FFmpeg coverage

Directory: ../../../ffmpeg/
File: src/libavcodec/me_cmp.c
Date: 2024-07-26 21:54:09
Exec Total Coverage
Lines: 396 528 75.0%
Functions: 29 43 67.4%
Branches: 101 158 63.9%
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 <stddef.h>
24
25 #include "libavutil/attributes.h"
26 #include "libavutil/internal.h"
27 #include "libavutil/mem_internal.h"
28 #include "avcodec.h"
29 #include "copy_block.h"
30 #include "simple_idct.h"
31 #include "me_cmp.h"
32 #include "mpegvideoenc.h"
33 #include "config.h"
34 #include "config_components.h"
35
36 /* (i - 256) * (i - 256) */
37 const uint32_t ff_square_tab[512] = {
38 65536, 65025, 64516, 64009, 63504, 63001, 62500, 62001, 61504, 61009, 60516, 60025, 59536, 59049, 58564, 58081,
39 57600, 57121, 56644, 56169, 55696, 55225, 54756, 54289, 53824, 53361, 52900, 52441, 51984, 51529, 51076, 50625,
40 50176, 49729, 49284, 48841, 48400, 47961, 47524, 47089, 46656, 46225, 45796, 45369, 44944, 44521, 44100, 43681,
41 43264, 42849, 42436, 42025, 41616, 41209, 40804, 40401, 40000, 39601, 39204, 38809, 38416, 38025, 37636, 37249,
42 36864, 36481, 36100, 35721, 35344, 34969, 34596, 34225, 33856, 33489, 33124, 32761, 32400, 32041, 31684, 31329,
43 30976, 30625, 30276, 29929, 29584, 29241, 28900, 28561, 28224, 27889, 27556, 27225, 26896, 26569, 26244, 25921,
44 25600, 25281, 24964, 24649, 24336, 24025, 23716, 23409, 23104, 22801, 22500, 22201, 21904, 21609, 21316, 21025,
45 20736, 20449, 20164, 19881, 19600, 19321, 19044, 18769, 18496, 18225, 17956, 17689, 17424, 17161, 16900, 16641,
46 16384, 16129, 15876, 15625, 15376, 15129, 14884, 14641, 14400, 14161, 13924, 13689, 13456, 13225, 12996, 12769,
47 12544, 12321, 12100, 11881, 11664, 11449, 11236, 11025, 10816, 10609, 10404, 10201, 10000, 9801, 9604, 9409,
48 9216, 9025, 8836, 8649, 8464, 8281, 8100, 7921, 7744, 7569, 7396, 7225, 7056, 6889, 6724, 6561,
49 6400, 6241, 6084, 5929, 5776, 5625, 5476, 5329, 5184, 5041, 4900, 4761, 4624, 4489, 4356, 4225,
50 4096, 3969, 3844, 3721, 3600, 3481, 3364, 3249, 3136, 3025, 2916, 2809, 2704, 2601, 2500, 2401,
51 2304, 2209, 2116, 2025, 1936, 1849, 1764, 1681, 1600, 1521, 1444, 1369, 1296, 1225, 1156, 1089,
52 1024, 961, 900, 841, 784, 729, 676, 625, 576, 529, 484, 441, 400, 361, 324, 289,
53 256, 225, 196, 169, 144, 121, 100, 81, 64, 49, 36, 25, 16, 9, 4, 1,
54 0, 1, 4, 9, 16, 25, 36, 49, 64, 81, 100, 121, 144, 169, 196, 225,
55 256, 289, 324, 361, 400, 441, 484, 529, 576, 625, 676, 729, 784, 841, 900, 961,
56 1024, 1089, 1156, 1225, 1296, 1369, 1444, 1521, 1600, 1681, 1764, 1849, 1936, 2025, 2116, 2209,
57 2304, 2401, 2500, 2601, 2704, 2809, 2916, 3025, 3136, 3249, 3364, 3481, 3600, 3721, 3844, 3969,
58 4096, 4225, 4356, 4489, 4624, 4761, 4900, 5041, 5184, 5329, 5476, 5625, 5776, 5929, 6084, 6241,
59 6400, 6561, 6724, 6889, 7056, 7225, 7396, 7569, 7744, 7921, 8100, 8281, 8464, 8649, 8836, 9025,
60 9216, 9409, 9604, 9801, 10000, 10201, 10404, 10609, 10816, 11025, 11236, 11449, 11664, 11881, 12100, 12321,
61 12544, 12769, 12996, 13225, 13456, 13689, 13924, 14161, 14400, 14641, 14884, 15129, 15376, 15625, 15876, 16129,
62 16384, 16641, 16900, 17161, 17424, 17689, 17956, 18225, 18496, 18769, 19044, 19321, 19600, 19881, 20164, 20449,
63 20736, 21025, 21316, 21609, 21904, 22201, 22500, 22801, 23104, 23409, 23716, 24025, 24336, 24649, 24964, 25281,
64 25600, 25921, 26244, 26569, 26896, 27225, 27556, 27889, 28224, 28561, 28900, 29241, 29584, 29929, 30276, 30625,
65 30976, 31329, 31684, 32041, 32400, 32761, 33124, 33489, 33856, 34225, 34596, 34969, 35344, 35721, 36100, 36481,
66 36864, 37249, 37636, 38025, 38416, 38809, 39204, 39601, 40000, 40401, 40804, 41209, 41616, 42025, 42436, 42849,
67 43264, 43681, 44100, 44521, 44944, 45369, 45796, 46225, 46656, 47089, 47524, 47961, 48400, 48841, 49284, 49729,
68 50176, 50625, 51076, 51529, 51984, 52441, 52900, 53361, 53824, 54289, 54756, 55225, 55696, 56169, 56644, 57121,
69 57600, 58081, 58564, 59049, 59536, 60025, 60516, 61009, 61504, 62001, 62500, 63001, 63504, 64009, 64516, 65025,
70 };
71
72 32 static int sse4_c(MpegEncContext *v, const uint8_t *pix1, const uint8_t *pix2,
73 ptrdiff_t stride, int h)
74 {
75 32 int s = 0, i;
76 32 const uint32_t *sq = ff_square_tab + 256;
77
78
2/2
✓ Branch 0 taken 296 times.
✓ Branch 1 taken 32 times.
 328 for (i = 0; i < h; i++) {
79 296 s += sq[pix1[0] - pix2[0]];
80 296 s += sq[pix1[1] - pix2[1]];
81 296 s += sq[pix1[2] - pix2[2]];
82 296 s += sq[pix1[3] - pix2[3]];
83 296 pix1 += stride;
84 296 pix2 += stride;
85 }
86 32 return s;
87 }
88
89 4998886 static int sse8_c(MpegEncContext *v, const uint8_t *pix1, const uint8_t *pix2,
90 ptrdiff_t stride, int h)
91 {
92 4998886 int s = 0, i;
93 4998886 const uint32_t *sq = ff_square_tab + 256;
94
95
2/2
✓ Branch 0 taken 46132574 times.
✓ Branch 1 taken 4998886 times.
 51131460 for (i = 0; i < h; i++) {
96 46132574 s += sq[pix1[0] - pix2[0]];
97 46132574 s += sq[pix1[1] - pix2[1]];
98 46132574 s += sq[pix1[2] - pix2[2]];
99 46132574 s += sq[pix1[3] - pix2[3]];
100 46132574 s += sq[pix1[4] - pix2[4]];
101 46132574 s += sq[pix1[5] - pix2[5]];
102 46132574 s += sq[pix1[6] - pix2[6]];
103 46132574 s += sq[pix1[7] - pix2[7]];
104 46132574 pix1 += stride;
105 46132574 pix2 += stride;
106 }
107 4998886 return s;
108 }
109
110 7479543 static int sse16_c(MpegEncContext *v, const uint8_t *pix1, const uint8_t *pix2,
111 ptrdiff_t stride, int h)
112 {
113 7479543 int s = 0, i;
114 7479543 const uint32_t *sq = ff_square_tab + 256;
115
116
2/2
✓ Branch 0 taken 119672430 times.
✓ Branch 1 taken 7479543 times.
 127151973 for (i = 0; i < h; i++) {
117 119672430 s += sq[pix1[0] - pix2[0]];
118 119672430 s += sq[pix1[1] - pix2[1]];
119 119672430 s += sq[pix1[2] - pix2[2]];
120 119672430 s += sq[pix1[3] - pix2[3]];
121 119672430 s += sq[pix1[4] - pix2[4]];
122 119672430 s += sq[pix1[5] - pix2[5]];
123 119672430 s += sq[pix1[6] - pix2[6]];
124 119672430 s += sq[pix1[7] - pix2[7]];
125 119672430 s += sq[pix1[8] - pix2[8]];
126 119672430 s += sq[pix1[9] - pix2[9]];
127 119672430 s += sq[pix1[10] - pix2[10]];
128 119672430 s += sq[pix1[11] - pix2[11]];
129 119672430 s += sq[pix1[12] - pix2[12]];
130 119672430 s += sq[pix1[13] - pix2[13]];
131 119672430 s += sq[pix1[14] - pix2[14]];
132 119672430 s += sq[pix1[15] - pix2[15]];
133
134 119672430 pix1 += stride;
135 119672430 pix2 += stride;
136 }
137 7479543 return s;
138 }
139
140 ✗ static int sum_abs_dctelem_c(const int16_t *block)
141 {
142 ✗ int sum = 0, i;
143
144 ✗ for (i = 0; i < 64; i++)
145 ✗ sum += FFABS(block[i]);
146 ✗ return sum;
147 }
148
149 #define avg2(a, b) (((a) + (b) + 1) >> 1)
150 #define avg4(a, b, c, d) (((a) + (b) + (c) + (d) + 2) >> 2)
151
152 44203458 static inline int pix_abs16_c(MpegEncContext *v, const uint8_t *pix1, const uint8_t *pix2,
153 ptrdiff_t stride, int h)
154 {
155 44203458 int s = 0, i;
156
157
2/2
✓ Branch 0 taken 626886232 times.
✓ Branch 1 taken 44203458 times.
 671089690 for (i = 0; i < h; i++) {
158 626886232 s += abs(pix1[0] - pix2[0]);
159 626886232 s += abs(pix1[1] - pix2[1]);
160 626886232 s += abs(pix1[2] - pix2[2]);
161 626886232 s += abs(pix1[3] - pix2[3]);
162 626886232 s += abs(pix1[4] - pix2[4]);
163 626886232 s += abs(pix1[5] - pix2[5]);
164 626886232 s += abs(pix1[6] - pix2[6]);
165 626886232 s += abs(pix1[7] - pix2[7]);
166 626886232 s += abs(pix1[8] - pix2[8]);
167 626886232 s += abs(pix1[9] - pix2[9]);
168 626886232 s += abs(pix1[10] - pix2[10]);
169 626886232 s += abs(pix1[11] - pix2[11]);
170 626886232 s += abs(pix1[12] - pix2[12]);
171 626886232 s += abs(pix1[13] - pix2[13]);
172 626886232 s += abs(pix1[14] - pix2[14]);
173 626886232 s += abs(pix1[15] - pix2[15]);
174 626886232 pix1 += stride;
175 626886232 pix2 += stride;
176 }
177 44203458 return s;
178 }
179
180 32 static inline int pix_median_abs16_c(MpegEncContext *v, const uint8_t *pix1, const uint8_t *pix2,
181 ptrdiff_t stride, int h)
182 {
183 32 int s = 0, i, j;
184
185 #define V(x) (pix1[x] - pix2[x])
186
187 32 s += abs(V(0));
188 32 s += abs(V(1) - V(0));
189 32 s += abs(V(2) - V(1));
190 32 s += abs(V(3) - V(2));
191 32 s += abs(V(4) - V(3));
192 32 s += abs(V(5) - V(4));
193 32 s += abs(V(6) - V(5));
194 32 s += abs(V(7) - V(6));
195 32 s += abs(V(8) - V(7));
196 32 s += abs(V(9) - V(8));
197 32 s += abs(V(10) - V(9));
198 32 s += abs(V(11) - V(10));
199 32 s += abs(V(12) - V(11));
200 32 s += abs(V(13) - V(12));
201 32 s += abs(V(14) - V(13));
202 32 s += abs(V(15) - V(14));
203
204 32 pix1 += stride;
205 32 pix2 += stride;
206
207
2/2
✓ Branch 0 taken 336 times.
✓ Branch 1 taken 32 times.
 368 for (i = 1; i < h; i++) {
208 336 s += abs(V(0) - V(-stride));
209
2/2
✓ Branch 0 taken 5040 times.
✓ Branch 1 taken 336 times.
 5376 for (j = 1; j < 16; j++)
210 5040 s += abs(V(j) - mid_pred(V(j-stride), V(j-1), V(j-stride) + V(j-1) - V(j-stride-1)));
211 336 pix1 += stride;
212 336 pix2 += stride;
213
214 }
215 #undef V
216 32 return s;
217 }
218
219 3286935 static int pix_abs16_x2_c(MpegEncContext *v, const uint8_t *pix1, const uint8_t *pix2,
220 ptrdiff_t stride, int h)
221 {
222 3286935 int s = 0, i;
223
224
2/2
✓ Branch 0 taken 43281106 times.
✓ Branch 1 taken 3286935 times.
 46568041 for (i = 0; i < h; i++) {
225 43281106 s += abs(pix1[0] - avg2(pix2[0], pix2[1]));
226 43281106 s += abs(pix1[1] - avg2(pix2[1], pix2[2]));
227 43281106 s += abs(pix1[2] - avg2(pix2[2], pix2[3]));
228 43281106 s += abs(pix1[3] - avg2(pix2[3], pix2[4]));
229 43281106 s += abs(pix1[4] - avg2(pix2[4], pix2[5]));
230 43281106 s += abs(pix1[5] - avg2(pix2[5], pix2[6]));
231 43281106 s += abs(pix1[6] - avg2(pix2[6], pix2[7]));
232 43281106 s += abs(pix1[7] - avg2(pix2[7], pix2[8]));
233 43281106 s += abs(pix1[8] - avg2(pix2[8], pix2[9]));
234 43281106 s += abs(pix1[9] - avg2(pix2[9], pix2[10]));
235 43281106 s += abs(pix1[10] - avg2(pix2[10], pix2[11]));
236 43281106 s += abs(pix1[11] - avg2(pix2[11], pix2[12]));
237 43281106 s += abs(pix1[12] - avg2(pix2[12], pix2[13]));
238 43281106 s += abs(pix1[13] - avg2(pix2[13], pix2[14]));
239 43281106 s += abs(pix1[14] - avg2(pix2[14], pix2[15]));
240 43281106 s += abs(pix1[15] - avg2(pix2[15], pix2[16]));
241 43281106 pix1 += stride;
242 43281106 pix2 += stride;
243 }
244 3286935 return s;
245 }
246
247 3286935 static int pix_abs16_y2_c(MpegEncContext *v, const uint8_t *pix1, const uint8_t *pix2,
248 ptrdiff_t stride, int h)
249 {
250 3286935 int s = 0, i;
251 3286935 const uint8_t *pix3 = pix2 + stride;
252
253
2/2
✓ Branch 0 taken 43281144 times.
✓ Branch 1 taken 3286935 times.
 46568079 for (i = 0; i < h; i++) {
254 43281144 s += abs(pix1[0] - avg2(pix2[0], pix3[0]));
255 43281144 s += abs(pix1[1] - avg2(pix2[1], pix3[1]));
256 43281144 s += abs(pix1[2] - avg2(pix2[2], pix3[2]));
257 43281144 s += abs(pix1[3] - avg2(pix2[3], pix3[3]));
258 43281144 s += abs(pix1[4] - avg2(pix2[4], pix3[4]));
259 43281144 s += abs(pix1[5] - avg2(pix2[5], pix3[5]));
260 43281144 s += abs(pix1[6] - avg2(pix2[6], pix3[6]));
261 43281144 s += abs(pix1[7] - avg2(pix2[7], pix3[7]));
262 43281144 s += abs(pix1[8] - avg2(pix2[8], pix3[8]));
263 43281144 s += abs(pix1[9] - avg2(pix2[9], pix3[9]));
264 43281144 s += abs(pix1[10] - avg2(pix2[10], pix3[10]));
265 43281144 s += abs(pix1[11] - avg2(pix2[11], pix3[11]));
266 43281144 s += abs(pix1[12] - avg2(pix2[12], pix3[12]));
267 43281144 s += abs(pix1[13] - avg2(pix2[13], pix3[13]));
268 43281144 s += abs(pix1[14] - avg2(pix2[14], pix3[14]));
269 43281144 s += abs(pix1[15] - avg2(pix2[15], pix3[15]));
270 43281144 pix1 += stride;
271 43281144 pix2 += stride;
272 43281144 pix3 += stride;
273 }
274 3286935 return s;
275 }
276
277 6573822 static int pix_abs16_xy2_c(MpegEncContext *v, const uint8_t *pix1, const uint8_t *pix2,
278 ptrdiff_t stride, int h)
279 {
280 6573822 int s = 0, i;
281 6573822 const uint8_t *pix3 = pix2 + stride;
282
283
2/2
✓ Branch 0 taken 86561688 times.
✓ Branch 1 taken 6573822 times.
 93135510 for (i = 0; i < h; i++) {
284 86561688 s += abs(pix1[0] - avg4(pix2[0], pix2[1], pix3[0], pix3[1]));
285 86561688 s += abs(pix1[1] - avg4(pix2[1], pix2[2], pix3[1], pix3[2]));
286 86561688 s += abs(pix1[2] - avg4(pix2[2], pix2[3], pix3[2], pix3[3]));
287 86561688 s += abs(pix1[3] - avg4(pix2[3], pix2[4], pix3[3], pix3[4]));
288 86561688 s += abs(pix1[4] - avg4(pix2[4], pix2[5], pix3[4], pix3[5]));
289 86561688 s += abs(pix1[5] - avg4(pix2[5], pix2[6], pix3[5], pix3[6]));
290 86561688 s += abs(pix1[6] - avg4(pix2[6], pix2[7], pix3[6], pix3[7]));
291 86561688 s += abs(pix1[7] - avg4(pix2[7], pix2[8], pix3[7], pix3[8]));
292 86561688 s += abs(pix1[8] - avg4(pix2[8], pix2[9], pix3[8], pix3[9]));
293 86561688 s += abs(pix1[9] - avg4(pix2[9], pix2[10], pix3[9], pix3[10]));
294 86561688 s += abs(pix1[10] - avg4(pix2[10], pix2[11], pix3[10], pix3[11]));
295 86561688 s += abs(pix1[11] - avg4(pix2[11], pix2[12], pix3[11], pix3[12]));
296 86561688 s += abs(pix1[12] - avg4(pix2[12], pix2[13], pix3[12], pix3[13]));
297 86561688 s += abs(pix1[13] - avg4(pix2[13], pix2[14], pix3[13], pix3[14]));
298 86561688 s += abs(pix1[14] - avg4(pix2[14], pix2[15], pix3[14], pix3[15]));
299 86561688 s += abs(pix1[15] - avg4(pix2[15], pix2[16], pix3[15], pix3[16]));
300 86561688 pix1 += stride;
301 86561688 pix2 += stride;
302 86561688 pix3 += stride;
303 }
304 6573822 return s;
305 }
306
307 33676325 static inline int pix_abs8_c(MpegEncContext *v, const uint8_t *pix1, const uint8_t *pix2,
308 ptrdiff_t stride, int h)
309 {
310 33676325 int s = 0, i;
311
312
2/2
✓ Branch 0 taken 269410786 times.
✓ Branch 1 taken 33676325 times.
 303087111 for (i = 0; i < h; i++) {
313 269410786 s += abs(pix1[0] - pix2[0]);
314 269410786 s += abs(pix1[1] - pix2[1]);
315 269410786 s += abs(pix1[2] - pix2[2]);
316 269410786 s += abs(pix1[3] - pix2[3]);
317 269410786 s += abs(pix1[4] - pix2[4]);
318 269410786 s += abs(pix1[5] - pix2[5]);
319 269410786 s += abs(pix1[6] - pix2[6]);
320 269410786 s += abs(pix1[7] - pix2[7]);
321 269410786 pix1 += stride;
322 269410786 pix2 += stride;
323 }
324 33676325 return s;
325 }
326
327 32 static inline int pix_median_abs8_c(MpegEncContext *v, const uint8_t *pix1, const uint8_t *pix2,
328 ptrdiff_t stride, int h)
329 {
330 32 int s = 0, i, j;
331
332 #define V(x) (pix1[x] - pix2[x])
333
334 32 s += abs(V(0));
335 32 s += abs(V(1) - V(0));
336 32 s += abs(V(2) - V(1));
337 32 s += abs(V(3) - V(2));
338 32 s += abs(V(4) - V(3));
339 32 s += abs(V(5) - V(4));
340 32 s += abs(V(6) - V(5));
341 32 s += abs(V(7) - V(6));
342
343 32 pix1 += stride;
344 32 pix2 += stride;
345
346
2/2
✓ Branch 0 taken 232 times.
✓ Branch 1 taken 32 times.
 264 for (i = 1; i < h; i++) {
347 232 s += abs(V(0) - V(-stride));
348
2/2
✓ Branch 0 taken 1624 times.
✓ Branch 1 taken 232 times.
 1856 for (j = 1; j < 8; j++)
349 1624 s += abs(V(j) - mid_pred(V(j-stride), V(j-1), V(j-stride) + V(j-1) - V(j-stride-1)));
350 232 pix1 += stride;
351 232 pix2 += stride;
352
353 }
354 #undef V
355 32 return s;
356 }
357
358 837597 static int pix_abs8_x2_c(MpegEncContext *v, const uint8_t *pix1, const uint8_t *pix2,
359 ptrdiff_t stride, int h)
360 {
361 837597 int s = 0, i;
362
363
2/2
✓ Branch 0 taken 6700848 times.
✓ Branch 1 taken 837597 times.
 7538445 for (i = 0; i < h; i++) {
364 6700848 s += abs(pix1[0] - avg2(pix2[0], pix2[1]));
365 6700848 s += abs(pix1[1] - avg2(pix2[1], pix2[2]));
366 6700848 s += abs(pix1[2] - avg2(pix2[2], pix2[3]));
367 6700848 s += abs(pix1[3] - avg2(pix2[3], pix2[4]));
368 6700848 s += abs(pix1[4] - avg2(pix2[4], pix2[5]));
369 6700848 s += abs(pix1[5] - avg2(pix2[5], pix2[6]));
370 6700848 s += abs(pix1[6] - avg2(pix2[6], pix2[7]));
371 6700848 s += abs(pix1[7] - avg2(pix2[7], pix2[8]));
372 6700848 pix1 += stride;
373 6700848 pix2 += stride;
374 }
375 837597 return s;
376 }
377
378 837597 static int pix_abs8_y2_c(MpegEncContext *v, const uint8_t *pix1, const uint8_t *pix2,
379 ptrdiff_t stride, int h)
380 {
381 837597 int s = 0, i;
382 837597 const uint8_t *pix3 = pix2 + stride;
383
384
2/2
✓ Branch 0 taken 6700838 times.
✓ Branch 1 taken 837597 times.
 7538435 for (i = 0; i < h; i++) {
385 6700838 s += abs(pix1[0] - avg2(pix2[0], pix3[0]));
386 6700838 s += abs(pix1[1] - avg2(pix2[1], pix3[1]));
387 6700838 s += abs(pix1[2] - avg2(pix2[2], pix3[2]));
388 6700838 s += abs(pix1[3] - avg2(pix2[3], pix3[3]));
389 6700838 s += abs(pix1[4] - avg2(pix2[4], pix3[4]));
390 6700838 s += abs(pix1[5] - avg2(pix2[5], pix3[5]));
391 6700838 s += abs(pix1[6] - avg2(pix2[6], pix3[6]));
392 6700838 s += abs(pix1[7] - avg2(pix2[7], pix3[7]));
393 6700838 pix1 += stride;
394 6700838 pix2 += stride;
395 6700838 pix3 += stride;
396 }
397 837597 return s;
398 }
399
400 1675146 static int pix_abs8_xy2_c(MpegEncContext *v, const uint8_t *pix1, const uint8_t *pix2,
401 ptrdiff_t stride, int h)
402 {
403 1675146 int s = 0, i;
404 1675146 const uint8_t *pix3 = pix2 + stride;
405
406
2/2
✓ Branch 0 taken 13401218 times.
✓ Branch 1 taken 1675146 times.
 15076364 for (i = 0; i < h; i++) {
407 13401218 s += abs(pix1[0] - avg4(pix2[0], pix2[1], pix3[0], pix3[1]));
408 13401218 s += abs(pix1[1] - avg4(pix2[1], pix2[2], pix3[1], pix3[2]));
409 13401218 s += abs(pix1[2] - avg4(pix2[2], pix2[3], pix3[2], pix3[3]));
410 13401218 s += abs(pix1[3] - avg4(pix2[3], pix2[4], pix3[3], pix3[4]));
411 13401218 s += abs(pix1[4] - avg4(pix2[4], pix2[5], pix3[4], pix3[5]));
412 13401218 s += abs(pix1[5] - avg4(pix2[5], pix2[6], pix3[5], pix3[6]));
413 13401218 s += abs(pix1[6] - avg4(pix2[6], pix2[7], pix3[6], pix3[7]));
414 13401218 s += abs(pix1[7] - avg4(pix2[7], pix2[8], pix3[7], pix3[8]));
415 13401218 pix1 += stride;
416 13401218 pix2 += stride;
417 13401218 pix3 += stride;
418 }
419 1675146 return s;
420 }
421
422 748189 static int nsse16_c(MpegEncContext *c, const uint8_t *s1, const uint8_t *s2,
423 ptrdiff_t stride, int h)
424 {
425 748189 int score1 = 0, score2 = 0, x, y;
426
427
2/2
✓ Branch 0 taken 11970726 times.
✓ Branch 1 taken 748189 times.
 12718915 for (y = 0; y < h; y++) {
428
2/2
✓ Branch 0 taken 191531616 times.
✓ Branch 1 taken 11970726 times.
 203502342 for (x = 0; x < 16; x++)
429 191531616 score1 += (s1[x] - s2[x]) * (s1[x] - s2[x]);
430
2/2
✓ Branch 0 taken 11222537 times.
✓ Branch 1 taken 748189 times.
 11970726 if (y + 1 < h) {
431
2/2
✓ Branch 0 taken 168338055 times.
✓ Branch 1 taken 11222537 times.
 179560592 for (x = 0; x < 15; x++)
432 168338055 score2 += FFABS(s1[x] - s1[x + stride] -
433 168338055 s1[x + 1] + s1[x + stride + 1]) -
434 168338055 FFABS(s2[x] - s2[x + stride] -
435 s2[x + 1] + s2[x + stride + 1]);
436 }
437 11970726 s1 += stride;
438 11970726 s2 += stride;
439 }
440
441
2/2
✓ Branch 0 taken 748141 times.
✓ Branch 1 taken 48 times.
 748189 if (c)
442 748141 return score1 + FFABS(score2) * c->avctx->nsse_weight;
443 else
444 48 return score1 + FFABS(score2) * 8;
445 }
446
447 48 static int nsse8_c(MpegEncContext *c, const uint8_t *s1, const uint8_t *s2,
448 ptrdiff_t stride, int h)
449 {
450 48 int score1 = 0, score2 = 0, x, y;
451
452
2/2
✓ Branch 0 taken 468 times.
✓ Branch 1 taken 48 times.
 516 for (y = 0; y < h; y++) {
453
2/2
✓ Branch 0 taken 3744 times.
✓ Branch 1 taken 468 times.
 4212 for (x = 0; x < 8; x++)
454 3744 score1 += (s1[x] - s2[x]) * (s1[x] - s2[x]);
455
2/2
✓ Branch 0 taken 420 times.
✓ Branch 1 taken 48 times.
 468 if (y + 1 < h) {
456
2/2
✓ Branch 0 taken 2940 times.
✓ Branch 1 taken 420 times.
 3360 for (x = 0; x < 7; x++)
457 2940 score2 += FFABS(s1[x] - s1[x + stride] -
458 2940 s1[x + 1] + s1[x + stride + 1]) -
459 2940 FFABS(s2[x] - s2[x + stride] -
460 s2[x + 1] + s2[x + stride + 1]);
461 }
462 468 s1 += stride;
463 468 s2 += stride;
464 }
465
466
1/2
✗ Branch 0 not taken.
✓ Branch 1 taken 48 times.
 48 if (c)
467 ✗ return score1 + FFABS(score2) * c->avctx->nsse_weight;
468 else
469 48 return score1 + FFABS(score2) * 8;
470 }
471
472 ✗ static int zero_cmp(MpegEncContext *s, const uint8_t *a, const uint8_t *b,
473 ptrdiff_t stride, int h)
474 {
475 ✗ return 0;
476 }
477
478 1087 av_cold int ff_set_cmp(const MECmpContext *c, me_cmp_func *cmp, int type, int mpvenc)
479 {
480 #define ENTRY(CMP_FLAG, ARRAY, MPVENC_ONLY) \
481 [FF_CMP_ ## CMP_FLAG] = { \
482 .offset = offsetof(MECmpContext, ARRAY), \
483 .mpv_only = MPVENC_ONLY, \
484 .available = 1, \
485 }
486 static const struct {
487 char available;
488 char mpv_only;
489 uint16_t offset;
490 } cmp_func_list[] = {
491 ENTRY(SAD, sad, 0),
492 ENTRY(SSE, sse, 0),
493 ENTRY(SATD, hadamard8_diff, 0),
494 ENTRY(DCT, dct_sad, 1),
495 ENTRY(PSNR, quant_psnr, 1),
496 ENTRY(BIT, bit, 1),
497 ENTRY(RD, rd, 1),
498 ENTRY(VSAD, vsad, 0),
499 ENTRY(VSSE, vsse, 0),
500 ENTRY(NSSE, nsse, 0),
501 #if CONFIG_SNOW_DECODER || CONFIG_SNOW_ENCODER
502 ENTRY(W53, w53, 0),
503 ENTRY(W97, w97, 0),
504 #endif
505 ENTRY(DCTMAX, dct_max, 1),
506 #if CONFIG_GPL
507 ENTRY(DCT264, dct264_sad, 1),
508 #endif
509 ENTRY(MEDIAN_SAD, median_sad, 0),
510 };
511 const me_cmp_func *me_cmp_func_array;
512
513 1087 type &= 0xFF;
514
515
1/2
✗ Branch 0 not taken.
✓ Branch 1 taken 1087 times.
 1087 if (type == FF_CMP_ZERO) {
516 ✗ for (int i = 0; i < 6; i++)
517 ✗ cmp[i] = zero_cmp;
518 ✗ return 0;
519 }
520
1/2
✓ Branch 0 taken 1087 times.
✗ Branch 1 not taken.
 1087 if (type >= FF_ARRAY_ELEMS(cmp_func_list) ||
521
3/4
✓ Branch 0 taken 1087 times.
✗ Branch 1 not taken.
✓ Branch 2 taken 60 times.
✓ Branch 3 taken 1027 times.
 1087 !cmp_func_list[type].available ||
522
1/2
✗ Branch 0 not taken.
✓ Branch 1 taken 60 times.
 60 !mpvenc && cmp_func_list[type].mpv_only) {
523 ✗ av_log(NULL, AV_LOG_ERROR,
524 "invalid cmp function selection\n");
525 ✗ return AVERROR(EINVAL);
526 }
527 1087 me_cmp_func_array = (const me_cmp_func*)(((const char*)c) + cmp_func_list[type].offset);
528
2/2
✓ Branch 0 taken 6522 times.
✓ Branch 1 taken 1087 times.
 7609 for (int i = 0; i < 6; i++)
529 6522 cmp[i] = me_cmp_func_array[i];
530
531 1087 return 0;
532 }
533
534 #define BUTTERFLY2(o1, o2, i1, i2) \
535 o1 = (i1) + (i2); \
536 o2 = (i1) - (i2);
537
538 #define BUTTERFLY1(x, y) \
539 { \
540 int a, b; \
541 a = x; \
542 b = y; \
543 x = a + b; \
544 y = a - b; \
545 }
546
547 #define BUTTERFLYA(x, y) (FFABS((x) + (y)) + FFABS((x) - (y)))
548
549 15480840 static int hadamard8_diff8x8_c(MpegEncContext *s, const uint8_t *dst,
550 const uint8_t *src, ptrdiff_t stride, int h)
551 {
552 15480840 int i, temp[64], sum = 0;
553
554
2/2
✓ Branch 0 taken 123846720 times.
✓ Branch 1 taken 15480840 times.
 139327560 for (i = 0; i < 8; i++) {
555 // FIXME: try pointer walks
556 123846720 BUTTERFLY2(temp[8 * i + 0], temp[8 * i + 1],
557 src[stride * i + 0] - dst[stride * i + 0],
558 src[stride * i + 1] - dst[stride * i + 1]);
559 123846720 BUTTERFLY2(temp[8 * i + 2], temp[8 * i + 3],
560 src[stride * i + 2] - dst[stride * i + 2],
561 src[stride * i + 3] - dst[stride * i + 3]);
562 123846720 BUTTERFLY2(temp[8 * i + 4], temp[8 * i + 5],
563 src[stride * i + 4] - dst[stride * i + 4],
564 src[stride * i + 5] - dst[stride * i + 5]);
565 123846720 BUTTERFLY2(temp[8 * i + 6], temp[8 * i + 7],
566 src[stride * i + 6] - dst[stride * i + 6],
567 src[stride * i + 7] - dst[stride * i + 7]);
568
569 123846720 BUTTERFLY1(temp[8 * i + 0], temp[8 * i + 2]);
570 123846720 BUTTERFLY1(temp[8 * i + 1], temp[8 * i + 3]);
571 123846720 BUTTERFLY1(temp[8 * i + 4], temp[8 * i + 6]);
572 123846720 BUTTERFLY1(temp[8 * i + 5], temp[8 * i + 7]);
573
574 123846720 BUTTERFLY1(temp[8 * i + 0], temp[8 * i + 4]);
575 123846720 BUTTERFLY1(temp[8 * i + 1], temp[8 * i + 5]);
576 123846720 BUTTERFLY1(temp[8 * i + 2], temp[8 * i + 6]);
577 123846720 BUTTERFLY1(temp[8 * i + 3], temp[8 * i + 7]);
578 }
579
580
2/2
✓ Branch 0 taken 123846720 times.
✓ Branch 1 taken 15480840 times.
 139327560 for (i = 0; i < 8; i++) {
581 123846720 BUTTERFLY1(temp[8 * 0 + i], temp[8 * 1 + i]);
582 123846720 BUTTERFLY1(temp[8 * 2 + i], temp[8 * 3 + i]);
583 123846720 BUTTERFLY1(temp[8 * 4 + i], temp[8 * 5 + i]);
584 123846720 BUTTERFLY1(temp[8 * 6 + i], temp[8 * 7 + i]);
585
586 123846720 BUTTERFLY1(temp[8 * 0 + i], temp[8 * 2 + i]);
587 123846720 BUTTERFLY1(temp[8 * 1 + i], temp[8 * 3 + i]);
588 123846720 BUTTERFLY1(temp[8 * 4 + i], temp[8 * 6 + i]);
589 123846720 BUTTERFLY1(temp[8 * 5 + i], temp[8 * 7 + i]);
590
591 123846720 sum += BUTTERFLYA(temp[8 * 0 + i], temp[8 * 4 + i]) +
592 123846720 BUTTERFLYA(temp[8 * 1 + i], temp[8 * 5 + i]) +
593 123846720 BUTTERFLYA(temp[8 * 2 + i], temp[8 * 6 + i]) +
594 123846720 BUTTERFLYA(temp[8 * 3 + i], temp[8 * 7 + i]);
595 }
596 15480840 return sum;
597 }
598
599 100 static int hadamard8_intra8x8_c(MpegEncContext *s, const uint8_t *src,
600 const uint8_t *dummy, ptrdiff_t stride, int h)
601 {
602 100 int i, temp[64], sum = 0;
603
604
2/2
✓ Branch 0 taken 800 times.
✓ Branch 1 taken 100 times.
 900 for (i = 0; i < 8; i++) {
605 // FIXME: try pointer walks
606 800 BUTTERFLY2(temp[8 * i + 0], temp[8 * i + 1],
607 src[stride * i + 0], src[stride * i + 1]);
608 800 BUTTERFLY2(temp[8 * i + 2], temp[8 * i + 3],
609 src[stride * i + 2], src[stride * i + 3]);
610 800 BUTTERFLY2(temp[8 * i + 4], temp[8 * i + 5],
611 src[stride * i + 4], src[stride * i + 5]);
612 800 BUTTERFLY2(temp[8 * i + 6], temp[8 * i + 7],
613 src[stride * i + 6], src[stride * i + 7]);
614
615 800 BUTTERFLY1(temp[8 * i + 0], temp[8 * i + 2]);
616 800 BUTTERFLY1(temp[8 * i + 1], temp[8 * i + 3]);
617 800 BUTTERFLY1(temp[8 * i + 4], temp[8 * i + 6]);
618 800 BUTTERFLY1(temp[8 * i + 5], temp[8 * i + 7]);
619
620 800 BUTTERFLY1(temp[8 * i + 0], temp[8 * i + 4]);
621 800 BUTTERFLY1(temp[8 * i + 1], temp[8 * i + 5]);
622 800 BUTTERFLY1(temp[8 * i + 2], temp[8 * i + 6]);
623 800 BUTTERFLY1(temp[8 * i + 3], temp[8 * i + 7]);
624 }
625
626
2/2
✓ Branch 0 taken 800 times.
✓ Branch 1 taken 100 times.
 900 for (i = 0; i < 8; i++) {
627 800 BUTTERFLY1(temp[8 * 0 + i], temp[8 * 1 + i]);
628 800 BUTTERFLY1(temp[8 * 2 + i], temp[8 * 3 + i]);
629 800 BUTTERFLY1(temp[8 * 4 + i], temp[8 * 5 + i]);
630 800 BUTTERFLY1(temp[8 * 6 + i], temp[8 * 7 + i]);
631
632 800 BUTTERFLY1(temp[8 * 0 + i], temp[8 * 2 + i]);
633 800 BUTTERFLY1(temp[8 * 1 + i], temp[8 * 3 + i]);
634 800 BUTTERFLY1(temp[8 * 4 + i], temp[8 * 6 + i]);
635 800 BUTTERFLY1(temp[8 * 5 + i], temp[8 * 7 + i]);
636
637 800 sum +=
638 800 BUTTERFLYA(temp[8 * 0 + i], temp[8 * 4 + i])
639 800 + BUTTERFLYA(temp[8 * 1 + i], temp[8 * 5 + i])
640 800 + BUTTERFLYA(temp[8 * 2 + i], temp[8 * 6 + i])
641 800 + BUTTERFLYA(temp[8 * 3 + i], temp[8 * 7 + i]);
642 }
643
644 100 sum -= FFABS(temp[8 * 0] + temp[8 * 4]); // -mean
645
646 100 return sum;
647 }
648
649 ✗ static int dct_sad8x8_c(MpegEncContext *s, const uint8_t *src1,
650 const uint8_t *src2, ptrdiff_t stride, int h)
651 {
652 ✗ LOCAL_ALIGNED_16(int16_t, temp, [64]);
653
654 ✗ s->pdsp.diff_pixels_unaligned(temp, src1, src2, stride);
655 ✗ s->fdsp.fdct(temp);
656 ✗ return s->sum_abs_dctelem(temp);
657 }
658
659 #if CONFIG_GPL
660 #define DCT8_1D \
661 { \
662 const int s07 = SRC(0) + SRC(7); \
663 const int s16 = SRC(1) + SRC(6); \
664 const int s25 = SRC(2) + SRC(5); \
665 const int s34 = SRC(3) + SRC(4); \
666 const int a0 = s07 + s34; \
667 const int a1 = s16 + s25; \
668 const int a2 = s07 - s34; \
669 const int a3 = s16 - s25; \
670 const int d07 = SRC(0) - SRC(7); \
671 const int d16 = SRC(1) - SRC(6); \
672 const int d25 = SRC(2) - SRC(5); \
673 const int d34 = SRC(3) - SRC(4); \
674 const int a4 = d16 + d25 + (d07 + (d07 >> 1)); \
675 const int a5 = d07 - d34 - (d25 + (d25 >> 1)); \
676 const int a6 = d07 + d34 - (d16 + (d16 >> 1)); \
677 const int a7 = d16 - d25 + (d34 + (d34 >> 1)); \
678 DST(0, a0 + a1); \
679 DST(1, a4 + (a7 >> 2)); \
680 DST(2, a2 + (a3 >> 1)); \
681 DST(3, a5 + (a6 >> 2)); \
682 DST(4, a0 - a1); \
683 DST(5, a6 - (a5 >> 2)); \
684 DST(6, (a2 >> 1) - a3); \
685 DST(7, (a4 >> 2) - a7); \
686 }
687
688 ✗ static int dct264_sad8x8_c(MpegEncContext *s, const uint8_t *src1,
689 const uint8_t *src2, ptrdiff_t stride, int h)
690 {
691 int16_t dct[8][8];
692 ✗ int i, sum = 0;
693
694 ✗ s->pdsp.diff_pixels_unaligned(dct[0], src1, src2, stride);
695
696 #define SRC(x) dct[i][x]
697 #define DST(x, v) dct[i][x] = v
698 ✗ for (i = 0; i < 8; i++)
699 ✗ DCT8_1D
700 #undef SRC
701 #undef DST
702
703 #define SRC(x) dct[x][i]
704 #define DST(x, v) sum += FFABS(v)
705 ✗ for (i = 0; i < 8; i++)
706 ✗ DCT8_1D
707 #undef SRC
708 #undef DST
709 ✗ return sum;
710 }
711 #endif
712
713 ✗ static int dct_max8x8_c(MpegEncContext *s, const uint8_t *src1,
714 const uint8_t *src2, ptrdiff_t stride, int h)
715 {
716 ✗ LOCAL_ALIGNED_16(int16_t, temp, [64]);
717 ✗ int sum = 0, i;
718
719 ✗ s->pdsp.diff_pixels_unaligned(temp, src1, src2, stride);
720 ✗ s->fdsp.fdct(temp);
721
722 ✗ for (i = 0; i < 64; i++)
723 ✗ sum = FFMAX(sum, FFABS(temp[i]));
724
725 ✗ return sum;
726 }
727
728 ✗ static int quant_psnr8x8_c(MpegEncContext *s, const uint8_t *src1,
729 const uint8_t *src2, ptrdiff_t stride, int h)
730 {
731 ✗ LOCAL_ALIGNED_16(int16_t, temp, [64 * 2]);
732 ✗ int16_t *const bak = temp + 64;
733 ✗ int sum = 0, i;
734
735 ✗ s->mb_intra = 0;
736
737 ✗ s->pdsp.diff_pixels_unaligned(temp, src1, src2, stride);
738
739 ✗ memcpy(bak, temp, 64 * sizeof(int16_t));
740
741 ✗ s->block_last_index[0 /* FIXME */] =
742 ✗ s->dct_quantize(s, temp, 0 /* FIXME */, s->qscale, &i);
743 ✗ s->dct_unquantize_inter(s, temp, 0, s->qscale);
744 ✗ ff_simple_idct_int16_8bit(temp); // FIXME
745
746 ✗ for (i = 0; i < 64; i++)
747 ✗ sum += (temp[i] - bak[i]) * (temp[i] - bak[i]);
748
749 ✗ return sum;
750 }
751
752 ✗ static int rd8x8_c(MpegEncContext *s, const uint8_t *src1, const uint8_t *src2,
753 ptrdiff_t stride, int h)
754 {
755 ✗ const uint8_t *scantable = s->intra_scantable.permutated;
756 ✗ LOCAL_ALIGNED_16(int16_t, temp, [64]);
757 ✗ LOCAL_ALIGNED_16(uint8_t, lsrc1, [64]);
758 ✗ LOCAL_ALIGNED_16(uint8_t, lsrc2, [64]);
759 int i, last, run, bits, level, distortion, start_i;
760 ✗ const int esc_length = s->ac_esc_length;
761 uint8_t *length, *last_length;
762
763 ✗ copy_block8(lsrc1, src1, 8, stride, 8);
764 ✗ copy_block8(lsrc2, src2, 8, stride, 8);
765
766 ✗ s->pdsp.diff_pixels(temp, lsrc1, lsrc2, 8);
767
768 ✗ s->block_last_index[0 /* FIXME */] =
769 last =
770 ✗ s->dct_quantize(s, temp, 0 /* FIXME */, s->qscale, &i);
771
772 ✗ bits = 0;
773
774 ✗ if (s->mb_intra) {
775 ✗ start_i = 1;
776 ✗ length = s->intra_ac_vlc_length;
777 ✗ last_length = s->intra_ac_vlc_last_length;
778 ✗ bits += s->luma_dc_vlc_length[temp[0] + 256]; // FIXME: chroma
779 } else {
780 ✗ start_i = 0;
781 ✗ length = s->inter_ac_vlc_length;
782 ✗ last_length = s->inter_ac_vlc_last_length;
783 }
784
785 ✗ if (last >= start_i) {
786 ✗ run = 0;
787 ✗ for (i = start_i; i < last; i++) {
788 ✗ int j = scantable[i];
789 ✗ level = temp[j];
790
791 ✗ if (level) {
792 ✗ level += 64;
793 ✗ if ((level & (~127)) == 0)
794 ✗ bits += length[UNI_AC_ENC_INDEX(run, level)];
795 else
796 ✗ bits += esc_length;
797 ✗ run = 0;
798 } else
799 ✗ run++;
800 }
801 ✗ i = scantable[last];
802
803 ✗ level = temp[i] + 64;
804
805 av_assert2(level - 64);
806
807 ✗ if ((level & (~127)) == 0) {
808 ✗ bits += last_length[UNI_AC_ENC_INDEX(run, level)];
809 } else
810 ✗ bits += esc_length;
811 }
812
813 ✗ if (last >= 0) {
814 ✗ if (s->mb_intra)
815 ✗ s->dct_unquantize_intra(s, temp, 0, s->qscale);
816 else
817 ✗ s->dct_unquantize_inter(s, temp, 0, s->qscale);
818 }
819
820 ✗ s->idsp.idct_add(lsrc2, 8, temp);
821
822 ✗ distortion = s->sse_cmp[1](NULL, lsrc2, lsrc1, 8, 8);
823
824 ✗ return distortion + ((bits * s->qscale * s->qscale * 109 + 64) >> 7);
825 }
826
827 ✗ static int bit8x8_c(MpegEncContext *s, const uint8_t *src1, const uint8_t *src2,
828 ptrdiff_t stride, int h)
829 {
830 ✗ const uint8_t *scantable = s->intra_scantable.permutated;
831 ✗ LOCAL_ALIGNED_16(int16_t, temp, [64]);
832 int i, last, run, bits, level, start_i;
833 ✗ const int esc_length = s->ac_esc_length;
834 uint8_t *length, *last_length;
835
836 ✗ s->pdsp.diff_pixels_unaligned(temp, src1, src2, stride);
837
838 ✗ s->block_last_index[0 /* FIXME */] =
839 last =
840 ✗ s->dct_quantize(s, temp, 0 /* FIXME */, s->qscale, &i);
841
842 ✗ bits = 0;
843
844 ✗ if (s->mb_intra) {
845 ✗ start_i = 1;
846 ✗ length = s->intra_ac_vlc_length;
847 ✗ last_length = s->intra_ac_vlc_last_length;
848 ✗ bits += s->luma_dc_vlc_length[temp[0] + 256]; // FIXME: chroma
849 } else {
850 ✗ start_i = 0;
851 ✗ length = s->inter_ac_vlc_length;
852 ✗ last_length = s->inter_ac_vlc_last_length;
853 }
854
855 ✗ if (last >= start_i) {
856 ✗ run = 0;
857 ✗ for (i = start_i; i < last; i++) {
858 ✗ int j = scantable[i];
859 ✗ level = temp[j];
860
861 ✗ if (level) {
862 ✗ level += 64;
863 ✗ if ((level & (~127)) == 0)
864 ✗ bits += length[UNI_AC_ENC_INDEX(run, level)];
865 else
866 ✗ bits += esc_length;
867 ✗ run = 0;
868 } else
869 ✗ run++;
870 }
871 ✗ i = scantable[last];
872
873 ✗ level = temp[i] + 64;
874
875 av_assert2(level - 64);
876
877 ✗ if ((level & (~127)) == 0)
878 ✗ bits += last_length[UNI_AC_ENC_INDEX(run, level)];
879 else
880 ✗ bits += esc_length;
881 }
882
883 ✗ return bits;
884 }
885
886 #define VSAD_INTRA(size) \
887 static int vsad_intra ## size ## _c(MpegEncContext *c, \
888 const uint8_t *s, const uint8_t *dummy, \
889 ptrdiff_t stride, int h) \
890 { \
891 int score = 0, x, y; \
892 \
893 for (y = 1; y < h; y++) { \
894 for (x = 0; x < size; x += 4) { \
895 score += FFABS(s[x] - s[x + stride]) + \
896 FFABS(s[x + 1] - s[x + stride + 1]) + \
897 FFABS(s[x + 2] - s[x + 2 + stride]) + \
898 FFABS(s[x + 3] - s[x + 3 + stride]); \
899 } \
900 s += stride; \
901 } \
902 \
903 return score; \
904 }
905
4/4
✓ Branch 0 taken 796 times.
✓ Branch 1 taken 398 times.
✓ Branch 2 taken 398 times.
✓ Branch 3 taken 48 times.
 1242 VSAD_INTRA(8)
906
4/4
✓ Branch 0 taken 34475328 times.
✓ Branch 1 taken 8618832 times.
✓ Branch 2 taken 8618832 times.
✓ Branch 3 taken 1231246 times.
 44325406 VSAD_INTRA(16)
907
908 #define VSAD(size) \
909 static int vsad ## size ## _c(MpegEncContext *c, \
910 const uint8_t *s1, const uint8_t *s2, \
911 ptrdiff_t stride, int h) \
912 { \
913 int score = 0, x, y; \
914 \
915 for (y = 1; y < h; y++) { \
916 for (x = 0; x < size; x++) \
917 score += FFABS(s1[x] - s2[x] - s1[x + stride] + s2[x + stride]); \
918 s1 += stride; \
919 s2 += stride; \
920 } \
921 \
922 return score; \
923 }
924
4/4
✓ Branch 0 taken 2592 times.
✓ Branch 1 taken 324 times.
✓ Branch 2 taken 324 times.
✓ Branch 3 taken 32 times.
 2948 VSAD(8)
925
4/4
✓ Branch 0 taken 213086304 times.
✓ Branch 1 taken 13317894 times.
✓ Branch 2 taken 13317894 times.
✓ Branch 3 taken 1902546 times.
 228306744 VSAD(16)
926
927 #define SQ(a) ((a) * (a))
928 #define VSSE_INTRA(size) \
929 static int vsse_intra ## size ## _c(MpegEncContext *c, \
930 const uint8_t *s, const uint8_t *dummy, \
931 ptrdiff_t stride, int h) \
932 { \
933 int score = 0, x, y; \
934 \
935 for (y = 1; y < h; y++) { \
936 for (x = 0; x < size; x += 4) { \
937 score += SQ(s[x] - s[x + stride]) + \
938 SQ(s[x + 1] - s[x + stride + 1]) + \
939 SQ(s[x + 2] - s[x + stride + 2]) + \
940 SQ(s[x + 3] - s[x + stride + 3]); \
941 } \
942 s += stride; \
943 } \
944 \
945 return score; \
946 }
947
4/4
✓ Branch 0 taken 600 times.
✓ Branch 1 taken 300 times.
✓ Branch 2 taken 300 times.
✓ Branch 3 taken 32 times.
 932 VSSE_INTRA(8)
948
4/4
✓ Branch 0 taken 1120 times.
✓ Branch 1 taken 280 times.
✓ Branch 2 taken 280 times.
✓ Branch 3 taken 32 times.
 1432 VSSE_INTRA(16)
949
950 #define VSSE(size) \
951 static int vsse ## size ## _c(MpegEncContext *c, const uint8_t *s1, const uint8_t *s2, \
952 ptrdiff_t stride, int h) \
953 { \
954 int score = 0, x, y; \
955 \
956 for (y = 1; y < h; y++) { \
957 for (x = 0; x < size; x++) \
958 score += SQ(s1[x] - s2[x] - s1[x + stride] + s2[x + stride]); \
959 s1 += stride; \
960 s2 += stride; \
961 } \
962 \
963 return score; \
964 }
965
4/4
✓ Branch 0 taken 2080 times.
✓ Branch 1 taken 260 times.
✓ Branch 2 taken 260 times.
✓ Branch 3 taken 32 times.
 2372 VSSE(8)
966
4/4
✓ Branch 0 taken 3840 times.
✓ Branch 1 taken 240 times.
✓ Branch 2 taken 240 times.
✓ Branch 3 taken 32 times.
 4112 VSSE(16)
967
968 #define WRAPPER8_16_SQ(name8, name16) \
969 static int name16(MpegEncContext *s, const uint8_t *dst, const uint8_t *src, \
970 ptrdiff_t stride, int h) \
971 { \
972 int score = 0; \
973 \
974 score += name8(s, dst, src, stride, 8); \
975 score += name8(s, dst + 8, src + 8, stride, 8); \
976 if (h == 16) { \
977 dst += 8 * stride; \
978 src += 8 * stride; \
979 score += name8(s, dst, src, stride, 8); \
980 score += name8(s, dst + 8, src + 8, stride, 8); \
981 } \
982 return score; \
983 }
984
985
2/2
✓ Branch 2 taken 3613183 times.
✓ Branch 3 taken 46 times.
 3613229 WRAPPER8_16_SQ(hadamard8_diff8x8_c, hadamard8_diff16_c)
986
2/2
✓ Branch 2 taken 2 times.
✓ Branch 3 taken 30 times.
 32 WRAPPER8_16_SQ(hadamard8_intra8x8_c, hadamard8_intra16_c)
987 ✗ WRAPPER8_16_SQ(dct_sad8x8_c, dct_sad16_c)
988 #if CONFIG_GPL
989 ✗ WRAPPER8_16_SQ(dct264_sad8x8_c, dct264_sad16_c)
990 #endif
991 ✗ WRAPPER8_16_SQ(dct_max8x8_c, dct_max16_c)
992 ✗ WRAPPER8_16_SQ(quant_psnr8x8_c, quant_psnr16_c)
993 ✗ WRAPPER8_16_SQ(rd8x8_c, rd16_c)
994 ✗ WRAPPER8_16_SQ(bit8x8_c, bit16_c)
995
996 1051 av_cold void ff_me_cmp_init(MECmpContext *c, AVCodecContext *avctx)
997 {
998 1051 memset(c, 0, sizeof(*c));
999
1000 1051 c->sum_abs_dctelem = sum_abs_dctelem_c;
1001
1002 /* TODO [0] 16 [1] 8 */
1003 1051 c->pix_abs[0][0] = pix_abs16_c;
1004 1051 c->pix_abs[0][1] = pix_abs16_x2_c;
1005 1051 c->pix_abs[0][2] = pix_abs16_y2_c;
1006 1051 c->pix_abs[0][3] = pix_abs16_xy2_c;
1007 1051 c->pix_abs[1][0] = pix_abs8_c;
1008 1051 c->pix_abs[1][1] = pix_abs8_x2_c;
1009 1051 c->pix_abs[1][2] = pix_abs8_y2_c;
1010 1051 c->pix_abs[1][3] = pix_abs8_xy2_c;
1011
1012 #define SET_CMP_FUNC(name) \
1013 c->name[0] = name ## 16_c; \
1014 c->name[1] = name ## 8x8_c;
1015
1016 1051 SET_CMP_FUNC(hadamard8_diff)
1017 1051 c->hadamard8_diff[4] = hadamard8_intra16_c;
1018 1051 c->hadamard8_diff[5] = hadamard8_intra8x8_c;
1019 1051 SET_CMP_FUNC(dct_sad)
1020 1051 SET_CMP_FUNC(dct_max)
1021 #if CONFIG_GPL
1022 1051 SET_CMP_FUNC(dct264_sad)
1023 #endif
1024 1051 c->sad[0] = pix_abs16_c;
1025 1051 c->sad[1] = pix_abs8_c;
1026 1051 c->sse[0] = sse16_c;
1027 1051 c->sse[1] = sse8_c;
1028 1051 c->sse[2] = sse4_c;
1029 1051 SET_CMP_FUNC(quant_psnr)
1030 1051 SET_CMP_FUNC(rd)
1031 1051 SET_CMP_FUNC(bit)
1032 1051 c->vsad[0] = vsad16_c;
1033 1051 c->vsad[1] = vsad8_c;
1034 1051 c->vsad[4] = vsad_intra16_c;
1035 1051 c->vsad[5] = vsad_intra8_c;
1036 1051 c->vsse[0] = vsse16_c;
1037 1051 c->vsse[1] = vsse8_c;
1038 1051 c->vsse[4] = vsse_intra16_c;
1039 1051 c->vsse[5] = vsse_intra8_c;
1040 1051 c->nsse[0] = nsse16_c;
1041 1051 c->nsse[1] = nsse8_c;
1042 #if CONFIG_SNOW_DECODER || CONFIG_SNOW_ENCODER
1043 1051 ff_dsputil_init_dwt(c);
1044 #endif
1045
1046 1051 c->median_sad[0] = pix_median_abs16_c;
1047 1051 c->median_sad[1] = pix_median_abs8_c;
1048
1049 #if ARCH_AARCH64
1050 ff_me_cmp_init_aarch64(c, avctx);
1051 #elif ARCH_ARM
1052 ff_me_cmp_init_arm(c, avctx);
1053 #elif ARCH_PPC
1054 ff_me_cmp_init_ppc(c, avctx);
1055 #elif ARCH_RISCV
1056 ff_me_cmp_init_riscv(c, avctx);
1057 #elif ARCH_X86
1058 1051 ff_me_cmp_init_x86(c, avctx);
1059 #elif ARCH_MIPS
1060 ff_me_cmp_init_mips(c, avctx);
1061 #endif
1062
1063 1051 }
1064