FFmpeg coverage


Directory: ../../../ffmpeg/
File: src/libavcodec/me_cmp.c
Date: 2022-11-26 13:19:19
Exec Total Coverage
Lines: 411 567 72.5%
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 336 times.
✓ Branch 1 taken 32 times.
368 for (i = 0; i < h; i++) {
77 336 s += sq[pix1[0] - pix2[0]];
78 336 s += sq[pix1[1] - pix2[1]];
79 336 s += sq[pix1[2] - pix2[2]];
80 336 s += sq[pix1[3] - pix2[3]];
81 336 pix1 += stride;
82 336 pix2 += stride;
83 }
84 32 return s;
85 }
86
87 4810891 static int sse8_c(MpegEncContext *v, const uint8_t *pix1, const uint8_t *pix2,
88 ptrdiff_t stride, int h)
89 {
90 4810891 int s = 0, i;
91 4810891 const uint32_t *sq = ff_square_tab + 256;
92
93
2/2
✓ Branch 0 taken 44628578 times.
✓ Branch 1 taken 4810891 times.
49439469 for (i = 0; i < h; i++) {
94 44628578 s += sq[pix1[0] - pix2[0]];
95 44628578 s += sq[pix1[1] - pix2[1]];
96 44628578 s += sq[pix1[2] - pix2[2]];
97 44628578 s += sq[pix1[3] - pix2[3]];
98 44628578 s += sq[pix1[4] - pix2[4]];
99 44628578 s += sq[pix1[5] - pix2[5]];
100 44628578 s += sq[pix1[6] - pix2[6]];
101 44628578 s += sq[pix1[7] - pix2[7]];
102 44628578 pix1 += stride;
103 44628578 pix2 += stride;
104 }
105 4810891 return s;
106 }
107
108 7186385 static int sse16_c(MpegEncContext *v, const uint8_t *pix1, const uint8_t *pix2,
109 ptrdiff_t stride, int h)
110 {
111 7186385 int s = 0, i;
112 7186385 const uint32_t *sq = ff_square_tab + 256;
113
114
2/2
✓ Branch 0 taken 114981868 times.
✓ Branch 1 taken 7186385 times.
122168253 for (i = 0; i < h; i++) {
115 114981868 s += sq[pix1[0] - pix2[0]];
116 114981868 s += sq[pix1[1] - pix2[1]];
117 114981868 s += sq[pix1[2] - pix2[2]];
118 114981868 s += sq[pix1[3] - pix2[3]];
119 114981868 s += sq[pix1[4] - pix2[4]];
120 114981868 s += sq[pix1[5] - pix2[5]];
121 114981868 s += sq[pix1[6] - pix2[6]];
122 114981868 s += sq[pix1[7] - pix2[7]];
123 114981868 s += sq[pix1[8] - pix2[8]];
124 114981868 s += sq[pix1[9] - pix2[9]];
125 114981868 s += sq[pix1[10] - pix2[10]];
126 114981868 s += sq[pix1[11] - pix2[11]];
127 114981868 s += sq[pix1[12] - pix2[12]];
128 114981868 s += sq[pix1[13] - pix2[13]];
129 114981868 s += sq[pix1[14] - pix2[14]];
130 114981868 s += sq[pix1[15] - pix2[15]];
131
132 114981868 pix1 += stride;
133 114981868 pix2 += stride;
134 }
135 7186385 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 39571611 static inline int pix_abs16_c(MpegEncContext *v, const uint8_t *pix1, const uint8_t *pix2,
151 ptrdiff_t stride, int h)
152 {
153 39571611 int s = 0, i;
154
155
2/2
✓ Branch 0 taken 552776772 times.
✓ Branch 1 taken 39571611 times.
592348383 for (i = 0; i < h; i++) {
156 552776772 s += abs(pix1[0] - pix2[0]);
157 552776772 s += abs(pix1[1] - pix2[1]);
158 552776772 s += abs(pix1[2] - pix2[2]);
159 552776772 s += abs(pix1[3] - pix2[3]);
160 552776772 s += abs(pix1[4] - pix2[4]);
161 552776772 s += abs(pix1[5] - pix2[5]);
162 552776772 s += abs(pix1[6] - pix2[6]);
163 552776772 s += abs(pix1[7] - pix2[7]);
164 552776772 s += abs(pix1[8] - pix2[8]);
165 552776772 s += abs(pix1[9] - pix2[9]);
166 552776772 s += abs(pix1[10] - pix2[10]);
167 552776772 s += abs(pix1[11] - pix2[11]);
168 552776772 s += abs(pix1[12] - pix2[12]);
169 552776772 s += abs(pix1[13] - pix2[13]);
170 552776772 s += abs(pix1[14] - pix2[14]);
171 552776772 s += abs(pix1[15] - pix2[15]);
172 552776772 pix1 += stride;
173 552776772 pix2 += stride;
174 }
175 39571611 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 304 times.
✓ Branch 1 taken 32 times.
336 for (i = 1; i < h; i++) {
206 304 s += abs(V(0) - V(-stride));
207
2/2
✓ Branch 0 taken 4560 times.
✓ Branch 1 taken 304 times.
4864 for (j = 1; j < 16; j++)
208 4560 s += abs(V(j) - mid_pred(V(j-stride), V(j-1), V(j-stride) + V(j-1) - V(j-stride-1)));
209 304 pix1 += stride;
210 304 pix2 += stride;
211
212 }
213 #undef V
214 32 return s;
215 }
216
217 3118769 static int pix_abs16_x2_c(MpegEncContext *v, const uint8_t *pix1, const uint8_t *pix2,
218 ptrdiff_t stride, int h)
219 {
220 3118769 int s = 0, i;
221
222
2/2
✓ Branch 0 taken 40590458 times.
✓ Branch 1 taken 3118769 times.
43709227 for (i = 0; i < h; i++) {
223 40590458 s += abs(pix1[0] - avg2(pix2[0], pix2[1]));
224 40590458 s += abs(pix1[1] - avg2(pix2[1], pix2[2]));
225 40590458 s += abs(pix1[2] - avg2(pix2[2], pix2[3]));
226 40590458 s += abs(pix1[3] - avg2(pix2[3], pix2[4]));
227 40590458 s += abs(pix1[4] - avg2(pix2[4], pix2[5]));
228 40590458 s += abs(pix1[5] - avg2(pix2[5], pix2[6]));
229 40590458 s += abs(pix1[6] - avg2(pix2[6], pix2[7]));
230 40590458 s += abs(pix1[7] - avg2(pix2[7], pix2[8]));
231 40590458 s += abs(pix1[8] - avg2(pix2[8], pix2[9]));
232 40590458 s += abs(pix1[9] - avg2(pix2[9], pix2[10]));
233 40590458 s += abs(pix1[10] - avg2(pix2[10], pix2[11]));
234 40590458 s += abs(pix1[11] - avg2(pix2[11], pix2[12]));
235 40590458 s += abs(pix1[12] - avg2(pix2[12], pix2[13]));
236 40590458 s += abs(pix1[13] - avg2(pix2[13], pix2[14]));
237 40590458 s += abs(pix1[14] - avg2(pix2[14], pix2[15]));
238 40590458 s += abs(pix1[15] - avg2(pix2[15], pix2[16]));
239 40590458 pix1 += stride;
240 40590458 pix2 += stride;
241 }
242 3118769 return s;
243 }
244
245 3118769 static int pix_abs16_y2_c(MpegEncContext *v, const uint8_t *pix1, const uint8_t *pix2,
246 ptrdiff_t stride, int h)
247 {
248 3118769 int s = 0, i;
249 3118769 const uint8_t *pix3 = pix2 + stride;
250
251
2/2
✓ Branch 0 taken 40590444 times.
✓ Branch 1 taken 3118769 times.
43709213 for (i = 0; i < h; i++) {
252 40590444 s += abs(pix1[0] - avg2(pix2[0], pix3[0]));
253 40590444 s += abs(pix1[1] - avg2(pix2[1], pix3[1]));
254 40590444 s += abs(pix1[2] - avg2(pix2[2], pix3[2]));
255 40590444 s += abs(pix1[3] - avg2(pix2[3], pix3[3]));
256 40590444 s += abs(pix1[4] - avg2(pix2[4], pix3[4]));
257 40590444 s += abs(pix1[5] - avg2(pix2[5], pix3[5]));
258 40590444 s += abs(pix1[6] - avg2(pix2[6], pix3[6]));
259 40590444 s += abs(pix1[7] - avg2(pix2[7], pix3[7]));
260 40590444 s += abs(pix1[8] - avg2(pix2[8], pix3[8]));
261 40590444 s += abs(pix1[9] - avg2(pix2[9], pix3[9]));
262 40590444 s += abs(pix1[10] - avg2(pix2[10], pix3[10]));
263 40590444 s += abs(pix1[11] - avg2(pix2[11], pix3[11]));
264 40590444 s += abs(pix1[12] - avg2(pix2[12], pix3[12]));
265 40590444 s += abs(pix1[13] - avg2(pix2[13], pix3[13]));
266 40590444 s += abs(pix1[14] - avg2(pix2[14], pix3[14]));
267 40590444 s += abs(pix1[15] - avg2(pix2[15], pix3[15]));
268 40590444 pix1 += stride;
269 40590444 pix2 += stride;
270 40590444 pix3 += stride;
271 }
272 3118769 return s;
273 }
274
275 6237490 static int pix_abs16_xy2_c(MpegEncContext *v, const uint8_t *pix1, const uint8_t *pix2,
276 ptrdiff_t stride, int h)
277 {
278 6237490 int s = 0, i;
279 6237490 const uint8_t *pix3 = pix2 + stride;
280
281
2/2
✓ Branch 0 taken 81180430 times.
✓ Branch 1 taken 6237490 times.
87417920 for (i = 0; i < h; i++) {
282 81180430 s += abs(pix1[0] - avg4(pix2[0], pix2[1], pix3[0], pix3[1]));
283 81180430 s += abs(pix1[1] - avg4(pix2[1], pix2[2], pix3[1], pix3[2]));
284 81180430 s += abs(pix1[2] - avg4(pix2[2], pix2[3], pix3[2], pix3[3]));
285 81180430 s += abs(pix1[3] - avg4(pix2[3], pix2[4], pix3[3], pix3[4]));
286 81180430 s += abs(pix1[4] - avg4(pix2[4], pix2[5], pix3[4], pix3[5]));
287 81180430 s += abs(pix1[5] - avg4(pix2[5], pix2[6], pix3[5], pix3[6]));
288 81180430 s += abs(pix1[6] - avg4(pix2[6], pix2[7], pix3[6], pix3[7]));
289 81180430 s += abs(pix1[7] - avg4(pix2[7], pix2[8], pix3[7], pix3[8]));
290 81180430 s += abs(pix1[8] - avg4(pix2[8], pix2[9], pix3[8], pix3[9]));
291 81180430 s += abs(pix1[9] - avg4(pix2[9], pix2[10], pix3[9], pix3[10]));
292 81180430 s += abs(pix1[10] - avg4(pix2[10], pix2[11], pix3[10], pix3[11]));
293 81180430 s += abs(pix1[11] - avg4(pix2[11], pix2[12], pix3[11], pix3[12]));
294 81180430 s += abs(pix1[12] - avg4(pix2[12], pix2[13], pix3[12], pix3[13]));
295 81180430 s += abs(pix1[13] - avg4(pix2[13], pix2[14], pix3[13], pix3[14]));
296 81180430 s += abs(pix1[14] - avg4(pix2[14], pix2[15], pix3[14], pix3[15]));
297 81180430 s += abs(pix1[15] - avg4(pix2[15], pix2[16], pix3[15], pix3[16]));
298 81180430 pix1 += stride;
299 81180430 pix2 += stride;
300 81180430 pix3 += stride;
301 }
302 6237490 return s;
303 }
304
305 27666461 static inline int pix_abs8_c(MpegEncContext *v, const uint8_t *pix1, const uint8_t *pix2,
306 ptrdiff_t stride, int h)
307 {
308 27666461 int s = 0, i;
309
310
2/2
✓ Branch 0 taken 221331830 times.
✓ Branch 1 taken 27666461 times.
248998291 for (i = 0; i < h; i++) {
311 221331830 s += abs(pix1[0] - pix2[0]);
312 221331830 s += abs(pix1[1] - pix2[1]);
313 221331830 s += abs(pix1[2] - pix2[2]);
314 221331830 s += abs(pix1[3] - pix2[3]);
315 221331830 s += abs(pix1[4] - pix2[4]);
316 221331830 s += abs(pix1[5] - pix2[5]);
317 221331830 s += abs(pix1[6] - pix2[6]);
318 221331830 s += abs(pix1[7] - pix2[7]);
319 221331830 pix1 += stride;
320 221331830 pix2 += stride;
321 }
322 27666461 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 284 times.
✓ Branch 1 taken 32 times.
316 for (i = 1; i < h; i++) {
345 284 s += abs(V(0) - V(-stride));
346
2/2
✓ Branch 0 taken 1988 times.
✓ Branch 1 taken 284 times.
2272 for (j = 1; j < 8; j++)
347 1988 s += abs(V(j) - mid_pred(V(j-stride), V(j-1), V(j-stride) + V(j-1) - V(j-stride-1)));
348 284 pix1 += stride;
349 284 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 6700848 times.
✓ Branch 1 taken 837597 times.
7538445 for (i = 0; i < h; i++) {
362 6700848 s += abs(pix1[0] - avg2(pix2[0], pix2[1]));
363 6700848 s += abs(pix1[1] - avg2(pix2[1], pix2[2]));
364 6700848 s += abs(pix1[2] - avg2(pix2[2], pix2[3]));
365 6700848 s += abs(pix1[3] - avg2(pix2[3], pix2[4]));
366 6700848 s += abs(pix1[4] - avg2(pix2[4], pix2[5]));
367 6700848 s += abs(pix1[5] - avg2(pix2[5], pix2[6]));
368 6700848 s += abs(pix1[6] - avg2(pix2[6], pix2[7]));
369 6700848 s += abs(pix1[7] - avg2(pix2[7], pix2[8]));
370 6700848 pix1 += stride;
371 6700848 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 6700880 times.
✓ Branch 1 taken 837597 times.
7538477 for (i = 0; i < h; i++) {
383 6700880 s += abs(pix1[0] - avg2(pix2[0], pix3[0]));
384 6700880 s += abs(pix1[1] - avg2(pix2[1], pix3[1]));
385 6700880 s += abs(pix1[2] - avg2(pix2[2], pix3[2]));
386 6700880 s += abs(pix1[3] - avg2(pix2[3], pix3[3]));
387 6700880 s += abs(pix1[4] - avg2(pix2[4], pix3[4]));
388 6700880 s += abs(pix1[5] - avg2(pix2[5], pix3[5]));
389 6700880 s += abs(pix1[6] - avg2(pix2[6], pix3[6]));
390 6700880 s += abs(pix1[7] - avg2(pix2[7], pix3[7]));
391 6700880 pix1 += stride;
392 6700880 pix2 += stride;
393 6700880 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 13401276 times.
✓ Branch 1 taken 1675146 times.
15076422 for (i = 0; i < h; i++) {
405 13401276 s += abs(pix1[0] - avg4(pix2[0], pix2[1], pix3[0], pix3[1]));
406 13401276 s += abs(pix1[1] - avg4(pix2[1], pix2[2], pix3[1], pix3[2]));
407 13401276 s += abs(pix1[2] - avg4(pix2[2], pix2[3], pix3[2], pix3[3]));
408 13401276 s += abs(pix1[3] - avg4(pix2[3], pix2[4], pix3[3], pix3[4]));
409 13401276 s += abs(pix1[4] - avg4(pix2[4], pix2[5], pix3[4], pix3[5]));
410 13401276 s += abs(pix1[5] - avg4(pix2[5], pix2[6], pix3[5], pix3[6]));
411 13401276 s += abs(pix1[6] - avg4(pix2[6], pix2[7], pix3[6], pix3[7]));
412 13401276 s += abs(pix1[7] - avg4(pix2[7], pix2[8], pix3[7], pix3[8]));
413 13401276 pix1 += stride;
414 13401276 pix2 += stride;
415 13401276 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 11970690 times.
✓ Branch 1 taken 748189 times.
12718879 for (y = 0; y < h; y++) {
426
2/2
✓ Branch 0 taken 191531040 times.
✓ Branch 1 taken 11970690 times.
203501730 for (x = 0; x < 16; x++)
427 191531040 score1 += (s1[x] - s2[x]) * (s1[x] - s2[x]);
428
2/2
✓ Branch 0 taken 11222501 times.
✓ Branch 1 taken 748189 times.
11970690 if (y + 1 < h) {
429
2/2
✓ Branch 0 taken 168337515 times.
✓ Branch 1 taken 11222501 times.
179560016 for (x = 0; x < 15; x++)
430 168337515 score2 += FFABS(s1[x] - s1[x + stride] -
431 168337515 s1[x + 1] + s1[x + stride + 1]) -
432 168337515 FFABS(s2[x] - s2[x + stride] -
433 s2[x + 1] + s2[x + stride + 1]);
434 }
435 11970690 s1 += stride;
436 11970690 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 492 times.
✓ Branch 1 taken 48 times.
540 for (y = 0; y < h; y++) {
451
2/2
✓ Branch 0 taken 3936 times.
✓ Branch 1 taken 492 times.
4428 for (x = 0; x < 8; x++)
452 3936 score1 += (s1[x] - s2[x]) * (s1[x] - s2[x]);
453
2/2
✓ Branch 0 taken 444 times.
✓ Branch 1 taken 48 times.
492 if (y + 1 < h) {
454
2/2
✓ Branch 0 taken 3108 times.
✓ Branch 1 taken 444 times.
3552 for (x = 0; x < 7; x++)
455 3108 score2 += FFABS(s1[x] - s1[x + stride] -
456 3108 s1[x + 1] + s1[x + stride + 1]) -
457 3108 FFABS(s2[x] - s2[x + stride] -
458 s2[x + 1] + s2[x + stride + 1]);
459 }
460 492 s1 += stride;
461 492 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 40768 void ff_set_cmp(MECmpContext *c, me_cmp_func *cmp, int type)
477 {
478 int i;
479
480 40768 memset(cmp, 0, sizeof(void *) * 6);
481
482
2/2
✓ Branch 0 taken 244608 times.
✓ Branch 1 taken 40768 times.
285376 for (i = 0; i < 6; i++) {
483
7/17
✓ Branch 0 taken 224412 times.
✗ Branch 1 not taken.
✓ Branch 2 taken 7200 times.
✓ Branch 3 taken 2400 times.
✗ Branch 4 not taken.
✗ Branch 5 not taken.
✓ Branch 6 taken 1152 times.
✗ Branch 7 not taken.
✗ Branch 8 not taken.
✗ Branch 9 not taken.
✓ Branch 10 taken 1308 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.
244608 switch (type & 0xFF) {
484 224412 case FF_CMP_SAD:
485 224412 cmp[i] = c->sad[i];
486 224412 break;
487 case FF_CMP_MEDIAN_SAD:
488 cmp[i] = c->median_sad[i];
489 break;
490 7200 case FF_CMP_SATD:
491 7200 cmp[i] = c->hadamard8_diff[i];
492 7200 break;
493 2400 case FF_CMP_SSE:
494 2400 cmp[i] = c->sse[i];
495 2400 break;
496 case FF_CMP_DCT:
497 cmp[i] = c->dct_sad[i];
498 break;
499 case FF_CMP_DCT264:
500 cmp[i] = c->dct264_sad[i];
501 break;
502 1152 case FF_CMP_DCTMAX:
503 1152 cmp[i] = c->dct_max[i];
504 1152 break;
505 case FF_CMP_PSNR:
506 cmp[i] = c->quant_psnr[i];
507 break;
508 case FF_CMP_BIT:
509 cmp[i] = c->bit[i];
510 break;
511 case FF_CMP_RD:
512 cmp[i] = c->rd[i];
513 break;
514 1308 case FF_CMP_VSAD:
515 1308 cmp[i] = c->vsad[i];
516 1308 break;
517 case FF_CMP_VSSE:
518 cmp[i] = c->vsse[i];
519 break;
520 case FF_CMP_ZERO:
521 cmp[i] = zero_cmp;
522 break;
523 4824 case FF_CMP_NSSE:
524 4824 cmp[i] = c->nsse[i];
525 4824 break;
526 #if CONFIG_DWT
527 case FF_CMP_W53:
528 cmp[i]= c->w53[i];
529 break;
530 3312 case FF_CMP_W97:
531 3312 cmp[i]= c->w97[i];
532 3312 break;
533 #endif
534 default:
535 av_log(NULL, AV_LOG_ERROR,
536 "internal error in cmp function selection\n");
537 }
538 }
539 40768 }
540
541 #define BUTTERFLY2(o1, o2, i1, i2) \
542 o1 = (i1) + (i2); \
543 o2 = (i1) - (i2);
544
545 #define BUTTERFLY1(x, y) \
546 { \
547 int a, b; \
548 a = x; \
549 b = y; \
550 x = a + b; \
551 y = a - b; \
552 }
553
554 #define BUTTERFLYA(x, y) (FFABS((x) + (y)) + FFABS((x) - (y)))
555
556 15481128 static int hadamard8_diff8x8_c(MpegEncContext *s, const uint8_t *dst,
557 const uint8_t *src, ptrdiff_t stride, int h)
558 {
559 15481128 int i, temp[64], sum = 0;
560
561
2/2
✓ Branch 0 taken 123849024 times.
✓ Branch 1 taken 15481128 times.
139330152 for (i = 0; i < 8; i++) {
562 // FIXME: try pointer walks
563 123849024 BUTTERFLY2(temp[8 * i + 0], temp[8 * i + 1],
564 src[stride * i + 0] - dst[stride * i + 0],
565 src[stride * i + 1] - dst[stride * i + 1]);
566 123849024 BUTTERFLY2(temp[8 * i + 2], temp[8 * i + 3],
567 src[stride * i + 2] - dst[stride * i + 2],
568 src[stride * i + 3] - dst[stride * i + 3]);
569 123849024 BUTTERFLY2(temp[8 * i + 4], temp[8 * i + 5],
570 src[stride * i + 4] - dst[stride * i + 4],
571 src[stride * i + 5] - dst[stride * i + 5]);
572 123849024 BUTTERFLY2(temp[8 * i + 6], temp[8 * i + 7],
573 src[stride * i + 6] - dst[stride * i + 6],
574 src[stride * i + 7] - dst[stride * i + 7]);
575
576 123849024 BUTTERFLY1(temp[8 * i + 0], temp[8 * i + 2]);
577 123849024 BUTTERFLY1(temp[8 * i + 1], temp[8 * i + 3]);
578 123849024 BUTTERFLY1(temp[8 * i + 4], temp[8 * i + 6]);
579 123849024 BUTTERFLY1(temp[8 * i + 5], temp[8 * i + 7]);
580
581 123849024 BUTTERFLY1(temp[8 * i + 0], temp[8 * i + 4]);
582 123849024 BUTTERFLY1(temp[8 * i + 1], temp[8 * i + 5]);
583 123849024 BUTTERFLY1(temp[8 * i + 2], temp[8 * i + 6]);
584 123849024 BUTTERFLY1(temp[8 * i + 3], temp[8 * i + 7]);
585 }
586
587
2/2
✓ Branch 0 taken 123849024 times.
✓ Branch 1 taken 15481128 times.
139330152 for (i = 0; i < 8; i++) {
588 123849024 BUTTERFLY1(temp[8 * 0 + i], temp[8 * 1 + i]);
589 123849024 BUTTERFLY1(temp[8 * 2 + i], temp[8 * 3 + i]);
590 123849024 BUTTERFLY1(temp[8 * 4 + i], temp[8 * 5 + i]);
591 123849024 BUTTERFLY1(temp[8 * 6 + i], temp[8 * 7 + i]);
592
593 123849024 BUTTERFLY1(temp[8 * 0 + i], temp[8 * 2 + i]);
594 123849024 BUTTERFLY1(temp[8 * 1 + i], temp[8 * 3 + i]);
595 123849024 BUTTERFLY1(temp[8 * 4 + i], temp[8 * 6 + i]);
596 123849024 BUTTERFLY1(temp[8 * 5 + i], temp[8 * 7 + i]);
597
598 123849024 sum += BUTTERFLYA(temp[8 * 0 + i], temp[8 * 4 + i]) +
599 123849024 BUTTERFLYA(temp[8 * 1 + i], temp[8 * 5 + i]) +
600 123849024 BUTTERFLYA(temp[8 * 2 + i], temp[8 * 6 + i]) +
601 123849024 BUTTERFLYA(temp[8 * 3 + i], temp[8 * 7 + i]);
602 }
603 15481128 return sum;
604 }
605
606 112 static int hadamard8_intra8x8_c(MpegEncContext *s, const uint8_t *src,
607 const uint8_t *dummy, ptrdiff_t stride, int h)
608 {
609 112 int i, temp[64], sum = 0;
610
611
2/2
✓ Branch 0 taken 896 times.
✓ Branch 1 taken 112 times.
1008 for (i = 0; i < 8; i++) {
612 // FIXME: try pointer walks
613 896 BUTTERFLY2(temp[8 * i + 0], temp[8 * i + 1],
614 src[stride * i + 0], src[stride * i + 1]);
615 896 BUTTERFLY2(temp[8 * i + 2], temp[8 * i + 3],
616 src[stride * i + 2], src[stride * i + 3]);
617 896 BUTTERFLY2(temp[8 * i + 4], temp[8 * i + 5],
618 src[stride * i + 4], src[stride * i + 5]);
619 896 BUTTERFLY2(temp[8 * i + 6], temp[8 * i + 7],
620 src[stride * i + 6], src[stride * i + 7]);
621
622 896 BUTTERFLY1(temp[8 * i + 0], temp[8 * i + 2]);
623 896 BUTTERFLY1(temp[8 * i + 1], temp[8 * i + 3]);
624 896 BUTTERFLY1(temp[8 * i + 4], temp[8 * i + 6]);
625 896 BUTTERFLY1(temp[8 * i + 5], temp[8 * i + 7]);
626
627 896 BUTTERFLY1(temp[8 * i + 0], temp[8 * i + 4]);
628 896 BUTTERFLY1(temp[8 * i + 1], temp[8 * i + 5]);
629 896 BUTTERFLY1(temp[8 * i + 2], temp[8 * i + 6]);
630 896 BUTTERFLY1(temp[8 * i + 3], temp[8 * i + 7]);
631 }
632
633
2/2
✓ Branch 0 taken 896 times.
✓ Branch 1 taken 112 times.
1008 for (i = 0; i < 8; i++) {
634 896 BUTTERFLY1(temp[8 * 0 + i], temp[8 * 1 + i]);
635 896 BUTTERFLY1(temp[8 * 2 + i], temp[8 * 3 + i]);
636 896 BUTTERFLY1(temp[8 * 4 + i], temp[8 * 5 + i]);
637 896 BUTTERFLY1(temp[8 * 6 + i], temp[8 * 7 + i]);
638
639 896 BUTTERFLY1(temp[8 * 0 + i], temp[8 * 2 + i]);
640 896 BUTTERFLY1(temp[8 * 1 + i], temp[8 * 3 + i]);
641 896 BUTTERFLY1(temp[8 * 4 + i], temp[8 * 6 + i]);
642 896 BUTTERFLY1(temp[8 * 5 + i], temp[8 * 7 + i]);
643
644 896 sum +=
645 896 BUTTERFLYA(temp[8 * 0 + i], temp[8 * 4 + i])
646 896 + BUTTERFLYA(temp[8 * 1 + i], temp[8 * 5 + i])
647 896 + BUTTERFLYA(temp[8 * 2 + i], temp[8 * 6 + i])
648 896 + BUTTERFLYA(temp[8 * 3 + i], temp[8 * 7 + i]);
649 }
650
651 112 sum -= FFABS(temp[8 * 0] + temp[8 * 4]); // -mean
652
653 112 return sum;
654 }
655
656 static int dct_sad8x8_c(MpegEncContext *s, const uint8_t *src1,
657 const uint8_t *src2, ptrdiff_t stride, int h)
658 {
659 LOCAL_ALIGNED_16(int16_t, temp, [64]);
660
661 s->pdsp.diff_pixels_unaligned(temp, src1, src2, stride);
662 s->fdsp.fdct(temp);
663 return s->mecc.sum_abs_dctelem(temp);
664 }
665
666 #if CONFIG_GPL
667 #define DCT8_1D \
668 { \
669 const int s07 = SRC(0) + SRC(7); \
670 const int s16 = SRC(1) + SRC(6); \
671 const int s25 = SRC(2) + SRC(5); \
672 const int s34 = SRC(3) + SRC(4); \
673 const int a0 = s07 + s34; \
674 const int a1 = s16 + s25; \
675 const int a2 = s07 - s34; \
676 const int a3 = s16 - s25; \
677 const int d07 = SRC(0) - SRC(7); \
678 const int d16 = SRC(1) - SRC(6); \
679 const int d25 = SRC(2) - SRC(5); \
680 const int d34 = SRC(3) - SRC(4); \
681 const int a4 = d16 + d25 + (d07 + (d07 >> 1)); \
682 const int a5 = d07 - d34 - (d25 + (d25 >> 1)); \
683 const int a6 = d07 + d34 - (d16 + (d16 >> 1)); \
684 const int a7 = d16 - d25 + (d34 + (d34 >> 1)); \
685 DST(0, a0 + a1); \
686 DST(1, a4 + (a7 >> 2)); \
687 DST(2, a2 + (a3 >> 1)); \
688 DST(3, a5 + (a6 >> 2)); \
689 DST(4, a0 - a1); \
690 DST(5, a6 - (a5 >> 2)); \
691 DST(6, (a2 >> 1) - a3); \
692 DST(7, (a4 >> 2) - a7); \
693 }
694
695 static int dct264_sad8x8_c(MpegEncContext *s, const uint8_t *src1,
696 const uint8_t *src2, ptrdiff_t stride, int h)
697 {
698 int16_t dct[8][8];
699 int i, sum = 0;
700
701 s->pdsp.diff_pixels_unaligned(dct[0], src1, src2, stride);
702
703 #define SRC(x) dct[i][x]
704 #define DST(x, v) dct[i][x] = v
705 for (i = 0; i < 8; i++)
706 DCT8_1D
707 #undef SRC
708 #undef DST
709
710 #define SRC(x) dct[x][i]
711 #define DST(x, v) sum += FFABS(v)
712 for (i = 0; i < 8; i++)
713 DCT8_1D
714 #undef SRC
715 #undef DST
716 return sum;
717 }
718 #endif
719
720 static int dct_max8x8_c(MpegEncContext *s, const uint8_t *src1,
721 const uint8_t *src2, ptrdiff_t stride, int h)
722 {
723 LOCAL_ALIGNED_16(int16_t, temp, [64]);
724 int sum = 0, i;
725
726 s->pdsp.diff_pixels_unaligned(temp, src1, src2, stride);
727 s->fdsp.fdct(temp);
728
729 for (i = 0; i < 64; i++)
730 sum = FFMAX(sum, FFABS(temp[i]));
731
732 return sum;
733 }
734
735 static int quant_psnr8x8_c(MpegEncContext *s, const uint8_t *src1,
736 const uint8_t *src2, ptrdiff_t stride, int h)
737 {
738 LOCAL_ALIGNED_16(int16_t, temp, [64 * 2]);
739 int16_t *const bak = temp + 64;
740 int sum = 0, i;
741
742 s->mb_intra = 0;
743
744 s->pdsp.diff_pixels_unaligned(temp, src1, src2, stride);
745
746 memcpy(bak, temp, 64 * sizeof(int16_t));
747
748 s->block_last_index[0 /* FIXME */] =
749 s->fast_dct_quantize(s, temp, 0 /* FIXME */, s->qscale, &i);
750 s->dct_unquantize_inter(s, temp, 0, s->qscale);
751 ff_simple_idct_int16_8bit(temp); // FIXME
752
753 for (i = 0; i < 64; i++)
754 sum += (temp[i] - bak[i]) * (temp[i] - bak[i]);
755
756 return sum;
757 }
758
759 static int rd8x8_c(MpegEncContext *s, const uint8_t *src1, const uint8_t *src2,
760 ptrdiff_t stride, int h)
761 {
762 const uint8_t *scantable = s->intra_scantable.permutated;
763 LOCAL_ALIGNED_16(int16_t, temp, [64]);
764 LOCAL_ALIGNED_16(uint8_t, lsrc1, [64]);
765 LOCAL_ALIGNED_16(uint8_t, lsrc2, [64]);
766 int i, last, run, bits, level, distortion, start_i;
767 const int esc_length = s->ac_esc_length;
768 uint8_t *length, *last_length;
769
770 copy_block8(lsrc1, src1, 8, stride, 8);
771 copy_block8(lsrc2, src2, 8, stride, 8);
772
773 s->pdsp.diff_pixels(temp, lsrc1, lsrc2, 8);
774
775 s->block_last_index[0 /* FIXME */] =
776 last =
777 s->fast_dct_quantize(s, temp, 0 /* FIXME */, s->qscale, &i);
778
779 bits = 0;
780
781 if (s->mb_intra) {
782 start_i = 1;
783 length = s->intra_ac_vlc_length;
784 last_length = s->intra_ac_vlc_last_length;
785 bits += s->luma_dc_vlc_length[temp[0] + 256]; // FIXME: chroma
786 } else {
787 start_i = 0;
788 length = s->inter_ac_vlc_length;
789 last_length = s->inter_ac_vlc_last_length;
790 }
791
792 if (last >= start_i) {
793 run = 0;
794 for (i = start_i; i < last; i++) {
795 int j = scantable[i];
796 level = temp[j];
797
798 if (level) {
799 level += 64;
800 if ((level & (~127)) == 0)
801 bits += length[UNI_AC_ENC_INDEX(run, level)];
802 else
803 bits += esc_length;
804 run = 0;
805 } else
806 run++;
807 }
808 i = scantable[last];
809
810 level = temp[i] + 64;
811
812 av_assert2(level - 64);
813
814 if ((level & (~127)) == 0) {
815 bits += last_length[UNI_AC_ENC_INDEX(run, level)];
816 } else
817 bits += esc_length;
818 }
819
820 if (last >= 0) {
821 if (s->mb_intra)
822 s->dct_unquantize_intra(s, temp, 0, s->qscale);
823 else
824 s->dct_unquantize_inter(s, temp, 0, s->qscale);
825 }
826
827 s->idsp.idct_add(lsrc2, 8, temp);
828
829 distortion = s->mecc.sse[1](NULL, lsrc2, lsrc1, 8, 8);
830
831 return distortion + ((bits * s->qscale * s->qscale * 109 + 64) >> 7);
832 }
833
834 static int bit8x8_c(MpegEncContext *s, const uint8_t *src1, const uint8_t *src2,
835 ptrdiff_t stride, int h)
836 {
837 const uint8_t *scantable = s->intra_scantable.permutated;
838 LOCAL_ALIGNED_16(int16_t, temp, [64]);
839 int i, last, run, bits, level, start_i;
840 const int esc_length = s->ac_esc_length;
841 uint8_t *length, *last_length;
842
843 s->pdsp.diff_pixels_unaligned(temp, src1, src2, stride);
844
845 s->block_last_index[0 /* FIXME */] =
846 last =
847 s->fast_dct_quantize(s, temp, 0 /* FIXME */, s->qscale, &i);
848
849 bits = 0;
850
851 if (s->mb_intra) {
852 start_i = 1;
853 length = s->intra_ac_vlc_length;
854 last_length = s->intra_ac_vlc_last_length;
855 bits += s->luma_dc_vlc_length[temp[0] + 256]; // FIXME: chroma
856 } else {
857 start_i = 0;
858 length = s->inter_ac_vlc_length;
859 last_length = s->inter_ac_vlc_last_length;
860 }
861
862 if (last >= start_i) {
863 run = 0;
864 for (i = start_i; i < last; i++) {
865 int j = scantable[i];
866 level = temp[j];
867
868 if (level) {
869 level += 64;
870 if ((level & (~127)) == 0)
871 bits += length[UNI_AC_ENC_INDEX(run, level)];
872 else
873 bits += esc_length;
874 run = 0;
875 } else
876 run++;
877 }
878 i = scantable[last];
879
880 level = temp[i] + 64;
881
882 av_assert2(level - 64);
883
884 if ((level & (~127)) == 0)
885 bits += last_length[UNI_AC_ENC_INDEX(run, level)];
886 else
887 bits += esc_length;
888 }
889
890 return bits;
891 }
892
893 #define VSAD_INTRA(size) \
894 static int vsad_intra ## size ## _c(MpegEncContext *c, \
895 const uint8_t *s, const uint8_t *dummy, \
896 ptrdiff_t stride, int h) \
897 { \
898 int score = 0, x, y; \
899 \
900 for (y = 1; y < h; y++) { \
901 for (x = 0; x < size; x += 4) { \
902 score += FFABS(s[x] - s[x + stride]) + \
903 FFABS(s[x + 1] - s[x + stride + 1]) + \
904 FFABS(s[x + 2] - s[x + 2 + stride]) + \
905 FFABS(s[x + 3] - s[x + 3 + stride]); \
906 } \
907 s += stride; \
908 } \
909 \
910 return score; \
911 }
912
4/4
✓ Branch 0 taken 728 times.
✓ Branch 1 taken 364 times.
✓ Branch 2 taken 364 times.
✓ Branch 3 taken 48 times.
1140 VSAD_INTRA(8)
913
4/4
✓ Branch 0 taken 34475096 times.
✓ Branch 1 taken 8618774 times.
✓ Branch 2 taken 8618774 times.
✓ Branch 3 taken 1231246 times.
44325116 VSAD_INTRA(16)
914
915 #define VSAD(size) \
916 static int vsad ## size ## _c(MpegEncContext *c, \
917 const uint8_t *s1, const uint8_t *s2, \
918 ptrdiff_t stride, int h) \
919 { \
920 int score = 0, x, y; \
921 \
922 for (y = 1; y < h; y++) { \
923 for (x = 0; x < size; x++) \
924 score += FFABS(s1[x] - s2[x] - s1[x + stride] + s2[x + stride]); \
925 s1 += stride; \
926 s2 += stride; \
927 } \
928 \
929 return score; \
930 }
931
4/4
✓ Branch 0 taken 2048 times.
✓ Branch 1 taken 256 times.
✓ Branch 2 taken 256 times.
✓ Branch 3 taken 32 times.
2336 VSAD(8)
932
4/4
✓ Branch 0 taken 213085792 times.
✓ Branch 1 taken 13317862 times.
✓ Branch 2 taken 13317862 times.
✓ Branch 3 taken 1902546 times.
228306200 VSAD(16)
933
934 #define SQ(a) ((a) * (a))
935 #define VSSE_INTRA(size) \
936 static int vsse_intra ## size ## _c(MpegEncContext *c, \
937 const uint8_t *s, const uint8_t *dummy, \
938 ptrdiff_t stride, int h) \
939 { \
940 int score = 0, x, y; \
941 \
942 for (y = 1; y < h; y++) { \
943 for (x = 0; x < size; x += 4) { \
944 score += SQ(s[x] - s[x + stride]) + \
945 SQ(s[x + 1] - s[x + stride + 1]) + \
946 SQ(s[x + 2] - s[x + stride + 2]) + \
947 SQ(s[x + 3] - s[x + stride + 3]); \
948 } \
949 s += stride; \
950 } \
951 \
952 return score; \
953 }
954
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)
955
4/4
✓ Branch 0 taken 1248 times.
✓ Branch 1 taken 312 times.
✓ Branch 2 taken 312 times.
✓ Branch 3 taken 32 times.
1592 VSSE_INTRA(16)
956
957 #define VSSE(size) \
958 static int vsse ## size ## _c(MpegEncContext *c, const uint8_t *s1, const uint8_t *s2, \
959 ptrdiff_t stride, int h) \
960 { \
961 int score = 0, x, y; \
962 \
963 for (y = 1; y < h; y++) { \
964 for (x = 0; x < size; x++) \
965 score += SQ(s1[x] - s2[x] - s1[x + stride] + s2[x + stride]); \
966 s1 += stride; \
967 s2 += stride; \
968 } \
969 \
970 return score; \
971 }
972
4/4
✓ Branch 0 taken 1952 times.
✓ Branch 1 taken 244 times.
✓ Branch 2 taken 244 times.
✓ Branch 3 taken 32 times.
2228 VSSE(8)
973
4/4
✓ Branch 0 taken 4736 times.
✓ Branch 1 taken 296 times.
✓ Branch 2 taken 296 times.
✓ Branch 3 taken 32 times.
5064 VSSE(16)
974
975 #define WRAPPER8_16_SQ(name8, name16) \
976 static int name16(MpegEncContext *s, const uint8_t *dst, const uint8_t *src, \
977 ptrdiff_t stride, int h) \
978 { \
979 int score = 0; \
980 \
981 score += name8(s, dst, src, stride, 8); \
982 score += name8(s, dst + 8, src + 8, stride, 8); \
983 if (h == 16) { \
984 dst += 8 * stride; \
985 src += 8 * stride; \
986 score += name8(s, dst, src, stride, 8); \
987 score += name8(s, dst + 8, src + 8, stride, 8); \
988 } \
989 return score; \
990 }
991
992
2/2
✓ Branch 2 taken 3613240 times.
✓ Branch 3 taken 44 times.
3613284 WRAPPER8_16_SQ(hadamard8_diff8x8_c, hadamard8_diff16_c)
993
2/2
✓ Branch 2 taken 8 times.
✓ Branch 3 taken 24 times.
32 WRAPPER8_16_SQ(hadamard8_intra8x8_c, hadamard8_intra16_c)
994 WRAPPER8_16_SQ(dct_sad8x8_c, dct_sad16_c)
995 #if CONFIG_GPL
996 WRAPPER8_16_SQ(dct264_sad8x8_c, dct264_sad16_c)
997 #endif
998 WRAPPER8_16_SQ(dct_max8x8_c, dct_max16_c)
999 WRAPPER8_16_SQ(quant_psnr8x8_c, quant_psnr16_c)
1000 WRAPPER8_16_SQ(rd8x8_c, rd16_c)
1001 WRAPPER8_16_SQ(bit8x8_c, bit16_c)
1002
1003 1029 av_cold void ff_me_cmp_init(MECmpContext *c, AVCodecContext *avctx)
1004 {
1005 1029 c->sum_abs_dctelem = sum_abs_dctelem_c;
1006
1007 /* TODO [0] 16 [1] 8 */
1008 1029 c->pix_abs[0][0] = pix_abs16_c;
1009 1029 c->pix_abs[0][1] = pix_abs16_x2_c;
1010 1029 c->pix_abs[0][2] = pix_abs16_y2_c;
1011 1029 c->pix_abs[0][3] = pix_abs16_xy2_c;
1012 1029 c->pix_abs[1][0] = pix_abs8_c;
1013 1029 c->pix_abs[1][1] = pix_abs8_x2_c;
1014 1029 c->pix_abs[1][2] = pix_abs8_y2_c;
1015 1029 c->pix_abs[1][3] = pix_abs8_xy2_c;
1016
1017 #define SET_CMP_FUNC(name) \
1018 c->name[0] = name ## 16_c; \
1019 c->name[1] = name ## 8x8_c;
1020
1021 1029 SET_CMP_FUNC(hadamard8_diff)
1022 1029 c->hadamard8_diff[4] = hadamard8_intra16_c;
1023 1029 c->hadamard8_diff[5] = hadamard8_intra8x8_c;
1024 1029 SET_CMP_FUNC(dct_sad)
1025 1029 SET_CMP_FUNC(dct_max)
1026 #if CONFIG_GPL
1027 1029 SET_CMP_FUNC(dct264_sad)
1028 #endif
1029 1029 c->sad[0] = pix_abs16_c;
1030 1029 c->sad[1] = pix_abs8_c;
1031 1029 c->sse[0] = sse16_c;
1032 1029 c->sse[1] = sse8_c;
1033 1029 c->sse[2] = sse4_c;
1034 1029 SET_CMP_FUNC(quant_psnr)
1035 1029 SET_CMP_FUNC(rd)
1036 1029 SET_CMP_FUNC(bit)
1037 1029 c->vsad[0] = vsad16_c;
1038 1029 c->vsad[1] = vsad8_c;
1039 1029 c->vsad[4] = vsad_intra16_c;
1040 1029 c->vsad[5] = vsad_intra8_c;
1041 1029 c->vsse[0] = vsse16_c;
1042 1029 c->vsse[1] = vsse8_c;
1043 1029 c->vsse[4] = vsse_intra16_c;
1044 1029 c->vsse[5] = vsse_intra8_c;
1045 1029 c->nsse[0] = nsse16_c;
1046 1029 c->nsse[1] = nsse8_c;
1047 #if CONFIG_SNOW_DECODER || CONFIG_SNOW_ENCODER
1048 1029 ff_dsputil_init_dwt(c);
1049 #endif
1050
1051 1029 c->median_sad[0] = pix_median_abs16_c;
1052 1029 c->median_sad[1] = pix_median_abs8_c;
1053
1054 #if ARCH_AARCH64
1055 ff_me_cmp_init_aarch64(c, avctx);
1056 #elif ARCH_ALPHA
1057 ff_me_cmp_init_alpha(c, avctx);
1058 #elif ARCH_ARM
1059 ff_me_cmp_init_arm(c, avctx);
1060 #elif ARCH_PPC
1061 ff_me_cmp_init_ppc(c, avctx);
1062 #elif ARCH_X86
1063 1029 ff_me_cmp_init_x86(c, avctx);
1064 #elif ARCH_MIPS
1065 ff_me_cmp_init_mips(c, avctx);
1066 #endif
1067
1068 1029 }
1069