GCC Code Coverage Report
Directory: ../../../ffmpeg/ Exec Total Coverage
File: src/libavcodec/simple_idct.c Lines: 96 101 95.0 %
Date: 2020-09-25 14:59:26 Branches: 26 30 86.7 %

Line Branch Exec Source
1
/*
2
 * Simple IDCT
3
 *
4
 * Copyright (c) 2001 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
/**
24
 * @file
25
 * simpleidct in C.
26
 */
27
28
#include "libavutil/intreadwrite.h"
29
#include "avcodec.h"
30
#include "mathops.h"
31
#include "simple_idct.h"
32
33
#define IN_IDCT_DEPTH 16
34
35
#define BIT_DEPTH 8
36
#include "simple_idct_template.c"
37
#undef BIT_DEPTH
38
39
#define BIT_DEPTH 10
40
#include "simple_idct_template.c"
41
42
#define EXTRA_SHIFT  2
43
#include "simple_idct_template.c"
44
45
#undef EXTRA_SHIFT
46
#undef BIT_DEPTH
47
48
#define BIT_DEPTH 12
49
#include "simple_idct_template.c"
50
#undef BIT_DEPTH
51
#undef IN_IDCT_DEPTH
52
53
#define IN_IDCT_DEPTH 32
54
#define BIT_DEPTH 10
55
#include "simple_idct_template.c"
56
#undef BIT_DEPTH
57
#undef IN_IDCT_DEPTH
58
59
/* 2x4x8 idct */
60
61
#define CN_SHIFT 12
62
#define C_FIX(x) ((int)((x) * (1 << CN_SHIFT) + 0.5))
63
#define C1 C_FIX(0.6532814824)
64
#define C2 C_FIX(0.2705980501)
65
66
/* row idct is multiple by 16 * sqrt(2.0), col idct4 is normalized,
67
   and the butterfly must be multiplied by 0.5 * sqrt(2.0) */
68
#define C_SHIFT (4+1+12)
69
70
324544
static inline void idct4col_put(uint8_t *dest, ptrdiff_t line_size, const int16_t *col)
71
{
72
    int c0, c1, c2, c3, a0, a1, a2, a3;
73
74
324544
    a0 = col[8*0];
75
324544
    a1 = col[8*2];
76
324544
    a2 = col[8*4];
77
324544
    a3 = col[8*6];
78
324544
    c0 = ((a0 + a2) * (1 << CN_SHIFT - 1)) + (1 << (C_SHIFT - 1));
79
324544
    c2 = ((a0 - a2) * (1 << CN_SHIFT - 1)) + (1 << (C_SHIFT - 1));
80
324544
    c1 = a1 * C1 + a3 * C2;
81
324544
    c3 = a1 * C2 - a3 * C1;
82
324544
    dest[0] = av_clip_uint8((c0 + c1) >> C_SHIFT);
83
324544
    dest += line_size;
84
324544
    dest[0] = av_clip_uint8((c2 + c3) >> C_SHIFT);
85
324544
    dest += line_size;
86
324544
    dest[0] = av_clip_uint8((c2 - c3) >> C_SHIFT);
87
324544
    dest += line_size;
88
324544
    dest[0] = av_clip_uint8((c0 - c1) >> C_SHIFT);
89
324544
}
90
91
#define BF(k) \
92
{\
93
    int a0, a1;\
94
    a0 = ptr[k];\
95
    a1 = ptr[8 + k];\
96
    ptr[k] = a0 + a1;\
97
    ptr[8 + k] = a0 - a1;\
98
}
99
100
/* only used by DV codec. The input must be interlaced. 128 is added
101
   to the pixels before clamping to avoid systematic error
102
   (1024*sqrt(2)) offset would be needed otherwise. */
103
/* XXX: I think a 1.0/sqrt(2) normalization should be needed to
104
   compensate the extra butterfly stage - I don't have the full DV
105
   specification */
106
20284
void ff_simple_idct248_put(uint8_t *dest, ptrdiff_t line_size, int16_t *block)
107
{
108
    int i;
109
    int16_t *ptr;
110
111
    /* butterfly */
112
20284
    ptr = block;
113
101420
    for(i=0;i<4;i++) {
114
81136
        BF(0);
115
81136
        BF(1);
116
81136
        BF(2);
117
81136
        BF(3);
118
81136
        BF(4);
119
81136
        BF(5);
120
81136
        BF(6);
121
81136
        BF(7);
122
81136
        ptr += 2 * 8;
123
    }
124
125
    /* IDCT8 on each line */
126
182556
    for(i=0; i<8; i++) {
127
162272
        idctRowCondDC_int16_8bit(block + i*8, 0);
128
    }
129
130
    /* IDCT4 and store */
131
182556
    for(i=0;i<8;i++) {
132
162272
        idct4col_put(dest + i, 2 * line_size, block + i);
133
162272
        idct4col_put(dest + line_size + i, 2 * line_size, block + 8 + i);
134
    }
135
20284
}
136
137
/* 8x4 & 4x8 WMV2 IDCT */
138
#undef CN_SHIFT
139
#undef C_SHIFT
140
#undef C_FIX
141
#undef C1
142
#undef C2
143
#define CN_SHIFT 12
144
#define C_FIX(x) ((int)((x) * M_SQRT2 * (1 << CN_SHIFT) + 0.5))
145
#define C1 C_FIX(0.6532814824)
146
#define C2 C_FIX(0.2705980501)
147
#define C3 C_FIX(0.5)
148
#define C_SHIFT (4+1+12)
149
263504
static inline void idct4col_add(uint8_t *dest, ptrdiff_t line_size, const int16_t *col)
150
{
151
    int c0, c1, c2, c3, a0, a1, a2, a3;
152
153
263504
    a0 = col[8*0];
154
263504
    a1 = col[8*1];
155
263504
    a2 = col[8*2];
156
263504
    a3 = col[8*3];
157
263504
    c0 = (a0 + a2)*C3 + (1 << (C_SHIFT - 1));
158
263504
    c2 = (a0 - a2)*C3 + (1 << (C_SHIFT - 1));
159
263504
    c1 = a1 * C1 + a3 * C2;
160
263504
    c3 = a1 * C2 - a3 * C1;
161
263504
    dest[0] = av_clip_uint8(dest[0] + ((c0 + c1) >> C_SHIFT));
162
263504
    dest += line_size;
163
263504
    dest[0] = av_clip_uint8(dest[0] + ((c2 + c3) >> C_SHIFT));
164
263504
    dest += line_size;
165
263504
    dest[0] = av_clip_uint8(dest[0] + ((c2 - c3) >> C_SHIFT));
166
263504
    dest += line_size;
167
263504
    dest[0] = av_clip_uint8(dest[0] + ((c0 - c1) >> C_SHIFT));
168
263504
}
169
170
#define RN_SHIFT 15
171
#define R_FIX(x) ((int)((x) * M_SQRT2 * (1 << RN_SHIFT) + 0.5))
172
#define R1 R_FIX(0.6532814824)
173
#define R2 R_FIX(0.2705980501)
174
#define R3 R_FIX(0.5)
175
#define R_SHIFT 11
176
239648
static inline void idct4row(int16_t *row)
177
{
178
    int c0, c1, c2, c3, a0, a1, a2, a3;
179
180
239648
    a0 = row[0];
181
239648
    a1 = row[1];
182
239648
    a2 = row[2];
183
239648
    a3 = row[3];
184
239648
    c0 = (a0 + a2)*R3 + (1 << (R_SHIFT - 1));
185
239648
    c2 = (a0 - a2)*R3 + (1 << (R_SHIFT - 1));
186
239648
    c1 = a1 * R1 + a3 * R2;
187
239648
    c3 = a1 * R2 - a3 * R1;
188
239648
    row[0]= (c0 + c1) >> R_SHIFT;
189
239648
    row[1]= (c2 + c3) >> R_SHIFT;
190
239648
    row[2]= (c2 - c3) >> R_SHIFT;
191
239648
    row[3]= (c0 - c1) >> R_SHIFT;
192
239648
}
193
194
32938
void ff_simple_idct84_add(uint8_t *dest, ptrdiff_t line_size, int16_t *block)
195
{
196
    int i;
197
198
    /* IDCT8 on each line */
199
164690
    for(i=0; i<4; i++) {
200
131752
        idctRowCondDC_int16_8bit(block + i*8, 0);
201
    }
202
203
    /* IDCT4 and store */
204
296442
    for(i=0;i<8;i++) {
205
263504
        idct4col_add(dest + i, line_size, block + i);
206
    }
207
32938
}
208
209
29956
void ff_simple_idct48_add(uint8_t *dest, ptrdiff_t line_size, int16_t *block)
210
{
211
    int i;
212
213
    /* IDCT4 on each line */
214
269604
    for(i=0; i<8; i++) {
215
239648
        idct4row(block + i*8);
216
    }
217
218
    /* IDCT8 and store */
219
149780
    for(i=0; i<4; i++){
220
119824
        idctSparseColAdd_int16_8bit(dest + i, line_size, block + i);
221
    }
222
29956
}
223
224
void ff_simple_idct44_add(uint8_t *dest, ptrdiff_t line_size, int16_t *block)
225
{
226
    int i;
227
228
    /* IDCT4 on each line */
229
    for(i=0; i<4; i++) {
230
        idct4row(block + i*8);
231
    }
232
233
    /* IDCT4 and store */
234
    for(i=0; i<4; i++){
235
        idct4col_add(dest + i, line_size, block + i);
236
    }
237
}
238
239
2481312
void ff_prores_idct_10(int16_t *block, const int16_t *qmat)
240
{
241
    int i;
242
243
161285280
    for (i = 0; i < 64; i++)
244
158803968
        block[i] *= qmat[i];
245
246
22331808
    for (i = 0; i < 8; i++)
247
19850496
        idctRowCondDC_extrashift_10(block + i*8, 2);
248
249
22331808
    for (i = 0; i < 8; i++) {
250
19850496
        block[i] += 8192;
251
19850496
        idctSparseCol_extrashift_10(block + i);
252
    }
253
2481312
}
254
255
2446164
void ff_prores_idct_12(int16_t *block, const int16_t *qmat)
256
{
257
    int i;
258
259
159000660
    for (i = 0; i < 64; i++)
260
156554496
        block[i] *= qmat[i];
261
262
22015476
    for (i = 0; i < 8; i++)
263
19569312
        idctRowCondDC_int16_12bit(block + i*8, 0);
264
265
22015476
    for (i = 0; i < 8; i++) {
266
19569312
        block[i] += 8192;
267
19569312
        idctSparseCol_int16_12bit(block + i);
268
    }
269
2446164
}