FFmpeg coverage


Directory: ../../../ffmpeg/
File: src/libavcodec/me_cmp.c
Date: 2024-04-20 14:10:07
Exec Total Coverage
Lines: 412 570 72.3%
Functions: 29 43 67.4%
Branches: 102 163 62.6%

Line Branch Exec Source
1 /*
2 * DSP utils
3 * Copyright (c) 2000, 2001 Fabrice Bellard
4 * Copyright (c) 2002-2004 Michael Niedermayer <michaelni@gmx.at>
5 *
6 * This file is part of FFmpeg.
7 *
8 * FFmpeg is free software; you can redistribute it and/or
9 * modify it under the terms of the GNU Lesser General Public
10 * License as published by the Free Software Foundation; either
11 * version 2.1 of the License, or (at your option) any later version.
12 *
13 * FFmpeg is distributed in the hope that it will be useful,
14 * but WITHOUT ANY WARRANTY; without even the implied warranty of
15 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
16 * Lesser General Public License for more details.
17 *
18 * You should have received a copy of the GNU Lesser General Public
19 * License along with FFmpeg; if not, write to the Free Software
20 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
21 */
22
23 #include "libavutil/attributes.h"
24 #include "libavutil/internal.h"
25 #include "libavutil/mem_internal.h"
26 #include "avcodec.h"
27 #include "copy_block.h"
28 #include "simple_idct.h"
29 #include "me_cmp.h"
30 #include "mpegvideoenc.h"
31 #include "config.h"
32 #include "config_components.h"
33
34 /* (i - 256) * (i - 256) */
35 const uint32_t ff_square_tab[512] = {
36 65536, 65025, 64516, 64009, 63504, 63001, 62500, 62001, 61504, 61009, 60516, 60025, 59536, 59049, 58564, 58081,
37 57600, 57121, 56644, 56169, 55696, 55225, 54756, 54289, 53824, 53361, 52900, 52441, 51984, 51529, 51076, 50625,
38 50176, 49729, 49284, 48841, 48400, 47961, 47524, 47089, 46656, 46225, 45796, 45369, 44944, 44521, 44100, 43681,
39 43264, 42849, 42436, 42025, 41616, 41209, 40804, 40401, 40000, 39601, 39204, 38809, 38416, 38025, 37636, 37249,
40 36864, 36481, 36100, 35721, 35344, 34969, 34596, 34225, 33856, 33489, 33124, 32761, 32400, 32041, 31684, 31329,
41 30976, 30625, 30276, 29929, 29584, 29241, 28900, 28561, 28224, 27889, 27556, 27225, 26896, 26569, 26244, 25921,
42 25600, 25281, 24964, 24649, 24336, 24025, 23716, 23409, 23104, 22801, 22500, 22201, 21904, 21609, 21316, 21025,
43 20736, 20449, 20164, 19881, 19600, 19321, 19044, 18769, 18496, 18225, 17956, 17689, 17424, 17161, 16900, 16641,
44 16384, 16129, 15876, 15625, 15376, 15129, 14884, 14641, 14400, 14161, 13924, 13689, 13456, 13225, 12996, 12769,
45 12544, 12321, 12100, 11881, 11664, 11449, 11236, 11025, 10816, 10609, 10404, 10201, 10000, 9801, 9604, 9409,
46 9216, 9025, 8836, 8649, 8464, 8281, 8100, 7921, 7744, 7569, 7396, 7225, 7056, 6889, 6724, 6561,
47 6400, 6241, 6084, 5929, 5776, 5625, 5476, 5329, 5184, 5041, 4900, 4761, 4624, 4489, 4356, 4225,
48 4096, 3969, 3844, 3721, 3600, 3481, 3364, 3249, 3136, 3025, 2916, 2809, 2704, 2601, 2500, 2401,
49 2304, 2209, 2116, 2025, 1936, 1849, 1764, 1681, 1600, 1521, 1444, 1369, 1296, 1225, 1156, 1089,
50 1024, 961, 900, 841, 784, 729, 676, 625, 576, 529, 484, 441, 400, 361, 324, 289,
51 256, 225, 196, 169, 144, 121, 100, 81, 64, 49, 36, 25, 16, 9, 4, 1,
52 0, 1, 4, 9, 16, 25, 36, 49, 64, 81, 100, 121, 144, 169, 196, 225,
53 256, 289, 324, 361, 400, 441, 484, 529, 576, 625, 676, 729, 784, 841, 900, 961,
54 1024, 1089, 1156, 1225, 1296, 1369, 1444, 1521, 1600, 1681, 1764, 1849, 1936, 2025, 2116, 2209,
55 2304, 2401, 2500, 2601, 2704, 2809, 2916, 3025, 3136, 3249, 3364, 3481, 3600, 3721, 3844, 3969,
56 4096, 4225, 4356, 4489, 4624, 4761, 4900, 5041, 5184, 5329, 5476, 5625, 5776, 5929, 6084, 6241,
57 6400, 6561, 6724, 6889, 7056, 7225, 7396, 7569, 7744, 7921, 8100, 8281, 8464, 8649, 8836, 9025,
58 9216, 9409, 9604, 9801, 10000, 10201, 10404, 10609, 10816, 11025, 11236, 11449, 11664, 11881, 12100, 12321,
59 12544, 12769, 12996, 13225, 13456, 13689, 13924, 14161, 14400, 14641, 14884, 15129, 15376, 15625, 15876, 16129,
60 16384, 16641, 16900, 17161, 17424, 17689, 17956, 18225, 18496, 18769, 19044, 19321, 19600, 19881, 20164, 20449,
61 20736, 21025, 21316, 21609, 21904, 22201, 22500, 22801, 23104, 23409, 23716, 24025, 24336, 24649, 24964, 25281,
62 25600, 25921, 26244, 26569, 26896, 27225, 27556, 27889, 28224, 28561, 28900, 29241, 29584, 29929, 30276, 30625,
63 30976, 31329, 31684, 32041, 32400, 32761, 33124, 33489, 33856, 34225, 34596, 34969, 35344, 35721, 36100, 36481,
64 36864, 37249, 37636, 38025, 38416, 38809, 39204, 39601, 40000, 40401, 40804, 41209, 41616, 42025, 42436, 42849,
65 43264, 43681, 44100, 44521, 44944, 45369, 45796, 46225, 46656, 47089, 47524, 47961, 48400, 48841, 49284, 49729,
66 50176, 50625, 51076, 51529, 51984, 52441, 52900, 53361, 53824, 54289, 54756, 55225, 55696, 56169, 56644, 57121,
67 57600, 58081, 58564, 59049, 59536, 60025, 60516, 61009, 61504, 62001, 62500, 63001, 63504, 64009, 64516, 65025,
68 };
69
70 32 static int sse4_c(MpegEncContext *v, const uint8_t *pix1, const uint8_t *pix2,
71 ptrdiff_t stride, int h)
72 {
73 32 int s = 0, i;
74 32 const uint32_t *sq = ff_square_tab + 256;
75
76
2/2
✓ Branch 0 taken 260 times.
✓ Branch 1 taken 32 times.
292 for (i = 0; i < h; i++) {
77 260 s += sq[pix1[0] - pix2[0]];
78 260 s += sq[pix1[1] - pix2[1]];
79 260 s += sq[pix1[2] - pix2[2]];
80 260 s += sq[pix1[3] - pix2[3]];
81 260 pix1 += stride;
82 260 pix2 += stride;
83 }
84 32 return s;
85 }
86
87 4998811 static int sse8_c(MpegEncContext *v, const uint8_t *pix1, const uint8_t *pix2,
88 ptrdiff_t stride, int h)
89 {
90 4998811 int s = 0, i;
91 4998811 const uint32_t *sq = ff_square_tab + 256;
92
93
2/2
✓ Branch 0 taken 46132016 times.
✓ Branch 1 taken 4998811 times.
51130827 for (i = 0; i < h; i++) {
94 46132016 s += sq[pix1[0] - pix2[0]];
95 46132016 s += sq[pix1[1] - pix2[1]];
96 46132016 s += sq[pix1[2] - pix2[2]];
97 46132016 s += sq[pix1[3] - pix2[3]];
98 46132016 s += sq[pix1[4] - pix2[4]];
99 46132016 s += sq[pix1[5] - pix2[5]];
100 46132016 s += sq[pix1[6] - pix2[6]];
101 46132016 s += sq[pix1[7] - pix2[7]];
102 46132016 pix1 += stride;
103 46132016 pix2 += stride;
104 }
105 4998811 return s;
106 }
107
108 7470774 static int sse16_c(MpegEncContext *v, const uint8_t *pix1, const uint8_t *pix2,
109 ptrdiff_t stride, int h)
110 {
111 7470774 int s = 0, i;
112 7470774 const uint32_t *sq = ff_square_tab + 256;
113
114
2/2
✓ Branch 0 taken 119532040 times.
✓ Branch 1 taken 7470774 times.
127002814 for (i = 0; i < h; i++) {
115 119532040 s += sq[pix1[0] - pix2[0]];
116 119532040 s += sq[pix1[1] - pix2[1]];
117 119532040 s += sq[pix1[2] - pix2[2]];
118 119532040 s += sq[pix1[3] - pix2[3]];
119 119532040 s += sq[pix1[4] - pix2[4]];
120 119532040 s += sq[pix1[5] - pix2[5]];
121 119532040 s += sq[pix1[6] - pix2[6]];
122 119532040 s += sq[pix1[7] - pix2[7]];
123 119532040 s += sq[pix1[8] - pix2[8]];
124 119532040 s += sq[pix1[9] - pix2[9]];
125 119532040 s += sq[pix1[10] - pix2[10]];
126 119532040 s += sq[pix1[11] - pix2[11]];
127 119532040 s += sq[pix1[12] - pix2[12]];
128 119532040 s += sq[pix1[13] - pix2[13]];
129 119532040 s += sq[pix1[14] - pix2[14]];
130 119532040 s += sq[pix1[15] - pix2[15]];
131
132 119532040 pix1 += stride;
133 119532040 pix2 += stride;
134 }
135 7470774 return s;
136 }
137
138 static int sum_abs_dctelem_c(const int16_t *block)
139 {
140 int sum = 0, i;
141
142 for (i = 0; i < 64; i++)
143 sum += FFABS(block[i]);
144 return sum;
145 }
146
147 #define avg2(a, b) (((a) + (b) + 1) >> 1)
148 #define avg4(a, b, c, d) (((a) + (b) + (c) + (d) + 2) >> 2)
149
150 44106372 static inline int pix_abs16_c(MpegEncContext *v, const uint8_t *pix1, const uint8_t *pix2,
151 ptrdiff_t stride, int h)
152 {
153 44106372 int s = 0, i;
154
155
2/2
✓ Branch 0 taken 625332812 times.
✓ Branch 1 taken 44106372 times.
669439184 for (i = 0; i < h; i++) {
156 625332812 s += abs(pix1[0] - pix2[0]);
157 625332812 s += abs(pix1[1] - pix2[1]);
158 625332812 s += abs(pix1[2] - pix2[2]);
159 625332812 s += abs(pix1[3] - pix2[3]);
160 625332812 s += abs(pix1[4] - pix2[4]);
161 625332812 s += abs(pix1[5] - pix2[5]);
162 625332812 s += abs(pix1[6] - pix2[6]);
163 625332812 s += abs(pix1[7] - pix2[7]);
164 625332812 s += abs(pix1[8] - pix2[8]);
165 625332812 s += abs(pix1[9] - pix2[9]);
166 625332812 s += abs(pix1[10] - pix2[10]);
167 625332812 s += abs(pix1[11] - pix2[11]);
168 625332812 s += abs(pix1[12] - pix2[12]);
169 625332812 s += abs(pix1[13] - pix2[13]);
170 625332812 s += abs(pix1[14] - pix2[14]);
171 625332812 s += abs(pix1[15] - pix2[15]);
172 625332812 pix1 += stride;
173 625332812 pix2 += stride;
174 }
175 44106372 return s;
176 }
177
178 32 static inline int pix_median_abs16_c(MpegEncContext *v, const uint8_t *pix1, const uint8_t *pix2,
179 ptrdiff_t stride, int h)
180 {
181 32 int s = 0, i, j;
182
183 #define V(x) (pix1[x] - pix2[x])
184
185 32 s += abs(V(0));
186 32 s += abs(V(1) - V(0));
187 32 s += abs(V(2) - V(1));
188 32 s += abs(V(3) - V(2));
189 32 s += abs(V(4) - V(3));
190 32 s += abs(V(5) - V(4));
191 32 s += abs(V(6) - V(5));
192 32 s += abs(V(7) - V(6));
193 32 s += abs(V(8) - V(7));
194 32 s += abs(V(9) - V(8));
195 32 s += abs(V(10) - V(9));
196 32 s += abs(V(11) - V(10));
197 32 s += abs(V(12) - V(11));
198 32 s += abs(V(13) - V(12));
199 32 s += abs(V(14) - V(13));
200 32 s += abs(V(15) - V(14));
201
202 32 pix1 += stride;
203 32 pix2 += stride;
204
205
2/2
✓ Branch 0 taken 224 times.
✓ Branch 1 taken 32 times.
256 for (i = 1; i < h; i++) {
206 224 s += abs(V(0) - V(-stride));
207
2/2
✓ Branch 0 taken 3360 times.
✓ Branch 1 taken 224 times.
3584 for (j = 1; j < 16; j++)
208 3360 s += abs(V(j) - mid_pred(V(j-stride), V(j-1), V(j-stride) + V(j-1) - V(j-stride-1)));
209 224 pix1 += stride;
210 224 pix2 += stride;
211
212 }
213 #undef V
214 32 return s;
215 }
216
217 3278223 static int pix_abs16_x2_c(MpegEncContext *v, const uint8_t *pix1, const uint8_t *pix2,
218 ptrdiff_t stride, int h)
219 {
220 3278223 int s = 0, i;
221
222
2/2
✓ Branch 0 taken 43141702 times.
✓ Branch 1 taken 3278223 times.
46419925 for (i = 0; i < h; i++) {
223 43141702 s += abs(pix1[0] - avg2(pix2[0], pix2[1]));
224 43141702 s += abs(pix1[1] - avg2(pix2[1], pix2[2]));
225 43141702 s += abs(pix1[2] - avg2(pix2[2], pix2[3]));
226 43141702 s += abs(pix1[3] - avg2(pix2[3], pix2[4]));
227 43141702 s += abs(pix1[4] - avg2(pix2[4], pix2[5]));
228 43141702 s += abs(pix1[5] - avg2(pix2[5], pix2[6]));
229 43141702 s += abs(pix1[6] - avg2(pix2[6], pix2[7]));
230 43141702 s += abs(pix1[7] - avg2(pix2[7], pix2[8]));
231 43141702 s += abs(pix1[8] - avg2(pix2[8], pix2[9]));
232 43141702 s += abs(pix1[9] - avg2(pix2[9], pix2[10]));
233 43141702 s += abs(pix1[10] - avg2(pix2[10], pix2[11]));
234 43141702 s += abs(pix1[11] - avg2(pix2[11], pix2[12]));
235 43141702 s += abs(pix1[12] - avg2(pix2[12], pix2[13]));
236 43141702 s += abs(pix1[13] - avg2(pix2[13], pix2[14]));
237 43141702 s += abs(pix1[14] - avg2(pix2[14], pix2[15]));
238 43141702 s += abs(pix1[15] - avg2(pix2[15], pix2[16]));
239 43141702 pix1 += stride;
240 43141702 pix2 += stride;
241 }
242 3278223 return s;
243 }
244
245 3278223 static int pix_abs16_y2_c(MpegEncContext *v, const uint8_t *pix1, const uint8_t *pix2,
246 ptrdiff_t stride, int h)
247 {
248 3278223 int s = 0, i;
249 3278223 const uint8_t *pix3 = pix2 + stride;
250
251
2/2
✓ Branch 0 taken 43141724 times.
✓ Branch 1 taken 3278223 times.
46419947 for (i = 0; i < h; i++) {
252 43141724 s += abs(pix1[0] - avg2(pix2[0], pix3[0]));
253 43141724 s += abs(pix1[1] - avg2(pix2[1], pix3[1]));
254 43141724 s += abs(pix1[2] - avg2(pix2[2], pix3[2]));
255 43141724 s += abs(pix1[3] - avg2(pix2[3], pix3[3]));
256 43141724 s += abs(pix1[4] - avg2(pix2[4], pix3[4]));
257 43141724 s += abs(pix1[5] - avg2(pix2[5], pix3[5]));
258 43141724 s += abs(pix1[6] - avg2(pix2[6], pix3[6]));
259 43141724 s += abs(pix1[7] - avg2(pix2[7], pix3[7]));
260 43141724 s += abs(pix1[8] - avg2(pix2[8], pix3[8]));
261 43141724 s += abs(pix1[9] - avg2(pix2[9], pix3[9]));
262 43141724 s += abs(pix1[10] - avg2(pix2[10], pix3[10]));
263 43141724 s += abs(pix1[11] - avg2(pix2[11], pix3[11]));
264 43141724 s += abs(pix1[12] - avg2(pix2[12], pix3[12]));
265 43141724 s += abs(pix1[13] - avg2(pix2[13], pix3[13]));
266 43141724 s += abs(pix1[14] - avg2(pix2[14], pix3[14]));
267 43141724 s += abs(pix1[15] - avg2(pix2[15], pix3[15]));
268 43141724 pix1 += stride;
269 43141724 pix2 += stride;
270 43141724 pix3 += stride;
271 }
272 3278223 return s;
273 }
274
275 6556398 static int pix_abs16_xy2_c(MpegEncContext *v, const uint8_t *pix1, const uint8_t *pix2,
276 ptrdiff_t stride, int h)
277 {
278 6556398 int s = 0, i;
279 6556398 const uint8_t *pix3 = pix2 + stride;
280
281
2/2
✓ Branch 0 taken 86282974 times.
✓ Branch 1 taken 6556398 times.
92839372 for (i = 0; i < h; i++) {
282 86282974 s += abs(pix1[0] - avg4(pix2[0], pix2[1], pix3[0], pix3[1]));
283 86282974 s += abs(pix1[1] - avg4(pix2[1], pix2[2], pix3[1], pix3[2]));
284 86282974 s += abs(pix1[2] - avg4(pix2[2], pix2[3], pix3[2], pix3[3]));
285 86282974 s += abs(pix1[3] - avg4(pix2[3], pix2[4], pix3[3], pix3[4]));
286 86282974 s += abs(pix1[4] - avg4(pix2[4], pix2[5], pix3[4], pix3[5]));
287 86282974 s += abs(pix1[5] - avg4(pix2[5], pix2[6], pix3[5], pix3[6]));
288 86282974 s += abs(pix1[6] - avg4(pix2[6], pix2[7], pix3[6], pix3[7]));
289 86282974 s += abs(pix1[7] - avg4(pix2[7], pix2[8], pix3[7], pix3[8]));
290 86282974 s += abs(pix1[8] - avg4(pix2[8], pix2[9], pix3[8], pix3[9]));
291 86282974 s += abs(pix1[9] - avg4(pix2[9], pix2[10], pix3[9], pix3[10]));
292 86282974 s += abs(pix1[10] - avg4(pix2[10], pix2[11], pix3[10], pix3[11]));
293 86282974 s += abs(pix1[11] - avg4(pix2[11], pix2[12], pix3[11], pix3[12]));
294 86282974 s += abs(pix1[12] - avg4(pix2[12], pix2[13], pix3[12], pix3[13]));
295 86282974 s += abs(pix1[13] - avg4(pix2[13], pix2[14], pix3[13], pix3[14]));
296 86282974 s += abs(pix1[14] - avg4(pix2[14], pix2[15], pix3[14], pix3[15]));
297 86282974 s += abs(pix1[15] - avg4(pix2[15], pix2[16], pix3[15], pix3[16]));
298 86282974 pix1 += stride;
299 86282974 pix2 += stride;
300 86282974 pix3 += stride;
301 }
302 6556398 return s;
303 }
304
305 33643253 static inline int pix_abs8_c(MpegEncContext *v, const uint8_t *pix1, const uint8_t *pix2,
306 ptrdiff_t stride, int h)
307 {
308 33643253 int s = 0, i;
309
310
2/2
✓ Branch 0 taken 269146252 times.
✓ Branch 1 taken 33643253 times.
302789505 for (i = 0; i < h; i++) {
311 269146252 s += abs(pix1[0] - pix2[0]);
312 269146252 s += abs(pix1[1] - pix2[1]);
313 269146252 s += abs(pix1[2] - pix2[2]);
314 269146252 s += abs(pix1[3] - pix2[3]);
315 269146252 s += abs(pix1[4] - pix2[4]);
316 269146252 s += abs(pix1[5] - pix2[5]);
317 269146252 s += abs(pix1[6] - pix2[6]);
318 269146252 s += abs(pix1[7] - pix2[7]);
319 269146252 pix1 += stride;
320 269146252 pix2 += stride;
321 }
322 33643253 return s;
323 }
324
325 32 static inline int pix_median_abs8_c(MpegEncContext *v, const uint8_t *pix1, const uint8_t *pix2,
326 ptrdiff_t stride, int h)
327 {
328 32 int s = 0, i, j;
329
330 #define V(x) (pix1[x] - pix2[x])
331
332 32 s += abs(V(0));
333 32 s += abs(V(1) - V(0));
334 32 s += abs(V(2) - V(1));
335 32 s += abs(V(3) - V(2));
336 32 s += abs(V(4) - V(3));
337 32 s += abs(V(5) - V(4));
338 32 s += abs(V(6) - V(5));
339 32 s += abs(V(7) - V(6));
340
341 32 pix1 += stride;
342 32 pix2 += stride;
343
344
2/2
✓ Branch 0 taken 296 times.
✓ Branch 1 taken 32 times.
328 for (i = 1; i < h; i++) {
345 296 s += abs(V(0) - V(-stride));
346
2/2
✓ Branch 0 taken 2072 times.
✓ Branch 1 taken 296 times.
2368 for (j = 1; j < 8; j++)
347 2072 s += abs(V(j) - mid_pred(V(j-stride), V(j-1), V(j-stride) + V(j-1) - V(j-stride-1)));
348 296 pix1 += stride;
349 296 pix2 += stride;
350
351 }
352 #undef V
353 32 return s;
354 }
355
356 837597 static int pix_abs8_x2_c(MpegEncContext *v, const uint8_t *pix1, const uint8_t *pix2,
357 ptrdiff_t stride, int h)
358 {
359 837597 int s = 0, i;
360
361
2/2
✓ Branch 0 taken 6700810 times.
✓ Branch 1 taken 837597 times.
7538407 for (i = 0; i < h; i++) {
362 6700810 s += abs(pix1[0] - avg2(pix2[0], pix2[1]));
363 6700810 s += abs(pix1[1] - avg2(pix2[1], pix2[2]));
364 6700810 s += abs(pix1[2] - avg2(pix2[2], pix2[3]));
365 6700810 s += abs(pix1[3] - avg2(pix2[3], pix2[4]));
366 6700810 s += abs(pix1[4] - avg2(pix2[4], pix2[5]));
367 6700810 s += abs(pix1[5] - avg2(pix2[5], pix2[6]));
368 6700810 s += abs(pix1[6] - avg2(pix2[6], pix2[7]));
369 6700810 s += abs(pix1[7] - avg2(pix2[7], pix2[8]));
370 6700810 pix1 += stride;
371 6700810 pix2 += stride;
372 }
373 837597 return s;
374 }
375
376 837597 static int pix_abs8_y2_c(MpegEncContext *v, const uint8_t *pix1, const uint8_t *pix2,
377 ptrdiff_t stride, int h)
378 {
379 837597 int s = 0, i;
380 837597 const uint8_t *pix3 = pix2 + stride;
381
382
2/2
✓ Branch 0 taken 6700850 times.
✓ Branch 1 taken 837597 times.
7538447 for (i = 0; i < h; i++) {
383 6700850 s += abs(pix1[0] - avg2(pix2[0], pix3[0]));
384 6700850 s += abs(pix1[1] - avg2(pix2[1], pix3[1]));
385 6700850 s += abs(pix1[2] - avg2(pix2[2], pix3[2]));
386 6700850 s += abs(pix1[3] - avg2(pix2[3], pix3[3]));
387 6700850 s += abs(pix1[4] - avg2(pix2[4], pix3[4]));
388 6700850 s += abs(pix1[5] - avg2(pix2[5], pix3[5]));
389 6700850 s += abs(pix1[6] - avg2(pix2[6], pix3[6]));
390 6700850 s += abs(pix1[7] - avg2(pix2[7], pix3[7]));
391 6700850 pix1 += stride;
392 6700850 pix2 += stride;
393 6700850 pix3 += stride;
394 }
395 837597 return s;
396 }
397
398 1675146 static int pix_abs8_xy2_c(MpegEncContext *v, const uint8_t *pix1, const uint8_t *pix2,
399 ptrdiff_t stride, int h)
400 {
401 1675146 int s = 0, i;
402 1675146 const uint8_t *pix3 = pix2 + stride;
403
404
2/2
✓ Branch 0 taken 13401256 times.
✓ Branch 1 taken 1675146 times.
15076402 for (i = 0; i < h; i++) {
405 13401256 s += abs(pix1[0] - avg4(pix2[0], pix2[1], pix3[0], pix3[1]));
406 13401256 s += abs(pix1[1] - avg4(pix2[1], pix2[2], pix3[1], pix3[2]));
407 13401256 s += abs(pix1[2] - avg4(pix2[2], pix2[3], pix3[2], pix3[3]));
408 13401256 s += abs(pix1[3] - avg4(pix2[3], pix2[4], pix3[3], pix3[4]));
409 13401256 s += abs(pix1[4] - avg4(pix2[4], pix2[5], pix3[4], pix3[5]));
410 13401256 s += abs(pix1[5] - avg4(pix2[5], pix2[6], pix3[5], pix3[6]));
411 13401256 s += abs(pix1[6] - avg4(pix2[6], pix2[7], pix3[6], pix3[7]));
412 13401256 s += abs(pix1[7] - avg4(pix2[7], pix2[8], pix3[7], pix3[8]));
413 13401256 pix1 += stride;
414 13401256 pix2 += stride;
415 13401256 pix3 += stride;
416 }
417 1675146 return s;
418 }
419
420 748189 static int nsse16_c(MpegEncContext *c, const uint8_t *s1, const uint8_t *s2,
421 ptrdiff_t stride, int h)
422 {
423 748189 int score1 = 0, score2 = 0, x, y;
424
425
2/2
✓ Branch 0 taken 11970658 times.
✓ Branch 1 taken 748189 times.
12718847 for (y = 0; y < h; y++) {
426
2/2
✓ Branch 0 taken 191530528 times.
✓ Branch 1 taken 11970658 times.
203501186 for (x = 0; x < 16; x++)
427 191530528 score1 += (s1[x] - s2[x]) * (s1[x] - s2[x]);
428
2/2
✓ Branch 0 taken 11222469 times.
✓ Branch 1 taken 748189 times.
11970658 if (y + 1 < h) {
429
2/2
✓ Branch 0 taken 168337035 times.
✓ Branch 1 taken 11222469 times.
179559504 for (x = 0; x < 15; x++)
430 168337035 score2 += FFABS(s1[x] - s1[x + stride] -
431 168337035 s1[x + 1] + s1[x + stride + 1]) -
432 168337035 FFABS(s2[x] - s2[x + stride] -
433 s2[x + 1] + s2[x + stride + 1]);
434 }
435 11970658 s1 += stride;
436 11970658 s2 += stride;
437 }
438
439
2/2
✓ Branch 0 taken 748141 times.
✓ Branch 1 taken 48 times.
748189 if (c)
440 748141 return score1 + FFABS(score2) * c->avctx->nsse_weight;
441 else
442 48 return score1 + FFABS(score2) * 8;
443 }
444
445 48 static int nsse8_c(MpegEncContext *c, const uint8_t *s1, const uint8_t *s2,
446 ptrdiff_t stride, int h)
447 {
448 48 int score1 = 0, score2 = 0, x, y;
449
450
2/2
✓ Branch 0 taken 446 times.
✓ Branch 1 taken 48 times.
494 for (y = 0; y < h; y++) {
451
2/2
✓ Branch 0 taken 3568 times.
✓ Branch 1 taken 446 times.
4014 for (x = 0; x < 8; x++)
452 3568 score1 += (s1[x] - s2[x]) * (s1[x] - s2[x]);
453
2/2
✓ Branch 0 taken 398 times.
✓ Branch 1 taken 48 times.
446 if (y + 1 < h) {
454
2/2
✓ Branch 0 taken 2786 times.
✓ Branch 1 taken 398 times.
3184 for (x = 0; x < 7; x++)
455 2786 score2 += FFABS(s1[x] - s1[x + stride] -
456 2786 s1[x + 1] + s1[x + stride + 1]) -
457 2786 FFABS(s2[x] - s2[x + stride] -
458 s2[x + 1] + s2[x + stride + 1]);
459 }
460 446 s1 += stride;
461 446 s2 += stride;
462 }
463
464
1/2
✗ Branch 0 not taken.
✓ Branch 1 taken 48 times.
48 if (c)
465 return score1 + FFABS(score2) * c->avctx->nsse_weight;
466 else
467 48 return score1 + FFABS(score2) * 8;
468 }
469
470 static int zero_cmp(MpegEncContext *s, const uint8_t *a, const uint8_t *b,
471 ptrdiff_t stride, int h)
472 {
473 return 0;
474 }
475
476 45839 int ff_set_cmp(MECmpContext *c, me_cmp_func *cmp, int type)
477 {
478 45839 int ret = 0;
479 int i;
480
481 45839 memset(cmp, 0, sizeof(void *) * 6);
482
483
2/2
✓ Branch 0 taken 275034 times.
✓ Branch 1 taken 45839 times.
320873 for (i = 0; i < 6; i++) {
484
7/17
✓ Branch 0 taken 254424 times.
✗ Branch 1 not taken.
✓ Branch 2 taken 7200 times.
✓ Branch 3 taken 2748 times.
✗ Branch 4 not taken.
✗ Branch 5 not taken.
✓ Branch 6 taken 1182 times.
✗ Branch 7 not taken.
✗ Branch 8 not taken.
✗ Branch 9 not taken.
✓ Branch 10 taken 1344 times.
✗ Branch 11 not taken.
✗ Branch 12 not taken.
✓ Branch 13 taken 4824 times.
✗ Branch 14 not taken.
✓ Branch 15 taken 3312 times.
✗ Branch 16 not taken.
275034 switch (type & 0xFF) {
485 254424 case FF_CMP_SAD:
486 254424 cmp[i] = c->sad[i];
487 254424 break;
488 case FF_CMP_MEDIAN_SAD:
489 cmp[i] = c->median_sad[i];
490 break;
491 7200 case FF_CMP_SATD:
492 7200 cmp[i] = c->hadamard8_diff[i];
493 7200 break;
494 2748 case FF_CMP_SSE:
495 2748 cmp[i] = c->sse[i];
496 2748 break;
497 case FF_CMP_DCT:
498 cmp[i] = c->dct_sad[i];
499 break;
500 case FF_CMP_DCT264:
501 cmp[i] = c->dct264_sad[i];
502 break;
503 1182 case FF_CMP_DCTMAX:
504 1182 cmp[i] = c->dct_max[i];
505 1182 break;
506 case FF_CMP_PSNR:
507 cmp[i] = c->quant_psnr[i];
508 break;
509 case FF_CMP_BIT:
510 cmp[i] = c->bit[i];
511 break;
512 case FF_CMP_RD:
513 cmp[i] = c->rd[i];
514 break;
515 1344 case FF_CMP_VSAD:
516 1344 cmp[i] = c->vsad[i];
517 1344 break;
518 case FF_CMP_VSSE:
519 cmp[i] = c->vsse[i];
520 break;
521 case FF_CMP_ZERO:
522 cmp[i] = zero_cmp;
523 break;
524 4824 case FF_CMP_NSSE:
525 4824 cmp[i] = c->nsse[i];
526 4824 break;
527 #if CONFIG_DWT
528 case FF_CMP_W53:
529 cmp[i]= c->w53[i];
530 break;
531 3312 case FF_CMP_W97:
532 3312 cmp[i]= c->w97[i];
533 3312 break;
534 #endif
535 default:
536 av_log(NULL, AV_LOG_ERROR,
537 "invalid cmp function selection\n");
538 ret = -1;
539 break;
540 }
541 }
542
543 45839 return ret;
544 }
545
546 #define BUTTERFLY2(o1, o2, i1, i2) \
547 o1 = (i1) + (i2); \
548 o2 = (i1) - (i2);
549
550 #define BUTTERFLY1(x, y) \
551 { \
552 int a, b; \
553 a = x; \
554 b = y; \
555 x = a + b; \
556 y = a - b; \
557 }
558
559 #define BUTTERFLYA(x, y) (FFABS((x) + (y)) + FFABS((x) - (y)))
560
561 15481146 static int hadamard8_diff8x8_c(MpegEncContext *s, const uint8_t *dst,
562 const uint8_t *src, ptrdiff_t stride, int h)
563 {
564 15481146 int i, temp[64], sum = 0;
565
566
2/2
✓ Branch 0 taken 123849168 times.
✓ Branch 1 taken 15481146 times.
139330314 for (i = 0; i < 8; i++) {
567 // FIXME: try pointer walks
568 123849168 BUTTERFLY2(temp[8 * i + 0], temp[8 * i + 1],
569 src[stride * i + 0] - dst[stride * i + 0],
570 src[stride * i + 1] - dst[stride * i + 1]);
571 123849168 BUTTERFLY2(temp[8 * i + 2], temp[8 * i + 3],
572 src[stride * i + 2] - dst[stride * i + 2],
573 src[stride * i + 3] - dst[stride * i + 3]);
574 123849168 BUTTERFLY2(temp[8 * i + 4], temp[8 * i + 5],
575 src[stride * i + 4] - dst[stride * i + 4],
576 src[stride * i + 5] - dst[stride * i + 5]);
577 123849168 BUTTERFLY2(temp[8 * i + 6], temp[8 * i + 7],
578 src[stride * i + 6] - dst[stride * i + 6],
579 src[stride * i + 7] - dst[stride * i + 7]);
580
581 123849168 BUTTERFLY1(temp[8 * i + 0], temp[8 * i + 2]);
582 123849168 BUTTERFLY1(temp[8 * i + 1], temp[8 * i + 3]);
583 123849168 BUTTERFLY1(temp[8 * i + 4], temp[8 * i + 6]);
584 123849168 BUTTERFLY1(temp[8 * i + 5], temp[8 * i + 7]);
585
586 123849168 BUTTERFLY1(temp[8 * i + 0], temp[8 * i + 4]);
587 123849168 BUTTERFLY1(temp[8 * i + 1], temp[8 * i + 5]);
588 123849168 BUTTERFLY1(temp[8 * i + 2], temp[8 * i + 6]);
589 123849168 BUTTERFLY1(temp[8 * i + 3], temp[8 * i + 7]);
590 }
591
592
2/2
✓ Branch 0 taken 123849168 times.
✓ Branch 1 taken 15481146 times.
139330314 for (i = 0; i < 8; i++) {
593 123849168 BUTTERFLY1(temp[8 * 0 + i], temp[8 * 1 + i]);
594 123849168 BUTTERFLY1(temp[8 * 2 + i], temp[8 * 3 + i]);
595 123849168 BUTTERFLY1(temp[8 * 4 + i], temp[8 * 5 + i]);
596 123849168 BUTTERFLY1(temp[8 * 6 + i], temp[8 * 7 + i]);
597
598 123849168 BUTTERFLY1(temp[8 * 0 + i], temp[8 * 2 + i]);
599 123849168 BUTTERFLY1(temp[8 * 1 + i], temp[8 * 3 + i]);
600 123849168 BUTTERFLY1(temp[8 * 4 + i], temp[8 * 6 + i]);
601 123849168 BUTTERFLY1(temp[8 * 5 + i], temp[8 * 7 + i]);
602
603 123849168 sum += BUTTERFLYA(temp[8 * 0 + i], temp[8 * 4 + i]) +
604 123849168 BUTTERFLYA(temp[8 * 1 + i], temp[8 * 5 + i]) +
605 123849168 BUTTERFLYA(temp[8 * 2 + i], temp[8 * 6 + i]) +
606 123849168 BUTTERFLYA(temp[8 * 3 + i], temp[8 * 7 + i]);
607 }
608 15481146 return sum;
609 }
610
611 104 static int hadamard8_intra8x8_c(MpegEncContext *s, const uint8_t *src,
612 const uint8_t *dummy, ptrdiff_t stride, int h)
613 {
614 104 int i, temp[64], sum = 0;
615
616
2/2
✓ Branch 0 taken 832 times.
✓ Branch 1 taken 104 times.
936 for (i = 0; i < 8; i++) {
617 // FIXME: try pointer walks
618 832 BUTTERFLY2(temp[8 * i + 0], temp[8 * i + 1],
619 src[stride * i + 0], src[stride * i + 1]);
620 832 BUTTERFLY2(temp[8 * i + 2], temp[8 * i + 3],
621 src[stride * i + 2], src[stride * i + 3]);
622 832 BUTTERFLY2(temp[8 * i + 4], temp[8 * i + 5],
623 src[stride * i + 4], src[stride * i + 5]);
624 832 BUTTERFLY2(temp[8 * i + 6], temp[8 * i + 7],
625 src[stride * i + 6], src[stride * i + 7]);
626
627 832 BUTTERFLY1(temp[8 * i + 0], temp[8 * i + 2]);
628 832 BUTTERFLY1(temp[8 * i + 1], temp[8 * i + 3]);
629 832 BUTTERFLY1(temp[8 * i + 4], temp[8 * i + 6]);
630 832 BUTTERFLY1(temp[8 * i + 5], temp[8 * i + 7]);
631
632 832 BUTTERFLY1(temp[8 * i + 0], temp[8 * i + 4]);
633 832 BUTTERFLY1(temp[8 * i + 1], temp[8 * i + 5]);
634 832 BUTTERFLY1(temp[8 * i + 2], temp[8 * i + 6]);
635 832 BUTTERFLY1(temp[8 * i + 3], temp[8 * i + 7]);
636 }
637
638
2/2
✓ Branch 0 taken 832 times.
✓ Branch 1 taken 104 times.
936 for (i = 0; i < 8; i++) {
639 832 BUTTERFLY1(temp[8 * 0 + i], temp[8 * 1 + i]);
640 832 BUTTERFLY1(temp[8 * 2 + i], temp[8 * 3 + i]);
641 832 BUTTERFLY1(temp[8 * 4 + i], temp[8 * 5 + i]);
642 832 BUTTERFLY1(temp[8 * 6 + i], temp[8 * 7 + i]);
643
644 832 BUTTERFLY1(temp[8 * 0 + i], temp[8 * 2 + i]);
645 832 BUTTERFLY1(temp[8 * 1 + i], temp[8 * 3 + i]);
646 832 BUTTERFLY1(temp[8 * 4 + i], temp[8 * 6 + i]);
647 832 BUTTERFLY1(temp[8 * 5 + i], temp[8 * 7 + i]);
648
649 832 sum +=
650 832 BUTTERFLYA(temp[8 * 0 + i], temp[8 * 4 + i])
651 832 + BUTTERFLYA(temp[8 * 1 + i], temp[8 * 5 + i])
652 832 + BUTTERFLYA(temp[8 * 2 + i], temp[8 * 6 + i])
653 832 + BUTTERFLYA(temp[8 * 3 + i], temp[8 * 7 + i]);
654 }
655
656 104 sum -= FFABS(temp[8 * 0] + temp[8 * 4]); // -mean
657
658 104 return sum;
659 }
660
661 static int dct_sad8x8_c(MpegEncContext *s, const uint8_t *src1,
662 const uint8_t *src2, ptrdiff_t stride, int h)
663 {
664 LOCAL_ALIGNED_16(int16_t, temp, [64]);
665
666 s->pdsp.diff_pixels_unaligned(temp, src1, src2, stride);
667 s->fdsp.fdct(temp);
668 return s->mecc.sum_abs_dctelem(temp);
669 }
670
671 #if CONFIG_GPL
672 #define DCT8_1D \
673 { \
674 const int s07 = SRC(0) + SRC(7); \
675 const int s16 = SRC(1) + SRC(6); \
676 const int s25 = SRC(2) + SRC(5); \
677 const int s34 = SRC(3) + SRC(4); \
678 const int a0 = s07 + s34; \
679 const int a1 = s16 + s25; \
680 const int a2 = s07 - s34; \
681 const int a3 = s16 - s25; \
682 const int d07 = SRC(0) - SRC(7); \
683 const int d16 = SRC(1) - SRC(6); \
684 const int d25 = SRC(2) - SRC(5); \
685 const int d34 = SRC(3) - SRC(4); \
686 const int a4 = d16 + d25 + (d07 + (d07 >> 1)); \
687 const int a5 = d07 - d34 - (d25 + (d25 >> 1)); \
688 const int a6 = d07 + d34 - (d16 + (d16 >> 1)); \
689 const int a7 = d16 - d25 + (d34 + (d34 >> 1)); \
690 DST(0, a0 + a1); \
691 DST(1, a4 + (a7 >> 2)); \
692 DST(2, a2 + (a3 >> 1)); \
693 DST(3, a5 + (a6 >> 2)); \
694 DST(4, a0 - a1); \
695 DST(5, a6 - (a5 >> 2)); \
696 DST(6, (a2 >> 1) - a3); \
697 DST(7, (a4 >> 2) - a7); \
698 }
699
700 static int dct264_sad8x8_c(MpegEncContext *s, const uint8_t *src1,
701 const uint8_t *src2, ptrdiff_t stride, int h)
702 {
703 int16_t dct[8][8];
704 int i, sum = 0;
705
706 s->pdsp.diff_pixels_unaligned(dct[0], src1, src2, stride);
707
708 #define SRC(x) dct[i][x]
709 #define DST(x, v) dct[i][x] = v
710 for (i = 0; i < 8; i++)
711 DCT8_1D
712 #undef SRC
713 #undef DST
714
715 #define SRC(x) dct[x][i]
716 #define DST(x, v) sum += FFABS(v)
717 for (i = 0; i < 8; i++)
718 DCT8_1D
719 #undef SRC
720 #undef DST
721 return sum;
722 }
723 #endif
724
725 static int dct_max8x8_c(MpegEncContext *s, const uint8_t *src1,
726 const uint8_t *src2, ptrdiff_t stride, int h)
727 {
728 LOCAL_ALIGNED_16(int16_t, temp, [64]);
729 int sum = 0, i;
730
731 s->pdsp.diff_pixels_unaligned(temp, src1, src2, stride);
732 s->fdsp.fdct(temp);
733
734 for (i = 0; i < 64; i++)
735 sum = FFMAX(sum, FFABS(temp[i]));
736
737 return sum;
738 }
739
740 static int quant_psnr8x8_c(MpegEncContext *s, const uint8_t *src1,
741 const uint8_t *src2, ptrdiff_t stride, int h)
742 {
743 LOCAL_ALIGNED_16(int16_t, temp, [64 * 2]);
744 int16_t *const bak = temp + 64;
745 int sum = 0, i;
746
747 s->mb_intra = 0;
748
749 s->pdsp.diff_pixels_unaligned(temp, src1, src2, stride);
750
751 memcpy(bak, temp, 64 * sizeof(int16_t));
752
753 s->block_last_index[0 /* FIXME */] =
754 s->fast_dct_quantize(s, temp, 0 /* FIXME */, s->qscale, &i);
755 s->dct_unquantize_inter(s, temp, 0, s->qscale);
756 ff_simple_idct_int16_8bit(temp); // FIXME
757
758 for (i = 0; i < 64; i++)
759 sum += (temp[i] - bak[i]) * (temp[i] - bak[i]);
760
761 return sum;
762 }
763
764 static int rd8x8_c(MpegEncContext *s, const uint8_t *src1, const uint8_t *src2,
765 ptrdiff_t stride, int h)
766 {
767 const uint8_t *scantable = s->intra_scantable.permutated;
768 LOCAL_ALIGNED_16(int16_t, temp, [64]);
769 LOCAL_ALIGNED_16(uint8_t, lsrc1, [64]);
770 LOCAL_ALIGNED_16(uint8_t, lsrc2, [64]);
771 int i, last, run, bits, level, distortion, start_i;
772 const int esc_length = s->ac_esc_length;
773 uint8_t *length, *last_length;
774
775 copy_block8(lsrc1, src1, 8, stride, 8);
776 copy_block8(lsrc2, src2, 8, stride, 8);
777
778 s->pdsp.diff_pixels(temp, lsrc1, lsrc2, 8);
779
780 s->block_last_index[0 /* FIXME */] =
781 last =
782 s->fast_dct_quantize(s, temp, 0 /* FIXME */, s->qscale, &i);
783
784 bits = 0;
785
786 if (s->mb_intra) {
787 start_i = 1;
788 length = s->intra_ac_vlc_length;
789 last_length = s->intra_ac_vlc_last_length;
790 bits += s->luma_dc_vlc_length[temp[0] + 256]; // FIXME: chroma
791 } else {
792 start_i = 0;
793 length = s->inter_ac_vlc_length;
794 last_length = s->inter_ac_vlc_last_length;
795 }
796
797 if (last >= start_i) {
798 run = 0;
799 for (i = start_i; i < last; i++) {
800 int j = scantable[i];
801 level = temp[j];
802
803 if (level) {
804 level += 64;
805 if ((level & (~127)) == 0)
806 bits += length[UNI_AC_ENC_INDEX(run, level)];
807 else
808 bits += esc_length;
809 run = 0;
810 } else
811 run++;
812 }
813 i = scantable[last];
814
815 level = temp[i] + 64;
816
817 av_assert2(level - 64);
818
819 if ((level & (~127)) == 0) {
820 bits += last_length[UNI_AC_ENC_INDEX(run, level)];
821 } else
822 bits += esc_length;
823 }
824
825 if (last >= 0) {
826 if (s->mb_intra)
827 s->dct_unquantize_intra(s, temp, 0, s->qscale);
828 else
829 s->dct_unquantize_inter(s, temp, 0, s->qscale);
830 }
831
832 s->idsp.idct_add(lsrc2, 8, temp);
833
834 distortion = s->mecc.sse[1](NULL, lsrc2, lsrc1, 8, 8);
835
836 return distortion + ((bits * s->qscale * s->qscale * 109 + 64) >> 7);
837 }
838
839 static int bit8x8_c(MpegEncContext *s, const uint8_t *src1, const uint8_t *src2,
840 ptrdiff_t stride, int h)
841 {
842 const uint8_t *scantable = s->intra_scantable.permutated;
843 LOCAL_ALIGNED_16(int16_t, temp, [64]);
844 int i, last, run, bits, level, start_i;
845 const int esc_length = s->ac_esc_length;
846 uint8_t *length, *last_length;
847
848 s->pdsp.diff_pixels_unaligned(temp, src1, src2, stride);
849
850 s->block_last_index[0 /* FIXME */] =
851 last =
852 s->fast_dct_quantize(s, temp, 0 /* FIXME */, s->qscale, &i);
853
854 bits = 0;
855
856 if (s->mb_intra) {
857 start_i = 1;
858 length = s->intra_ac_vlc_length;
859 last_length = s->intra_ac_vlc_last_length;
860 bits += s->luma_dc_vlc_length[temp[0] + 256]; // FIXME: chroma
861 } else {
862 start_i = 0;
863 length = s->inter_ac_vlc_length;
864 last_length = s->inter_ac_vlc_last_length;
865 }
866
867 if (last >= start_i) {
868 run = 0;
869 for (i = start_i; i < last; i++) {
870 int j = scantable[i];
871 level = temp[j];
872
873 if (level) {
874 level += 64;
875 if ((level & (~127)) == 0)
876 bits += length[UNI_AC_ENC_INDEX(run, level)];
877 else
878 bits += esc_length;
879 run = 0;
880 } else
881 run++;
882 }
883 i = scantable[last];
884
885 level = temp[i] + 64;
886
887 av_assert2(level - 64);
888
889 if ((level & (~127)) == 0)
890 bits += last_length[UNI_AC_ENC_INDEX(run, level)];
891 else
892 bits += esc_length;
893 }
894
895 return bits;
896 }
897
898 #define VSAD_INTRA(size) \
899 static int vsad_intra ## size ## _c(MpegEncContext *c, \
900 const uint8_t *s, const uint8_t *dummy, \
901 ptrdiff_t stride, int h) \
902 { \
903 int score = 0, x, y; \
904 \
905 for (y = 1; y < h; y++) { \
906 for (x = 0; x < size; x += 4) { \
907 score += FFABS(s[x] - s[x + stride]) + \
908 FFABS(s[x + 1] - s[x + stride + 1]) + \
909 FFABS(s[x + 2] - s[x + 2 + stride]) + \
910 FFABS(s[x + 3] - s[x + 3 + stride]); \
911 } \
912 s += stride; \
913 } \
914 \
915 return score; \
916 }
917
4/4
✓ Branch 0 taken 700 times.
✓ Branch 1 taken 350 times.
✓ Branch 2 taken 350 times.
✓ Branch 3 taken 48 times.
1098 VSAD_INTRA(8)
918
4/4
✓ Branch 0 taken 34475192 times.
✓ Branch 1 taken 8618798 times.
✓ Branch 2 taken 8618798 times.
✓ Branch 3 taken 1231246 times.
44325236 VSAD_INTRA(16)
919
920 #define VSAD(size) \
921 static int vsad ## size ## _c(MpegEncContext *c, \
922 const uint8_t *s1, const uint8_t *s2, \
923 ptrdiff_t stride, int h) \
924 { \
925 int score = 0, x, y; \
926 \
927 for (y = 1; y < h; y++) { \
928 for (x = 0; x < size; x++) \
929 score += FFABS(s1[x] - s2[x] - s1[x + stride] + s2[x + stride]); \
930 s1 += stride; \
931 s2 += stride; \
932 } \
933 \
934 return score; \
935 }
936
4/4
✓ Branch 0 taken 1952 times.
✓ Branch 1 taken 244 times.
✓ Branch 2 taken 244 times.
✓ Branch 3 taken 32 times.
2228 VSAD(8)
937
4/4
✓ Branch 0 taken 213086048 times.
✓ Branch 1 taken 13317878 times.
✓ Branch 2 taken 13317878 times.
✓ Branch 3 taken 1902546 times.
228306472 VSAD(16)
938
939 #define SQ(a) ((a) * (a))
940 #define VSSE_INTRA(size) \
941 static int vsse_intra ## size ## _c(MpegEncContext *c, \
942 const uint8_t *s, const uint8_t *dummy, \
943 ptrdiff_t stride, int h) \
944 { \
945 int score = 0, x, y; \
946 \
947 for (y = 1; y < h; y++) { \
948 for (x = 0; x < size; x += 4) { \
949 score += SQ(s[x] - s[x + stride]) + \
950 SQ(s[x + 1] - s[x + stride + 1]) + \
951 SQ(s[x + 2] - s[x + stride + 2]) + \
952 SQ(s[x + 3] - s[x + stride + 3]); \
953 } \
954 s += stride; \
955 } \
956 \
957 return score; \
958 }
959
4/4
✓ Branch 0 taken 528 times.
✓ Branch 1 taken 264 times.
✓ Branch 2 taken 264 times.
✓ Branch 3 taken 32 times.
824 VSSE_INTRA(8)
960
4/4
✓ Branch 0 taken 1008 times.
✓ Branch 1 taken 252 times.
✓ Branch 2 taken 252 times.
✓ Branch 3 taken 32 times.
1292 VSSE_INTRA(16)
961
962 #define VSSE(size) \
963 static int vsse ## size ## _c(MpegEncContext *c, const uint8_t *s1, const uint8_t *s2, \
964 ptrdiff_t stride, int h) \
965 { \
966 int score = 0, x, y; \
967 \
968 for (y = 1; y < h; y++) { \
969 for (x = 0; x < size; x++) \
970 score += SQ(s1[x] - s2[x] - s1[x + stride] + s2[x + stride]); \
971 s1 += stride; \
972 s2 += stride; \
973 } \
974 \
975 return score; \
976 }
977
4/4
✓ Branch 0 taken 2176 times.
✓ Branch 1 taken 272 times.
✓ Branch 2 taken 272 times.
✓ Branch 3 taken 32 times.
2480 VSSE(8)
978
4/4
✓ Branch 0 taken 4800 times.
✓ Branch 1 taken 300 times.
✓ Branch 2 taken 300 times.
✓ Branch 3 taken 32 times.
5132 VSSE(16)
979
980 #define WRAPPER8_16_SQ(name8, name16) \
981 static int name16(MpegEncContext *s, const uint8_t *dst, const uint8_t *src, \
982 ptrdiff_t stride, int h) \
983 { \
984 int score = 0; \
985 \
986 score += name8(s, dst, src, stride, 8); \
987 score += name8(s, dst + 8, src + 8, stride, 8); \
988 if (h == 16) { \
989 dst += 8 * stride; \
990 src += 8 * stride; \
991 score += name8(s, dst, src, stride, 8); \
992 score += name8(s, dst + 8, src + 8, stride, 8); \
993 } \
994 return score; \
995 }
996
997
2/2
✓ Branch 2 taken 3613249 times.
✓ Branch 3 taken 35 times.
3613284 WRAPPER8_16_SQ(hadamard8_diff8x8_c, hadamard8_diff16_c)
998
2/2
✓ Branch 2 taken 4 times.
✓ Branch 3 taken 28 times.
32 WRAPPER8_16_SQ(hadamard8_intra8x8_c, hadamard8_intra16_c)
999 WRAPPER8_16_SQ(dct_sad8x8_c, dct_sad16_c)
1000 #if CONFIG_GPL
1001 WRAPPER8_16_SQ(dct264_sad8x8_c, dct264_sad16_c)
1002 #endif
1003 WRAPPER8_16_SQ(dct_max8x8_c, dct_max16_c)
1004 WRAPPER8_16_SQ(quant_psnr8x8_c, quant_psnr16_c)
1005 WRAPPER8_16_SQ(rd8x8_c, rd16_c)
1006 WRAPPER8_16_SQ(bit8x8_c, bit16_c)
1007
1008 1073 av_cold void ff_me_cmp_init(MECmpContext *c, AVCodecContext *avctx)
1009 {
1010 1073 c->sum_abs_dctelem = sum_abs_dctelem_c;
1011
1012 /* TODO [0] 16 [1] 8 */
1013 1073 c->pix_abs[0][0] = pix_abs16_c;
1014 1073 c->pix_abs[0][1] = pix_abs16_x2_c;
1015 1073 c->pix_abs[0][2] = pix_abs16_y2_c;
1016 1073 c->pix_abs[0][3] = pix_abs16_xy2_c;
1017 1073 c->pix_abs[1][0] = pix_abs8_c;
1018 1073 c->pix_abs[1][1] = pix_abs8_x2_c;
1019 1073 c->pix_abs[1][2] = pix_abs8_y2_c;
1020 1073 c->pix_abs[1][3] = pix_abs8_xy2_c;
1021
1022 #define SET_CMP_FUNC(name) \
1023 c->name[0] = name ## 16_c; \
1024 c->name[1] = name ## 8x8_c;
1025
1026 1073 SET_CMP_FUNC(hadamard8_diff)
1027 1073 c->hadamard8_diff[4] = hadamard8_intra16_c;
1028 1073 c->hadamard8_diff[5] = hadamard8_intra8x8_c;
1029 1073 SET_CMP_FUNC(dct_sad)
1030 1073 SET_CMP_FUNC(dct_max)
1031 #if CONFIG_GPL
1032 1073 SET_CMP_FUNC(dct264_sad)
1033 #endif
1034 1073 c->sad[0] = pix_abs16_c;
1035 1073 c->sad[1] = pix_abs8_c;
1036 1073 c->sse[0] = sse16_c;
1037 1073 c->sse[1] = sse8_c;
1038 1073 c->sse[2] = sse4_c;
1039 1073 SET_CMP_FUNC(quant_psnr)
1040 1073 SET_CMP_FUNC(rd)
1041 1073 SET_CMP_FUNC(bit)
1042 1073 c->vsad[0] = vsad16_c;
1043 1073 c->vsad[1] = vsad8_c;
1044 1073 c->vsad[4] = vsad_intra16_c;
1045 1073 c->vsad[5] = vsad_intra8_c;
1046 1073 c->vsse[0] = vsse16_c;
1047 1073 c->vsse[1] = vsse8_c;
1048 1073 c->vsse[4] = vsse_intra16_c;
1049 1073 c->vsse[5] = vsse_intra8_c;
1050 1073 c->nsse[0] = nsse16_c;
1051 1073 c->nsse[1] = nsse8_c;
1052 #if CONFIG_SNOW_DECODER || CONFIG_SNOW_ENCODER
1053 1073 ff_dsputil_init_dwt(c);
1054 #endif
1055
1056 1073 c->median_sad[0] = pix_median_abs16_c;
1057 1073 c->median_sad[1] = pix_median_abs8_c;
1058
1059 #if ARCH_AARCH64
1060 ff_me_cmp_init_aarch64(c, avctx);
1061 #elif ARCH_ALPHA
1062 ff_me_cmp_init_alpha(c, avctx);
1063 #elif ARCH_ARM
1064 ff_me_cmp_init_arm(c, avctx);
1065 #elif ARCH_PPC
1066 ff_me_cmp_init_ppc(c, avctx);
1067 #elif ARCH_RISCV
1068 ff_me_cmp_init_riscv(c, avctx);
1069 #elif ARCH_X86
1070 1073 ff_me_cmp_init_x86(c, avctx);
1071 #elif ARCH_MIPS
1072 ff_me_cmp_init_mips(c, avctx);
1073 #endif
1074
1075 1073 }
1076