FFmpeg coverage


Directory: ../../../ffmpeg/
File: src/libavcodec/me_cmp.c
Date: 2022-07-07 01:21:54
Exec Total Coverage
Lines: 411 567 72.5%
Branches: 101 163 62.0%

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, uint8_t *pix1, 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 256 times.
✓ Branch 1 taken 32 times.
288 for (i = 0; i < h; i++) {
77 256 s += sq[pix1[0] - pix2[0]];
78 256 s += sq[pix1[1] - pix2[1]];
79 256 s += sq[pix1[2] - pix2[2]];
80 256 s += sq[pix1[3] - pix2[3]];
81 256 pix1 += stride;
82 256 pix2 += stride;
83 }
84 32 return s;
85 }
86
87 4814382 static int sse8_c(MpegEncContext *v, uint8_t *pix1, uint8_t *pix2,
88 ptrdiff_t stride, int h)
89 {
90 4814382 int s = 0, i;
91 4814382 const uint32_t *sq = ff_square_tab + 256;
92
93
2/2
✓ Branch 0 taken 38515056 times.
✓ Branch 1 taken 4814382 times.
43329438 for (i = 0; i < h; i++) {
94 38515056 s += sq[pix1[0] - pix2[0]];
95 38515056 s += sq[pix1[1] - pix2[1]];
96 38515056 s += sq[pix1[2] - pix2[2]];
97 38515056 s += sq[pix1[3] - pix2[3]];
98 38515056 s += sq[pix1[4] - pix2[4]];
99 38515056 s += sq[pix1[5] - pix2[5]];
100 38515056 s += sq[pix1[6] - pix2[6]];
101 38515056 s += sq[pix1[7] - pix2[7]];
102 38515056 pix1 += stride;
103 38515056 pix2 += stride;
104 }
105 4814382 return s;
106 }
107
108 7187308 static int sse16_c(MpegEncContext *v, uint8_t *pix1, uint8_t *pix2,
109 ptrdiff_t stride, int h)
110 {
111 7187308 int s = 0, i;
112 7187308 const uint32_t *sq = ff_square_tab + 256;
113
114
2/2
✓ Branch 0 taken 114996544 times.
✓ Branch 1 taken 7187308 times.
122183852 for (i = 0; i < h; i++) {
115 114996544 s += sq[pix1[0] - pix2[0]];
116 114996544 s += sq[pix1[1] - pix2[1]];
117 114996544 s += sq[pix1[2] - pix2[2]];
118 114996544 s += sq[pix1[3] - pix2[3]];
119 114996544 s += sq[pix1[4] - pix2[4]];
120 114996544 s += sq[pix1[5] - pix2[5]];
121 114996544 s += sq[pix1[6] - pix2[6]];
122 114996544 s += sq[pix1[7] - pix2[7]];
123 114996544 s += sq[pix1[8] - pix2[8]];
124 114996544 s += sq[pix1[9] - pix2[9]];
125 114996544 s += sq[pix1[10] - pix2[10]];
126 114996544 s += sq[pix1[11] - pix2[11]];
127 114996544 s += sq[pix1[12] - pix2[12]];
128 114996544 s += sq[pix1[13] - pix2[13]];
129 114996544 s += sq[pix1[14] - pix2[14]];
130 114996544 s += sq[pix1[15] - pix2[15]];
131
132 114996544 pix1 += stride;
133 114996544 pix2 += stride;
134 }
135 7187308 return s;
136 }
137
138 static int sum_abs_dctelem_c(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 39613834 static inline int pix_abs16_c(MpegEncContext *v, uint8_t *pix1, uint8_t *pix2,
151 ptrdiff_t stride, int h)
152 {
153 39613834 int s = 0, i;
154
155
2/2
✓ Branch 0 taken 553274408 times.
✓ Branch 1 taken 39613834 times.
592888242 for (i = 0; i < h; i++) {
156 553274408 s += abs(pix1[0] - pix2[0]);
157 553274408 s += abs(pix1[1] - pix2[1]);
158 553274408 s += abs(pix1[2] - pix2[2]);
159 553274408 s += abs(pix1[3] - pix2[3]);
160 553274408 s += abs(pix1[4] - pix2[4]);
161 553274408 s += abs(pix1[5] - pix2[5]);
162 553274408 s += abs(pix1[6] - pix2[6]);
163 553274408 s += abs(pix1[7] - pix2[7]);
164 553274408 s += abs(pix1[8] - pix2[8]);
165 553274408 s += abs(pix1[9] - pix2[9]);
166 553274408 s += abs(pix1[10] - pix2[10]);
167 553274408 s += abs(pix1[11] - pix2[11]);
168 553274408 s += abs(pix1[12] - pix2[12]);
169 553274408 s += abs(pix1[13] - pix2[13]);
170 553274408 s += abs(pix1[14] - pix2[14]);
171 553274408 s += abs(pix1[15] - pix2[15]);
172 553274408 pix1 += stride;
173 553274408 pix2 += stride;
174 }
175 39613834 return s;
176 }
177
178 32 static inline int pix_median_abs16_c(MpegEncContext *v, uint8_t *pix1, 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 3119040 static int pix_abs16_x2_c(MpegEncContext *v, uint8_t *pix1, uint8_t *pix2,
218 ptrdiff_t stride, int h)
219 {
220 3119040 int s = 0, i;
221
222
2/2
✓ Branch 0 taken 40592712 times.
✓ Branch 1 taken 3119040 times.
43711752 for (i = 0; i < h; i++) {
223 40592712 s += abs(pix1[0] - avg2(pix2[0], pix2[1]));
224 40592712 s += abs(pix1[1] - avg2(pix2[1], pix2[2]));
225 40592712 s += abs(pix1[2] - avg2(pix2[2], pix2[3]));
226 40592712 s += abs(pix1[3] - avg2(pix2[3], pix2[4]));
227 40592712 s += abs(pix1[4] - avg2(pix2[4], pix2[5]));
228 40592712 s += abs(pix1[5] - avg2(pix2[5], pix2[6]));
229 40592712 s += abs(pix1[6] - avg2(pix2[6], pix2[7]));
230 40592712 s += abs(pix1[7] - avg2(pix2[7], pix2[8]));
231 40592712 s += abs(pix1[8] - avg2(pix2[8], pix2[9]));
232 40592712 s += abs(pix1[9] - avg2(pix2[9], pix2[10]));
233 40592712 s += abs(pix1[10] - avg2(pix2[10], pix2[11]));
234 40592712 s += abs(pix1[11] - avg2(pix2[11], pix2[12]));
235 40592712 s += abs(pix1[12] - avg2(pix2[12], pix2[13]));
236 40592712 s += abs(pix1[13] - avg2(pix2[13], pix2[14]));
237 40592712 s += abs(pix1[14] - avg2(pix2[14], pix2[15]));
238 40592712 s += abs(pix1[15] - avg2(pix2[15], pix2[16]));
239 40592712 pix1 += stride;
240 40592712 pix2 += stride;
241 }
242 3119040 return s;
243 }
244
245 3119040 static int pix_abs16_y2_c(MpegEncContext *v, uint8_t *pix1, uint8_t *pix2,
246 ptrdiff_t stride, int h)
247 {
248 3119040 int s = 0, i;
249 3119040 uint8_t *pix3 = pix2 + stride;
250
251
2/2
✓ Branch 0 taken 40592712 times.
✓ Branch 1 taken 3119040 times.
43711752 for (i = 0; i < h; i++) {
252 40592712 s += abs(pix1[0] - avg2(pix2[0], pix3[0]));
253 40592712 s += abs(pix1[1] - avg2(pix2[1], pix3[1]));
254 40592712 s += abs(pix1[2] - avg2(pix2[2], pix3[2]));
255 40592712 s += abs(pix1[3] - avg2(pix2[3], pix3[3]));
256 40592712 s += abs(pix1[4] - avg2(pix2[4], pix3[4]));
257 40592712 s += abs(pix1[5] - avg2(pix2[5], pix3[5]));
258 40592712 s += abs(pix1[6] - avg2(pix2[6], pix3[6]));
259 40592712 s += abs(pix1[7] - avg2(pix2[7], pix3[7]));
260 40592712 s += abs(pix1[8] - avg2(pix2[8], pix3[8]));
261 40592712 s += abs(pix1[9] - avg2(pix2[9], pix3[9]));
262 40592712 s += abs(pix1[10] - avg2(pix2[10], pix3[10]));
263 40592712 s += abs(pix1[11] - avg2(pix2[11], pix3[11]));
264 40592712 s += abs(pix1[12] - avg2(pix2[12], pix3[12]));
265 40592712 s += abs(pix1[13] - avg2(pix2[13], pix3[13]));
266 40592712 s += abs(pix1[14] - avg2(pix2[14], pix3[14]));
267 40592712 s += abs(pix1[15] - avg2(pix2[15], pix3[15]));
268 40592712 pix1 += stride;
269 40592712 pix2 += stride;
270 40592712 pix3 += stride;
271 }
272 3119040 return s;
273 }
274
275 6238032 static int pix_abs16_xy2_c(MpegEncContext *v, uint8_t *pix1, uint8_t *pix2,
276 ptrdiff_t stride, int h)
277 {
278 6238032 int s = 0, i;
279 6238032 uint8_t *pix3 = pix2 + stride;
280
281
2/2
✓ Branch 0 taken 81185040 times.
✓ Branch 1 taken 6238032 times.
87423072 for (i = 0; i < h; i++) {
282 81185040 s += abs(pix1[0] - avg4(pix2[0], pix2[1], pix3[0], pix3[1]));
283 81185040 s += abs(pix1[1] - avg4(pix2[1], pix2[2], pix3[1], pix3[2]));
284 81185040 s += abs(pix1[2] - avg4(pix2[2], pix2[3], pix3[2], pix3[3]));
285 81185040 s += abs(pix1[3] - avg4(pix2[3], pix2[4], pix3[3], pix3[4]));
286 81185040 s += abs(pix1[4] - avg4(pix2[4], pix2[5], pix3[4], pix3[5]));
287 81185040 s += abs(pix1[5] - avg4(pix2[5], pix2[6], pix3[5], pix3[6]));
288 81185040 s += abs(pix1[6] - avg4(pix2[6], pix2[7], pix3[6], pix3[7]));
289 81185040 s += abs(pix1[7] - avg4(pix2[7], pix2[8], pix3[7], pix3[8]));
290 81185040 s += abs(pix1[8] - avg4(pix2[8], pix2[9], pix3[8], pix3[9]));
291 81185040 s += abs(pix1[9] - avg4(pix2[9], pix2[10], pix3[9], pix3[10]));
292 81185040 s += abs(pix1[10] - avg4(pix2[10], pix2[11], pix3[10], pix3[11]));
293 81185040 s += abs(pix1[11] - avg4(pix2[11], pix2[12], pix3[11], pix3[12]));
294 81185040 s += abs(pix1[12] - avg4(pix2[12], pix2[13], pix3[12], pix3[13]));
295 81185040 s += abs(pix1[13] - avg4(pix2[13], pix2[14], pix3[13], pix3[14]));
296 81185040 s += abs(pix1[14] - avg4(pix2[14], pix2[15], pix3[14], pix3[15]));
297 81185040 s += abs(pix1[15] - avg4(pix2[15], pix2[16], pix3[15], pix3[16]));
298 81185040 pix1 += stride;
299 81185040 pix2 += stride;
300 81185040 pix3 += stride;
301 }
302 6238032 return s;
303 }
304
305 27725567 static inline int pix_abs8_c(MpegEncContext *v, uint8_t *pix1, uint8_t *pix2,
306 ptrdiff_t stride, int h)
307 {
308 27725567 int s = 0, i;
309
310
2/2
✓ Branch 0 taken 221804536 times.
✓ Branch 1 taken 27725567 times.
249530103 for (i = 0; i < h; i++) {
311 221804536 s += abs(pix1[0] - pix2[0]);
312 221804536 s += abs(pix1[1] - pix2[1]);
313 221804536 s += abs(pix1[2] - pix2[2]);
314 221804536 s += abs(pix1[3] - pix2[3]);
315 221804536 s += abs(pix1[4] - pix2[4]);
316 221804536 s += abs(pix1[5] - pix2[5]);
317 221804536 s += abs(pix1[6] - pix2[6]);
318 221804536 s += abs(pix1[7] - pix2[7]);
319 221804536 pix1 += stride;
320 221804536 pix2 += stride;
321 }
322 27725567 return s;
323 }
324
325 32 static inline int pix_median_abs8_c(MpegEncContext *v, uint8_t *pix1, 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 224 times.
✓ Branch 1 taken 32 times.
256 for (i = 1; i < h; i++) {
345 224 s += abs(V(0) - V(-stride));
346
2/2
✓ Branch 0 taken 1568 times.
✓ Branch 1 taken 224 times.
1792 for (j = 1; j < 8; j++)
347 1568 s += abs(V(j) - mid_pred(V(j-stride), V(j-1), V(j-stride) + V(j-1) - V(j-stride-1)));
348 224 pix1 += stride;
349 224 pix2 += stride;
350
351 }
352 #undef V
353 32 return s;
354 }
355
356 837597 static int pix_abs8_x2_c(MpegEncContext *v, uint8_t *pix1, uint8_t *pix2,
357 ptrdiff_t stride, int h)
358 {
359 837597 int s = 0, i;
360
361
2/2
✓ Branch 0 taken 6700776 times.
✓ Branch 1 taken 837597 times.
7538373 for (i = 0; i < h; i++) {
362 6700776 s += abs(pix1[0] - avg2(pix2[0], pix2[1]));
363 6700776 s += abs(pix1[1] - avg2(pix2[1], pix2[2]));
364 6700776 s += abs(pix1[2] - avg2(pix2[2], pix2[3]));
365 6700776 s += abs(pix1[3] - avg2(pix2[3], pix2[4]));
366 6700776 s += abs(pix1[4] - avg2(pix2[4], pix2[5]));
367 6700776 s += abs(pix1[5] - avg2(pix2[5], pix2[6]));
368 6700776 s += abs(pix1[6] - avg2(pix2[6], pix2[7]));
369 6700776 s += abs(pix1[7] - avg2(pix2[7], pix2[8]));
370 6700776 pix1 += stride;
371 6700776 pix2 += stride;
372 }
373 837597 return s;
374 }
375
376 837597 static int pix_abs8_y2_c(MpegEncContext *v, uint8_t *pix1, uint8_t *pix2,
377 ptrdiff_t stride, int h)
378 {
379 837597 int s = 0, i;
380 837597 uint8_t *pix3 = pix2 + stride;
381
382
2/2
✓ Branch 0 taken 6700776 times.
✓ Branch 1 taken 837597 times.
7538373 for (i = 0; i < h; i++) {
383 6700776 s += abs(pix1[0] - avg2(pix2[0], pix3[0]));
384 6700776 s += abs(pix1[1] - avg2(pix2[1], pix3[1]));
385 6700776 s += abs(pix1[2] - avg2(pix2[2], pix3[2]));
386 6700776 s += abs(pix1[3] - avg2(pix2[3], pix3[3]));
387 6700776 s += abs(pix1[4] - avg2(pix2[4], pix3[4]));
388 6700776 s += abs(pix1[5] - avg2(pix2[5], pix3[5]));
389 6700776 s += abs(pix1[6] - avg2(pix2[6], pix3[6]));
390 6700776 s += abs(pix1[7] - avg2(pix2[7], pix3[7]));
391 6700776 pix1 += stride;
392 6700776 pix2 += stride;
393 6700776 pix3 += stride;
394 }
395 837597 return s;
396 }
397
398 1675146 static int pix_abs8_xy2_c(MpegEncContext *v, uint8_t *pix1, uint8_t *pix2,
399 ptrdiff_t stride, int h)
400 {
401 1675146 int s = 0, i;
402 1675146 uint8_t *pix3 = pix2 + stride;
403
404
2/2
✓ Branch 0 taken 13401168 times.
✓ Branch 1 taken 1675146 times.
15076314 for (i = 0; i < h; i++) {
405 13401168 s += abs(pix1[0] - avg4(pix2[0], pix2[1], pix3[0], pix3[1]));
406 13401168 s += abs(pix1[1] - avg4(pix2[1], pix2[2], pix3[1], pix3[2]));
407 13401168 s += abs(pix1[2] - avg4(pix2[2], pix2[3], pix3[2], pix3[3]));
408 13401168 s += abs(pix1[3] - avg4(pix2[3], pix2[4], pix3[3], pix3[4]));
409 13401168 s += abs(pix1[4] - avg4(pix2[4], pix2[5], pix3[4], pix3[5]));
410 13401168 s += abs(pix1[5] - avg4(pix2[5], pix2[6], pix3[5], pix3[6]));
411 13401168 s += abs(pix1[6] - avg4(pix2[6], pix2[7], pix3[6], pix3[7]));
412 13401168 s += abs(pix1[7] - avg4(pix2[7], pix2[8], pix3[7], pix3[8]));
413 13401168 pix1 += stride;
414 13401168 pix2 += stride;
415 13401168 pix3 += stride;
416 }
417 1675146 return s;
418 }
419
420 748189 static int nsse16_c(MpegEncContext *c, uint8_t *s1, 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 11970640 times.
✓ Branch 1 taken 748189 times.
12718829 for (y = 0; y < h; y++) {
426
2/2
✓ Branch 0 taken 191530240 times.
✓ Branch 1 taken 11970640 times.
203500880 for (x = 0; x < 16; x++)
427 191530240 score1 += (s1[x] - s2[x]) * (s1[x] - s2[x]);
428
2/2
✓ Branch 0 taken 11222451 times.
✓ Branch 1 taken 748189 times.
11970640 if (y + 1 < h) {
429
2/2
✓ Branch 0 taken 168336765 times.
✓ Branch 1 taken 11222451 times.
179559216 for (x = 0; x < 15; x++)
430 168336765 score2 += FFABS(s1[x] - s1[x + stride] -
431 168336765 s1[x + 1] + s1[x + stride + 1]) -
432 168336765 FFABS(s2[x] - s2[x + stride] -
433 s2[x + 1] + s2[x + stride + 1]);
434 }
435 11970640 s1 += stride;
436 11970640 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, uint8_t *s1, 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 384 times.
✓ Branch 1 taken 48 times.
432 for (y = 0; y < h; y++) {
451
2/2
✓ Branch 0 taken 3072 times.
✓ Branch 1 taken 384 times.
3456 for (x = 0; x < 8; x++)
452 3072 score1 += (s1[x] - s2[x]) * (s1[x] - s2[x]);
453
2/2
✓ Branch 0 taken 336 times.
✓ Branch 1 taken 48 times.
384 if (y + 1 < h) {
454
2/2
✓ Branch 0 taken 2352 times.
✓ Branch 1 taken 336 times.
2688 for (x = 0; x < 7; x++)
455 2352 score2 += FFABS(s1[x] - s1[x + stride] -
456 2352 s1[x + 1] + s1[x + stride + 1]) -
457 2352 FFABS(s2[x] - s2[x + stride] -
458 s2[x + 1] + s2[x + stride + 1]);
459 }
460 384 s1 += stride;
461 384 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, uint8_t *a, uint8_t *b,
471 ptrdiff_t stride, int h)
472 {
473 return 0;
474 }
475
476 38950 void ff_set_cmp(MECmpContext *c, me_cmp_func *cmp, int type)
477 {
478 int i;
479
480 38950 memset(cmp, 0, sizeof(void *) * 6);
481
482
2/2
✓ Branch 0 taken 233700 times.
✓ Branch 1 taken 38950 times.
272650 for (i = 0; i < 6; i++) {
483
7/17
✓ Branch 0 taken 213612 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 1098 times.
✗ Branch 7 not taken.
✗ Branch 8 not taken.
✗ Branch 9 not taken.
✓ Branch 10 taken 1254 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.
233700 switch (type & 0xFF) {
484 213612 case FF_CMP_SAD:
485 213612 cmp[i] = c->sad[i];
486 213612 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 1098 case FF_CMP_DCTMAX:
503 1098 cmp[i] = c->dct_max[i];
504 1098 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 1254 case FF_CMP_VSAD:
515 1254 cmp[i] = c->vsad[i];
516 1254 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 38950 }
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 15481940 static int hadamard8_diff8x8_c(MpegEncContext *s, uint8_t *dst,
557 uint8_t *src, ptrdiff_t stride, int h)
558 {
559 15481940 int i, temp[64], sum = 0;
560
561 av_assert2(h == 8);
562
563
2/2
✓ Branch 0 taken 123855520 times.
✓ Branch 1 taken 15481940 times.
139337460 for (i = 0; i < 8; i++) {
564 // FIXME: try pointer walks
565 123855520 BUTTERFLY2(temp[8 * i + 0], temp[8 * i + 1],
566 src[stride * i + 0] - dst[stride * i + 0],
567 src[stride * i + 1] - dst[stride * i + 1]);
568 123855520 BUTTERFLY2(temp[8 * i + 2], temp[8 * i + 3],
569 src[stride * i + 2] - dst[stride * i + 2],
570 src[stride * i + 3] - dst[stride * i + 3]);
571 123855520 BUTTERFLY2(temp[8 * i + 4], temp[8 * i + 5],
572 src[stride * i + 4] - dst[stride * i + 4],
573 src[stride * i + 5] - dst[stride * i + 5]);
574 123855520 BUTTERFLY2(temp[8 * i + 6], temp[8 * i + 7],
575 src[stride * i + 6] - dst[stride * i + 6],
576 src[stride * i + 7] - dst[stride * i + 7]);
577
578 123855520 BUTTERFLY1(temp[8 * i + 0], temp[8 * i + 2]);
579 123855520 BUTTERFLY1(temp[8 * i + 1], temp[8 * i + 3]);
580 123855520 BUTTERFLY1(temp[8 * i + 4], temp[8 * i + 6]);
581 123855520 BUTTERFLY1(temp[8 * i + 5], temp[8 * i + 7]);
582
583 123855520 BUTTERFLY1(temp[8 * i + 0], temp[8 * i + 4]);
584 123855520 BUTTERFLY1(temp[8 * i + 1], temp[8 * i + 5]);
585 123855520 BUTTERFLY1(temp[8 * i + 2], temp[8 * i + 6]);
586 123855520 BUTTERFLY1(temp[8 * i + 3], temp[8 * i + 7]);
587 }
588
589
2/2
✓ Branch 0 taken 123855520 times.
✓ Branch 1 taken 15481940 times.
139337460 for (i = 0; i < 8; i++) {
590 123855520 BUTTERFLY1(temp[8 * 0 + i], temp[8 * 1 + i]);
591 123855520 BUTTERFLY1(temp[8 * 2 + i], temp[8 * 3 + i]);
592 123855520 BUTTERFLY1(temp[8 * 4 + i], temp[8 * 5 + i]);
593 123855520 BUTTERFLY1(temp[8 * 6 + i], temp[8 * 7 + i]);
594
595 123855520 BUTTERFLY1(temp[8 * 0 + i], temp[8 * 2 + i]);
596 123855520 BUTTERFLY1(temp[8 * 1 + i], temp[8 * 3 + i]);
597 123855520 BUTTERFLY1(temp[8 * 4 + i], temp[8 * 6 + i]);
598 123855520 BUTTERFLY1(temp[8 * 5 + i], temp[8 * 7 + i]);
599
600 123855520 sum += BUTTERFLYA(temp[8 * 0 + i], temp[8 * 4 + i]) +
601 123855520 BUTTERFLYA(temp[8 * 1 + i], temp[8 * 5 + i]) +
602 123855520 BUTTERFLYA(temp[8 * 2 + i], temp[8 * 6 + i]) +
603 123855520 BUTTERFLYA(temp[8 * 3 + i], temp[8 * 7 + i]);
604 }
605 15481940 return sum;
606 }
607
608 96 static int hadamard8_intra8x8_c(MpegEncContext *s, uint8_t *src,
609 uint8_t *dummy, ptrdiff_t stride, int h)
610 {
611 96 int i, temp[64], sum = 0;
612
613 av_assert2(h == 8);
614
615
2/2
✓ Branch 0 taken 768 times.
✓ Branch 1 taken 96 times.
864 for (i = 0; i < 8; i++) {
616 // FIXME: try pointer walks
617 768 BUTTERFLY2(temp[8 * i + 0], temp[8 * i + 1],
618 src[stride * i + 0], src[stride * i + 1]);
619 768 BUTTERFLY2(temp[8 * i + 2], temp[8 * i + 3],
620 src[stride * i + 2], src[stride * i + 3]);
621 768 BUTTERFLY2(temp[8 * i + 4], temp[8 * i + 5],
622 src[stride * i + 4], src[stride * i + 5]);
623 768 BUTTERFLY2(temp[8 * i + 6], temp[8 * i + 7],
624 src[stride * i + 6], src[stride * i + 7]);
625
626 768 BUTTERFLY1(temp[8 * i + 0], temp[8 * i + 2]);
627 768 BUTTERFLY1(temp[8 * i + 1], temp[8 * i + 3]);
628 768 BUTTERFLY1(temp[8 * i + 4], temp[8 * i + 6]);
629 768 BUTTERFLY1(temp[8 * i + 5], temp[8 * i + 7]);
630
631 768 BUTTERFLY1(temp[8 * i + 0], temp[8 * i + 4]);
632 768 BUTTERFLY1(temp[8 * i + 1], temp[8 * i + 5]);
633 768 BUTTERFLY1(temp[8 * i + 2], temp[8 * i + 6]);
634 768 BUTTERFLY1(temp[8 * i + 3], temp[8 * i + 7]);
635 }
636
637
2/2
✓ Branch 0 taken 768 times.
✓ Branch 1 taken 96 times.
864 for (i = 0; i < 8; i++) {
638 768 BUTTERFLY1(temp[8 * 0 + i], temp[8 * 1 + i]);
639 768 BUTTERFLY1(temp[8 * 2 + i], temp[8 * 3 + i]);
640 768 BUTTERFLY1(temp[8 * 4 + i], temp[8 * 5 + i]);
641 768 BUTTERFLY1(temp[8 * 6 + i], temp[8 * 7 + i]);
642
643 768 BUTTERFLY1(temp[8 * 0 + i], temp[8 * 2 + i]);
644 768 BUTTERFLY1(temp[8 * 1 + i], temp[8 * 3 + i]);
645 768 BUTTERFLY1(temp[8 * 4 + i], temp[8 * 6 + i]);
646 768 BUTTERFLY1(temp[8 * 5 + i], temp[8 * 7 + i]);
647
648 768 sum +=
649 768 BUTTERFLYA(temp[8 * 0 + i], temp[8 * 4 + i])
650 768 + BUTTERFLYA(temp[8 * 1 + i], temp[8 * 5 + i])
651 768 + BUTTERFLYA(temp[8 * 2 + i], temp[8 * 6 + i])
652 768 + BUTTERFLYA(temp[8 * 3 + i], temp[8 * 7 + i]);
653 }
654
655 96 sum -= FFABS(temp[8 * 0] + temp[8 * 4]); // -mean
656
657 96 return sum;
658 }
659
660 static int dct_sad8x8_c(MpegEncContext *s, uint8_t *src1,
661 uint8_t *src2, ptrdiff_t stride, int h)
662 {
663 LOCAL_ALIGNED_16(int16_t, temp, [64]);
664
665 av_assert2(h == 8);
666
667 s->pdsp.diff_pixels_unaligned(temp, src1, src2, stride);
668 s->fdsp.fdct(temp);
669 return s->mecc.sum_abs_dctelem(temp);
670 }
671
672 #if CONFIG_GPL
673 #define DCT8_1D \
674 { \
675 const int s07 = SRC(0) + SRC(7); \
676 const int s16 = SRC(1) + SRC(6); \
677 const int s25 = SRC(2) + SRC(5); \
678 const int s34 = SRC(3) + SRC(4); \
679 const int a0 = s07 + s34; \
680 const int a1 = s16 + s25; \
681 const int a2 = s07 - s34; \
682 const int a3 = s16 - s25; \
683 const int d07 = SRC(0) - SRC(7); \
684 const int d16 = SRC(1) - SRC(6); \
685 const int d25 = SRC(2) - SRC(5); \
686 const int d34 = SRC(3) - SRC(4); \
687 const int a4 = d16 + d25 + (d07 + (d07 >> 1)); \
688 const int a5 = d07 - d34 - (d25 + (d25 >> 1)); \
689 const int a6 = d07 + d34 - (d16 + (d16 >> 1)); \
690 const int a7 = d16 - d25 + (d34 + (d34 >> 1)); \
691 DST(0, a0 + a1); \
692 DST(1, a4 + (a7 >> 2)); \
693 DST(2, a2 + (a3 >> 1)); \
694 DST(3, a5 + (a6 >> 2)); \
695 DST(4, a0 - a1); \
696 DST(5, a6 - (a5 >> 2)); \
697 DST(6, (a2 >> 1) - a3); \
698 DST(7, (a4 >> 2) - a7); \
699 }
700
701 static int dct264_sad8x8_c(MpegEncContext *s, uint8_t *src1,
702 uint8_t *src2, ptrdiff_t stride, int h)
703 {
704 int16_t dct[8][8];
705 int i, sum = 0;
706
707 s->pdsp.diff_pixels_unaligned(dct[0], src1, src2, stride);
708
709 #define SRC(x) dct[i][x]
710 #define DST(x, v) dct[i][x] = v
711 for (i = 0; i < 8; i++)
712 DCT8_1D
713 #undef SRC
714 #undef DST
715
716 #define SRC(x) dct[x][i]
717 #define DST(x, v) sum += FFABS(v)
718 for (i = 0; i < 8; i++)
719 DCT8_1D
720 #undef SRC
721 #undef DST
722 return sum;
723 }
724 #endif
725
726 static int dct_max8x8_c(MpegEncContext *s, uint8_t *src1,
727 uint8_t *src2, ptrdiff_t stride, int h)
728 {
729 LOCAL_ALIGNED_16(int16_t, temp, [64]);
730 int sum = 0, i;
731
732 av_assert2(h == 8);
733
734 s->pdsp.diff_pixels_unaligned(temp, src1, src2, stride);
735 s->fdsp.fdct(temp);
736
737 for (i = 0; i < 64; i++)
738 sum = FFMAX(sum, FFABS(temp[i]));
739
740 return sum;
741 }
742
743 static int quant_psnr8x8_c(MpegEncContext *s, uint8_t *src1,
744 uint8_t *src2, ptrdiff_t stride, int h)
745 {
746 LOCAL_ALIGNED_16(int16_t, temp, [64 * 2]);
747 int16_t *const bak = temp + 64;
748 int sum = 0, i;
749
750 av_assert2(h == 8);
751 s->mb_intra = 0;
752
753 s->pdsp.diff_pixels_unaligned(temp, src1, src2, stride);
754
755 memcpy(bak, temp, 64 * sizeof(int16_t));
756
757 s->block_last_index[0 /* FIXME */] =
758 s->fast_dct_quantize(s, temp, 0 /* FIXME */, s->qscale, &i);
759 s->dct_unquantize_inter(s, temp, 0, s->qscale);
760 ff_simple_idct_int16_8bit(temp); // FIXME
761
762 for (i = 0; i < 64; i++)
763 sum += (temp[i] - bak[i]) * (temp[i] - bak[i]);
764
765 return sum;
766 }
767
768 static int rd8x8_c(MpegEncContext *s, uint8_t *src1, uint8_t *src2,
769 ptrdiff_t stride, int h)
770 {
771 const uint8_t *scantable = s->intra_scantable.permutated;
772 LOCAL_ALIGNED_16(int16_t, temp, [64]);
773 LOCAL_ALIGNED_16(uint8_t, lsrc1, [64]);
774 LOCAL_ALIGNED_16(uint8_t, lsrc2, [64]);
775 int i, last, run, bits, level, distortion, start_i;
776 const int esc_length = s->ac_esc_length;
777 uint8_t *length, *last_length;
778
779 av_assert2(h == 8);
780
781 copy_block8(lsrc1, src1, 8, stride, 8);
782 copy_block8(lsrc2, src2, 8, stride, 8);
783
784 s->pdsp.diff_pixels(temp, lsrc1, lsrc2, 8);
785
786 s->block_last_index[0 /* FIXME */] =
787 last =
788 s->fast_dct_quantize(s, temp, 0 /* FIXME */, s->qscale, &i);
789
790 bits = 0;
791
792 if (s->mb_intra) {
793 start_i = 1;
794 length = s->intra_ac_vlc_length;
795 last_length = s->intra_ac_vlc_last_length;
796 bits += s->luma_dc_vlc_length[temp[0] + 256]; // FIXME: chroma
797 } else {
798 start_i = 0;
799 length = s->inter_ac_vlc_length;
800 last_length = s->inter_ac_vlc_last_length;
801 }
802
803 if (last >= start_i) {
804 run = 0;
805 for (i = start_i; i < last; i++) {
806 int j = scantable[i];
807 level = temp[j];
808
809 if (level) {
810 level += 64;
811 if ((level & (~127)) == 0)
812 bits += length[UNI_AC_ENC_INDEX(run, level)];
813 else
814 bits += esc_length;
815 run = 0;
816 } else
817 run++;
818 }
819 i = scantable[last];
820
821 level = temp[i] + 64;
822
823 av_assert2(level - 64);
824
825 if ((level & (~127)) == 0) {
826 bits += last_length[UNI_AC_ENC_INDEX(run, level)];
827 } else
828 bits += esc_length;
829 }
830
831 if (last >= 0) {
832 if (s->mb_intra)
833 s->dct_unquantize_intra(s, temp, 0, s->qscale);
834 else
835 s->dct_unquantize_inter(s, temp, 0, s->qscale);
836 }
837
838 s->idsp.idct_add(lsrc2, 8, temp);
839
840 distortion = s->mecc.sse[1](NULL, lsrc2, lsrc1, 8, 8);
841
842 return distortion + ((bits * s->qscale * s->qscale * 109 + 64) >> 7);
843 }
844
845 static int bit8x8_c(MpegEncContext *s, uint8_t *src1, uint8_t *src2,
846 ptrdiff_t stride, int h)
847 {
848 const uint8_t *scantable = s->intra_scantable.permutated;
849 LOCAL_ALIGNED_16(int16_t, temp, [64]);
850 int i, last, run, bits, level, start_i;
851 const int esc_length = s->ac_esc_length;
852 uint8_t *length, *last_length;
853
854 av_assert2(h == 8);
855
856 s->pdsp.diff_pixels_unaligned(temp, src1, src2, stride);
857
858 s->block_last_index[0 /* FIXME */] =
859 last =
860 s->fast_dct_quantize(s, temp, 0 /* FIXME */, s->qscale, &i);
861
862 bits = 0;
863
864 if (s->mb_intra) {
865 start_i = 1;
866 length = s->intra_ac_vlc_length;
867 last_length = s->intra_ac_vlc_last_length;
868 bits += s->luma_dc_vlc_length[temp[0] + 256]; // FIXME: chroma
869 } else {
870 start_i = 0;
871 length = s->inter_ac_vlc_length;
872 last_length = s->inter_ac_vlc_last_length;
873 }
874
875 if (last >= start_i) {
876 run = 0;
877 for (i = start_i; i < last; i++) {
878 int j = scantable[i];
879 level = temp[j];
880
881 if (level) {
882 level += 64;
883 if ((level & (~127)) == 0)
884 bits += length[UNI_AC_ENC_INDEX(run, level)];
885 else
886 bits += esc_length;
887 run = 0;
888 } else
889 run++;
890 }
891 i = scantable[last];
892
893 level = temp[i] + 64;
894
895 av_assert2(level - 64);
896
897 if ((level & (~127)) == 0)
898 bits += last_length[UNI_AC_ENC_INDEX(run, level)];
899 else
900 bits += esc_length;
901 }
902
903 return bits;
904 }
905
906 #define VSAD_INTRA(size) \
907 static int vsad_intra ## size ## _c(MpegEncContext *c, \
908 uint8_t *s, uint8_t *dummy, \
909 ptrdiff_t stride, int h) \
910 { \
911 int score = 0, x, y; \
912 \
913 for (y = 1; y < h; y++) { \
914 for (x = 0; x < size; x += 4) { \
915 score += FFABS(s[x] - s[x + stride]) + \
916 FFABS(s[x + 1] - s[x + stride + 1]) + \
917 FFABS(s[x + 2] - s[x + 2 + stride]) + \
918 FFABS(s[x + 3] - s[x + 3 + stride]); \
919 } \
920 s += stride; \
921 } \
922 \
923 return score; \
924 }
925
4/4
✓ Branch 0 taken 672 times.
✓ Branch 1 taken 336 times.
✓ Branch 2 taken 336 times.
✓ Branch 3 taken 48 times.
1056 VSAD_INTRA(8)
926
4/4
✓ Branch 0 taken 34481776 times.
✓ Branch 1 taken 8620444 times.
✓ Branch 2 taken 8620444 times.
✓ Branch 3 taken 1231492 times.
44333712 VSAD_INTRA(16)
927
928 #define VSAD(size) \
929 static int vsad ## size ## _c(MpegEncContext *c, \
930 uint8_t *s1, uint8_t *s2, \
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++) \
937 score += FFABS(s1[x] - s2[x] - s1[x + stride] + s2[x + stride]); \
938 s1 += stride; \
939 s2 += stride; \
940 } \
941 \
942 return score; \
943 }
944
4/4
✓ Branch 0 taken 1792 times.
✓ Branch 1 taken 224 times.
✓ Branch 2 taken 224 times.
✓ Branch 3 taken 32 times.
2048 VSAD(8)
945
4/4
✓ Branch 0 taken 214486048 times.
✓ Branch 1 taken 13405378 times.
✓ Branch 2 taken 13405378 times.
✓ Branch 3 taken 1915054 times.
229806480 VSAD(16)
946
947 #define SQ(a) ((a) * (a))
948 #define VSSE_INTRA(size) \
949 static int vsse_intra ## size ## _c(MpegEncContext *c, \
950 uint8_t *s, uint8_t *dummy, \
951 ptrdiff_t stride, int h) \
952 { \
953 int score = 0, x, y; \
954 \
955 for (y = 1; y < h; y++) { \
956 for (x = 0; x < size; x += 4) { \
957 score += SQ(s[x] - s[x + stride]) + \
958 SQ(s[x + 1] - s[x + stride + 1]) + \
959 SQ(s[x + 2] - s[x + stride + 2]) + \
960 SQ(s[x + 3] - s[x + stride + 3]); \
961 } \
962 s += stride; \
963 } \
964 \
965 return score; \
966 }
967
4/4
✓ Branch 0 taken 448 times.
✓ Branch 1 taken 224 times.
✓ Branch 2 taken 224 times.
✓ Branch 3 taken 32 times.
704 VSSE_INTRA(8)
968
4/4
✓ Branch 0 taken 896 times.
✓ Branch 1 taken 224 times.
✓ Branch 2 taken 224 times.
✓ Branch 3 taken 32 times.
1152 VSSE_INTRA(16)
969
970 #define VSSE(size) \
971 static int vsse ## size ## _c(MpegEncContext *c, uint8_t *s1, uint8_t *s2, \
972 ptrdiff_t stride, int h) \
973 { \
974 int score = 0, x, y; \
975 \
976 for (y = 1; y < h; y++) { \
977 for (x = 0; x < size; x++) \
978 score += SQ(s1[x] - s2[x] - s1[x + stride] + s2[x + stride]); \
979 s1 += stride; \
980 s2 += stride; \
981 } \
982 \
983 return score; \
984 }
985
4/4
✓ Branch 0 taken 1792 times.
✓ Branch 1 taken 224 times.
✓ Branch 2 taken 224 times.
✓ Branch 3 taken 32 times.
2048 VSSE(8)
986
4/4
✓ Branch 0 taken 3584 times.
✓ Branch 1 taken 224 times.
✓ Branch 2 taken 224 times.
✓ Branch 3 taken 32 times.
3840 VSSE(16)
987
988 #define WRAPPER8_16_SQ(name8, name16) \
989 static int name16(MpegEncContext *s, uint8_t *dst, uint8_t *src, \
990 ptrdiff_t stride, int h) \
991 { \
992 int score = 0; \
993 \
994 score += name8(s, dst, src, stride, 8); \
995 score += name8(s, dst + 8, src + 8, stride, 8); \
996 if (h == 16) { \
997 dst += 8 * stride; \
998 src += 8 * stride; \
999 score += name8(s, dst, src, stride, 8); \
1000 score += name8(s, dst + 8, src + 8, stride, 8); \
1001 } \
1002 return score; \
1003 }
1004
1005
2/2
✓ Branch 2 taken 3613091 times.
✓ Branch 3 taken 48 times.
3613139 WRAPPER8_16_SQ(hadamard8_diff8x8_c, hadamard8_diff16_c)
1006
1/2
✗ Branch 2 not taken.
✓ Branch 3 taken 32 times.
32 WRAPPER8_16_SQ(hadamard8_intra8x8_c, hadamard8_intra16_c)
1007 WRAPPER8_16_SQ(dct_sad8x8_c, dct_sad16_c)
1008 #if CONFIG_GPL
1009 WRAPPER8_16_SQ(dct264_sad8x8_c, dct264_sad16_c)
1010 #endif
1011 WRAPPER8_16_SQ(dct_max8x8_c, dct_max16_c)
1012 WRAPPER8_16_SQ(quant_psnr8x8_c, quant_psnr16_c)
1013 WRAPPER8_16_SQ(rd8x8_c, rd16_c)
1014 WRAPPER8_16_SQ(bit8x8_c, bit16_c)
1015
1016 1015 av_cold void ff_me_cmp_init(MECmpContext *c, AVCodecContext *avctx)
1017 {
1018 1015 c->sum_abs_dctelem = sum_abs_dctelem_c;
1019
1020 /* TODO [0] 16 [1] 8 */
1021 1015 c->pix_abs[0][0] = pix_abs16_c;
1022 1015 c->pix_abs[0][1] = pix_abs16_x2_c;
1023 1015 c->pix_abs[0][2] = pix_abs16_y2_c;
1024 1015 c->pix_abs[0][3] = pix_abs16_xy2_c;
1025 1015 c->pix_abs[1][0] = pix_abs8_c;
1026 1015 c->pix_abs[1][1] = pix_abs8_x2_c;
1027 1015 c->pix_abs[1][2] = pix_abs8_y2_c;
1028 1015 c->pix_abs[1][3] = pix_abs8_xy2_c;
1029
1030 #define SET_CMP_FUNC(name) \
1031 c->name[0] = name ## 16_c; \
1032 c->name[1] = name ## 8x8_c;
1033
1034 1015 SET_CMP_FUNC(hadamard8_diff)
1035 1015 c->hadamard8_diff[4] = hadamard8_intra16_c;
1036 1015 c->hadamard8_diff[5] = hadamard8_intra8x8_c;
1037 1015 SET_CMP_FUNC(dct_sad)
1038 1015 SET_CMP_FUNC(dct_max)
1039 #if CONFIG_GPL
1040 1015 SET_CMP_FUNC(dct264_sad)
1041 #endif
1042 1015 c->sad[0] = pix_abs16_c;
1043 1015 c->sad[1] = pix_abs8_c;
1044 1015 c->sse[0] = sse16_c;
1045 1015 c->sse[1] = sse8_c;
1046 1015 c->sse[2] = sse4_c;
1047 1015 SET_CMP_FUNC(quant_psnr)
1048 1015 SET_CMP_FUNC(rd)
1049 1015 SET_CMP_FUNC(bit)
1050 1015 c->vsad[0] = vsad16_c;
1051 1015 c->vsad[1] = vsad8_c;
1052 1015 c->vsad[4] = vsad_intra16_c;
1053 1015 c->vsad[5] = vsad_intra8_c;
1054 1015 c->vsse[0] = vsse16_c;
1055 1015 c->vsse[1] = vsse8_c;
1056 1015 c->vsse[4] = vsse_intra16_c;
1057 1015 c->vsse[5] = vsse_intra8_c;
1058 1015 c->nsse[0] = nsse16_c;
1059 1015 c->nsse[1] = nsse8_c;
1060 #if CONFIG_SNOW_DECODER || CONFIG_SNOW_ENCODER
1061 1015 ff_dsputil_init_dwt(c);
1062 #endif
1063
1064 #if ARCH_AARCH64
1065 ff_me_cmp_init_aarch64(c, avctx);
1066 #elif ARCH_ALPHA
1067 ff_me_cmp_init_alpha(c, avctx);
1068 #elif ARCH_ARM
1069 ff_me_cmp_init_arm(c, avctx);
1070 #elif ARCH_PPC
1071 ff_me_cmp_init_ppc(c, avctx);
1072 #elif ARCH_X86
1073 1015 ff_me_cmp_init_x86(c, avctx);
1074 #elif ARCH_MIPS
1075 ff_me_cmp_init_mips(c, avctx);
1076 #endif
1077
1078 1015 c->median_sad[0] = pix_median_abs16_c;
1079 1015 c->median_sad[1] = pix_median_abs8_c;
1080 1015 }
1081