FFmpeg coverage

Directory: ../../../ffmpeg/
File: src/libavcodec/cavsdsp.c
Date: 2024-04-19 17:50:32
Exec Total Coverage
Lines: 181 181 100.0%
Functions: 118 118 100.0%
Branches: 156 156 100.0%
Line Branch Exec Source
1 /*
2 * Chinese AVS video (AVS1-P2, JiZhun profile) decoder.
3 *
4 * DSP functions
5 *
6 * Copyright (c) 2006 Stefan Gehrer <stefan.gehrer@gmx.de>
7 *
8 * This file is part of FFmpeg.
9 *
10 * FFmpeg 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 * FFmpeg 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 FFmpeg; if not, write to the Free Software
22 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
23 */
24
25 #include "idctdsp.h"
26 #include "mathops.h"
27 #include "cavsdsp.h"
28 #include "libavutil/common.h"
29
30 /*****************************************************************************
31 *
32 * in-loop deblocking filter
33 *
34 ****************************************************************************/
35
36 #define P2 p0_p[-3*stride]
37 #define P1 p0_p[-2*stride]
38 #define P0 p0_p[-1*stride]
39 #define Q0 p0_p[ 0*stride]
40 #define Q1 p0_p[ 1*stride]
41 #define Q2 p0_p[ 2*stride]
42
43 1997152 static inline void loop_filter_l2(uint8_t *p0_p, ptrdiff_t stride, int alpha, int beta)
44 {
45 1997152 int p0 = P0;
46 1997152 int q0 = Q0;
47
48
6/6
✓ Branch 0 taken 1924897 times.
✓ Branch 1 taken 72255 times.
✓ Branch 2 taken 1853634 times.
✓ Branch 3 taken 71263 times.
✓ Branch 4 taken 1800900 times.
✓ Branch 5 taken 52734 times.
 1997152 if(abs(p0-q0)<alpha && abs(P1-p0)<beta && abs(Q1-q0)<beta) {
49 1800900 int s = p0 + q0 + 2;
50 1800900 alpha = (alpha>>2) + 2;
51
4/4
✓ Branch 0 taken 1750579 times.
✓ Branch 1 taken 50321 times.
✓ Branch 2 taken 1688025 times.
✓ Branch 3 taken 62554 times.
 1800900 if(abs(P2-p0) < beta && abs(p0-q0) < alpha) {
52 1688025 P0 = (P1 + p0 + s) >> 2;
53 1688025 P1 = (2*P1 + s) >> 2;
54 } else
55 112875 P0 = (2*P1 + s) >> 2;
56
4/4
✓ Branch 0 taken 1749501 times.
✓ Branch 1 taken 51399 times.
✓ Branch 2 taken 1687034 times.
✓ Branch 3 taken 62467 times.
 1800900 if(abs(Q2-q0) < beta && abs(q0-p0) < alpha) {
57 1687034 Q0 = (Q1 + q0 + s) >> 2;
58 1687034 Q1 = (2*Q1 + s) >> 2;
59 } else
60 113866 Q0 = (2*Q1 + s) >> 2;
61 }
62 1997152 }
63
64 2780432 static inline void loop_filter_l1(uint8_t *p0_p, ptrdiff_t stride, int alpha, int beta, int tc)
65 {
66 2780432 int p0 = P0;
67 2780432 int q0 = Q0;
68
69
6/6
✓ Branch 0 taken 2467535 times.
✓ Branch 1 taken 312897 times.
✓ Branch 2 taken 2273533 times.
✓ Branch 3 taken 194002 times.
✓ Branch 4 taken 2156397 times.
✓ Branch 5 taken 117136 times.
 2780432 if(abs(p0-q0)<alpha && abs(P1-p0)<beta && abs(Q1-q0)<beta) {
70 2156397 int delta = av_clip(((q0-p0)*3+P1-Q1+4)>>3,-tc, tc);
71 2156397 P0 = av_clip_uint8(p0+delta);
72 2156397 Q0 = av_clip_uint8(q0-delta);
73
2/2
✓ Branch 0 taken 2076123 times.
✓ Branch 1 taken 80274 times.
 2156397 if(abs(P2-p0)<beta) {
74 2076123 delta = av_clip(((P0-P1)*3+P2-Q0+4)>>3, -tc, tc);
75 2076123 P1 = av_clip_uint8(P1+delta);
76 }
77
2/2
✓ Branch 0 taken 2072458 times.
✓ Branch 1 taken 83939 times.
 2156397 if(abs(Q2-q0)<beta) {
78 2072458 delta = av_clip(((Q1-Q0)*3+P0-Q2+4)>>3, -tc, tc);
79 2072458 Q1 = av_clip_uint8(Q1-delta);
80 }
81 }
82 2780432 }
83
84 1054816 static inline void loop_filter_c2(uint8_t *p0_p, ptrdiff_t stride, int alpha, int beta)
85 {
86 1054816 int p0 = P0;
87 1054816 int q0 = Q0;
88
89
6/6
✓ Branch 0 taken 1032928 times.
✓ Branch 1 taken 21888 times.
✓ Branch 2 taken 1008833 times.
✓ Branch 3 taken 24095 times.
✓ Branch 4 taken 987840 times.
✓ Branch 5 taken 20993 times.
 1054816 if(abs(p0-q0)<alpha && abs(P1-p0)<beta && abs(Q1-q0)<beta) {
90 987840 int s = p0 + q0 + 2;
91 987840 alpha = (alpha>>2) + 2;
92
4/4
✓ Branch 0 taken 967641 times.
✓ Branch 1 taken 20199 times.
✓ Branch 2 taken 946591 times.
✓ Branch 3 taken 21050 times.
 987840 if(abs(P2-p0) < beta && abs(p0-q0) < alpha) {
93 946591 P0 = (P1 + p0 + s) >> 2;
94 } else
95 41249 P0 = (2*P1 + s) >> 2;
96
4/4
✓ Branch 0 taken 967180 times.
✓ Branch 1 taken 20660 times.
✓ Branch 2 taken 946635 times.
✓ Branch 3 taken 20545 times.
 987840 if(abs(Q2-q0) < beta && abs(q0-p0) < alpha) {
97 946635 Q0 = (Q1 + q0 + s) >> 2;
98 } else
99 41205 Q0 = (2*Q1 + s) >> 2;
100 }
101 1054816 }
102
103 2003616 static inline void loop_filter_c1(uint8_t *p0_p, ptrdiff_t stride, int alpha, int beta,
104 int tc)
105 {
106
6/6
✓ Branch 0 taken 1889843 times.
✓ Branch 1 taken 113773 times.
✓ Branch 2 taken 1780422 times.
✓ Branch 3 taken 109421 times.
✓ Branch 4 taken 1704791 times.
✓ Branch 5 taken 75631 times.
 2003616 if(abs(P0-Q0)<alpha && abs(P1-P0)<beta && abs(Q1-Q0)<beta) {
107 1704791 int delta = av_clip(((Q0-P0)*3+P1-Q1+4)>>3, -tc, tc);
108 1704791 P0 = av_clip_uint8(P0+delta);
109 1704791 Q0 = av_clip_uint8(Q0-delta);
110 }
111 2003616 }
112
113 #undef P0
114 #undef P1
115 #undef P2
116 #undef Q0
117 #undef Q1
118 #undef Q2
119
120 272600 static void cavs_filter_lv_c(uint8_t *d, ptrdiff_t stride, int alpha, int beta, int tc,
121 int bs1, int bs2)
122 {
123 int i;
124
2/2
✓ Branch 0 taken 62142 times.
✓ Branch 1 taken 210458 times.
 272600 if(bs1==2)
125
2/2
✓ Branch 0 taken 994272 times.
✓ Branch 1 taken 62142 times.
 1056414 for(i=0;i<16;i++)
126 994272 loop_filter_l2(d + i*stride,1,alpha,beta);
127 else {
128
2/2
✓ Branch 0 taken 87189 times.
✓ Branch 1 taken 123269 times.
 210458 if(bs1)
129
2/2
✓ Branch 0 taken 697512 times.
✓ Branch 1 taken 87189 times.
 784701 for(i=0;i<8;i++)
130 697512 loop_filter_l1(d + i*stride,1,alpha,beta,tc);
131
2/2
✓ Branch 0 taken 85048 times.
✓ Branch 1 taken 125410 times.
 210458 if (bs2)
132
2/2
✓ Branch 0 taken 680384 times.
✓ Branch 1 taken 85048 times.
 765432 for(i=8;i<16;i++)
133 680384 loop_filter_l1(d + i*stride,1,alpha,beta,tc);
134 }
135 272600 }
136
137 270884 static void cavs_filter_lh_c(uint8_t *d, ptrdiff_t stride, int alpha, int beta, int tc,
138 int bs1, int bs2)
139 {
140 int i;
141
2/2
✓ Branch 0 taken 62680 times.
✓ Branch 1 taken 208204 times.
 270884 if(bs1==2)
142
2/2
✓ Branch 0 taken 1002880 times.
✓ Branch 1 taken 62680 times.
 1065560 for(i=0;i<16;i++)
143 1002880 loop_filter_l2(d + i,stride,alpha,beta);
144 else {
145
2/2
✓ Branch 0 taken 87029 times.
✓ Branch 1 taken 121175 times.
 208204 if(bs1)
146
2/2
✓ Branch 0 taken 696232 times.
✓ Branch 1 taken 87029 times.
 783261 for(i=0;i<8;i++)
147 696232 loop_filter_l1(d + i,stride,alpha,beta,tc);
148
2/2
✓ Branch 0 taken 88288 times.
✓ Branch 1 taken 119916 times.
 208204 if (bs2)
149
2/2
✓ Branch 0 taken 706304 times.
✓ Branch 1 taken 88288 times.
 794592 for(i=8;i<16;i++)
150 706304 loop_filter_l1(d + i,stride,alpha,beta,tc);
151 }
152 270884 }
153
154 265866 static void cavs_filter_cv_c(uint8_t *d, ptrdiff_t stride, int alpha, int beta, int tc,
155 int bs1, int bs2)
156 {
157 int i;
158
2/2
✓ Branch 0 taken 65388 times.
✓ Branch 1 taken 200478 times.
 265866 if(bs1==2)
159
2/2
✓ Branch 0 taken 523104 times.
✓ Branch 1 taken 65388 times.
 588492 for(i=0;i<8;i++)
160 523104 loop_filter_c2(d + i*stride,1,alpha,beta);
161 else {
162
2/2
✓ Branch 0 taken 126658 times.
✓ Branch 1 taken 73820 times.
 200478 if(bs1)
163
2/2
✓ Branch 0 taken 506632 times.
✓ Branch 1 taken 126658 times.
 633290 for(i=0;i<4;i++)
164 506632 loop_filter_c1(d + i*stride,1,alpha,beta,tc);
165
2/2
✓ Branch 0 taken 123392 times.
✓ Branch 1 taken 77086 times.
 200478 if (bs2)
166
2/2
✓ Branch 0 taken 493568 times.
✓ Branch 1 taken 123392 times.
 616960 for(i=4;i<8;i++)
167 493568 loop_filter_c1(d + i*stride,1,alpha,beta,tc);
168 }
169 265866 }
170
171 262434 static void cavs_filter_ch_c(uint8_t *d, ptrdiff_t stride, int alpha, int beta, int tc,
172 int bs1, int bs2)
173 {
174 int i;
175
2/2
✓ Branch 0 taken 66464 times.
✓ Branch 1 taken 195970 times.
 262434 if(bs1==2)
176
2/2
✓ Branch 0 taken 531712 times.
✓ Branch 1 taken 66464 times.
 598176 for(i=0;i<8;i++)
177 531712 loop_filter_c2(d + i,stride,alpha,beta);
178 else {
179
2/2
✓ Branch 0 taken 124080 times.
✓ Branch 1 taken 71890 times.
 195970 if(bs1)
180
2/2
✓ Branch 0 taken 496320 times.
✓ Branch 1 taken 124080 times.
 620400 for(i=0;i<4;i++)
181 496320 loop_filter_c1(d + i,stride,alpha,beta,tc);
182
2/2
✓ Branch 0 taken 126774 times.
✓ Branch 1 taken 69196 times.
 195970 if (bs2)
183
2/2
✓ Branch 0 taken 507096 times.
✓ Branch 1 taken 126774 times.
 633870 for(i=4;i<8;i++)
184 507096 loop_filter_c1(d + i,stride,alpha,beta,tc);
185 }
186 262434 }
187
188 /*****************************************************************************
189 *
190 * inverse transform
191 *
192 ****************************************************************************/
193
194 356707 static void cavs_idct8_add_c(uint8_t *dst, int16_t *block, ptrdiff_t stride)
195 {
196 int i;
197 356707 int16_t (*src)[8] = (int16_t(*)[8])block;
198
199 356707 src[0][0] += 8;
200
201
2/2
✓ Branch 0 taken 2853656 times.
✓ Branch 1 taken 356707 times.
 3210363 for( i = 0; i < 8; i++ ) {
202 2853656 const int a0 = 3 * src[i][1] - 2 * src[i][7];
203 2853656 const int a1 = 3 * src[i][3] + 2 * src[i][5];
204 2853656 const int a2 = 2 * src[i][3] - 3 * src[i][5];
205 2853656 const int a3 = 2 * src[i][1] + 3 * src[i][7];
206
207 2853656 const int b4 = 2 * (a0 + a1 + a3) + a1;
208 2853656 const int b5 = 2 * (a0 - a1 + a2) + a0;
209 2853656 const int b6 = 2 * (a3 - a2 - a1) + a3;
210 2853656 const int b7 = 2 * (a0 - a2 - a3) - a2;
211
212 2853656 const int a7 = 4 * src[i][2] - 10 * src[i][6];
213 2853656 const int a6 = 4 * src[i][6] + 10 * src[i][2];
214 2853656 const int a5 = 8 * (src[i][0] - src[i][4]) + 4;
215 2853656 const int a4 = 8 * (src[i][0] + src[i][4]) + 4;
216
217 2853656 const int b0 = a4 + a6;
218 2853656 const int b1 = a5 + a7;
219 2853656 const int b2 = a5 - a7;
220 2853656 const int b3 = a4 - a6;
221
222 2853656 src[i][0] = (b0 + b4) >> 3;
223 2853656 src[i][1] = (b1 + b5) >> 3;
224 2853656 src[i][2] = (b2 + b6) >> 3;
225 2853656 src[i][3] = (b3 + b7) >> 3;
226 2853656 src[i][4] = (b3 - b7) >> 3;
227 2853656 src[i][5] = (b2 - b6) >> 3;
228 2853656 src[i][6] = (b1 - b5) >> 3;
229 2853656 src[i][7] = (b0 - b4) >> 3;
230 }
231
2/2
✓ Branch 0 taken 2853656 times.
✓ Branch 1 taken 356707 times.
 3210363 for( i = 0; i < 8; i++ ) {
232 2853656 const int a0 = 3 * src[1][i] - 2 * src[7][i];
233 2853656 const int a1 = 3 * src[3][i] + 2 * src[5][i];
234 2853656 const int a2 = 2 * src[3][i] - 3 * src[5][i];
235 2853656 const int a3 = 2 * src[1][i] + 3 * src[7][i];
236
237 2853656 const int b4 = 2 * (a0 + a1 + a3) + a1;
238 2853656 const int b5 = 2 * (a0 - a1 + a2) + a0;
239 2853656 const int b6 = 2 * (a3 - a2 - a1) + a3;
240 2853656 const int b7 = 2 * (a0 - a2 - a3) - a2;
241
242 2853656 const int a7 = 4 * src[2][i] - 10 * src[6][i];
243 2853656 const int a6 = 4 * src[6][i] + 10 * src[2][i];
244 2853656 const int a5 = 8 * (src[0][i] - src[4][i]);
245 2853656 const int a4 = 8 * (src[0][i] + src[4][i]);
246
247 2853656 const int b0 = a4 + a6;
248 2853656 const int b1 = a5 + a7;
249 2853656 const int b2 = a5 - a7;
250 2853656 const int b3 = a4 - a6;
251
252 2853656 dst[i + 0*stride] = av_clip_uint8( dst[i + 0*stride] + ((b0 + b4) >> 7));
253 2853656 dst[i + 1*stride] = av_clip_uint8( dst[i + 1*stride] + ((b1 + b5) >> 7));
254 2853656 dst[i + 2*stride] = av_clip_uint8( dst[i + 2*stride] + ((b2 + b6) >> 7));
255 2853656 dst[i + 3*stride] = av_clip_uint8( dst[i + 3*stride] + ((b3 + b7) >> 7));
256 2853656 dst[i + 4*stride] = av_clip_uint8( dst[i + 4*stride] + ((b3 - b7) >> 7));
257 2853656 dst[i + 5*stride] = av_clip_uint8( dst[i + 5*stride] + ((b2 - b6) >> 7));
258 2853656 dst[i + 6*stride] = av_clip_uint8( dst[i + 6*stride] + ((b1 - b5) >> 7));
259 2853656 dst[i + 7*stride] = av_clip_uint8( dst[i + 7*stride] + ((b0 - b4) >> 7));
260 }
261 356707 }
262
263 /*****************************************************************************
264 *
265 * motion compensation
266 *
267 ****************************************************************************/
268
269 #define CAVS_SUBPIX(OPNAME, OP, NAME, A, B, C, D, E, F) \
270 static void OPNAME ## cavs_filt8_h_ ## NAME(uint8_t *dst, const uint8_t *src, ptrdiff_t dstStride, ptrdiff_t srcStride)\
271 { \
272 const int h=8;\
273 const uint8_t *cm = ff_crop_tab + MAX_NEG_CROP;\
274 int i;\
275 for(i=0; i<h; i++)\
276 {\
277 OP(dst[0], A*src[-2] + B*src[-1] + C*src[0] + D*src[1] + E*src[2] + F*src[3]);\
278 OP(dst[1], A*src[-1] + B*src[ 0] + C*src[1] + D*src[2] + E*src[3] + F*src[4]);\
279 OP(dst[2], A*src[ 0] + B*src[ 1] + C*src[2] + D*src[3] + E*src[4] + F*src[5]);\
280 OP(dst[3], A*src[ 1] + B*src[ 2] + C*src[3] + D*src[4] + E*src[5] + F*src[6]);\
281 OP(dst[4], A*src[ 2] + B*src[ 3] + C*src[4] + D*src[5] + E*src[6] + F*src[7]);\
282 OP(dst[5], A*src[ 3] + B*src[ 4] + C*src[5] + D*src[6] + E*src[7] + F*src[8]);\
283 OP(dst[6], A*src[ 4] + B*src[ 5] + C*src[6] + D*src[7] + E*src[8] + F*src[9]);\
284 OP(dst[7], A*src[ 5] + B*src[ 6] + C*src[7] + D*src[8] + E*src[9] + F*src[10]);\
285 dst+=dstStride;\
286 src+=srcStride;\
287 }\
288 }\
289 \
290 static void OPNAME ## cavs_filt8_v_ ## NAME(uint8_t *dst, const uint8_t *src, ptrdiff_t dstStride, ptrdiff_t srcStride)\
291 { \
292 const int w=8;\
293 const uint8_t *cm = ff_crop_tab + MAX_NEG_CROP;\
294 int i;\
295 for(i=0; i<w; i++)\
296 {\
297 const int srcB= src[-2*srcStride];\
298 const int srcA= src[-1*srcStride];\
299 const int src0= src[0 *srcStride];\
300 const int src1= src[1 *srcStride];\
301 const int src2= src[2 *srcStride];\
302 const int src3= src[3 *srcStride];\
303 const int src4= src[4 *srcStride];\
304 const int src5= src[5 *srcStride];\
305 const int src6= src[6 *srcStride];\
306 const int src7= src[7 *srcStride];\
307 const int src8= src[8 *srcStride];\
308 const int src9= src[9 *srcStride];\
309 const int src10= src[10 *srcStride];\
310 OP(dst[0*dstStride], A*srcB + B*srcA + C*src0 + D*src1 + E*src2 + F*src3);\
311 OP(dst[1*dstStride], A*srcA + B*src0 + C*src1 + D*src2 + E*src3 + F*src4);\
312 OP(dst[2*dstStride], A*src0 + B*src1 + C*src2 + D*src3 + E*src4 + F*src5);\
313 OP(dst[3*dstStride], A*src1 + B*src2 + C*src3 + D*src4 + E*src5 + F*src6);\
314 OP(dst[4*dstStride], A*src2 + B*src3 + C*src4 + D*src5 + E*src6 + F*src7);\
315 OP(dst[5*dstStride], A*src3 + B*src4 + C*src5 + D*src6 + E*src7 + F*src8);\
316 OP(dst[6*dstStride], A*src4 + B*src5 + C*src6 + D*src7 + E*src8 + F*src9);\
317 OP(dst[7*dstStride], A*src5 + B*src6 + C*src7 + D*src8 + E*src9 + F*src10);\
318 dst++;\
319 src++;\
320 }\
321 }\
322 \
323 static void OPNAME ## cavs_filt16_v_ ## NAME(uint8_t *dst, const uint8_t *src, ptrdiff_t dstStride, ptrdiff_t srcStride)\
324 { \
325 OPNAME ## cavs_filt8_v_ ## NAME(dst , src , dstStride, srcStride);\
326 OPNAME ## cavs_filt8_v_ ## NAME(dst+8, src+8, dstStride, srcStride);\
327 src += 8*srcStride;\
328 dst += 8*dstStride;\
329 OPNAME ## cavs_filt8_v_ ## NAME(dst , src , dstStride, srcStride);\
330 OPNAME ## cavs_filt8_v_ ## NAME(dst+8, src+8, dstStride, srcStride);\
331 }\
332 \
333 static void OPNAME ## cavs_filt16_h_ ## NAME(uint8_t *dst, const uint8_t *src, ptrdiff_t dstStride, ptrdiff_t srcStride)\
334 { \
335 OPNAME ## cavs_filt8_h_ ## NAME(dst , src , dstStride, srcStride);\
336 OPNAME ## cavs_filt8_h_ ## NAME(dst+8, src+8, dstStride, srcStride);\
337 src += 8*srcStride;\
338 dst += 8*dstStride;\
339 OPNAME ## cavs_filt8_h_ ## NAME(dst , src , dstStride, srcStride);\
340 OPNAME ## cavs_filt8_h_ ## NAME(dst+8, src+8, dstStride, srcStride);\
341 }\
342
343 #define CAVS_SUBPIX_HV(OPNAME, OP, NAME, AH, BH, CH, DH, EH, FH, AV, BV, CV, DV, EV, FV, FULL) \
344 static void OPNAME ## cavs_filt8_hv_ ## NAME(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, ptrdiff_t dstStride, ptrdiff_t srcStride)\
345 { \
346 int16_t temp[8*(8+5)];\
347 int16_t *tmp = temp;\
348 const int h=8;\
349 const int w=8;\
350 const uint8_t *cm = ff_crop_tab + MAX_NEG_CROP;\
351 int i;\
352 src1 -= 2*srcStride;\
353 for(i=0; i<h+5; i++)\
354 {\
355 tmp[0]= AH*src1[-2] + BH*src1[-1] + CH*src1[0] + DH*src1[1] + EH*src1[2] + FH*src1[3];\
356 tmp[1]= AH*src1[-1] + BH*src1[ 0] + CH*src1[1] + DH*src1[2] + EH*src1[3] + FH*src1[4];\
357 tmp[2]= AH*src1[ 0] + BH*src1[ 1] + CH*src1[2] + DH*src1[3] + EH*src1[4] + FH*src1[5];\
358 tmp[3]= AH*src1[ 1] + BH*src1[ 2] + CH*src1[3] + DH*src1[4] + EH*src1[5] + FH*src1[6];\
359 tmp[4]= AH*src1[ 2] + BH*src1[ 3] + CH*src1[4] + DH*src1[5] + EH*src1[6] + FH*src1[7];\
360 tmp[5]= AH*src1[ 3] + BH*src1[ 4] + CH*src1[5] + DH*src1[6] + EH*src1[7] + FH*src1[8];\
361 tmp[6]= AH*src1[ 4] + BH*src1[ 5] + CH*src1[6] + DH*src1[7] + EH*src1[8] + FH*src1[9];\
362 tmp[7]= AH*src1[ 5] + BH*src1[ 6] + CH*src1[7] + DH*src1[8] + EH*src1[9] + FH*src1[10];\
363 tmp+=8;\
364 src1+=srcStride;\
365 }\
366 if(FULL) {\
367 tmp = temp+8*2; \
368 for(i=0; i<w; i++) \
369 { \
370 const int tmpB= tmp[-2*8]; \
371 const int tmpA= tmp[-1*8]; \
372 const int tmp0= tmp[0 *8]; \
373 const int tmp1= tmp[1 *8]; \
374 const int tmp2= tmp[2 *8]; \
375 const int tmp3= tmp[3 *8]; \
376 const int tmp4= tmp[4 *8]; \
377 const int tmp5= tmp[5 *8]; \
378 const int tmp6= tmp[6 *8]; \
379 const int tmp7= tmp[7 *8]; \
380 const int tmp8= tmp[8 *8]; \
381 const int tmp9= tmp[9 *8]; \
382 const int tmp10=tmp[10*8]; \
383 OP(dst[0*dstStride], AV*tmpB + BV*tmpA + CV*tmp0 + DV*tmp1 + EV*tmp2 + FV*tmp3 + 64*src2[0*srcStride]); \
384 OP(dst[1*dstStride], AV*tmpA + BV*tmp0 + CV*tmp1 + DV*tmp2 + EV*tmp3 + FV*tmp4 + 64*src2[1*srcStride]); \
385 OP(dst[2*dstStride], AV*tmp0 + BV*tmp1 + CV*tmp2 + DV*tmp3 + EV*tmp4 + FV*tmp5 + 64*src2[2*srcStride]); \
386 OP(dst[3*dstStride], AV*tmp1 + BV*tmp2 + CV*tmp3 + DV*tmp4 + EV*tmp5 + FV*tmp6 + 64*src2[3*srcStride]); \
387 OP(dst[4*dstStride], AV*tmp2 + BV*tmp3 + CV*tmp4 + DV*tmp5 + EV*tmp6 + FV*tmp7 + 64*src2[4*srcStride]); \
388 OP(dst[5*dstStride], AV*tmp3 + BV*tmp4 + CV*tmp5 + DV*tmp6 + EV*tmp7 + FV*tmp8 + 64*src2[5*srcStride]); \
389 OP(dst[6*dstStride], AV*tmp4 + BV*tmp5 + CV*tmp6 + DV*tmp7 + EV*tmp8 + FV*tmp9 + 64*src2[6*srcStride]); \
390 OP(dst[7*dstStride], AV*tmp5 + BV*tmp6 + CV*tmp7 + DV*tmp8 + EV*tmp9 + FV*tmp10 + 64*src2[7*srcStride]); \
391 dst++; \
392 tmp++; \
393 src2++; \
394 } \
395 } else {\
396 tmp = temp+8*2; \
397 for(i=0; i<w; i++) \
398 { \
399 const int tmpB= tmp[-2*8]; \
400 const int tmpA= tmp[-1*8]; \
401 const int tmp0= tmp[0 *8]; \
402 const int tmp1= tmp[1 *8]; \
403 const int tmp2= tmp[2 *8]; \
404 const int tmp3= tmp[3 *8]; \
405 const int tmp4= tmp[4 *8]; \
406 const int tmp5= tmp[5 *8]; \
407 const int tmp6= tmp[6 *8]; \
408 const int tmp7= tmp[7 *8]; \
409 const int tmp8= tmp[8 *8]; \
410 const int tmp9= tmp[9 *8]; \
411 const int tmp10=tmp[10*8]; \
412 OP(dst[0*dstStride], AV*tmpB + BV*tmpA + CV*tmp0 + DV*tmp1 + EV*tmp2 + FV*tmp3); \
413 OP(dst[1*dstStride], AV*tmpA + BV*tmp0 + CV*tmp1 + DV*tmp2 + EV*tmp3 + FV*tmp4); \
414 OP(dst[2*dstStride], AV*tmp0 + BV*tmp1 + CV*tmp2 + DV*tmp3 + EV*tmp4 + FV*tmp5); \
415 OP(dst[3*dstStride], AV*tmp1 + BV*tmp2 + CV*tmp3 + DV*tmp4 + EV*tmp5 + FV*tmp6); \
416 OP(dst[4*dstStride], AV*tmp2 + BV*tmp3 + CV*tmp4 + DV*tmp5 + EV*tmp6 + FV*tmp7); \
417 OP(dst[5*dstStride], AV*tmp3 + BV*tmp4 + CV*tmp5 + DV*tmp6 + EV*tmp7 + FV*tmp8); \
418 OP(dst[6*dstStride], AV*tmp4 + BV*tmp5 + CV*tmp6 + DV*tmp7 + EV*tmp8 + FV*tmp9); \
419 OP(dst[7*dstStride], AV*tmp5 + BV*tmp6 + CV*tmp7 + DV*tmp8 + EV*tmp9 + FV*tmp10); \
420 dst++; \
421 tmp++; \
422 } \
423 }\
424 }\
425 \
426 static void OPNAME ## cavs_filt16_hv_ ## NAME(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, ptrdiff_t dstStride, ptrdiff_t srcStride)\
427 { \
428 OPNAME ## cavs_filt8_hv_ ## NAME(dst , src1, FULL ? src2 : NULL, dstStride, srcStride); \
429 OPNAME ## cavs_filt8_hv_ ## NAME(dst+8, src1+8, FULL ? src2 + 8 : NULL, dstStride, srcStride); \
430 src1 += 8*srcStride;\
431 if (FULL) \
432 src2 += 8*srcStride;\
433 dst += 8*dstStride;\
434 OPNAME ## cavs_filt8_hv_ ## NAME(dst , src1, FULL ? src2 : NULL, dstStride, srcStride); \
435 OPNAME ## cavs_filt8_hv_ ## NAME(dst+8, src1+8, FULL ? src2 + 8 : NULL, dstStride, srcStride); \
436 }\
437
438 #define CAVS_MC(OPNAME, SIZE) \
439 static void OPNAME ## cavs_qpel ## SIZE ## _mc10_c(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
440 {\
441 OPNAME ## cavs_filt ## SIZE ## _h_qpel_l(dst, src, stride, stride);\
442 }\
443 \
444 static void OPNAME ## cavs_qpel ## SIZE ## _mc20_c(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
445 {\
446 OPNAME ## cavs_filt ## SIZE ## _h_hpel(dst, src, stride, stride);\
447 }\
448 \
449 static void OPNAME ## cavs_qpel ## SIZE ## _mc30_c(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
450 {\
451 OPNAME ## cavs_filt ## SIZE ## _h_qpel_r(dst, src, stride, stride);\
452 }\
453 \
454 static void OPNAME ## cavs_qpel ## SIZE ## _mc01_c(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
455 {\
456 OPNAME ## cavs_filt ## SIZE ## _v_qpel_l(dst, src, stride, stride);\
457 }\
458 \
459 static void OPNAME ## cavs_qpel ## SIZE ## _mc02_c(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
460 {\
461 OPNAME ## cavs_filt ## SIZE ## _v_hpel(dst, src, stride, stride);\
462 }\
463 \
464 static void OPNAME ## cavs_qpel ## SIZE ## _mc03_c(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
465 {\
466 OPNAME ## cavs_filt ## SIZE ## _v_qpel_r(dst, src, stride, stride);\
467 }\
468 \
469 static void OPNAME ## cavs_qpel ## SIZE ## _mc22_c(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
470 {\
471 OPNAME ## cavs_filt ## SIZE ## _hv_jj(dst, src, NULL, stride, stride); \
472 }\
473 \
474 static void OPNAME ## cavs_qpel ## SIZE ## _mc11_c(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
475 {\
476 OPNAME ## cavs_filt ## SIZE ## _hv_egpr(dst, src, src, stride, stride); \
477 }\
478 \
479 static void OPNAME ## cavs_qpel ## SIZE ## _mc13_c(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
480 {\
481 OPNAME ## cavs_filt ## SIZE ## _hv_egpr(dst, src, src+stride, stride, stride); \
482 }\
483 \
484 static void OPNAME ## cavs_qpel ## SIZE ## _mc31_c(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
485 {\
486 OPNAME ## cavs_filt ## SIZE ## _hv_egpr(dst, src, src+1, stride, stride); \
487 }\
488 \
489 static void OPNAME ## cavs_qpel ## SIZE ## _mc33_c(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
490 {\
491 OPNAME ## cavs_filt ## SIZE ## _hv_egpr(dst, src, src+stride+1,stride, stride); \
492 }\
493 \
494 static void OPNAME ## cavs_qpel ## SIZE ## _mc21_c(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
495 {\
496 OPNAME ## cavs_filt ## SIZE ## _hv_ff(dst, src, NULL, stride, stride); \
497 }\
498 \
499 static void OPNAME ## cavs_qpel ## SIZE ## _mc12_c(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
500 {\
501 OPNAME ## cavs_filt ## SIZE ## _hv_ii(dst, src, NULL, stride, stride); \
502 }\
503 \
504 static void OPNAME ## cavs_qpel ## SIZE ## _mc32_c(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
505 {\
506 OPNAME ## cavs_filt ## SIZE ## _hv_kk(dst, src, NULL, stride, stride); \
507 }\
508 \
509 static void OPNAME ## cavs_qpel ## SIZE ## _mc23_c(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
510 {\
511 OPNAME ## cavs_filt ## SIZE ## _hv_qq(dst, src, NULL, stride, stride); \
512 }\
513
514 #define op_put1(a, b) a = cm[((b)+4)>>3]
515 #define op_put2(a, b) a = cm[((b)+64)>>7]
516 #define op_put3(a, b) a = cm[((b)+32)>>6]
517 #define op_put4(a, b) a = cm[((b)+512)>>10]
518 #define op_avg1(a, b) a = ((a)+cm[((b)+4)>>3] +1)>>1
519 #define op_avg2(a, b) a = ((a)+cm[((b)+64)>>7] +1)>>1
520 #define op_avg3(a, b) a = ((a)+cm[((b)+32)>>6] +1)>>1
521 #define op_avg4(a, b) a = ((a)+cm[((b)+512)>>10]+1)>>1
522
2/2
✓ Branch 0 taken 267944 times.
✓ Branch 1 taken 33493 times.
 606866 CAVS_SUBPIX(put_ , op_put1, hpel, 0, -1, 5, 5, -1, 0)
523
2/2
✓ Branch 0 taken 368392 times.
✓ Branch 1 taken 46049 times.
 835604 CAVS_SUBPIX(put_ , op_put2, qpel_l, -1, -2, 96, 42, -7, 0)
524
2/2
✓ Branch 0 taken 369304 times.
✓ Branch 1 taken 46163 times.
 837846 CAVS_SUBPIX(put_ , op_put2, qpel_r, 0, -7, 42, 96, -2, -1)
525
4/4
✓ Branch 0 taken 35529 times.
✓ Branch 1 taken 2733 times.
✓ Branch 2 taken 21864 times.
✓ Branch 3 taken 2733 times.
 120584 CAVS_SUBPIX_HV(put_, op_put3, jj, 0, -1, 5, 5, -1, 0, 0, -1, 5, 5, -1, 0, 0)
526
4/4
✓ Branch 0 taken 49491 times.
✓ Branch 1 taken 3807 times.
✓ Branch 2 taken 30456 times.
✓ Branch 3 taken 3807 times.
 167984 CAVS_SUBPIX_HV(put_, op_put4, ff, 0, -1, 5, 5, -1, 0, -1, -2, 96, 42, -7, 0, 0)
527
4/4
✓ Branch 0 taken 47606 times.
✓ Branch 1 taken 3662 times.
✓ Branch 2 taken 29296 times.
✓ Branch 3 taken 3662 times.
 161542 CAVS_SUBPIX_HV(put_, op_put4, ii, -1, -2, 96, 42, -7, 0, 0, -1, 5, 5, -1, 0, 0)
528
4/4
✓ Branch 0 taken 54821 times.
✓ Branch 1 taken 4217 times.
✓ Branch 2 taken 33736 times.
✓ Branch 3 taken 4217 times.
 186086 CAVS_SUBPIX_HV(put_, op_put4, kk, 0, -7, 42, 96, -2, -1, 0, -1, 5, 5, -1, 0, 0)
529
4/4
✓ Branch 0 taken 49218 times.
✓ Branch 1 taken 3786 times.
✓ Branch 2 taken 30288 times.
✓ Branch 3 taken 3786 times.
 167042 CAVS_SUBPIX_HV(put_, op_put4, qq, 0, -1, 5, 5, -1, 0, 0, -7, 42, 96, -2,-1, 0)
530
4/4
✓ Branch 0 taken 297648 times.
✓ Branch 1 taken 22896 times.
✓ Branch 2 taken 183168 times.
✓ Branch 3 taken 22896 times.
 1010754 CAVS_SUBPIX_HV(put_, op_put2, egpr, 0, -1, 5, 5, -1, 0, 0, -1, 5, 5, -1, 0, 1)
531
2/2
✓ Branch 0 taken 171352 times.
✓ Branch 1 taken 21419 times.
 386408 CAVS_SUBPIX(avg_ , op_avg1, hpel, 0, -1, 5, 5, -1, 0)
532
2/2
✓ Branch 0 taken 77712 times.
✓ Branch 1 taken 9714 times.
 175106 CAVS_SUBPIX(avg_ , op_avg2, qpel_l, -1, -2, 96, 42, -7, 0)
533
2/2
✓ Branch 0 taken 91392 times.
✓ Branch 1 taken 11424 times.
 206104 CAVS_SUBPIX(avg_ , op_avg2, qpel_r, 0, -7, 42, 96, -2, -1)
534
4/4
✓ Branch 0 taken 26364 times.
✓ Branch 1 taken 2028 times.
✓ Branch 2 taken 16224 times.
✓ Branch 3 taken 2028 times.
 89356 CAVS_SUBPIX_HV(avg_, op_avg3, jj, 0, -1, 5, 5, -1, 0, 0, -1, 5, 5, -1, 0, 0)
535
4/4
✓ Branch 0 taken 9165 times.
✓ Branch 1 taken 705 times.
✓ Branch 2 taken 5640 times.
✓ Branch 3 taken 705 times.
 31030 CAVS_SUBPIX_HV(avg_, op_avg4, ff, 0, -1, 5, 5, -1, 0, -1, -2, 96, 42, -7, 0, 0)
536
4/4
✓ Branch 0 taken 8242 times.
✓ Branch 1 taken 634 times.
✓ Branch 2 taken 5072 times.
✓ Branch 3 taken 634 times.
 27918 CAVS_SUBPIX_HV(avg_, op_avg4, ii, -1, -2, 96, 42, -7, 0, 0, -1, 5, 5, -1, 0, 0)
537
4/4
✓ Branch 0 taken 9386 times.
✓ Branch 1 taken 722 times.
✓ Branch 2 taken 5776 times.
✓ Branch 3 taken 722 times.
 31786 CAVS_SUBPIX_HV(avg_, op_avg4, kk, 0, -7, 42, 96, -2, -1, 0, -1, 5, 5, -1, 0, 0)
538
4/4
✓ Branch 0 taken 11882 times.
✓ Branch 1 taken 914 times.
✓ Branch 2 taken 7312 times.
✓ Branch 3 taken 914 times.
 40244 CAVS_SUBPIX_HV(avg_, op_avg4, qq, 0, -1, 5, 5, -1, 0, 0, -7, 42, 96, -2,-1, 0)
539
4/4
✓ Branch 0 taken 39650 times.
✓ Branch 1 taken 3050 times.
✓ Branch 2 taken 24400 times.
✓ Branch 3 taken 3050 times.
 134348 CAVS_SUBPIX_HV(avg_, op_avg2, egpr, 0, -1, 5, 5, -1, 0, 0, -1, 5, 5, -1, 0, 1)
540 240916 CAVS_MC(put_, 8)
541 23174 CAVS_MC(put_, 16)
542 93452 CAVS_MC(avg_, 8)
543 1942 CAVS_MC(avg_, 16)
544
545 #define put_cavs_qpel8_mc00_c ff_put_pixels8x8_c
546 #define avg_cavs_qpel8_mc00_c ff_avg_pixels8x8_c
547 #define put_cavs_qpel16_mc00_c ff_put_pixels16x16_c
548 #define avg_cavs_qpel16_mc00_c ff_avg_pixels16x16_c
549
550 6 av_cold void ff_cavsdsp_init(CAVSDSPContext* c)
551 {
552 #define dspfunc(PFX, IDX, NUM) \
553 c->PFX ## _pixels_tab[IDX][ 0] = PFX ## NUM ## _mc00_c; \
554 c->PFX ## _pixels_tab[IDX][ 1] = PFX ## NUM ## _mc10_c; \
555 c->PFX ## _pixels_tab[IDX][ 2] = PFX ## NUM ## _mc20_c; \
556 c->PFX ## _pixels_tab[IDX][ 3] = PFX ## NUM ## _mc30_c; \
557 c->PFX ## _pixels_tab[IDX][ 4] = PFX ## NUM ## _mc01_c; \
558 c->PFX ## _pixels_tab[IDX][ 5] = PFX ## NUM ## _mc11_c; \
559 c->PFX ## _pixels_tab[IDX][ 6] = PFX ## NUM ## _mc21_c; \
560 c->PFX ## _pixels_tab[IDX][ 7] = PFX ## NUM ## _mc31_c; \
561 c->PFX ## _pixels_tab[IDX][ 8] = PFX ## NUM ## _mc02_c; \
562 c->PFX ## _pixels_tab[IDX][ 9] = PFX ## NUM ## _mc12_c; \
563 c->PFX ## _pixels_tab[IDX][10] = PFX ## NUM ## _mc22_c; \
564 c->PFX ## _pixels_tab[IDX][11] = PFX ## NUM ## _mc32_c; \
565 c->PFX ## _pixels_tab[IDX][12] = PFX ## NUM ## _mc03_c; \
566 c->PFX ## _pixels_tab[IDX][13] = PFX ## NUM ## _mc13_c; \
567 c->PFX ## _pixels_tab[IDX][14] = PFX ## NUM ## _mc23_c; \
568 c->PFX ## _pixels_tab[IDX][15] = PFX ## NUM ## _mc33_c
569 6 dspfunc(put_cavs_qpel, 0, 16);
570 6 dspfunc(put_cavs_qpel, 1, 8);
571 6 dspfunc(avg_cavs_qpel, 0, 16);
572 6 dspfunc(avg_cavs_qpel, 1, 8);
573 6 c->cavs_filter_lv = cavs_filter_lv_c;
574 6 c->cavs_filter_lh = cavs_filter_lh_c;
575 6 c->cavs_filter_cv = cavs_filter_cv_c;
576 6 c->cavs_filter_ch = cavs_filter_ch_c;
577 6 c->cavs_idct8_add = cavs_idct8_add_c;
578 6 c->idct_perm = FF_IDCT_PERM_NONE;
579
580 #if ARCH_X86
581 6 ff_cavsdsp_init_x86(c);
582 #endif
583 6 }
584