1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
|
/*
* DSP utils
* Copyright (c) 2000, 2001, 2002 Fabrice Bellard
* Copyright (c) 2002-2004 Michael Niedermayer <michaelni@gmx.at>
*
* This file is part of FFmpeg.
*
* FFmpeg is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2.1 of the License, or (at your option) any later version.
*
* FFmpeg is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with FFmpeg; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
/**
* @file
* DSP utils.
* Note, many functions in here may use MMX which trashes the FPU state, it is
* absolutely necessary to call emms_c() between DSP & float/double code.
*/
#ifndef AVCODEC_DSPUTIL_H
#define AVCODEC_DSPUTIL_H
#include "avcodec.h"
extern uint32_t ff_square_tab[512];
/* minimum alignment rules ;)
* If you notice errors in the align stuff, need more alignment for some ASM code
* for some CPU or need to use a function with less aligned data then send a mail
* to the ffmpeg-devel mailing list, ...
*
* !warning These alignments might not match reality, (missing attribute((align))
* stuff somewhere possible).
* I (Michael) did not check them, these are just the alignments which I think
* could be reached easily ...
*
* !future video codecs might need functions with less strict alignment
*/
struct MpegEncContext;
/* Motion estimation:
* h is limited to { width / 2, width, 2 * width },
* but never larger than 16 and never smaller than 2.
* Although currently h < 4 is not used as functions with
* width < 8 are neither used nor implemented. */
typedef int (*me_cmp_func)(struct MpegEncContext *c,
uint8_t *blk1 /* align width (8 or 16) */,
uint8_t *blk2 /* align 1 */, int line_size, int h);
/**
* Scantable.
*/
typedef struct ScanTable {
const uint8_t *scantable;
uint8_t permutated[64];
uint8_t raster_end[64];
} ScanTable;
void ff_init_scantable(uint8_t *permutation, ScanTable *st,
const uint8_t *src_scantable);
void ff_init_scantable_permutation(uint8_t *idct_permutation,
int idct_permutation_type);
/**
* DSPContext.
*/
typedef struct DSPContext {
/* pixel ops : interface with DCT */
void (*get_pixels)(int16_t *block /* align 16 */,
const uint8_t *pixels /* align 8 */,
int line_size);
void (*diff_pixels)(int16_t *block /* align 16 */,
const uint8_t *s1 /* align 8 */,
const uint8_t *s2 /* align 8 */,
int stride);
void (*put_pixels_clamped)(const int16_t *block /* align 16 */,
uint8_t *pixels /* align 8 */,
int line_size);
void (*put_signed_pixels_clamped)(const int16_t *block /* align 16 */,
uint8_t *pixels /* align 8 */,
int line_size);
void (*add_pixels_clamped)(const int16_t *block /* align 16 */,
uint8_t *pixels /* align 8 */,
int line_size);
int (*sum_abs_dctelem)(int16_t *block /* align 16 */);
int (*pix_sum)(uint8_t *pix, int line_size);
int (*pix_norm1)(uint8_t *pix, int line_size);
me_cmp_func sad[6]; /* identical to pix_absAxA except additional void * */
me_cmp_func sse[6];
me_cmp_func hadamard8_diff[6];
me_cmp_func dct_sad[6];
me_cmp_func quant_psnr[6];
me_cmp_func bit[6];
me_cmp_func rd[6];
me_cmp_func vsad[6];
me_cmp_func vsse[6];
me_cmp_func nsse[6];
me_cmp_func w53[6];
me_cmp_func w97[6];
me_cmp_func dct_max[6];
me_cmp_func dct264_sad[6];
me_cmp_func me_pre_cmp[6];
me_cmp_func me_cmp[6];
me_cmp_func me_sub_cmp[6];
me_cmp_func mb_cmp[6];
me_cmp_func ildct_cmp[6]; // only width 16 used
me_cmp_func frame_skip_cmp[6]; // only width 8 used
me_cmp_func pix_abs[2][4];
/* (I)DCT */
void (*fdct)(int16_t *block /* align 16 */);
void (*fdct248)(int16_t *block /* align 16 */);
/* IDCT really */
void (*idct)(int16_t *block /* align 16 */);
/**
* block -> idct -> clip to unsigned 8 bit -> dest.
* (-1392, 0, 0, ...) -> idct -> (-174, -174, ...) -> put -> (0, 0, ...)
* @param line_size size in bytes of a horizontal line of dest
*/
void (*idct_put)(uint8_t *dest /* align 8 */,
int line_size, int16_t *block /* align 16 */);
/**
* block -> idct -> add dest -> clip to unsigned 8 bit -> dest.
* @param line_size size in bytes of a horizontal line of dest
*/
void (*idct_add)(uint8_t *dest /* align 8 */,
int line_size, int16_t *block /* align 16 */);
/**
* IDCT input permutation.
* Several optimized IDCTs need a permutated input (relative to the
* normal order of the reference IDCT).
* This permutation must be performed before the idct_put/add.
* Note, normally this can be merged with the zigzag/alternate scan<br>
* An example to avoid confusion:
* - (->decode coeffs -> zigzag reorder -> dequant -> reference IDCT -> ...)
* - (x -> reference DCT -> reference IDCT -> x)
* - (x -> reference DCT -> simple_mmx_perm = idct_permutation
* -> simple_idct_mmx -> x)
* - (-> decode coeffs -> zigzag reorder -> simple_mmx_perm -> dequant
* -> simple_idct_mmx -> ...)
*/
uint8_t idct_permutation[64];
int idct_permutation_type;
#define FF_NO_IDCT_PERM 1
#define FF_LIBMPEG2_IDCT_PERM 2
#define FF_SIMPLE_IDCT_PERM 3
#define FF_TRANSPOSE_IDCT_PERM 4
#define FF_PARTTRANS_IDCT_PERM 5
#define FF_SSE2_IDCT_PERM 6
int (*try_8x8basis)(int16_t rem[64], int16_t weight[64],
int16_t basis[64], int scale);
void (*add_8x8basis)(int16_t rem[64], int16_t basis[64], int scale);
#define BASIS_SHIFT 16
#define RECON_SHIFT 6
void (*draw_edges)(uint8_t *buf, int wrap, int width, int height,
int w, int h, int sides);
#define EDGE_WIDTH 16
#define EDGE_TOP 1
#define EDGE_BOTTOM 2
void (*shrink[4])(uint8_t *dst, int dst_wrap, const uint8_t *src,
int src_wrap, int width, int height);
} DSPContext;
void ff_dsputil_static_init(void);
void ff_dsputil_init(DSPContext *p, AVCodecContext *avctx);
void avpriv_dsputil_init(DSPContext* p, AVCodecContext *avctx);
attribute_deprecated void dsputil_init(DSPContext* c, AVCodecContext *avctx);
int ff_check_alignment(void);
void ff_set_cmp(DSPContext *c, me_cmp_func *cmp, int type);
void ff_dsputil_init_alpha(DSPContext* c, AVCodecContext *avctx);
void ff_dsputil_init_arm(DSPContext *c, AVCodecContext *avctx,
unsigned high_bit_depth);
void ff_dsputil_init_ppc(DSPContext *c, AVCodecContext *avctx,
unsigned high_bit_depth);
void ff_dsputil_init_x86(DSPContext *c, AVCodecContext *avctx,
unsigned high_bit_depth);
void ff_dsputil_init_dwt(DSPContext *c);
#endif /* AVCODEC_DSPUTIL_H */
|
