_pixels_tab[1] = avg_rv40_chroma_mc4_c;
580 
581  c->rv40_weight_pixels_tab[0] = rv40_weight_func_16;
582  c->rv40_weight_pixels_tab[1] = rv40_weight_func_8;
583 
590 
591  if (HAVE_MMX)
592  ff_rv40dsp_init_x86(c, dsp);
593  if (HAVE_NEON)
594  ff_rv40dsp_init_neon(c, dsp);
595 }