FFmpeg coverage

Directory: ../../../ffmpeg/
File: src/libavcodec/h264idct_template.c
Date: 2023-12-04 05:51:44
Exec Total Coverage
Lines: 203 203 100.0%
Functions: 36 60 60.0%
Branches: 70 70 100.0%
Line Branch Exec Source
1 /*
2 * H.264 IDCT
3 * Copyright (c) 2004-2011 Michael Niedermayer <michaelni@gmx.at>
4 *
5 * This file is part of FFmpeg.
6 *
7 * FFmpeg is free software; you can redistribute it and/or
8 * modify it under the terms of the GNU Lesser General Public
9 * License as published by the Free Software Foundation; either
10 * version 2.1 of the License, or (at your option) any later version.
11 *
12 * FFmpeg is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15 * Lesser General Public License for more details.
16 *
17 * You should have received a copy of the GNU Lesser General Public
18 * License along with FFmpeg; if not, write to the Free Software
19 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
20 */
21
22 /**
23 * @file
24 * H.264 IDCT.
25 * @author Michael Niedermayer <michaelni@gmx.at>
26 */
27
28 #include "bit_depth_template.c"
29 #include "libavutil/common.h"
30 #include "h264_parse.h"
31 #include "h264idct.h"
32
33 108321602 void FUNCC(ff_h264_idct_add)(uint8_t *_dst, int16_t *_block, int stride)
34 {
35 int i;
36 108321602 pixel *dst = (pixel*)_dst;
37 108321602 dctcoef *block = (dctcoef*)_block;
38 108321602 stride >>= sizeof(pixel)-1;
39
40 108321602 block[0] += 1 << 5;
41
42
2/2
✓ Branch 0 taken 216643204 times.
✓ Branch 1 taken 54160801 times.
 541608010 for(i=0; i<4; i++){
43 433286408 const SUINT z0= block[i + 4*0] + (unsigned)block[i + 4*2];
44 433286408 const SUINT z1= block[i + 4*0] - (unsigned)block[i + 4*2];
45 433286408 const SUINT z2= (block[i + 4*1]>>1) - (unsigned)block[i + 4*3];
46 433286408 const SUINT z3= block[i + 4*1] + (unsigned)(block[i + 4*3]>>1);
47
48 433286408 block[i + 4*0]= z0 + z3;
49 433286408 block[i + 4*1]= z1 + z2;
50 433286408 block[i + 4*2]= z1 - z2;
51 433286408 block[i + 4*3]= z0 - z3;
52 }
53
54
2/2
✓ Branch 0 taken 216643204 times.
✓ Branch 1 taken 54160801 times.
 541608010 for(i=0; i<4; i++){
55 433286408 const SUINT z0= block[0 + 4*i] + (SUINT)block[2 + 4*i];
56 433286408 const SUINT z1= block[0 + 4*i] - (SUINT)block[2 + 4*i];
57 433286408 const SUINT z2= (block[1 + 4*i]>>1) - (SUINT)block[3 + 4*i];
58 433286408 const SUINT z3= block[1 + 4*i] + (SUINT)(block[3 + 4*i]>>1);
59
60 433286408 dst[i + 0*stride]= av_clip_pixel(dst[i + 0*stride] + ((int)(z0 + z3) >> 6));
61 433286408 dst[i + 1*stride]= av_clip_pixel(dst[i + 1*stride] + ((int)(z1 + z2) >> 6));
62 433286408 dst[i + 2*stride]= av_clip_pixel(dst[i + 2*stride] + ((int)(z1 - z2) >> 6));
63 433286408 dst[i + 3*stride]= av_clip_pixel(dst[i + 3*stride] + ((int)(z0 - z3) >> 6));
64 }
65
66 108321602 memset(block, 0, 16 * sizeof(dctcoef));
67 108321602 }
68
69 13497310 void FUNCC(ff_h264_idct8_add)(uint8_t *_dst, int16_t *_block, int stride){
70 int i;
71 13497310 pixel *dst = (pixel*)_dst;
72 13497310 dctcoef *block = (dctcoef*)_block;
73 13497310 stride >>= sizeof(pixel)-1;
74
75 13497310 block[0] += 32;
76
77
2/2
✓ Branch 0 taken 53989240 times.
✓ Branch 1 taken 6748655 times.
 121475790 for( i = 0; i < 8; i++ )
78 {
79 107978480 const unsigned int a0 = block[i+0*8] + (unsigned)block[i+4*8];
80 107978480 const unsigned int a2 = block[i+0*8] - (unsigned)block[i+4*8];
81 107978480 const unsigned int a4 = (block[i+2*8]>>1) - (unsigned)block[i+6*8];
82 107978480 const unsigned int a6 = (block[i+6*8]>>1) + (unsigned)block[i+2*8];
83
84 107978480 const unsigned int b0 = a0 + a6;
85 107978480 const unsigned int b2 = a2 + a4;
86 107978480 const unsigned int b4 = a2 - a4;
87 107978480 const unsigned int b6 = a0 - a6;
88
89 107978480 const int a1 = -block[i+3*8] + (unsigned)block[i+5*8] - block[i+7*8] - (block[i+7*8]>>1);
90 107978480 const int a3 = block[i+1*8] + (unsigned)block[i+7*8] - block[i+3*8] - (block[i+3*8]>>1);
91 107978480 const int a5 = -block[i+1*8] + (unsigned)block[i+7*8] + block[i+5*8] + (block[i+5*8]>>1);
92 107978480 const int a7 = block[i+3*8] + (unsigned)block[i+5*8] + block[i+1*8] + (block[i+1*8]>>1);
93
94 107978480 const int b1 = (a7>>2) + (unsigned)a1;
95 107978480 const int b3 = (unsigned)a3 + (a5>>2);
96 107978480 const int b5 = (a3>>2) - (unsigned)a5;
97 107978480 const int b7 = (unsigned)a7 - (a1>>2);
98
99 107978480 block[i+0*8] = b0 + b7;
100 107978480 block[i+7*8] = b0 - b7;
101 107978480 block[i+1*8] = b2 + b5;
102 107978480 block[i+6*8] = b2 - b5;
103 107978480 block[i+2*8] = b4 + b3;
104 107978480 block[i+5*8] = b4 - b3;
105 107978480 block[i+3*8] = b6 + b1;
106 107978480 block[i+4*8] = b6 - b1;
107 }
108
2/2
✓ Branch 0 taken 53989240 times.
✓ Branch 1 taken 6748655 times.
 121475790 for( i = 0; i < 8; i++ )
109 {
110 107978480 const unsigned a0 = block[0+i*8] + (unsigned)block[4+i*8];
111 107978480 const unsigned a2 = block[0+i*8] - (unsigned)block[4+i*8];
112 107978480 const unsigned a4 = (block[2+i*8]>>1) - (unsigned)block[6+i*8];
113 107978480 const unsigned a6 = (block[6+i*8]>>1) + (unsigned)block[2+i*8];
114
115 107978480 const unsigned b0 = a0 + a6;
116 107978480 const unsigned b2 = a2 + a4;
117 107978480 const unsigned b4 = a2 - a4;
118 107978480 const unsigned b6 = a0 - a6;
119
120 107978480 const int a1 = -(unsigned)block[3+i*8] + block[5+i*8] - block[7+i*8] - (block[7+i*8]>>1);
121 107978480 const int a3 = (unsigned)block[1+i*8] + block[7+i*8] - block[3+i*8] - (block[3+i*8]>>1);
122 107978480 const int a5 = -(unsigned)block[1+i*8] + block[7+i*8] + block[5+i*8] + (block[5+i*8]>>1);
123 107978480 const int a7 = (unsigned)block[3+i*8] + block[5+i*8] + block[1+i*8] + (block[1+i*8]>>1);
124
125 107978480 const unsigned b1 = (a7>>2) + (unsigned)a1;
126 107978480 const unsigned b3 = (unsigned)a3 + (a5>>2);
127 107978480 const unsigned b5 = (a3>>2) - (unsigned)a5;
128 107978480 const unsigned b7 = (unsigned)a7 - (a1>>2);
129
130 107978480 dst[i + 0*stride] = av_clip_pixel( dst[i + 0*stride] + ((int)(b0 + b7) >> 6) );
131 107978480 dst[i + 1*stride] = av_clip_pixel( dst[i + 1*stride] + ((int)(b2 + b5) >> 6) );
132 107978480 dst[i + 2*stride] = av_clip_pixel( dst[i + 2*stride] + ((int)(b4 + b3) >> 6) );
133 107978480 dst[i + 3*stride] = av_clip_pixel( dst[i + 3*stride] + ((int)(b6 + b1) >> 6) );
134 107978480 dst[i + 4*stride] = av_clip_pixel( dst[i + 4*stride] + ((int)(b6 - b1) >> 6) );
135 107978480 dst[i + 5*stride] = av_clip_pixel( dst[i + 5*stride] + ((int)(b4 - b3) >> 6) );
136 107978480 dst[i + 6*stride] = av_clip_pixel( dst[i + 6*stride] + ((int)(b2 - b5) >> 6) );
137 107978480 dst[i + 7*stride] = av_clip_pixel( dst[i + 7*stride] + ((int)(b0 - b7) >> 6) );
138 }
139
140 13497310 memset(block, 0, 64 * sizeof(dctcoef));
141 13497310 }
142
143 // assumes all AC coefs are 0
144 58035780 void FUNCC(ff_h264_idct_dc_add)(uint8_t *_dst, int16_t *_block, int stride){
145 int i, j;
146 58035780 pixel *dst = (pixel*)_dst;
147 58035780 dctcoef *block = (dctcoef*)_block;
148 58035780 int dc = (block[0] + 32) >> 6;
149 58035780 stride /= sizeof(pixel);
150 58035780 block[0] = 0;
151
2/2
✓ Branch 0 taken 116071560 times.
✓ Branch 1 taken 29017890 times.
 290178900 for( j = 0; j < 4; j++ )
152 {
153
2/2
✓ Branch 0 taken 464286240 times.
✓ Branch 1 taken 116071560 times.
 1160715600 for( i = 0; i < 4; i++ )
154 928572480 dst[i] = av_clip_pixel( dst[i] + dc );
155 232143120 dst += stride;
156 }
157 58035780 }
158
159 1235612 void FUNCC(ff_h264_idct8_dc_add)(uint8_t *_dst, int16_t *_block, int stride){
160 int i, j;
161 1235612 pixel *dst = (pixel*)_dst;
162 1235612 dctcoef *block = (dctcoef*)_block;
163 1235612 int dc = (block[0] + 32) >> 6;
164 1235612 block[0] = 0;
165 1235612 stride /= sizeof(pixel);
166
2/2
✓ Branch 0 taken 4942448 times.
✓ Branch 1 taken 617806 times.
 11120508 for( j = 0; j < 8; j++ )
167 {
168
2/2
✓ Branch 0 taken 39539584 times.
✓ Branch 1 taken 4942448 times.
 88964064 for( i = 0; i < 8; i++ )
169 79079168 dst[i] = av_clip_pixel( dst[i] + dc );
170 9884896 dst += stride;
171 }
172 1235612 }
173
174 9672244 void FUNCC(ff_h264_idct_add16)(uint8_t *dst, const int *block_offset,
175 int16_t *block, int stride,
176 const uint8_t nnzc[5 * 8])
177 {
178 int i;
179
2/2
✓ Branch 0 taken 77377952 times.
✓ Branch 1 taken 4836122 times.
 164428148 for(i=0; i<16; i++){
180 154755904 int nnz = nnzc[ scan8[i] ];
181
2/2
✓ Branch 0 taken 28973214 times.
✓ Branch 1 taken 48404738 times.
 154755904 if(nnz){
182
4/4
✓ Branch 0 taken 13104848 times.
✓ Branch 1 taken 15868366 times.
✓ Branch 2 taken 4626326 times.
✓ Branch 3 taken 8478522 times.
 57946428 if(nnz==1 && ((dctcoef*)block)[i*16]) FUNCC(ff_h264_idct_dc_add)(dst + block_offset[i], block + i*16*sizeof(pixel), stride);
183 48693776 else FUNCC(ff_h264_idct_add )(dst + block_offset[i], block + i*16*sizeof(pixel), stride);
184 }
185 }
186 9672244 }
187
188 1869130 void FUNCC(ff_h264_idct_add16intra)(uint8_t *dst, const int *block_offset,
189 int16_t *block, int stride,
190 const uint8_t nnzc[5 * 8])
191 {
192 int i;
193
2/2
✓ Branch 0 taken 14953040 times.
✓ Branch 1 taken 934565 times.
 31775210 for(i=0; i<16; i++){
194
2/2
✓ Branch 0 taken 3741814 times.
✓ Branch 1 taken 11211226 times.
 29906080 if(nnzc[ scan8[i] ]) FUNCC(ff_h264_idct_add )(dst + block_offset[i], block + i*16*sizeof(pixel), stride);
195
2/2
✓ Branch 0 taken 4986491 times.
✓ Branch 1 taken 6224735 times.
 22422452 else if(((dctcoef*)block)[i*16]) FUNCC(ff_h264_idct_dc_add)(dst + block_offset[i], block + i*16*sizeof(pixel), stride);
196 }
197 1869130 }
198
199 1800042 void FUNCC(ff_h264_idct8_add4)(uint8_t *dst, const int *block_offset,
200 int16_t *block, int stride,
201 const uint8_t nnzc[5 * 8])
202 {
203 int i;
204
2/2
✓ Branch 0 taken 3600084 times.
✓ Branch 1 taken 900021 times.
 9000210 for(i=0; i<16; i+=4){
205 7200168 int nnz = nnzc[ scan8[i] ];
206
2/2
✓ Branch 0 taken 2288246 times.
✓ Branch 1 taken 1311838 times.
 7200168 if(nnz){
207
4/4
✓ Branch 0 taken 688126 times.
✓ Branch 1 taken 1600120 times.
✓ Branch 2 taken 374781 times.
✓ Branch 3 taken 313345 times.
 4576492 if(nnz==1 && ((dctcoef*)block)[i*16]) FUNCC(ff_h264_idct8_dc_add)(dst + block_offset[i], block + i*16*sizeof(pixel), stride);
208 3826930 else FUNCC(ff_h264_idct8_add )(dst + block_offset[i], block + i*16*sizeof(pixel), stride);
209 }
210 }
211 1800042 }
212
213 8966916 void FUNCC(ff_h264_idct_add8)(uint8_t **dest, const int *block_offset, int16_t *block, int stride, const uint8_t nnzc[15*8]){
214 int i, j;
215
2/2
✓ Branch 0 taken 8966916 times.
✓ Branch 1 taken 4483458 times.
 26900748 for(j=1; j<3; j++){
216
2/2
✓ Branch 0 taken 35867664 times.
✓ Branch 1 taken 8966916 times.
 89669160 for(i=j*16; i<j*16+4; i++){
217
2/2
✓ Branch 0 taken 8259106 times.
✓ Branch 1 taken 27608558 times.
 71735328 if(nnzc[ scan8[i] ])
218 16518212 FUNCC(ff_h264_idct_add )(dest[j-1] + block_offset[i], block + i*16*sizeof(pixel), stride);
219
2/2
✓ Branch 0 taken 12718305 times.
✓ Branch 1 taken 14890253 times.
 55217116 else if(((dctcoef*)block)[i*16])
220 25436610 FUNCC(ff_h264_idct_dc_add)(dest[j-1] + block_offset[i], block + i*16*sizeof(pixel), stride);
221 }
222 }
223 8966916 }
224
225 1350636 void FUNCC(ff_h264_idct_add8_422)(uint8_t **dest, const int *block_offset, int16_t *block, int stride, const uint8_t nnzc[15*8]){
226 int i, j;
227
228
2/2
✓ Branch 0 taken 1350636 times.
✓ Branch 1 taken 675318 times.
 4051908 for(j=1; j<3; j++){
229
2/2
✓ Branch 0 taken 5402544 times.
✓ Branch 1 taken 1350636 times.
 13506360 for(i=j*16; i<j*16+4; i++){
230
2/2
✓ Branch 0 taken 1820857 times.
✓ Branch 1 taken 3581687 times.
 10805088 if(nnzc[ scan8[i] ])
231 3641714 FUNCC(ff_h264_idct_add )(dest[j-1] + block_offset[i], block + i*16*sizeof(pixel), stride);
232
2/2
✓ Branch 0 taken 2290895 times.
✓ Branch 1 taken 1290792 times.
 7163374 else if(((dctcoef*)block)[i*16])
233 4581790 FUNCC(ff_h264_idct_dc_add)(dest[j-1] + block_offset[i], block + i*16*sizeof(pixel), stride);
234 }
235 }
236
237
2/2
✓ Branch 0 taken 1350636 times.
✓ Branch 1 taken 675318 times.
 4051908 for(j=1; j<3; j++){
238
2/2
✓ Branch 0 taken 5402544 times.
✓ Branch 1 taken 1350636 times.
 13506360 for(i=j*16+4; i<j*16+8; i++){
239
2/2
✓ Branch 0 taken 1845890 times.
✓ Branch 1 taken 3556654 times.
 10805088 if(nnzc[ scan8[i+4] ])
240 3691780 FUNCC(ff_h264_idct_add )(dest[j-1] + block_offset[i+4], block + i*16*sizeof(pixel), stride);
241
2/2
✓ Branch 0 taken 2323433 times.
✓ Branch 1 taken 1233221 times.
 7113308 else if(((dctcoef*)block)[i*16])
242 4646866 FUNCC(ff_h264_idct_dc_add)(dest[j-1] + block_offset[i+4], block + i*16*sizeof(pixel), stride);
243 }
244 }
245 1350636 }
246
247 /**
248 * IDCT transforms the 16 dc values and dequantizes them.
249 * @param qmul quantization parameter
250 */
251 1179660 void FUNCC(ff_h264_luma_dc_dequant_idct)(int16_t *_output, int16_t *_input, int qmul){
252 #define stride 16
253 int i;
254 int temp[16];
255 static const uint8_t x_offset[4]={0, 2*stride, 8*stride, 10*stride};
256 1179660 dctcoef *input = (dctcoef*)_input;
257 1179660 dctcoef *output = (dctcoef*)_output;
258
259
2/2
✓ Branch 0 taken 2359320 times.
✓ Branch 1 taken 589830 times.
 5898300 for(i=0; i<4; i++){
260 4718640 const int z0= input[4*i+0] + input[4*i+1];
261 4718640 const int z1= input[4*i+0] - input[4*i+1];
262 4718640 const int z2= input[4*i+2] - input[4*i+3];
263 4718640 const int z3= input[4*i+2] + input[4*i+3];
264
265 4718640 temp[4*i+0]= z0+z3;
266 4718640 temp[4*i+1]= z0-z3;
267 4718640 temp[4*i+2]= z1-z2;
268 4718640 temp[4*i+3]= z1+z2;
269 }
270
271
2/2
✓ Branch 0 taken 2359320 times.
✓ Branch 1 taken 589830 times.
 5898300 for(i=0; i<4; i++){
272 4718640 const int offset= x_offset[i];
273 4718640 const SUINT z0= temp[4*0+i] + temp[4*2+i];
274 4718640 const SUINT z1= temp[4*0+i] - temp[4*2+i];
275 4718640 const SUINT z2= temp[4*1+i] - temp[4*3+i];
276 4718640 const SUINT z3= temp[4*1+i] + temp[4*3+i];
277
278 4718640 output[stride* 0+offset]= (int)((z0 + z3)*qmul + 128 ) >> 8;
279 4718640 output[stride* 1+offset]= (int)((z1 + z2)*qmul + 128 ) >> 8;
280 4718640 output[stride* 4+offset]= (int)((z1 - z2)*qmul + 128 ) >> 8;
281 4718640 output[stride* 5+offset]= (int)((z0 - z3)*qmul + 128 ) >> 8;
282 }
283 #undef stride
284 1179660 }
285
286 2252062 void FUNCC(ff_h264_chroma422_dc_dequant_idct)(int16_t *_block, int qmul){
287 2252062 const int stride= 16*2;
288 2252062 const int xStride= 16;
289 int i;
290 unsigned temp[8];
291 static const uint8_t x_offset[2]={0, 16};
292 2252062 dctcoef *block = (dctcoef*)_block;
293
294
2/2
✓ Branch 0 taken 4504124 times.
✓ Branch 1 taken 1126031 times.
 11260310 for(i=0; i<4; i++){
295 9008248 temp[2*i+0] = block[stride*i + xStride*0] + (unsigned)block[stride*i + xStride*1];
296 9008248 temp[2*i+1] = block[stride*i + xStride*0] - (unsigned)block[stride*i + xStride*1];
297 }
298
299
2/2
✓ Branch 0 taken 2252062 times.
✓ Branch 1 taken 1126031 times.
 6756186 for(i=0; i<2; i++){
300 4504124 const int offset= x_offset[i];
301 4504124 const SUINT z0= temp[2*0+i] + temp[2*2+i];
302 4504124 const SUINT z1= temp[2*0+i] - temp[2*2+i];
303 4504124 const SUINT z2= temp[2*1+i] - temp[2*3+i];
304 4504124 const SUINT z3= temp[2*1+i] + temp[2*3+i];
305
306 4504124 block[stride*0+offset]= (int)((z0 + z3)*qmul + 128) >> 8;
307 4504124 block[stride*1+offset]= (int)((z1 + z2)*qmul + 128) >> 8;
308 4504124 block[stride*2+offset]= (int)((z1 - z2)*qmul + 128) >> 8;
309 4504124 block[stride*3+offset]= (int)((z0 - z3)*qmul + 128) >> 8;
310 }
311 2252062 }
312
313 13055704 void FUNCC(ff_h264_chroma_dc_dequant_idct)(int16_t *_block, int qmul){
314 13055704 const int stride= 16*2;
315 13055704 const int xStride= 16;
316 SUINT a,b,c,d,e;
317 13055704 dctcoef *block = (dctcoef*)_block;
318
319 13055704 a= block[stride*0 + xStride*0];
320 13055704 b= block[stride*0 + xStride*1];
321 13055704 c= block[stride*1 + xStride*0];
322 13055704 d= block[stride*1 + xStride*1];
323
324 13055704 e= a-b;
325 13055704 a= a+b;
326 13055704 b= c-d;
327 13055704 c= c+d;
328
329 13055704 block[stride*0 + xStride*0]= (int)((a+c)*qmul) >> 7;
330 13055704 block[stride*0 + xStride*1]= (int)((e+b)*qmul) >> 7;
331 13055704 block[stride*1 + xStride*0]= (int)((a-c)*qmul) >> 7;
332 13055704 block[stride*1 + xStride*1]= (int)((e-b)*qmul) >> 7;
333 13055704 }
334