annotate dsputil.h @ 11732:c40cc26cd23f libavcodec

aacenc: Add a rate only trellis for codebook selection for the TLS.
author alexc
date Fri, 14 May 2010 16:49:51 +0000
parents 7dd2a45249a9
children eaf3ffb5df56
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
429
718a22dc121f license/copyright change
glantau
parents: 403
diff changeset
1 /*
718a22dc121f license/copyright change
glantau
parents: 403
diff changeset
2 * DSP utils
8629
04423b2f6e0b cosmetics: Remove pointless period after copyright statement non-sentences.
diego
parents: 8590
diff changeset
3 * Copyright (c) 2000, 2001, 2002 Fabrice Bellard
1739
07a484280a82 copyright year update of the files i touched and remembered, things look annoyingly unmaintained otherwise
michael
parents: 1729
diff changeset
4 * Copyright (c) 2002-2004 Michael Niedermayer <michaelni@gmx.at>
429
718a22dc121f license/copyright change
glantau
parents: 403
diff changeset
5 *
3947
c8c591fe26f8 Change license headers to say 'FFmpeg' instead of 'this program/this library'
diego
parents: 3807
diff changeset
6 * This file is part of FFmpeg.
c8c591fe26f8 Change license headers to say 'FFmpeg' instead of 'this program/this library'
diego
parents: 3807
diff changeset
7 *
c8c591fe26f8 Change license headers to say 'FFmpeg' instead of 'this program/this library'
diego
parents: 3807
diff changeset
8 * FFmpeg is free software; you can redistribute it and/or
429
718a22dc121f license/copyright change
glantau
parents: 403
diff changeset
9 * modify it under the terms of the GNU Lesser General Public
718a22dc121f license/copyright change
glantau
parents: 403
diff changeset
10 * License as published by the Free Software Foundation; either
3947
c8c591fe26f8 Change license headers to say 'FFmpeg' instead of 'this program/this library'
diego
parents: 3807
diff changeset
11 * version 2.1 of the License, or (at your option) any later version.
429
718a22dc121f license/copyright change
glantau
parents: 403
diff changeset
12 *
3947
c8c591fe26f8 Change license headers to say 'FFmpeg' instead of 'this program/this library'
diego
parents: 3807
diff changeset
13 * FFmpeg is distributed in the hope that it will be useful,
429
718a22dc121f license/copyright change
glantau
parents: 403
diff changeset
14 * but WITHOUT ANY WARRANTY; without even the implied warranty of
718a22dc121f license/copyright change
glantau
parents: 403
diff changeset
15 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
718a22dc121f license/copyright change
glantau
parents: 403
diff changeset
16 * Lesser General Public License for more details.
718a22dc121f license/copyright change
glantau
parents: 403
diff changeset
17 *
718a22dc121f license/copyright change
glantau
parents: 403
diff changeset
18 * You should have received a copy of the GNU Lesser General Public
3947
c8c591fe26f8 Change license headers to say 'FFmpeg' instead of 'this program/this library'
diego
parents: 3807
diff changeset
19 * License along with FFmpeg; if not, write to the Free Software
3036
0b546eab515d Update licensing information: The FSF changed postal address.
diego
parents: 3029
diff changeset
20 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
429
718a22dc121f license/copyright change
glantau
parents: 403
diff changeset
21 */
1102
eeaa1e837efb more dox
michaelni
parents: 1092
diff changeset
22
eeaa1e837efb more dox
michaelni
parents: 1092
diff changeset
23 /**
11644
7dd2a45249a9 Remove explicit filename from Doxygen @file commands.
diego
parents: 11637
diff changeset
24 * @file
1106
1e39f273ecd6 per file doxy
michaelni
parents: 1104
diff changeset
25 * DSP utils.
1213
a71fe70b428b doc fixes (forgot to commit)
michaelni
parents: 1201
diff changeset
26 * note, many functions in here may use MMX which trashes the FPU state, it is
a71fe70b428b doc fixes (forgot to commit)
michaelni
parents: 1201
diff changeset
27 * absolutely necessary to call emms_c() between dsp & float/double code
1102
eeaa1e837efb more dox
michaelni
parents: 1092
diff changeset
28 */
eeaa1e837efb more dox
michaelni
parents: 1092
diff changeset
29
7760
c4a4495715dd Globally rename the header inclusion guard names.
stefano
parents: 7687
diff changeset
30 #ifndef AVCODEC_DSPUTIL_H
c4a4495715dd Globally rename the header inclusion guard names.
stefano
parents: 7687
diff changeset
31 #define AVCODEC_DSPUTIL_H
0
986e461dc072 Initial revision
glantau
parents:
diff changeset
32
8573
2acf0ae7b041 Fix build: Add intreadwrite.h and bswap.h #includes where necessary.
diego
parents: 8567
diff changeset
33 #include "libavutil/intreadwrite.h"
252
ddb1a0e94cf4 - Added PSNR feature to libavcodec and ffmpeg. By now just Y PSNR until I'm
pulento
parents: 214
diff changeset
34 #include "avcodec.h"
0
986e461dc072 Initial revision
glantau
parents:
diff changeset
35
1102
eeaa1e837efb more dox
michaelni
parents: 1092
diff changeset
36
255
db20b987c32d divx5-gmc support
michaelni
parents: 252
diff changeset
37 //#define DEBUG
0
986e461dc072 Initial revision
glantau
parents:
diff changeset
38 /* dct code */
986e461dc072 Initial revision
glantau
parents:
diff changeset
39 typedef short DCTELEM;
986e461dc072 Initial revision
glantau
parents:
diff changeset
40
474
11dbd00682fc avoid name clash with libjpeg - added missing externs
bellard
parents: 429
diff changeset
41 void fdct_ifast (DCTELEM *data);
1571
aa4dc16c0f18 * adding integer/floating point AAN implementations for DCT 2-4-8
romansh
parents: 1567
diff changeset
42 void fdct_ifast248 (DCTELEM *data);
625
bb6a69f9d409 slow but accurate integer dct from IJG (should be ok with the LGPL as the old DCT is the fast integer DCT from IJG)
michaelni
parents: 623
diff changeset
43 void ff_jpeg_fdct_islow (DCTELEM *data);
1567
e08df4d22d27 * introducing dct248 into the DSP context.
romansh
parents: 1565
diff changeset
44 void ff_fdct248_islow (DCTELEM *data);
0
986e461dc072 Initial revision
glantau
parents:
diff changeset
45
986e461dc072 Initial revision
glantau
parents:
diff changeset
46 void j_rev_dct (DCTELEM *data);
2256
7e0b2e86afa9 1/2 resolution decoding
michael
parents: 2185
diff changeset
47 void j_rev_dct4 (DCTELEM *data);
2257
5f64a30339e5 1/4 resolution decoding
michael
parents: 2256
diff changeset
48 void j_rev_dct2 (DCTELEM *data);
2259
12e75af1d44c 1/8 resolution decoding
michael
parents: 2257
diff changeset
49 void j_rev_dct1 (DCTELEM *data);
5887
83ac4620c6ed intrax8 decoder patch by "someone"
michael
parents: 5830
diff changeset
50 void ff_wmv2_idct_c(DCTELEM *data);
0
986e461dc072 Initial revision
glantau
parents:
diff changeset
51
687
9abb13c21fbe fdct_mmx -> ff_fdct_mmx (renamed to avoid namespace conflict with xvid)
arpi_esp
parents: 675
diff changeset
52 void ff_fdct_mmx(DCTELEM *block);
1565
1a9a63f59849 minor mmx2 optimization if the dct
michael
parents: 1527
diff changeset
53 void ff_fdct_mmx2(DCTELEM *block);
1765
e31754bc5b65 SSE2 fdct by (Balatoni Denes <pnis at coder dot hu>)
michael
parents: 1739
diff changeset
54 void ff_fdct_sse2(DCTELEM *block);
0
986e461dc072 Initial revision
glantau
parents:
diff changeset
55
2755
975074f04b95 decode H.264 with 8x8 transform.
lorenm
parents: 2725
diff changeset
56 void ff_h264_idct8_add_c(uint8_t *dst, DCTELEM *block, int stride);
2272
cd43603c46f9 move h264 idct to its own file and call via function pointer in DspContext
michael
parents: 2259
diff changeset
57 void ff_h264_idct_add_c(uint8_t *dst, DCTELEM *block, int stride);
3105
2d35fb3cb940 h264: special case dc-only idct. ~1% faster overall
lorenm
parents: 3089
diff changeset
58 void ff_h264_idct8_dc_add_c(uint8_t *dst, DCTELEM *block, int stride);
2d35fb3cb940 h264: special case dc-only idct. ~1% faster overall
lorenm
parents: 3089
diff changeset
59 void ff_h264_idct_dc_add_c(uint8_t *dst, DCTELEM *block, int stride);
2272
cd43603c46f9 move h264 idct to its own file and call via function pointer in DspContext
michael
parents: 2259
diff changeset
60 void ff_h264_lowres_idct_add_c(uint8_t *dst, int stride, DCTELEM *block);
cd43603c46f9 move h264 idct to its own file and call via function pointer in DspContext
michael
parents: 2259
diff changeset
61 void ff_h264_lowres_idct_put_c(uint8_t *dst, int stride, DCTELEM *block);
8375
de2509cf3c44 H.264 idct functions that include the chroma, inter luma and intra16 luma loops
michael
parents: 8359
diff changeset
62 void ff_h264_idct_add16_c(uint8_t *dst, const int *blockoffset, DCTELEM *block, int stride, const uint8_t nnzc[6*8]);
de2509cf3c44 H.264 idct functions that include the chroma, inter luma and intra16 luma loops
michael
parents: 8359
diff changeset
63 void ff_h264_idct_add16intra_c(uint8_t *dst, const int *blockoffset, DCTELEM *block, int stride, const uint8_t nnzc[6*8]);
de2509cf3c44 H.264 idct functions that include the chroma, inter luma and intra16 luma loops
michael
parents: 8359
diff changeset
64 void ff_h264_idct8_add4_c(uint8_t *dst, const int *blockoffset, DCTELEM *block, int stride, const uint8_t nnzc[6*8]);
de2509cf3c44 H.264 idct functions that include the chroma, inter luma and intra16 luma loops
michael
parents: 8359
diff changeset
65 void ff_h264_idct_add8_c(uint8_t **dest, const int *blockoffset, DCTELEM *block, int stride, const uint8_t nnzc[6*8]);
2272
cd43603c46f9 move h264 idct to its own file and call via function pointer in DspContext
michael
parents: 2259
diff changeset
66
7261
032a49f033e8 simplify vorbis windowing
lorenm
parents: 7232
diff changeset
67 void ff_vector_fmul_window_c(float *dst, const float *src0, const float *src1,
032a49f033e8 simplify vorbis windowing
lorenm
parents: 7232
diff changeset
68 const float *win, float add_bias, int len);
7218
7f3d6509628b Fix x86-64
michael
parents: 7203
diff changeset
69 void ff_float_to_int16_c(int16_t *dst, const float *src, long len);
7446
a4ebced5d4a2 r14205 broke vorbis float_to_int on ppc
lorenm
parents: 7327
diff changeset
70 void ff_float_to_int16_interleave_c(int16_t *dst, const float **src, long len, int channels);
3568
945caa35ee9a sse and 3dnow implementations of float->int conversion and mdct windowing.
lorenm
parents: 3555
diff changeset
71
34
2733a4c1c693 added block permutation functions
glantau
parents: 19
diff changeset
72 /* encoding scans */
1064
b32afefe7d33 * UINTX -> uintx_t INTX -> intx_t
kabi
parents: 1040
diff changeset
73 extern const uint8_t ff_alternate_horizontal_scan[64];
b32afefe7d33 * UINTX -> uintx_t INTX -> intx_t
kabi
parents: 1040
diff changeset
74 extern const uint8_t ff_alternate_vertical_scan[64];
b32afefe7d33 * UINTX -> uintx_t INTX -> intx_t
kabi
parents: 1040
diff changeset
75 extern const uint8_t ff_zigzag_direct[64];
1567
e08df4d22d27 * introducing dct248 into the DSP context.
romansh
parents: 1565
diff changeset
76 extern const uint8_t ff_zigzag248_direct[64];
190
9e0e56869d05 fix for non-mmx runtimedetect encoding bugs - patch by Michael Niedermayer <michaelni@gmx.at>
uid46427
parents: 174
diff changeset
77
0
986e461dc072 Initial revision
glantau
parents:
diff changeset
78 /* pixel operations */
2090
869805505b30 increase size of crop table
michael
parents: 2075
diff changeset
79 #define MAX_NEG_CROP 1024
0
986e461dc072 Initial revision
glantau
parents:
diff changeset
80
986e461dc072 Initial revision
glantau
parents:
diff changeset
81 /* temporary */
4179
46d38e3d7038 rename squareTbl -> ff_squareTbl
mru
parents: 4176
diff changeset
82 extern uint32_t ff_squareTbl[512];
4176
23da44e8fd05 rename cropTbl -> ff_cropTbl
mru
parents: 3947
diff changeset
83 extern uint8_t ff_cropTbl[256 + 2 * MAX_NEG_CROP];
0
986e461dc072 Initial revision
glantau
parents:
diff changeset
84
1866
1755f959ab7f seperated out the C-based VP3 DSP functions into a different file; also
melanson
parents: 1864
diff changeset
85 /* VP3 DSP functions */
2693
02925a3903b6 porting vp3 idct over to lavc idct api
michael
parents: 2633
diff changeset
86 void ff_vp3_idct_c(DCTELEM *block/* align 16*/);
02925a3903b6 porting vp3 idct over to lavc idct api
michael
parents: 2633
diff changeset
87 void ff_vp3_idct_put_c(uint8_t *dest/*align 8*/, int line_size, DCTELEM *block/*align 16*/);
02925a3903b6 porting vp3 idct over to lavc idct api
michael
parents: 2633
diff changeset
88 void ff_vp3_idct_add_c(uint8_t *dest/*align 8*/, int line_size, DCTELEM *block/*align 16*/);
11637
f7281af560fe vp3: DC-only IDCT
conrad
parents: 11499
diff changeset
89 void ff_vp3_idct_dc_add_c(uint8_t *dest/*align 8*/, int line_size, const DCTELEM *block/*align 16*/);
0
986e461dc072 Initial revision
glantau
parents:
diff changeset
90
7995
1fbfce20cb79 Move VP3 loop filter to DSPContext
conrad
parents: 7760
diff changeset
91 void ff_vp3_v_loop_filter_c(uint8_t *src, int stride, int *bounding_values);
1fbfce20cb79 Move VP3 loop filter to DSPContext
conrad
parents: 7760
diff changeset
92 void ff_vp3_h_loop_filter_c(uint8_t *src, int stride, int *bounding_values);
1fbfce20cb79 Move VP3 loop filter to DSPContext
conrad
parents: 7760
diff changeset
93
8785
bee83b3f9a6b move vp6_filter_diag4() to a new vp6dsp.c file and use it throught dsputil
aurel
parents: 8760
diff changeset
94 /* VP6 DSP functions */
bee83b3f9a6b move vp6_filter_diag4() to a new vp6dsp.c file and use it throught dsputil
aurel
parents: 8760
diff changeset
95 void ff_vp6_filter_diag4_c(uint8_t *dst, uint8_t *src, int stride,
bee83b3f9a6b move vp6_filter_diag4() to a new vp6dsp.c file and use it throught dsputil
aurel
parents: 8760
diff changeset
96 const int16_t *h_weights, const int16_t *v_weights);
bee83b3f9a6b move vp6_filter_diag4() to a new vp6dsp.c file and use it throught dsputil
aurel
parents: 8760
diff changeset
97
11379
4b0cc1b97def Move some dsp func prototypes to dsputil.h; they are defined in dsputil.c
mru
parents: 11378
diff changeset
98 /* Bink functions */
4b0cc1b97def Move some dsp func prototypes to dsputil.h; they are defined in dsputil.c
mru
parents: 11378
diff changeset
99 void ff_bink_idct_c (DCTELEM *block);
4b0cc1b97def Move some dsp func prototypes to dsputil.h; they are defined in dsputil.c
mru
parents: 11378
diff changeset
100 void ff_bink_idct_add_c(uint8_t *dest, int linesize, DCTELEM *block);
4b0cc1b97def Move some dsp func prototypes to dsputil.h; they are defined in dsputil.c
mru
parents: 11378
diff changeset
101 void ff_bink_idct_put_c(uint8_t *dest, int linesize, DCTELEM *block);
4b0cc1b97def Move some dsp func prototypes to dsputil.h; they are defined in dsputil.c
mru
parents: 11378
diff changeset
102
4b0cc1b97def Move some dsp func prototypes to dsputil.h; they are defined in dsputil.c
mru
parents: 11378
diff changeset
103 /* CAVS functions */
4b0cc1b97def Move some dsp func prototypes to dsputil.h; they are defined in dsputil.c
mru
parents: 11378
diff changeset
104 void ff_put_cavs_qpel8_mc00_c(uint8_t *dst, uint8_t *src, int stride);
4b0cc1b97def Move some dsp func prototypes to dsputil.h; they are defined in dsputil.c
mru
parents: 11378
diff changeset
105 void ff_avg_cavs_qpel8_mc00_c(uint8_t *dst, uint8_t *src, int stride);
4b0cc1b97def Move some dsp func prototypes to dsputil.h; they are defined in dsputil.c
mru
parents: 11378
diff changeset
106 void ff_put_cavs_qpel16_mc00_c(uint8_t *dst, uint8_t *src, int stride);
4b0cc1b97def Move some dsp func prototypes to dsputil.h; they are defined in dsputil.c
mru
parents: 11378
diff changeset
107 void ff_avg_cavs_qpel16_mc00_c(uint8_t *dst, uint8_t *src, int stride);
4b0cc1b97def Move some dsp func prototypes to dsputil.h; they are defined in dsputil.c
mru
parents: 11378
diff changeset
108
11378
f46b68960464 Move some VC1 dsp prototypes to dsputil.h; they are defined in dsputil.c
mru
parents: 11376
diff changeset
109 /* VC1 functions */
f46b68960464 Move some VC1 dsp prototypes to dsputil.h; they are defined in dsputil.c
mru
parents: 11376
diff changeset
110 void ff_put_vc1_mspel_mc00_c(uint8_t *dst, const uint8_t *src, int stride, int rnd);
f46b68960464 Move some VC1 dsp prototypes to dsputil.h; they are defined in dsputil.c
mru
parents: 11376
diff changeset
111 void ff_avg_vc1_mspel_mc00_c(uint8_t *dst, const uint8_t *src, int stride, int rnd);
f46b68960464 Move some VC1 dsp prototypes to dsputil.h; they are defined in dsputil.c
mru
parents: 11376
diff changeset
112
11379
4b0cc1b97def Move some dsp func prototypes to dsputil.h; they are defined in dsputil.c
mru
parents: 11378
diff changeset
113 /* EA functions */
4b0cc1b97def Move some dsp func prototypes to dsputil.h; they are defined in dsputil.c
mru
parents: 11378
diff changeset
114 void ff_ea_idct_put_c(uint8_t *dest, int linesize, DCTELEM *block);
4b0cc1b97def Move some dsp func prototypes to dsputil.h; they are defined in dsputil.c
mru
parents: 11378
diff changeset
115
3245
c2c29be6282e remove mpegvideo.c img resample dependancy
michael
parents: 3217
diff changeset
116 /* 1/2^n downscaling functions from imgconvert.c */
c2c29be6282e remove mpegvideo.c img resample dependancy
michael
parents: 3217
diff changeset
117 void ff_img_copy_plane(uint8_t *dst, int dst_wrap, const uint8_t *src, int src_wrap, int width, int height);
c2c29be6282e remove mpegvideo.c img resample dependancy
michael
parents: 3217
diff changeset
118 void ff_shrink22(uint8_t *dst, int dst_wrap, const uint8_t *src, int src_wrap, int width, int height);
c2c29be6282e remove mpegvideo.c img resample dependancy
michael
parents: 3217
diff changeset
119 void ff_shrink44(uint8_t *dst, int dst_wrap, const uint8_t *src, int src_wrap, int width, int height);
c2c29be6282e remove mpegvideo.c img resample dependancy
michael
parents: 3217
diff changeset
120 void ff_shrink88(uint8_t *dst, int dst_wrap, const uint8_t *src, int src_wrap, int width, int height);
c2c29be6282e remove mpegvideo.c img resample dependancy
michael
parents: 3217
diff changeset
121
3248
7aa9f80e7954 mmx implementation of 3-point GMC. (5x faster than C)
lorenm
parents: 3245
diff changeset
122 void ff_gmc_c(uint8_t *dst, uint8_t *src, int stride, int h, int ox, int oy,
7aa9f80e7954 mmx implementation of 3-point GMC. (5x faster than C)
lorenm
parents: 3245
diff changeset
123 int dxx, int dxy, int dyx, int dyy, int shift, int r, int width, int height);
7aa9f80e7954 mmx implementation of 3-point GMC. (5x faster than C)
lorenm
parents: 3245
diff changeset
124
675
9008cbe229ce alignment info
michaelni
parents: 651
diff changeset
125 /* minimum alignment rules ;)
6486
3f96dc62c4b9 misc spelling fixes
diego
parents: 6449
diff changeset
126 If you notice errors in the align stuff, need more alignment for some ASM code
3f96dc62c4b9 misc spelling fixes
diego
parents: 6449
diff changeset
127 for some CPU or need to use a function with less aligned data then send a mail
3f96dc62c4b9 misc spelling fixes
diego
parents: 6449
diff changeset
128 to the ffmpeg-devel mailing list, ...
675
9008cbe229ce alignment info
michaelni
parents: 651
diff changeset
129
6486
3f96dc62c4b9 misc spelling fixes
diego
parents: 6449
diff changeset
130 !warning These alignments might not match reality, (missing attribute((align))
3f96dc62c4b9 misc spelling fixes
diego
parents: 6449
diff changeset
131 stuff somewhere possible).
6488
ecbe9565dcae typo fixes
diego
parents: 6486
diff changeset
132 I (Michael) did not check them, these are just the alignments which I think
6486
3f96dc62c4b9 misc spelling fixes
diego
parents: 6449
diff changeset
133 could be reached easily ...
675
9008cbe229ce alignment info
michaelni
parents: 651
diff changeset
134
9008cbe229ce alignment info
michaelni
parents: 651
diff changeset
135 !future video codecs might need functions with less strict alignment
9008cbe229ce alignment info
michaelni
parents: 651
diff changeset
136 */
0
986e461dc072 Initial revision
glantau
parents:
diff changeset
137
853
eacc2dd8fd9d * using DSPContext - so each codec could use its local (sub)set of CPU extension
kabi
parents: 838
diff changeset
138 /*
1064
b32afefe7d33 * UINTX -> uintx_t INTX -> intx_t
kabi
parents: 1040
diff changeset
139 void get_pixels_c(DCTELEM *block, const uint8_t *pixels, int line_size);
b32afefe7d33 * UINTX -> uintx_t INTX -> intx_t
kabi
parents: 1040
diff changeset
140 void diff_pixels_c(DCTELEM *block, const uint8_t *s1, const uint8_t *s2, int stride);
b32afefe7d33 * UINTX -> uintx_t INTX -> intx_t
kabi
parents: 1040
diff changeset
141 void put_pixels_clamped_c(const DCTELEM *block, uint8_t *pixels, int line_size);
b32afefe7d33 * UINTX -> uintx_t INTX -> intx_t
kabi
parents: 1040
diff changeset
142 void add_pixels_clamped_c(const DCTELEM *block, uint8_t *pixels, int line_size);
296
c1a8a1b4a24b sizeof(s->block) isnt 64*6*2 anymore bugfix
michaelni
parents: 294
diff changeset
143 void clear_blocks_c(DCTELEM *blocks);
853
eacc2dd8fd9d * using DSPContext - so each codec could use its local (sub)set of CPU extension
kabi
parents: 838
diff changeset
144 */
0
986e461dc072 Initial revision
glantau
parents:
diff changeset
145
986e461dc072 Initial revision
glantau
parents:
diff changeset
146 /* add and put pixel (decoding) */
675
9008cbe229ce alignment info
michaelni
parents: 651
diff changeset
147 // blocksizes for op_pixels_func are 8x4,8x8 16x8 16x16
1709
39a209b0c82c info about h and tpel
michael
parents: 1708
diff changeset
148 //h for op_pixels_func is limited to {width/2, width} but never larger than 16 and never smaller then 4
1064
b32afefe7d33 * UINTX -> uintx_t INTX -> intx_t
kabi
parents: 1040
diff changeset
149 typedef void (*op_pixels_func)(uint8_t *block/*align width (8 or 16)*/, const uint8_t *pixels/*align 1*/, int line_size, int h);
1267
85b71f9f7450 moving the svq3 motion compensation stuff to dsputil (this also means that existing optimized halfpel code is used now ...)
michaelni
parents: 1264
diff changeset
150 typedef void (*tpel_mc_func)(uint8_t *block/*align width (8 or 16)*/, const uint8_t *pixels/*align 1*/, int line_size, int w, int h);
1064
b32afefe7d33 * UINTX -> uintx_t INTX -> intx_t
kabi
parents: 1040
diff changeset
151 typedef void (*qpel_mc_func)(uint8_t *dst/*align width (8 or 16)*/, uint8_t *src/*align 1*/, int stride);
1168
5af9aeadbdc3 H264 decoder & demuxer
michaelni
parents: 1136
diff changeset
152 typedef void (*h264_chroma_mc_func)(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int srcStride, int h, int x, int y);
0
986e461dc072 Initial revision
glantau
parents:
diff changeset
153
11231
0fc1cdd984b7 Bink video decoder
kostya
parents: 11196
diff changeset
154 typedef void (*op_fill_func)(uint8_t *block/*align width (8 or 16)*/, uint8_t value, int line_size, int h);
0fc1cdd984b7 Bink video decoder
kostya
parents: 11196
diff changeset
155
984
e162c09efbe7 qpel fix
michaelni
parents: 975
diff changeset
156 #define DEF_OLD_QPEL(name)\
1064
b32afefe7d33 * UINTX -> uintx_t INTX -> intx_t
kabi
parents: 1040
diff changeset
157 void ff_put_ ## name (uint8_t *dst/*align width (8 or 16)*/, uint8_t *src/*align 1*/, int stride);\
b32afefe7d33 * UINTX -> uintx_t INTX -> intx_t
kabi
parents: 1040
diff changeset
158 void ff_put_no_rnd_ ## name (uint8_t *dst/*align width (8 or 16)*/, uint8_t *src/*align 1*/, int stride);\
b32afefe7d33 * UINTX -> uintx_t INTX -> intx_t
kabi
parents: 1040
diff changeset
159 void ff_avg_ ## name (uint8_t *dst/*align width (8 or 16)*/, uint8_t *src/*align 1*/, int stride);
853
eacc2dd8fd9d * using DSPContext - so each codec could use its local (sub)set of CPU extension
kabi
parents: 838
diff changeset
160
984
e162c09efbe7 qpel fix
michaelni
parents: 975
diff changeset
161 DEF_OLD_QPEL(qpel16_mc11_old_c)
e162c09efbe7 qpel fix
michaelni
parents: 975
diff changeset
162 DEF_OLD_QPEL(qpel16_mc31_old_c)
e162c09efbe7 qpel fix
michaelni
parents: 975
diff changeset
163 DEF_OLD_QPEL(qpel16_mc12_old_c)
e162c09efbe7 qpel fix
michaelni
parents: 975
diff changeset
164 DEF_OLD_QPEL(qpel16_mc32_old_c)
e162c09efbe7 qpel fix
michaelni
parents: 975
diff changeset
165 DEF_OLD_QPEL(qpel16_mc13_old_c)
e162c09efbe7 qpel fix
michaelni
parents: 975
diff changeset
166 DEF_OLD_QPEL(qpel16_mc33_old_c)
e162c09efbe7 qpel fix
michaelni
parents: 975
diff changeset
167 DEF_OLD_QPEL(qpel8_mc11_old_c)
e162c09efbe7 qpel fix
michaelni
parents: 975
diff changeset
168 DEF_OLD_QPEL(qpel8_mc31_old_c)
e162c09efbe7 qpel fix
michaelni
parents: 975
diff changeset
169 DEF_OLD_QPEL(qpel8_mc12_old_c)
e162c09efbe7 qpel fix
michaelni
parents: 975
diff changeset
170 DEF_OLD_QPEL(qpel8_mc32_old_c)
e162c09efbe7 qpel fix
michaelni
parents: 975
diff changeset
171 DEF_OLD_QPEL(qpel8_mc13_old_c)
e162c09efbe7 qpel fix
michaelni
parents: 975
diff changeset
172 DEF_OLD_QPEL(qpel8_mc33_old_c)
651
45e8f39fda50 put/avg_pixels16
michaelni
parents: 638
diff changeset
173
45e8f39fda50 put/avg_pixels16
michaelni
parents: 638
diff changeset
174 #define CALL_2X_PIXELS(a, b, n)\
45e8f39fda50 put/avg_pixels16
michaelni
parents: 638
diff changeset
175 static void a(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
45e8f39fda50 put/avg_pixels16
michaelni
parents: 638
diff changeset
176 b(block , pixels , line_size, h);\
45e8f39fda50 put/avg_pixels16
michaelni
parents: 638
diff changeset
177 b(block+n, pixels+n, line_size, h);\
45e8f39fda50 put/avg_pixels16
michaelni
parents: 638
diff changeset
178 }
255
db20b987c32d divx5-gmc support
michaelni
parents: 252
diff changeset
179
0
986e461dc072 Initial revision
glantau
parents:
diff changeset
180 /* motion estimation */
1709
39a209b0c82c info about h and tpel
michael
parents: 1708
diff changeset
181 // h is limited to {width/2, width, 2*width} but never larger than 16 and never smaller then 2
5127
4dbe6578f811 misc spelling fixes
diego
parents: 4999
diff changeset
182 // although currently h<4 is not used as functions with width <8 are neither used nor implemented
1708
dea5b2946999 interlaced motion estimation
michael
parents: 1653
diff changeset
183 typedef int (*me_cmp_func)(void /*MpegEncContext*/ *s, uint8_t *blk1/*align width (8 or 16)*/, uint8_t *blk2/*align 1*/, int line_size, int h)/* __attribute__ ((const))*/;
936
caa77cd960c0 qpel encoding
michaelni
parents: 894
diff changeset
184
1102
eeaa1e837efb more dox
michaelni
parents: 1092
diff changeset
185 /**
6438
e1dd408a7864 move ff_init_scantable() into dsputil
aurel
parents: 6437
diff changeset
186 * Scantable.
e1dd408a7864 move ff_init_scantable() into dsputil
aurel
parents: 6437
diff changeset
187 */
e1dd408a7864 move ff_init_scantable() into dsputil
aurel
parents: 6437
diff changeset
188 typedef struct ScanTable{
e1dd408a7864 move ff_init_scantable() into dsputil
aurel
parents: 6437
diff changeset
189 const uint8_t *scantable;
e1dd408a7864 move ff_init_scantable() into dsputil
aurel
parents: 6437
diff changeset
190 uint8_t permutated[64];
e1dd408a7864 move ff_init_scantable() into dsputil
aurel
parents: 6437
diff changeset
191 uint8_t raster_end[64];
8590
7a463923ecd1 Change semantic of CONFIG_*, HAVE_* and ARCH_*.
aurel
parents: 8573
diff changeset
192 #if ARCH_PPC
6438
e1dd408a7864 move ff_init_scantable() into dsputil
aurel
parents: 6437
diff changeset
193 /** Used by dct_quantize_altivec to find last-non-zero */
10961
34a65026fa06 Move array specifiers outside DECLARE_ALIGNED() invocations
mru
parents: 10944
diff changeset
194 DECLARE_ALIGNED(16, uint8_t, inverse)[64];
6438
e1dd408a7864 move ff_init_scantable() into dsputil
aurel
parents: 6437
diff changeset
195 #endif
e1dd408a7864 move ff_init_scantable() into dsputil
aurel
parents: 6437
diff changeset
196 } ScanTable;
e1dd408a7864 move ff_init_scantable() into dsputil
aurel
parents: 6437
diff changeset
197
e1dd408a7864 move ff_init_scantable() into dsputil
aurel
parents: 6437
diff changeset
198 void ff_init_scantable(uint8_t *, ScanTable *st, const uint8_t *src_scantable);
e1dd408a7864 move ff_init_scantable() into dsputil
aurel
parents: 6437
diff changeset
199
6445
2b553c57ec51 move ff_emulated_edge_mc() to dsputil
aurel
parents: 6440
diff changeset
200 void ff_emulated_edge_mc(uint8_t *buf, uint8_t *src, int linesize,
2b553c57ec51 move ff_emulated_edge_mc() to dsputil
aurel
parents: 6440
diff changeset
201 int block_w, int block_h,
2b553c57ec51 move ff_emulated_edge_mc() to dsputil
aurel
parents: 6440
diff changeset
202 int src_x, int src_y, int w, int h);
2b553c57ec51 move ff_emulated_edge_mc() to dsputil
aurel
parents: 6440
diff changeset
203
6438
e1dd408a7864 move ff_init_scantable() into dsputil
aurel
parents: 6437
diff changeset
204 /**
1102
eeaa1e837efb more dox
michaelni
parents: 1092
diff changeset
205 * DSPContext.
eeaa1e837efb more dox
michaelni
parents: 1092
diff changeset
206 */
853
eacc2dd8fd9d * using DSPContext - so each codec could use its local (sub)set of CPU extension
kabi
parents: 838
diff changeset
207 typedef struct DSPContext {
eacc2dd8fd9d * using DSPContext - so each codec could use its local (sub)set of CPU extension
kabi
parents: 838
diff changeset
208 /* pixel ops : interface with DCT */
1064
b32afefe7d33 * UINTX -> uintx_t INTX -> intx_t
kabi
parents: 1040
diff changeset
209 void (*get_pixels)(DCTELEM *block/*align 16*/, const uint8_t *pixels/*align 8*/, int line_size);
b32afefe7d33 * UINTX -> uintx_t INTX -> intx_t
kabi
parents: 1040
diff changeset
210 void (*diff_pixels)(DCTELEM *block/*align 16*/, const uint8_t *s1/*align 8*/, const uint8_t *s2/*align 8*/, int stride);
b32afefe7d33 * UINTX -> uintx_t INTX -> intx_t
kabi
parents: 1040
diff changeset
211 void (*put_pixels_clamped)(const DCTELEM *block/*align 16*/, uint8_t *pixels/*align 8*/, int line_size);
1984
ef919e9ef73e separate out put_signed_pixels_clamped() into its own function and
melanson
parents: 1977
diff changeset
212 void (*put_signed_pixels_clamped)(const DCTELEM *block/*align 16*/, uint8_t *pixels/*align 8*/, int line_size);
11231
0fc1cdd984b7 Bink video decoder
kostya
parents: 11196
diff changeset
213 void (*put_pixels_nonclamped)(const DCTELEM *block/*align 16*/, uint8_t *pixels/*align 8*/, int line_size);
1064
b32afefe7d33 * UINTX -> uintx_t INTX -> intx_t
kabi
parents: 1040
diff changeset
214 void (*add_pixels_clamped)(const DCTELEM *block/*align 16*/, uint8_t *pixels/*align 8*/, int line_size);
2763
01ed2c3b8f08 H.264 lossless mode
lorenm
parents: 2755
diff changeset
215 void (*add_pixels8)(uint8_t *pixels, DCTELEM *block, int line_size);
01ed2c3b8f08 H.264 lossless mode
lorenm
parents: 2755
diff changeset
216 void (*add_pixels4)(uint8_t *pixels, DCTELEM *block, int line_size);
4988
689490842cf5 factor sum_abs_dctelem out of dct_sad, and simd it.
lorenm
parents: 4962
diff changeset
217 int (*sum_abs_dctelem)(DCTELEM *block/*align 16*/);
1136
9eb49e9e8dd9 halfpel MC doxy
michaelni
parents: 1111
diff changeset
218 /**
9eb49e9e8dd9 halfpel MC doxy
michaelni
parents: 1111
diff changeset
219 * translational global motion compensation.
9eb49e9e8dd9 halfpel MC doxy
michaelni
parents: 1111
diff changeset
220 */
1064
b32afefe7d33 * UINTX -> uintx_t INTX -> intx_t
kabi
parents: 1040
diff changeset
221 void (*gmc1)(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int srcStride, int h, int x16, int y16, int rounder);
1136
9eb49e9e8dd9 halfpel MC doxy
michaelni
parents: 1111
diff changeset
222 /**
9eb49e9e8dd9 halfpel MC doxy
michaelni
parents: 1111
diff changeset
223 * global motion compensation.
9eb49e9e8dd9 halfpel MC doxy
michaelni
parents: 1111
diff changeset
224 */
1064
b32afefe7d33 * UINTX -> uintx_t INTX -> intx_t
kabi
parents: 1040
diff changeset
225 void (*gmc )(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int stride, int h, int ox, int oy,
2979
bfabfdf9ce55 COSMETICS: tabs --> spaces, some prettyprinting
diego
parents: 2967
diff changeset
226 int dxx, int dxy, int dyx, int dyy, int shift, int r, int width, int height);
8288
800444234375 clear_block mmx
lorenm
parents: 8233
diff changeset
227 void (*clear_block)(DCTELEM *block/*align 16*/);
853
eacc2dd8fd9d * using DSPContext - so each codec could use its local (sub)set of CPU extension
kabi
parents: 838
diff changeset
228 void (*clear_blocks)(DCTELEM *blocks/*align 16*/);
1064
b32afefe7d33 * UINTX -> uintx_t INTX -> intx_t
kabi
parents: 1040
diff changeset
229 int (*pix_sum)(uint8_t * pix, int line_size);
b32afefe7d33 * UINTX -> uintx_t INTX -> intx_t
kabi
parents: 1040
diff changeset
230 int (*pix_norm1)(uint8_t * pix, int line_size);
1708
dea5b2946999 interlaced motion estimation
michael
parents: 1653
diff changeset
231 // 16x16 8x8 4x4 2x2 16x8 8x4 4x2 8x16 4x8 2x4
2967
ef2149182f1c COSMETICS: Remove all trailing whitespace.
diego
parents: 2887
diff changeset
232
8976
e7d87561b42b Making the arrays accomodate an extra intra 8x8 cmp function
romansh
parents: 8785
diff changeset
233 me_cmp_func sad[6]; /* identical to pix_absAxA except additional void * */
e7d87561b42b Making the arrays accomodate an extra intra 8x8 cmp function
romansh
parents: 8785
diff changeset
234 me_cmp_func sse[6];
e7d87561b42b Making the arrays accomodate an extra intra 8x8 cmp function
romansh
parents: 8785
diff changeset
235 me_cmp_func hadamard8_diff[6];
e7d87561b42b Making the arrays accomodate an extra intra 8x8 cmp function
romansh
parents: 8785
diff changeset
236 me_cmp_func dct_sad[6];
e7d87561b42b Making the arrays accomodate an extra intra 8x8 cmp function
romansh
parents: 8785
diff changeset
237 me_cmp_func quant_psnr[6];
e7d87561b42b Making the arrays accomodate an extra intra 8x8 cmp function
romansh
parents: 8785
diff changeset
238 me_cmp_func bit[6];
e7d87561b42b Making the arrays accomodate an extra intra 8x8 cmp function
romansh
parents: 8785
diff changeset
239 me_cmp_func rd[6];
e7d87561b42b Making the arrays accomodate an extra intra 8x8 cmp function
romansh
parents: 8785
diff changeset
240 me_cmp_func vsad[6];
e7d87561b42b Making the arrays accomodate an extra intra 8x8 cmp function
romansh
parents: 8785
diff changeset
241 me_cmp_func vsse[6];
e7d87561b42b Making the arrays accomodate an extra intra 8x8 cmp function
romansh
parents: 8785
diff changeset
242 me_cmp_func nsse[6];
e7d87561b42b Making the arrays accomodate an extra intra 8x8 cmp function
romansh
parents: 8785
diff changeset
243 me_cmp_func w53[6];
e7d87561b42b Making the arrays accomodate an extra intra 8x8 cmp function
romansh
parents: 8785
diff changeset
244 me_cmp_func w97[6];
e7d87561b42b Making the arrays accomodate an extra intra 8x8 cmp function
romansh
parents: 8785
diff changeset
245 me_cmp_func dct_max[6];
e7d87561b42b Making the arrays accomodate an extra intra 8x8 cmp function
romansh
parents: 8785
diff changeset
246 me_cmp_func dct264_sad[6];
936
caa77cd960c0 qpel encoding
michaelni
parents: 894
diff changeset
247
8976
e7d87561b42b Making the arrays accomodate an extra intra 8x8 cmp function
romansh
parents: 8785
diff changeset
248 me_cmp_func me_pre_cmp[6];
e7d87561b42b Making the arrays accomodate an extra intra 8x8 cmp function
romansh
parents: 8785
diff changeset
249 me_cmp_func me_cmp[6];
e7d87561b42b Making the arrays accomodate an extra intra 8x8 cmp function
romansh
parents: 8785
diff changeset
250 me_cmp_func me_sub_cmp[6];
e7d87561b42b Making the arrays accomodate an extra intra 8x8 cmp function
romansh
parents: 8785
diff changeset
251 me_cmp_func mb_cmp[6];
e7d87561b42b Making the arrays accomodate an extra intra 8x8 cmp function
romansh
parents: 8785
diff changeset
252 me_cmp_func ildct_cmp[6]; //only width 16 used
e7d87561b42b Making the arrays accomodate an extra intra 8x8 cmp function
romansh
parents: 8785
diff changeset
253 me_cmp_func frame_skip_cmp[6]; //only width 8 used
853
eacc2dd8fd9d * using DSPContext - so each codec could use its local (sub)set of CPU extension
kabi
parents: 838
diff changeset
254
5255
669a97223dc7 make arguments to ssd_int8_vs_int16() const
mru
parents: 5149
diff changeset
255 int (*ssd_int8_vs_int16)(const int8_t *pix1, const int16_t *pix2,
669a97223dc7 make arguments to ssd_int8_vs_int16() const
mru
parents: 5149
diff changeset
256 int size);
4749
7011f597e473 mmx 16-bit ssd. 2.3x faster svq1 encoding.
lorenm
parents: 4436
diff changeset
257
1136
9eb49e9e8dd9 halfpel MC doxy
michaelni
parents: 1111
diff changeset
258 /**
9eb49e9e8dd9 halfpel MC doxy
michaelni
parents: 1111
diff changeset
259 * Halfpel motion compensation with rounding (a+b+1)>>1.
4751
0d48b61e655d Typos in comments
lu_zero
parents: 4749
diff changeset
260 * this is an array[4][4] of motion compensation functions for 4
1713
49e46374baa4 put_pixels width comment fix
michael
parents: 1709
diff changeset
261 * horizontal blocksizes (8,16) and the 4 halfpel positions<br>
1213
a71fe70b428b doc fixes (forgot to commit)
michaelni
parents: 1201
diff changeset
262 * *pixels_tab[ 0->16xH 1->8xH ][ xhalfpel + 2*yhalfpel ]
1136
9eb49e9e8dd9 halfpel MC doxy
michaelni
parents: 1111
diff changeset
263 * @param block destination where the result is stored
9eb49e9e8dd9 halfpel MC doxy
michaelni
parents: 1111
diff changeset
264 * @param pixels source
9eb49e9e8dd9 halfpel MC doxy
michaelni
parents: 1111
diff changeset
265 * @param line_size number of bytes in a horizontal line of block
9eb49e9e8dd9 halfpel MC doxy
michaelni
parents: 1111
diff changeset
266 * @param h height
9eb49e9e8dd9 halfpel MC doxy
michaelni
parents: 1111
diff changeset
267 */
1267
85b71f9f7450 moving the svq3 motion compensation stuff to dsputil (this also means that existing optimized halfpel code is used now ...)
michaelni
parents: 1264
diff changeset
268 op_pixels_func put_pixels_tab[4][4];
1136
9eb49e9e8dd9 halfpel MC doxy
michaelni
parents: 1111
diff changeset
269
9eb49e9e8dd9 halfpel MC doxy
michaelni
parents: 1111
diff changeset
270 /**
9eb49e9e8dd9 halfpel MC doxy
michaelni
parents: 1111
diff changeset
271 * Halfpel motion compensation with rounding (a+b+1)>>1.
2967
ef2149182f1c COSMETICS: Remove all trailing whitespace.
diego
parents: 2887
diff changeset
272 * This is an array[4][4] of motion compensation functions for 4
1713
49e46374baa4 put_pixels width comment fix
michael
parents: 1709
diff changeset
273 * horizontal blocksizes (8,16) and the 4 halfpel positions<br>
1213
a71fe70b428b doc fixes (forgot to commit)
michaelni
parents: 1201
diff changeset
274 * *pixels_tab[ 0->16xH 1->8xH ][ xhalfpel + 2*yhalfpel ]
1136
9eb49e9e8dd9 halfpel MC doxy
michaelni
parents: 1111
diff changeset
275 * @param block destination into which the result is averaged (a+b+1)>>1
9eb49e9e8dd9 halfpel MC doxy
michaelni
parents: 1111
diff changeset
276 * @param pixels source
9eb49e9e8dd9 halfpel MC doxy
michaelni
parents: 1111
diff changeset
277 * @param line_size number of bytes in a horizontal line of block
9eb49e9e8dd9 halfpel MC doxy
michaelni
parents: 1111
diff changeset
278 * @param h height
9eb49e9e8dd9 halfpel MC doxy
michaelni
parents: 1111
diff changeset
279 */
1319
449f6e32b425 added support for B-frames and multiple slices
tmmm
parents: 1273
diff changeset
280 op_pixels_func avg_pixels_tab[4][4];
1136
9eb49e9e8dd9 halfpel MC doxy
michaelni
parents: 1111
diff changeset
281
9eb49e9e8dd9 halfpel MC doxy
michaelni
parents: 1111
diff changeset
282 /**
9eb49e9e8dd9 halfpel MC doxy
michaelni
parents: 1111
diff changeset
283 * Halfpel motion compensation with no rounding (a+b)>>1.
4751
0d48b61e655d Typos in comments
lu_zero
parents: 4749
diff changeset
284 * this is an array[2][4] of motion compensation functions for 2
1225
michaelni
parents: 1213
diff changeset
285 * horizontal blocksizes (8,16) and the 4 halfpel positions<br>
1213
a71fe70b428b doc fixes (forgot to commit)
michaelni
parents: 1201
diff changeset
286 * *pixels_tab[ 0->16xH 1->8xH ][ xhalfpel + 2*yhalfpel ]
1136
9eb49e9e8dd9 halfpel MC doxy
michaelni
parents: 1111
diff changeset
287 * @param block destination where the result is stored
9eb49e9e8dd9 halfpel MC doxy
michaelni
parents: 1111
diff changeset
288 * @param pixels source
9eb49e9e8dd9 halfpel MC doxy
michaelni
parents: 1111
diff changeset
289 * @param line_size number of bytes in a horizontal line of block
9eb49e9e8dd9 halfpel MC doxy
michaelni
parents: 1111
diff changeset
290 * @param h height
9eb49e9e8dd9 halfpel MC doxy
michaelni
parents: 1111
diff changeset
291 */
2075
aae3b4e66560 chroma me fix
michael
parents: 2065
diff changeset
292 op_pixels_func put_no_rnd_pixels_tab[4][4];
1136
9eb49e9e8dd9 halfpel MC doxy
michaelni
parents: 1111
diff changeset
293
9eb49e9e8dd9 halfpel MC doxy
michaelni
parents: 1111
diff changeset
294 /**
9eb49e9e8dd9 halfpel MC doxy
michaelni
parents: 1111
diff changeset
295 * Halfpel motion compensation with no rounding (a+b)>>1.
4751
0d48b61e655d Typos in comments
lu_zero
parents: 4749
diff changeset
296 * this is an array[2][4] of motion compensation functions for 2
1225
michaelni
parents: 1213
diff changeset
297 * horizontal blocksizes (8,16) and the 4 halfpel positions<br>
1213
a71fe70b428b doc fixes (forgot to commit)
michaelni
parents: 1201
diff changeset
298 * *pixels_tab[ 0->16xH 1->8xH ][ xhalfpel + 2*yhalfpel ]
1136
9eb49e9e8dd9 halfpel MC doxy
michaelni
parents: 1111
diff changeset
299 * @param block destination into which the result is averaged (a+b)>>1
9eb49e9e8dd9 halfpel MC doxy
michaelni
parents: 1111
diff changeset
300 * @param pixels source
9eb49e9e8dd9 halfpel MC doxy
michaelni
parents: 1111
diff changeset
301 * @param line_size number of bytes in a horizontal line of block
9eb49e9e8dd9 halfpel MC doxy
michaelni
parents: 1111
diff changeset
302 * @param h height
9eb49e9e8dd9 halfpel MC doxy
michaelni
parents: 1111
diff changeset
303 */
2075
aae3b4e66560 chroma me fix
michael
parents: 2065
diff changeset
304 op_pixels_func avg_no_rnd_pixels_tab[4][4];
2967
ef2149182f1c COSMETICS: Remove all trailing whitespace.
diego
parents: 2887
diff changeset
305
1864
9a73e6dc5d68 fix motion compensation with (x+1/2,y+1/2) MVs
michael
parents: 1784
diff changeset
306 void (*put_no_rnd_pixels_l2[2])(uint8_t *block/*align width (8 or 16)*/, const uint8_t *a/*align 1*/, const uint8_t *b/*align 1*/, int line_size, int h);
2967
ef2149182f1c COSMETICS: Remove all trailing whitespace.
diego
parents: 2887
diff changeset
307
1267
85b71f9f7450 moving the svq3 motion compensation stuff to dsputil (this also means that existing optimized halfpel code is used now ...)
michaelni
parents: 1264
diff changeset
308 /**
85b71f9f7450 moving the svq3 motion compensation stuff to dsputil (this also means that existing optimized halfpel code is used now ...)
michaelni
parents: 1264
diff changeset
309 * Thirdpel motion compensation with rounding (a+b+1)>>1.
4751
0d48b61e655d Typos in comments
lu_zero
parents: 4749
diff changeset
310 * this is an array[12] of motion compensation functions for the 9 thirdpe
0d48b61e655d Typos in comments
lu_zero
parents: 4749
diff changeset
311 * positions<br>
1267
85b71f9f7450 moving the svq3 motion compensation stuff to dsputil (this also means that existing optimized halfpel code is used now ...)
michaelni
parents: 1264
diff changeset
312 * *pixels_tab[ xthirdpel + 4*ythirdpel ]
85b71f9f7450 moving the svq3 motion compensation stuff to dsputil (this also means that existing optimized halfpel code is used now ...)
michaelni
parents: 1264
diff changeset
313 * @param block destination where the result is stored
85b71f9f7450 moving the svq3 motion compensation stuff to dsputil (this also means that existing optimized halfpel code is used now ...)
michaelni
parents: 1264
diff changeset
314 * @param pixels source
85b71f9f7450 moving the svq3 motion compensation stuff to dsputil (this also means that existing optimized halfpel code is used now ...)
michaelni
parents: 1264
diff changeset
315 * @param line_size number of bytes in a horizontal line of block
85b71f9f7450 moving the svq3 motion compensation stuff to dsputil (this also means that existing optimized halfpel code is used now ...)
michaelni
parents: 1264
diff changeset
316 * @param h height
85b71f9f7450 moving the svq3 motion compensation stuff to dsputil (this also means that existing optimized halfpel code is used now ...)
michaelni
parents: 1264
diff changeset
317 */
85b71f9f7450 moving the svq3 motion compensation stuff to dsputil (this also means that existing optimized halfpel code is used now ...)
michaelni
parents: 1264
diff changeset
318 tpel_mc_func put_tpel_pixels_tab[11]; //FIXME individual func ptr per width?
1319
449f6e32b425 added support for B-frames and multiple slices
tmmm
parents: 1273
diff changeset
319 tpel_mc_func avg_tpel_pixels_tab[11]; //FIXME individual func ptr per width?
449f6e32b425 added support for B-frames and multiple slices
tmmm
parents: 1273
diff changeset
320
853
eacc2dd8fd9d * using DSPContext - so each codec could use its local (sub)set of CPU extension
kabi
parents: 838
diff changeset
321 qpel_mc_func put_qpel_pixels_tab[2][16];
eacc2dd8fd9d * using DSPContext - so each codec could use its local (sub)set of CPU extension
kabi
parents: 838
diff changeset
322 qpel_mc_func avg_qpel_pixels_tab[2][16];
eacc2dd8fd9d * using DSPContext - so each codec could use its local (sub)set of CPU extension
kabi
parents: 838
diff changeset
323 qpel_mc_func put_no_rnd_qpel_pixels_tab[2][16];
eacc2dd8fd9d * using DSPContext - so each codec could use its local (sub)set of CPU extension
kabi
parents: 838
diff changeset
324 qpel_mc_func avg_no_rnd_qpel_pixels_tab[2][16];
936
caa77cd960c0 qpel encoding
michaelni
parents: 894
diff changeset
325 qpel_mc_func put_mspel_pixels_tab[8];
2967
ef2149182f1c COSMETICS: Remove all trailing whitespace.
diego
parents: 2887
diff changeset
326
1168
5af9aeadbdc3 H264 decoder & demuxer
michaelni
parents: 1136
diff changeset
327 /**
4751
0d48b61e655d Typos in comments
lu_zero
parents: 4749
diff changeset
328 * h264 Chroma MC
1168
5af9aeadbdc3 H264 decoder & demuxer
michaelni
parents: 1136
diff changeset
329 */
5af9aeadbdc3 H264 decoder & demuxer
michaelni
parents: 1136
diff changeset
330 h264_chroma_mc_func put_h264_chroma_pixels_tab[3];
9439
ef3a7b711cc0 Rename put_no_rnd_h264_chroma* to reflect its usage in VC1 only
conrad
parents: 9437
diff changeset
331 h264_chroma_mc_func avg_h264_chroma_pixels_tab[3];
3663
8e180a3276fe New function for chroma MC (will be used in VC-1)
kostya
parents: 3660
diff changeset
332 /* This is really one func used in VC-1 decoding */
9439
ef3a7b711cc0 Rename put_no_rnd_h264_chroma* to reflect its usage in VC1 only
conrad
parents: 9437
diff changeset
333 h264_chroma_mc_func put_no_rnd_vc1_chroma_pixels_tab[3];
9440
daee921fb6bb VC1: add and use avg_no_rnd chroma MC functions
conrad
parents: 9439
diff changeset
334 h264_chroma_mc_func avg_no_rnd_vc1_chroma_pixels_tab[3];
853
eacc2dd8fd9d * using DSPContext - so each codec could use its local (sub)set of CPU extension
kabi
parents: 838
diff changeset
335
3020
c75fb0747e74 use h264 MC functions for 2xX Xx2 blocks in snow too
michael
parents: 3010
diff changeset
336 qpel_mc_func put_h264_qpel_pixels_tab[4][16];
c75fb0747e74 use h264 MC functions for 2xX Xx2 blocks in snow too
michael
parents: 3010
diff changeset
337 qpel_mc_func avg_h264_qpel_pixels_tab[4][16];
2967
ef2149182f1c COSMETICS: Remove all trailing whitespace.
diego
parents: 2887
diff changeset
338
3807
6a40092eb9e6 approximate qpel functions: sacrifice some quality for some decoding speed. enabled on B-frames with -lavdopts fast.
lorenm
parents: 3746
diff changeset
339 qpel_mc_func put_2tap_qpel_pixels_tab[4][16];
6a40092eb9e6 approximate qpel functions: sacrifice some quality for some decoding speed. enabled on B-frames with -lavdopts fast.
lorenm
parents: 3746
diff changeset
340 qpel_mc_func avg_2tap_qpel_pixels_tab[4][16];
6a40092eb9e6 approximate qpel functions: sacrifice some quality for some decoding speed. enabled on B-frames with -lavdopts fast.
lorenm
parents: 3746
diff changeset
341
3395
adccbf4a1040 CAVS decoder by (Stefan Gehrer stefan.gehrer gmx.de)
michael
parents: 3279
diff changeset
342 /* AVS specific */
adccbf4a1040 CAVS decoder by (Stefan Gehrer stefan.gehrer gmx.de)
michael
parents: 3279
diff changeset
343 qpel_mc_func put_cavs_qpel_pixels_tab[2][16];
adccbf4a1040 CAVS decoder by (Stefan Gehrer stefan.gehrer gmx.de)
michael
parents: 3279
diff changeset
344 qpel_mc_func avg_cavs_qpel_pixels_tab[2][16];
adccbf4a1040 CAVS decoder by (Stefan Gehrer stefan.gehrer gmx.de)
michael
parents: 3279
diff changeset
345 void (*cavs_filter_lv)(uint8_t *pix, int stride, int alpha, int beta, int tc, int bs1, int bs2);
adccbf4a1040 CAVS decoder by (Stefan Gehrer stefan.gehrer gmx.de)
michael
parents: 3279
diff changeset
346 void (*cavs_filter_lh)(uint8_t *pix, int stride, int alpha, int beta, int tc, int bs1, int bs2);
adccbf4a1040 CAVS decoder by (Stefan Gehrer stefan.gehrer gmx.de)
michael
parents: 3279
diff changeset
347 void (*cavs_filter_cv)(uint8_t *pix, int stride, int alpha, int beta, int tc, int bs1, int bs2);
adccbf4a1040 CAVS decoder by (Stefan Gehrer stefan.gehrer gmx.de)
michael
parents: 3279
diff changeset
348 void (*cavs_filter_ch)(uint8_t *pix, int stride, int alpha, int beta, int tc, int bs1, int bs2);
adccbf4a1040 CAVS decoder by (Stefan Gehrer stefan.gehrer gmx.de)
michael
parents: 3279
diff changeset
349 void (*cavs_idct8_add)(uint8_t *dst, DCTELEM *block, int stride);
adccbf4a1040 CAVS decoder by (Stefan Gehrer stefan.gehrer gmx.de)
michael
parents: 3279
diff changeset
350
1708
dea5b2946999 interlaced motion estimation
michael
parents: 1653
diff changeset
351 me_cmp_func pix_abs[2][4];
2967
ef2149182f1c COSMETICS: Remove all trailing whitespace.
diego
parents: 2887
diff changeset
352
866
725ef4ea3ecc huffyuv
michaelni
parents: 862
diff changeset
353 /* huffyuv specific */
725ef4ea3ecc huffyuv
michaelni
parents: 862
diff changeset
354 void (*add_bytes)(uint8_t *dst/*align 16*/, uint8_t *src/*align 16*/, int w);
6384
0a403ade8c81 simd and unroll png_filter_row
lorenm
parents: 6363
diff changeset
355 void (*add_bytes_l2)(uint8_t *dst/*align 16*/, uint8_t *src1/*align 16*/, uint8_t *src2/*align 16*/, int w);
936
caa77cd960c0 qpel encoding
michaelni
parents: 894
diff changeset
356 void (*diff_bytes)(uint8_t *dst/*align 16*/, uint8_t *src1/*align 16*/, uint8_t *src2/*align 1*/,int w);
1527
8ffd0c00e6df mmx2 optimization of huffyuv median encoding
michael
parents: 1446
diff changeset
357 /**
8ffd0c00e6df mmx2 optimization of huffyuv median encoding
michael
parents: 1446
diff changeset
358 * subtract huffyuv's variant of median prediction
8ffd0c00e6df mmx2 optimization of huffyuv median encoding
michael
parents: 1446
diff changeset
359 * note, this might read from src1[-1], src2[-1]
8ffd0c00e6df mmx2 optimization of huffyuv median encoding
michael
parents: 1446
diff changeset
360 */
10431
546b7ebeaf07 huffyuv: add some const qualifiers
lorenm
parents: 10430
diff changeset
361 void (*sub_hfyu_median_prediction)(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int w, int *left, int *left_top);
546b7ebeaf07 huffyuv: add some const qualifiers
lorenm
parents: 10430
diff changeset
362 void (*add_hfyu_median_prediction)(uint8_t *dst, const uint8_t *top, const uint8_t *diff, int w, int *left, int *left_top);
10430
12c8175d6db5 simd add_hfyu_left_prediction
lorenm
parents: 10424
diff changeset
363 int (*add_hfyu_left_prediction)(uint8_t *dst, const uint8_t *src, int w, int left);
10878
a8620b001ed3 Implement alpha channel decoding for BGR HuffYUV.
astrange
parents: 10827
diff changeset
364 void (*add_hfyu_left_prediction_bgr32)(uint8_t *dst, const uint8_t *src, int w, int *red, int *green, int *blue, int *alpha);
6384
0a403ade8c81 simd and unroll png_filter_row
lorenm
parents: 6363
diff changeset
365 /* this might write to dst[w] */
0a403ade8c81 simd and unroll png_filter_row
lorenm
parents: 6363
diff changeset
366 void (*add_png_paeth_prediction)(uint8_t *dst, uint8_t *src, uint8_t *top, int w, int bpp);
6241
2799f65a24de const src for bswap
michael
parents: 6142
diff changeset
367 void (*bswap_buf)(uint32_t *dst, const uint32_t *src, int w);
2633
72e6ffa1f3a5 MMX for H.264 deblocking filter
lorenm
parents: 2415
diff changeset
368
1644
835cf346975e h263 loop filter
michael
parents: 1603
diff changeset
369 void (*h263_v_loop_filter)(uint8_t *src, int stride, int qscale);
835cf346975e h263 loop filter
michael
parents: 1603
diff changeset
370 void (*h263_h_loop_filter)(uint8_t *src, int stride, int qscale);
835cf346975e h263 loop filter
michael
parents: 1603
diff changeset
371
2045
9447bbd8a7e9 rewrite h261 loop filter
michael
parents: 2044
diff changeset
372 void (*h261_loop_filter)(uint8_t *src, int stride);
2044
b6f2add2511e h261 decoder by (Maarten Daniels <maarten.daniels at student dot luc dot ac dot be>)
michael
parents: 2024
diff changeset
373
5887
83ac4620c6ed intrax8 decoder patch by "someone"
michael
parents: 5830
diff changeset
374 void (*x8_v_loop_filter)(uint8_t *src, int stride, int qscale);
83ac4620c6ed intrax8 decoder patch by "someone"
michael
parents: 5830
diff changeset
375 void (*x8_h_loop_filter)(uint8_t *src, int stride, int qscale);
83ac4620c6ed intrax8 decoder patch by "someone"
michael
parents: 5830
diff changeset
376
11637
f7281af560fe vp3: DC-only IDCT
conrad
parents: 11499
diff changeset
377 void (*vp3_idct_dc_add)(uint8_t *dest/*align 8*/, int line_size, const DCTELEM *block/*align 16*/);
7995
1fbfce20cb79 Move VP3 loop filter to DSPContext
conrad
parents: 7760
diff changeset
378 void (*vp3_v_loop_filter)(uint8_t *src, int stride, int *bounding_values);
1fbfce20cb79 Move VP3 loop filter to DSPContext
conrad
parents: 7760
diff changeset
379 void (*vp3_h_loop_filter)(uint8_t *src, int stride, int *bounding_values);
1fbfce20cb79 Move VP3 loop filter to DSPContext
conrad
parents: 7760
diff changeset
380
8785
bee83b3f9a6b move vp6_filter_diag4() to a new vp6dsp.c file and use it throught dsputil
aurel
parents: 8760
diff changeset
381 void (*vp6_filter_diag4)(uint8_t *dst, uint8_t *src, int stride,
bee83b3f9a6b move vp6_filter_diag4() to a new vp6dsp.c file and use it throught dsputil
aurel
parents: 8760
diff changeset
382 const int16_t *h_weights,const int16_t *v_weights);
bee83b3f9a6b move vp6_filter_diag4() to a new vp6dsp.c file and use it throught dsputil
aurel
parents: 8760
diff changeset
383
3568
945caa35ee9a sse and 3dnow implementations of float->int conversion and mdct windowing.
lorenm
parents: 3555
diff changeset
384 /* assume len is a multiple of 4, and arrays are 16-byte aligned */
3536
545a15c19c91 sse & sse2 implementations of vorbis channel coupling.
lorenm
parents: 3526
diff changeset
385 void (*vorbis_inverse_coupling)(float *mag, float *ang, int blocksize);
7563
8390efaa0c03 simd downmix
lorenm
parents: 7547
diff changeset
386 void (*ac3_downmix)(float (*samples)[256], float (*matrix)[2], int out_ch, int in_ch, int len);
5737
efa3c1f9259a sse2 version of compute_autocorr().
lorenm
parents: 5728
diff changeset
387 /* no alignment needed */
10424
94595d0e617c Move autocorrelation function from flacenc.c to lpc.c. Also rename the
jbr
parents: 10420
diff changeset
388 void (*lpc_compute_autocorr)(const int32_t *data, int len, int lag, double *autoc);
3574
f549d1e685f7 vorbis simd tweaks
lorenm
parents: 3568
diff changeset
389 /* assume len is a multiple of 8, and arrays are 16-byte aligned */
3568
945caa35ee9a sse and 3dnow implementations of float->int conversion and mdct windowing.
lorenm
parents: 3555
diff changeset
390 void (*vector_fmul)(float *dst, const float *src, int len);
945caa35ee9a sse and 3dnow implementations of float->int conversion and mdct windowing.
lorenm
parents: 3555
diff changeset
391 void (*vector_fmul_reverse)(float *dst, const float *src0, const float *src1, int len);
945caa35ee9a sse and 3dnow implementations of float->int conversion and mdct windowing.
lorenm
parents: 3555
diff changeset
392 /* assume len is a multiple of 8, and src arrays are 16-byte aligned */
10300
4d1b9ca628fc Drop unused args from vector_fmul_add_add, simpify code, and rename
mru
parents: 10219
diff changeset
393 void (*vector_fmul_add)(float *dst, const float *src0, const float *src1, const float *src2, int len);
7261
032a49f033e8 simplify vorbis windowing
lorenm
parents: 7232
diff changeset
394 /* assume len is a multiple of 4, and arrays are 16-byte aligned */
032a49f033e8 simplify vorbis windowing
lorenm
parents: 7232
diff changeset
395 void (*vector_fmul_window)(float *dst, const float *src0, const float *src1, const float *win, float add_bias, int len);
7564
7cf793954871 simd int->float
lorenm
parents: 7563
diff changeset
396 /* assume len is a multiple of 8, and arrays are 16-byte aligned */
7cf793954871 simd int->float
lorenm
parents: 7563
diff changeset
397 void (*int32_to_float_fmul_scalar)(float *dst, const int *src, float mul, int len);
10105
7775f6627612 Mark parameter src of vector_clipf() as const
vitor
parents: 10104
diff changeset
398 void (*vector_clipf)(float *dst /* align 16 */, const float *src /* align 16 */, float min, float max, int len /* align 16 */);
10219
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
399 /**
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
400 * Multiply a vector of floats by a scalar float. Source and
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
401 * destination vectors must overlap exactly or not at all.
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
402 * @param dst result vector, 16-byte aligned
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
403 * @param src input vector, 16-byte aligned
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
404 * @param mul scalar value
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
405 * @param len length of vector, multiple of 4
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
406 */
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
407 void (*vector_fmul_scalar)(float *dst, const float *src, float mul,
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
408 int len);
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
409 /**
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
410 * Multiply a vector of floats by concatenated short vectors of
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
411 * floats and by a scalar float. Source and destination vectors
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
412 * must overlap exactly or not at all.
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
413 * [0]: short vectors of length 2, 8-byte aligned
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
414 * [1]: short vectors of length 4, 16-byte aligned
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
415 * @param dst output vector, 16-byte aligned
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
416 * @param src input vector, 16-byte aligned
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
417 * @param sv array of pointers to short vectors
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
418 * @param mul scalar value
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
419 * @param len number of elements in src and dst, multiple of 4
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
420 */
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
421 void (*vector_fmul_sv_scalar[2])(float *dst, const float *src,
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
422 const float **sv, float mul, int len);
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
423 /**
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
424 * Multiply short vectors of floats by a scalar float, store
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
425 * concatenated result.
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
426 * [0]: short vectors of length 2, 8-byte aligned
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
427 * [1]: short vectors of length 4, 16-byte aligned
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
428 * @param dst output vector, 16-byte aligned
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
429 * @param sv array of pointers to short vectors
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
430 * @param mul scalar value
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
431 * @param len number of output elements, multiple of 4
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
432 */
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
433 void (*sv_fmul_scalar[2])(float *dst, const float **sv,
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
434 float mul, int len);
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
435 /**
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
436 * Calculate the scalar product of two vectors of floats.
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
437 * @param v1 first vector, 16-byte aligned
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
438 * @param v2 second vector, 16-byte aligned
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
439 * @param len length of vectors, multiple of 4
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
440 */
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
441 float (*scalarproduct_float)(const float *v1, const float *v2, int len);
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
442 /**
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
443 * Calculate the sum and difference of two vectors of floats.
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
444 * @param v1 first input vector, sum output, 16-byte aligned
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
445 * @param v2 second input vector, difference output, 16-byte aligned
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
446 * @param len length of vectors, multiple of 4
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
447 */
87ae4a167138 Add some dsputil functions useful for AAC decoder
mru
parents: 10204
diff changeset
448 void (*butterflies_float)(float *restrict v1, float *restrict v2, int len);
3568
945caa35ee9a sse and 3dnow implementations of float->int conversion and mdct windowing.
lorenm
parents: 3555
diff changeset
449
945caa35ee9a sse and 3dnow implementations of float->int conversion and mdct windowing.
lorenm
parents: 3555
diff changeset
450 /* C version: convert floats from the range [384.0,386.0] to ints in [-32768,32767]
3660
7e1ee254a3ee Align the input buffer in ffplay, introduce a public macro for aligned declarations
lu_zero
parents: 3656
diff changeset
451 * simd versions: convert floats from [-32768.0,32767.0] without rescaling and arrays are 16byte aligned */
7218
7f3d6509628b Fix x86-64
michael
parents: 7203
diff changeset
452 void (*float_to_int16)(int16_t *dst, const float *src, long len);
7286
e267f2519248 float_to_int16_interleave: change src to an array of pointers instead of assuming it's contiguous.
lorenm
parents: 7263
diff changeset
453 void (*float_to_int16_interleave)(int16_t *dst, const float **src, long len, int channels);
3536
545a15c19c91 sse & sse2 implementations of vorbis channel coupling.
lorenm
parents: 3526
diff changeset
454
1092
f59c3f66363b MpegEncContext.(i)dct_* -> DspContext.(i)dct_*
michaelni
parents: 1065
diff changeset
455 /* (I)DCT */
f59c3f66363b MpegEncContext.(i)dct_* -> DspContext.(i)dct_*
michaelni
parents: 1065
diff changeset
456 void (*fdct)(DCTELEM *block/* align 16*/);
1567
e08df4d22d27 * introducing dct248 into the DSP context.
romansh
parents: 1565
diff changeset
457 void (*fdct248)(DCTELEM *block/* align 16*/);
2967
ef2149182f1c COSMETICS: Remove all trailing whitespace.
diego
parents: 2887
diff changeset
458
1324
7d328fd9d8a5 the return of the idct with 16bit output by ("Ivan Kalvachev" <ivan at cacad dot com>)
michaelni
parents: 1320
diff changeset
459 /* IDCT really*/
7d328fd9d8a5 the return of the idct with 16bit output by ("Ivan Kalvachev" <ivan at cacad dot com>)
michaelni
parents: 1320
diff changeset
460 void (*idct)(DCTELEM *block/* align 16*/);
2967
ef2149182f1c COSMETICS: Remove all trailing whitespace.
diego
parents: 2887
diff changeset
461
1102
eeaa1e837efb more dox
michaelni
parents: 1092
diff changeset
462 /**
1104
6a01ffc5351c cosmetics
michaelni
parents: 1102
diff changeset
463 * block -> idct -> clip to unsigned 8 bit -> dest.
1102
eeaa1e837efb more dox
michaelni
parents: 1092
diff changeset
464 * (-1392, 0, 0, ...) -> idct -> (-174, -174, ...) -> put -> (0, 0, ...)
5728
d7970c9e3049 cosmetics: typos
takis
parents: 5703
diff changeset
465 * @param line_size size in bytes of a horizontal line of dest
1102
eeaa1e837efb more dox
michaelni
parents: 1092
diff changeset
466 */
1092
f59c3f66363b MpegEncContext.(i)dct_* -> DspContext.(i)dct_*
michaelni
parents: 1065
diff changeset
467 void (*idct_put)(uint8_t *dest/*align 8*/, int line_size, DCTELEM *block/*align 16*/);
2967
ef2149182f1c COSMETICS: Remove all trailing whitespace.
diego
parents: 2887
diff changeset
468
1102
eeaa1e837efb more dox
michaelni
parents: 1092
diff changeset
469 /**
eeaa1e837efb more dox
michaelni
parents: 1092
diff changeset
470 * block -> idct -> add dest -> clip to unsigned 8 bit -> dest.
5728
d7970c9e3049 cosmetics: typos
takis
parents: 5703
diff changeset
471 * @param line_size size in bytes of a horizontal line of dest
1102
eeaa1e837efb more dox
michaelni
parents: 1092
diff changeset
472 */
1092
f59c3f66363b MpegEncContext.(i)dct_* -> DspContext.(i)dct_*
michaelni
parents: 1065
diff changeset
473 void (*idct_add)(uint8_t *dest/*align 8*/, int line_size, DCTELEM *block/*align 16*/);
2967
ef2149182f1c COSMETICS: Remove all trailing whitespace.
diego
parents: 2887
diff changeset
474
1102
eeaa1e837efb more dox
michaelni
parents: 1092
diff changeset
475 /**
1104
6a01ffc5351c cosmetics
michaelni
parents: 1102
diff changeset
476 * idct input permutation.
1241
6a873cad413e more idct_permute docs
michaelni
parents: 1225
diff changeset
477 * several optimized IDCTs need a permutated input (relative to the normal order of the reference
6a873cad413e more idct_permute docs
michaelni
parents: 1225
diff changeset
478 * IDCT)
6a873cad413e more idct_permute docs
michaelni
parents: 1225
diff changeset
479 * this permutation must be performed before the idct_put/add, note, normally this can be merged
6a873cad413e more idct_permute docs
michaelni
parents: 1225
diff changeset
480 * with the zigzag/alternate scan<br>
1102
eeaa1e837efb more dox
michaelni
parents: 1092
diff changeset
481 * an example to avoid confusion:
eeaa1e837efb more dox
michaelni
parents: 1092
diff changeset
482 * - (->decode coeffs -> zigzag reorder -> dequant -> reference idct ->...)
eeaa1e837efb more dox
michaelni
parents: 1092
diff changeset
483 * - (x -> referece dct -> reference idct -> x)
eeaa1e837efb more dox
michaelni
parents: 1092
diff changeset
484 * - (x -> referece dct -> simple_mmx_perm = idct_permutation -> simple_idct_mmx -> x)
eeaa1e837efb more dox
michaelni
parents: 1092
diff changeset
485 * - (->decode coeffs -> zigzag reorder -> simple_mmx_perm -> dequant -> simple_idct_mmx ->...)
eeaa1e837efb more dox
michaelni
parents: 1092
diff changeset
486 */
1092
f59c3f66363b MpegEncContext.(i)dct_* -> DspContext.(i)dct_*
michaelni
parents: 1065
diff changeset
487 uint8_t idct_permutation[64];
f59c3f66363b MpegEncContext.(i)dct_* -> DspContext.(i)dct_*
michaelni
parents: 1065
diff changeset
488 int idct_permutation_type;
f59c3f66363b MpegEncContext.(i)dct_* -> DspContext.(i)dct_*
michaelni
parents: 1065
diff changeset
489 #define FF_NO_IDCT_PERM 1
f59c3f66363b MpegEncContext.(i)dct_* -> DspContext.(i)dct_*
michaelni
parents: 1065
diff changeset
490 #define FF_LIBMPEG2_IDCT_PERM 2
f59c3f66363b MpegEncContext.(i)dct_* -> DspContext.(i)dct_*
michaelni
parents: 1065
diff changeset
491 #define FF_SIMPLE_IDCT_PERM 3
f59c3f66363b MpegEncContext.(i)dct_* -> DspContext.(i)dct_*
michaelni
parents: 1065
diff changeset
492 #define FF_TRANSPOSE_IDCT_PERM 4
2696
9699d325049d porting the mmx&sse2 (sse2 untested) vp3 idcts to the lavc idct API
michael
parents: 2693
diff changeset
493 #define FF_PARTTRANS_IDCT_PERM 5
6600
c3213c91124c Add a new IDCT permutation, used in xvid_sse2 and possibly future similar IDCTs.
astrange
parents: 6488
diff changeset
494 #define FF_SSE2_IDCT_PERM 6
1092
f59c3f66363b MpegEncContext.(i)dct_* -> DspContext.(i)dct_*
michaelni
parents: 1065
diff changeset
495
1784
65f7bd09f37b quantizer noise shaping optimization
michael
parents: 1765
diff changeset
496 int (*try_8x8basis)(int16_t rem[64], int16_t weight[64], int16_t basis[64], int scale);
65f7bd09f37b quantizer noise shaping optimization
michael
parents: 1765
diff changeset
497 void (*add_8x8basis)(int16_t rem[64], int16_t basis[64], int scale);
65f7bd09f37b quantizer noise shaping optimization
michael
parents: 1765
diff changeset
498 #define BASIS_SHIFT 16
65f7bd09f37b quantizer noise shaping optimization
michael
parents: 1765
diff changeset
499 #define RECON_SHIFT 6
2967
ef2149182f1c COSMETICS: Remove all trailing whitespace.
diego
parents: 2887
diff changeset
500
6437
5154ab444372 move draw_edges() into dsputil
aurel
parents: 6391
diff changeset
501 void (*draw_edges)(uint8_t *buf, int wrap, int width, int height, int w);
6449
208074826b34 move EDGE_WIDTH definition allong with draw_edges where it belongs
aurel
parents: 6445
diff changeset
502 #define EDGE_WIDTH 16
6437
5154ab444372 move draw_edges() into dsputil
aurel
parents: 6391
diff changeset
503
3215
06f98047ff26 prefetch pixels for future motion compensation. 2-5% faster h264.
lorenm
parents: 3198
diff changeset
504 void (*prefetch)(void *mem, int stride, int h);
3245
c2c29be6282e remove mpegvideo.c img resample dependancy
michael
parents: 3217
diff changeset
505
c2c29be6282e remove mpegvideo.c img resample dependancy
michael
parents: 3217
diff changeset
506 void (*shrink[4])(uint8_t *dst, int dst_wrap, const uint8_t *src, int src_wrap, int width, int height);
3526
7dc8e4a12105 New functions in DSPContext for VC-1 decoding
kostya
parents: 3395
diff changeset
507
9585
5e1d9508b62f mlpdec: Move MLP's filter_channel() to dsputils.
ramiro
parents: 9443
diff changeset
508 /* mlp/truehd functions */
9647
d0fe5dc427f0 mlp: Simplify adressing of state and coeffs arrays for both filters by making
ramiro
parents: 9585
diff changeset
509 void (*mlp_filter_channel)(int32_t *state, const int32_t *coeff,
d0fe5dc427f0 mlp: Simplify adressing of state and coeffs arrays for both filters by making
ramiro
parents: 9585
diff changeset
510 int firorder, int iirorder,
9585
5e1d9508b62f mlpdec: Move MLP's filter_channel() to dsputils.
ramiro
parents: 9443
diff changeset
511 unsigned int filter_shift, int32_t mask, int blocksize,
5e1d9508b62f mlpdec: Move MLP's filter_channel() to dsputils.
ramiro
parents: 9443
diff changeset
512 int32_t *sample_buffer);
5e1d9508b62f mlpdec: Move MLP's filter_channel() to dsputils.
ramiro
parents: 9443
diff changeset
513
3526
7dc8e4a12105 New functions in DSPContext for VC-1 decoding
kostya
parents: 3395
diff changeset
514 /* vc1 functions */
7dc8e4a12105 New functions in DSPContext for VC-1 decoding
kostya
parents: 3395
diff changeset
515 void (*vc1_inv_trans_8x8)(DCTELEM *b);
5997
90de28dfd8d6 Switch VC-1 decoder to output decoded residual immediately.
kostya
parents: 5907
diff changeset
516 void (*vc1_inv_trans_8x4)(uint8_t *dest, int line_size, DCTELEM *block);
90de28dfd8d6 Switch VC-1 decoder to output decoded residual immediately.
kostya
parents: 5907
diff changeset
517 void (*vc1_inv_trans_4x8)(uint8_t *dest, int line_size, DCTELEM *block);
90de28dfd8d6 Switch VC-1 decoder to output decoded residual immediately.
kostya
parents: 5907
diff changeset
518 void (*vc1_inv_trans_4x4)(uint8_t *dest, int line_size, DCTELEM *block);
9859
7a116de63777 idct_dc for VC-1/WMV3 decoder; ~11% faster decoding overall.
darkshikari
parents: 9658
diff changeset
519 void (*vc1_inv_trans_8x8_dc)(uint8_t *dest, int line_size, DCTELEM *block);
7a116de63777 idct_dc for VC-1/WMV3 decoder; ~11% faster decoding overall.
darkshikari
parents: 9658
diff changeset
520 void (*vc1_inv_trans_8x4_dc)(uint8_t *dest, int line_size, DCTELEM *block);
7a116de63777 idct_dc for VC-1/WMV3 decoder; ~11% faster decoding overall.
darkshikari
parents: 9658
diff changeset
521 void (*vc1_inv_trans_4x8_dc)(uint8_t *dest, int line_size, DCTELEM *block);
7a116de63777 idct_dc for VC-1/WMV3 decoder; ~11% faster decoding overall.
darkshikari
parents: 9658
diff changeset
522 void (*vc1_inv_trans_4x4_dc)(uint8_t *dest, int line_size, DCTELEM *block);
4239
30b14d0f2324 Correct rounding values in overlap filtering
kostya
parents: 4197
diff changeset
523 void (*vc1_v_overlap)(uint8_t* src, int stride);
30b14d0f2324 Correct rounding values in overlap filtering
kostya
parents: 4197
diff changeset
524 void (*vc1_h_overlap)(uint8_t* src, int stride);
9443
3970fe47fea3 Split VC1 loop filter into separate functions for h/v and size
conrad
parents: 9442
diff changeset
525 void (*vc1_v_loop_filter4)(uint8_t *src, int stride, int pq);
3970fe47fea3 Split VC1 loop filter into separate functions for h/v and size
conrad
parents: 9442
diff changeset
526 void (*vc1_h_loop_filter4)(uint8_t *src, int stride, int pq);
3970fe47fea3 Split VC1 loop filter into separate functions for h/v and size
conrad
parents: 9442
diff changeset
527 void (*vc1_v_loop_filter8)(uint8_t *src, int stride, int pq);
3970fe47fea3 Split VC1 loop filter into separate functions for h/v and size
conrad
parents: 9442
diff changeset
528 void (*vc1_h_loop_filter8)(uint8_t *src, int stride, int pq);
3970fe47fea3 Split VC1 loop filter into separate functions for h/v and size
conrad
parents: 9442
diff changeset
529 void (*vc1_v_loop_filter16)(uint8_t *src, int stride, int pq);
3970fe47fea3 Split VC1 loop filter into separate functions for h/v and size
conrad
parents: 9442
diff changeset
530 void (*vc1_h_loop_filter16)(uint8_t *src, int stride, int pq);
3526
7dc8e4a12105 New functions in DSPContext for VC-1 decoding
kostya
parents: 3395
diff changeset
531 /* put 8x8 block with bicubic interpolation and quarterpel precision
7dc8e4a12105 New functions in DSPContext for VC-1 decoding
kostya
parents: 3395
diff changeset
532 * last argument is actually round value instead of height
7dc8e4a12105 New functions in DSPContext for VC-1 decoding
kostya
parents: 3395
diff changeset
533 */
7dc8e4a12105 New functions in DSPContext for VC-1 decoding
kostya
parents: 3395
diff changeset
534 op_pixels_func put_vc1_mspel_pixels_tab[16];
9437
8aa2e86549cd VC1: Do qpel when needed for both MVs in a B frame
conrad
parents: 9436
diff changeset
535 op_pixels_func avg_vc1_mspel_pixels_tab[16];
5887
83ac4620c6ed intrax8 decoder patch by "someone"
michael
parents: 5830
diff changeset
536
83ac4620c6ed intrax8 decoder patch by "someone"
michael
parents: 5830
diff changeset
537 /* intrax8 functions */
5907
fbd10e6dfbe1 cosmetics: Fix spacial --> spatial typo in function names.
diego
parents: 5887
diff changeset
538 void (*x8_spatial_compensation[12])(uint8_t *src , uint8_t *dst, int linesize);
fbd10e6dfbe1 cosmetics: Fix spacial --> spatial typo in function names.
diego
parents: 5887
diff changeset
539 void (*x8_setup_spatial_compensation)(uint8_t *src, uint8_t *dst, int linesize,
5887
83ac4620c6ed intrax8 decoder patch by "someone"
michael
parents: 5830
diff changeset
540 int * range, int * sum, int edges);
83ac4620c6ed intrax8 decoder patch by "someone"
michael
parents: 5830
diff changeset
541
7203
87b1dfb5a98d Add several vector functions used by Monkey's Audio decoder to dsputil
kostya
parents: 7135
diff changeset
542 /**
87b1dfb5a98d Add several vector functions used by Monkey's Audio decoder to dsputil
kostya
parents: 7135
diff changeset
543 * Calculate scalar product of two vectors.
7232
cc55dd004819 Monkey's Audio decoder vector functions work on input with length
kostya
parents: 7218
diff changeset
544 * @param len length of vectors, should be multiple of 16
7203
87b1dfb5a98d Add several vector functions used by Monkey's Audio decoder to dsputil
kostya
parents: 7135
diff changeset
545 * @param shift number of bits to discard from product
87b1dfb5a98d Add several vector functions used by Monkey's Audio decoder to dsputil
kostya
parents: 7135
diff changeset
546 */
87b1dfb5a98d Add several vector functions used by Monkey's Audio decoder to dsputil
kostya
parents: 7135
diff changeset
547 int32_t (*scalarproduct_int16)(int16_t *v1, int16_t *v2/*align 16*/, int len, int shift);
10644
5da7180afadf refactor and optimize scalarproduct
lorenm
parents: 10496
diff changeset
548 /* ape functions */
5da7180afadf refactor and optimize scalarproduct
lorenm
parents: 10496
diff changeset
549 /**
5da7180afadf refactor and optimize scalarproduct
lorenm
parents: 10496
diff changeset
550 * Calculate scalar product of v1 and v2,
5da7180afadf refactor and optimize scalarproduct
lorenm
parents: 10496
diff changeset
551 * and v1[i] += v3[i] * mul
5da7180afadf refactor and optimize scalarproduct
lorenm
parents: 10496
diff changeset
552 * @param len length of vectors, should be multiple of 16
5da7180afadf refactor and optimize scalarproduct
lorenm
parents: 10496
diff changeset
553 */
5da7180afadf refactor and optimize scalarproduct
lorenm
parents: 10496
diff changeset
554 int32_t (*scalarproduct_and_madd_int16)(int16_t *v1/*align 16*/, int16_t *v2, int16_t *v3, int len, int mul);
8232
520a6e56b8c0 RV40 MC functions
kostya
parents: 8104
diff changeset
555
8233
ababfa151ced enable RV40 decoder
kostya
parents: 8232
diff changeset
556 /* rv30 functions */
ababfa151ced enable RV40 decoder
kostya
parents: 8232
diff changeset
557 qpel_mc_func put_rv30_tpel_pixels_tab[4][16];
ababfa151ced enable RV40 decoder
kostya
parents: 8232
diff changeset
558 qpel_mc_func avg_rv30_tpel_pixels_tab[4][16];
ababfa151ced enable RV40 decoder
kostya
parents: 8232
diff changeset
559
8232
520a6e56b8c0 RV40 MC functions
kostya
parents: 8104
diff changeset
560 /* rv40 functions */
520a6e56b8c0 RV40 MC functions
kostya
parents: 8104
diff changeset
561 qpel_mc_func put_rv40_qpel_pixels_tab[4][16];
520a6e56b8c0 RV40 MC functions
kostya
parents: 8104
diff changeset
562 qpel_mc_func avg_rv40_qpel_pixels_tab[4][16];
520a6e56b8c0 RV40 MC functions
kostya
parents: 8104
diff changeset
563 h264_chroma_mc_func put_rv40_chroma_pixels_tab[3];
520a6e56b8c0 RV40 MC functions
kostya
parents: 8104
diff changeset
564 h264_chroma_mc_func avg_rv40_chroma_pixels_tab[3];
11231
0fc1cdd984b7 Bink video decoder
kostya
parents: 11196
diff changeset
565
0fc1cdd984b7 Bink video decoder
kostya
parents: 11196
diff changeset
566 /* bink functions */
0fc1cdd984b7 Bink video decoder
kostya
parents: 11196
diff changeset
567 op_fill_func fill_block_tab[2];
0fc1cdd984b7 Bink video decoder
kostya
parents: 11196
diff changeset
568 void (*scale_block)(const uint8_t src[64]/*align 8*/, uint8_t *dst/*align 8*/, int linesize);
853
eacc2dd8fd9d * using DSPContext - so each codec could use its local (sub)set of CPU extension
kabi
parents: 838
diff changeset
569 } DSPContext;
eacc2dd8fd9d * using DSPContext - so each codec could use its local (sub)set of CPU extension
kabi
parents: 838
diff changeset
570
4197
bbe0bc387a19 revert bad checkin
mru
parents: 4196
diff changeset
571 void dsputil_static_init(void);
1092
f59c3f66363b MpegEncContext.(i)dct_* -> DspContext.(i)dct_*
michaelni
parents: 1065
diff changeset
572 void dsputil_init(DSPContext* p, AVCodecContext *avctx);
0
986e461dc072 Initial revision
glantau
parents:
diff changeset
573
4281
de525a2b41db ff_check_alignment to warn the user about a missaligned stack
michael
parents: 4279
diff changeset
574 int ff_check_alignment(void);
de525a2b41db ff_check_alignment to warn the user about a missaligned stack
michael
parents: 4279
diff changeset
575
764
d4726182dfd2 optimize block_permute()
michaelni
parents: 753
diff changeset
576 /**
d4726182dfd2 optimize block_permute()
michaelni
parents: 753
diff changeset
577 * permute block according to permuatation.
d4726182dfd2 optimize block_permute()
michaelni
parents: 753
diff changeset
578 * @param last last non zero element in scantable order
d4726182dfd2 optimize block_permute()
michaelni
parents: 753
diff changeset
579 */
1064
b32afefe7d33 * UINTX -> uintx_t INTX -> intx_t
kabi
parents: 1040
diff changeset
580 void ff_block_permute(DCTELEM *block, uint8_t *permutation, const uint8_t *scantable, int last);
34
2733a4c1c693 added block permutation functions
glantau
parents: 19
diff changeset
581
1729
a4a5e7521339 interlaced dct decision cleanup
michael
parents: 1713
diff changeset
582 void ff_set_cmp(DSPContext* c, me_cmp_func *cmp, int type);
a4a5e7521339 interlaced dct decision cleanup
michael
parents: 1713
diff changeset
583
2979
bfabfdf9ce55 COSMETICS: tabs --> spaces, some prettyprinting
diego
parents: 2967
diff changeset
584 #define BYTE_VEC32(c) ((c)*0x01010101UL)
1264
2fa34e615c76 cleanup
michaelni
parents: 1259
diff changeset
585
2fa34e615c76 cleanup
michaelni
parents: 1259
diff changeset
586 static inline uint32_t rnd_avg32(uint32_t a, uint32_t b)
2fa34e615c76 cleanup
michaelni
parents: 1259
diff changeset
587 {
2fa34e615c76 cleanup
michaelni
parents: 1259
diff changeset
588 return (a | b) - (((a ^ b) & ~BYTE_VEC32(0x01)) >> 1);
2fa34e615c76 cleanup
michaelni
parents: 1259
diff changeset
589 }
2fa34e615c76 cleanup
michaelni
parents: 1259
diff changeset
590
2fa34e615c76 cleanup
michaelni
parents: 1259
diff changeset
591 static inline uint32_t no_rnd_avg32(uint32_t a, uint32_t b)
2fa34e615c76 cleanup
michaelni
parents: 1259
diff changeset
592 {
2fa34e615c76 cleanup
michaelni
parents: 1259
diff changeset
593 return (a & b) + (((a ^ b) & ~BYTE_VEC32(0x01)) >> 1);
2fa34e615c76 cleanup
michaelni
parents: 1259
diff changeset
594 }
2fa34e615c76 cleanup
michaelni
parents: 1259
diff changeset
595
2184
3378d0677903 4x4 SSE compare function
michael
parents: 2090
diff changeset
596 static inline int get_penalty_factor(int lambda, int lambda2, int type){
3378d0677903 4x4 SSE compare function
michael
parents: 2090
diff changeset
597 switch(type&0xFF){
3378d0677903 4x4 SSE compare function
michael
parents: 2090
diff changeset
598 default:
3378d0677903 4x4 SSE compare function
michael
parents: 2090
diff changeset
599 case FF_CMP_SAD:
3378d0677903 4x4 SSE compare function
michael
parents: 2090
diff changeset
600 return lambda>>FF_LAMBDA_SHIFT;
3378d0677903 4x4 SSE compare function
michael
parents: 2090
diff changeset
601 case FF_CMP_DCT:
3378d0677903 4x4 SSE compare function
michael
parents: 2090
diff changeset
602 return (3*lambda)>>(FF_LAMBDA_SHIFT+1);
3378d0677903 4x4 SSE compare function
michael
parents: 2090
diff changeset
603 case FF_CMP_W53:
3378d0677903 4x4 SSE compare function
michael
parents: 2090
diff changeset
604 return (4*lambda)>>(FF_LAMBDA_SHIFT);
3378d0677903 4x4 SSE compare function
michael
parents: 2090
diff changeset
605 case FF_CMP_W97:
3378d0677903 4x4 SSE compare function
michael
parents: 2090
diff changeset
606 return (2*lambda)>>(FF_LAMBDA_SHIFT);
3378d0677903 4x4 SSE compare function
michael
parents: 2090
diff changeset
607 case FF_CMP_SATD:
3010
533c6386eca9 8x8 integer dct from x264 as cmp function (under CONFIG_GPL)
michael
parents: 2979
diff changeset
608 case FF_CMP_DCT264:
2184
3378d0677903 4x4 SSE compare function
michael
parents: 2090
diff changeset
609 return (2*lambda)>>FF_LAMBDA_SHIFT;
3378d0677903 4x4 SSE compare function
michael
parents: 2090
diff changeset
610 case FF_CMP_RD:
3378d0677903 4x4 SSE compare function
michael
parents: 2090
diff changeset
611 case FF_CMP_PSNR:
3378d0677903 4x4 SSE compare function
michael
parents: 2090
diff changeset
612 case FF_CMP_SSE:
3378d0677903 4x4 SSE compare function
michael
parents: 2090
diff changeset
613 case FF_CMP_NSSE:
3378d0677903 4x4 SSE compare function
michael
parents: 2090
diff changeset
614 return lambda2>>FF_LAMBDA_SHIFT;
3378d0677903 4x4 SSE compare function
michael
parents: 2090
diff changeset
615 case FF_CMP_BIT:
3378d0677903 4x4 SSE compare function
michael
parents: 2090
diff changeset
616 return 1;
3378d0677903 4x4 SSE compare function
michael
parents: 2090
diff changeset
617 }
3378d0677903 4x4 SSE compare function
michael
parents: 2090
diff changeset
618 }
3378d0677903 4x4 SSE compare function
michael
parents: 2090
diff changeset
619
1102
eeaa1e837efb more dox
michaelni
parents: 1092
diff changeset
620 /**
1104
6a01ffc5351c cosmetics
michaelni
parents: 1102
diff changeset
621 * Empty mmx state.
1102
eeaa1e837efb more dox
michaelni
parents: 1092
diff changeset
622 * this must be called between any dsp function and float/double code.
eeaa1e837efb more dox
michaelni
parents: 1092
diff changeset
623 * for example sin(); dsp->idct_put(); emms_c(); cos()
eeaa1e837efb more dox
michaelni
parents: 1092
diff changeset
624 */
853
eacc2dd8fd9d * using DSPContext - so each codec could use its local (sub)set of CPU extension
kabi
parents: 838
diff changeset
625 #define emms_c()
eacc2dd8fd9d * using DSPContext - so each codec could use its local (sub)set of CPU extension
kabi
parents: 838
diff changeset
626
995
edc10966b081 altivec jumbo patch by (Romain Dolbeau <dolbeaur at club-internet dot fr>)
michaelni
parents: 984
diff changeset
627 /* should be defined by architectures supporting
edc10966b081 altivec jumbo patch by (Romain Dolbeau <dolbeaur at club-internet dot fr>)
michaelni
parents: 984
diff changeset
628 one or more MultiMedia extension */
edc10966b081 altivec jumbo patch by (Romain Dolbeau <dolbeaur at club-internet dot fr>)
michaelni
parents: 984
diff changeset
629 int mm_support(void);
10115
0b76ca7e07b5 One declaration of mm_flags is enough
mru
parents: 10114
diff changeset
630 extern int mm_flags;
995
edc10966b081 altivec jumbo patch by (Romain Dolbeau <dolbeaur at club-internet dot fr>)
michaelni
parents: 984
diff changeset
631
5149
b7bae4ce8b07 move dsputil_init_foo() out from #ifdef
mru
parents: 5127
diff changeset
632 void dsputil_init_alpha(DSPContext* c, AVCodecContext *avctx);
8359
9281a8a9387a ARM: replace "armv4l" with "arm"
mru
parents: 8288
diff changeset
633 void dsputil_init_arm(DSPContext* c, AVCodecContext *avctx);
5149
b7bae4ce8b07 move dsputil_init_foo() out from #ifdef
mru
parents: 5127
diff changeset
634 void dsputil_init_bfin(DSPContext* c, AVCodecContext *avctx);
b7bae4ce8b07 move dsputil_init_foo() out from #ifdef
mru
parents: 5127
diff changeset
635 void dsputil_init_mlib(DSPContext* c, AVCodecContext *avctx);
b7bae4ce8b07 move dsputil_init_foo() out from #ifdef
mru
parents: 5127
diff changeset
636 void dsputil_init_mmi(DSPContext* c, AVCodecContext *avctx);
b7bae4ce8b07 move dsputil_init_foo() out from #ifdef
mru
parents: 5127
diff changeset
637 void dsputil_init_mmx(DSPContext* c, AVCodecContext *avctx);
b7bae4ce8b07 move dsputil_init_foo() out from #ifdef
mru
parents: 5127
diff changeset
638 void dsputil_init_ppc(DSPContext* c, AVCodecContext *avctx);
b7bae4ce8b07 move dsputil_init_foo() out from #ifdef
mru
parents: 5127
diff changeset
639 void dsputil_init_sh4(DSPContext* c, AVCodecContext *avctx);
b7bae4ce8b07 move dsputil_init_foo() out from #ifdef
mru
parents: 5127
diff changeset
640 void dsputil_init_vis(DSPContext* c, AVCodecContext *avctx);
b7bae4ce8b07 move dsputil_init_foo() out from #ifdef
mru
parents: 5127
diff changeset
641
11485
0f0cd6b5791f Separate DWT from snow and dsputil
mru
parents: 11379
diff changeset
642 void ff_dsputil_init_dwt(DSPContext *c);
11376
0b220468ba0d Move prototypes for various dsputil init functions to dsputil.h
mru
parents: 11370
diff changeset
643 void ff_cavsdsp_init(DSPContext* c, AVCodecContext *avctx);
0b220468ba0d Move prototypes for various dsputil init functions to dsputil.h
mru
parents: 11370
diff changeset
644 void ff_rv30dsp_init(DSPContext* c, AVCodecContext *avctx);
0b220468ba0d Move prototypes for various dsputil init functions to dsputil.h
mru
parents: 11370
diff changeset
645 void ff_rv40dsp_init(DSPContext* c, AVCodecContext *avctx);
0b220468ba0d Move prototypes for various dsputil init functions to dsputil.h
mru
parents: 11370
diff changeset
646 void ff_vc1dsp_init(DSPContext* c, AVCodecContext *avctx);
0b220468ba0d Move prototypes for various dsputil init functions to dsputil.h
mru
parents: 11370
diff changeset
647 void ff_intrax8dsp_init(DSPContext* c, AVCodecContext *avctx);
0b220468ba0d Move prototypes for various dsputil init functions to dsputil.h
mru
parents: 11370
diff changeset
648 void ff_mlp_init(DSPContext* c, AVCodecContext *avctx);
0b220468ba0d Move prototypes for various dsputil init functions to dsputil.h
mru
parents: 11370
diff changeset
649 void ff_mlp_init_x86(DSPContext* c, AVCodecContext *avctx);
0b220468ba0d Move prototypes for various dsputil init functions to dsputil.h
mru
parents: 11370
diff changeset
650
8590
7a463923ecd1 Change semantic of CONFIG_*, HAVE_* and ARCH_*.
aurel
parents: 8573
diff changeset
651 #if HAVE_MMX
0
986e461dc072 Initial revision
glantau
parents:
diff changeset
652
862
058194d7ade6 * fixing some minor const warnings
kabi
parents: 853
diff changeset
653 #undef emms_c
853
eacc2dd8fd9d * using DSPContext - so each codec could use its local (sub)set of CPU extension
kabi
parents: 838
diff changeset
654
0
986e461dc072 Initial revision
glantau
parents:
diff changeset
655 static inline void emms(void)
986e461dc072 Initial revision
glantau
parents:
diff changeset
656 {
8031
eebc7209c47f Convert asm keyword into __asm__.
flameeyes
parents: 7995
diff changeset
657 __asm__ volatile ("emms;":::"memory");
6
ec4642daa6fe added emms_c() macro which should can used in c code in both mmx/non mmx cases
glantau
parents: 2
diff changeset
658 }
ec4642daa6fe added emms_c() macro which should can used in c code in both mmx/non mmx cases
glantau
parents: 2
diff changeset
659
936
caa77cd960c0 qpel encoding
michaelni
parents: 894
diff changeset
660
6
ec4642daa6fe added emms_c() macro which should can used in c code in both mmx/non mmx cases
glantau
parents: 2
diff changeset
661 #define emms_c() \
ec4642daa6fe added emms_c() macro which should can used in c code in both mmx/non mmx cases
glantau
parents: 2
diff changeset
662 {\
8104
0d108ec85620 Remove duplicated MM_* macros for CPU capabilities from dsputil.h.
rathann
parents: 8031
diff changeset
663 if (mm_flags & FF_MM_MMX)\
6
ec4642daa6fe added emms_c() macro which should can used in c code in both mmx/non mmx cases
glantau
parents: 2
diff changeset
664 emms();\
0
986e461dc072 Initial revision
glantau
parents:
diff changeset
665 }
986e461dc072 Initial revision
glantau
parents:
diff changeset
666
8590
7a463923ecd1 Change semantic of CONFIG_*, HAVE_* and ARCH_*.
aurel
parents: 8573
diff changeset
667 #elif ARCH_ARM
62
4bfc845cdfea arm optimizations
glantau
parents: 50
diff changeset
668
8590
7a463923ecd1 Change semantic of CONFIG_*, HAVE_* and ARCH_*.
aurel
parents: 8573
diff changeset
669 #if HAVE_NEON
7687
e5b5a9af1b68 ARM: set STRIDE_ALIGN and DECLARE_ALIGNED_8 to 16 for NEON
mru
parents: 7678
diff changeset
670 # define STRIDE_ALIGN 16
e5b5a9af1b68 ARM: set STRIDE_ALIGN and DECLARE_ALIGNED_8 to 16 for NEON
mru
parents: 7678
diff changeset
671 #endif
e5b5a9af1b68 ARM: set STRIDE_ALIGN and DECLARE_ALIGNED_8 to 16 for NEON
mru
parents: 7678
diff changeset
672
8590
7a463923ecd1 Change semantic of CONFIG_*, HAVE_* and ARCH_*.
aurel
parents: 8573
diff changeset
673 #elif ARCH_PPC
623
92e99e506920 first cut at altivec support on darwin patch by (Brian Foley <bfoley at compsoc dot nuigalway dot ie>)
michaelni
parents: 612
diff changeset
674
2324
ba91c976c7aa stride align cleanup
michael
parents: 2272
diff changeset
675 #define STRIDE_ALIGN 16
623
92e99e506920 first cut at altivec support on darwin patch by (Brian Foley <bfoley at compsoc dot nuigalway dot ie>)
michaelni
parents: 612
diff changeset
676
8590
7a463923ecd1 Change semantic of CONFIG_*, HAVE_* and ARCH_*.
aurel
parents: 8573
diff changeset
677 #elif HAVE_MMI
689
efcbfbd18864 ps2 idct patch by (Leon van Stuivenberg <leonvs at iae dot nl>)
michaelni
parents: 687
diff changeset
678
2324
ba91c976c7aa stride align cleanup
michael
parents: 2272
diff changeset
679 #define STRIDE_ALIGN 16
689
efcbfbd18864 ps2 idct patch by (Leon van Stuivenberg <leonvs at iae dot nl>)
michaelni
parents: 687
diff changeset
680
7135
6bd6a2da306e Define mm_flags/support to be 0 on architectures where they don't exist.
astrange
parents: 7094
diff changeset
681 #else
6bd6a2da306e Define mm_flags/support to be 0 on architectures where they don't exist.
astrange
parents: 7094
diff changeset
682
6bd6a2da306e Define mm_flags/support to be 0 on architectures where they don't exist.
astrange
parents: 7094
diff changeset
683 #define mm_flags 0
6bd6a2da306e Define mm_flags/support to be 0 on architectures where they don't exist.
astrange
parents: 7094
diff changeset
684 #define mm_support() 0
6bd6a2da306e Define mm_flags/support to be 0 on architectures where they don't exist.
astrange
parents: 7094
diff changeset
685
6363
7ebd1cdb2142 clean up definition of DECLARE_ALIGNED_8 and STRIDE_ALIGN
mru
parents: 6241
diff changeset
686 #endif
3728
4d346abd723c ADI Blackfin optimizations
diego
parents: 3665
diff changeset
687
6363
7ebd1cdb2142 clean up definition of DECLARE_ALIGNED_8 and STRIDE_ALIGN
mru
parents: 6241
diff changeset
688 #ifndef STRIDE_ALIGN
7ebd1cdb2142 clean up definition of DECLARE_ALIGNED_8 and STRIDE_ALIGN
mru
parents: 6241
diff changeset
689 # define STRIDE_ALIGN 8
0
986e461dc072 Initial revision
glantau
parents:
diff changeset
690 #endif
986e461dc072 Initial revision
glantau
parents:
diff changeset
691
11194
7b3f6955462b Add LOCAL_ALIGNED() macro for declaring aligned local arrays
mru
parents: 10961
diff changeset
692 #define LOCAL_ALIGNED(a, t, v, s, ...) \
7b3f6955462b Add LOCAL_ALIGNED() macro for declaring aligned local arrays
mru
parents: 10961
diff changeset
693 uint8_t la_##v[sizeof(t s __VA_ARGS__) + (a)]; \
7b3f6955462b Add LOCAL_ALIGNED() macro for declaring aligned local arrays
mru
parents: 10961
diff changeset
694 t (*v) __VA_ARGS__ = (void *)FFALIGN((uintptr_t)la_##v, a)
7b3f6955462b Add LOCAL_ALIGNED() macro for declaring aligned local arrays
mru
parents: 10961
diff changeset
695
7b3f6955462b Add LOCAL_ALIGNED() macro for declaring aligned local arrays
mru
parents: 10961
diff changeset
696 #if HAVE_LOCAL_ALIGNED_8
11369
98970e51365a Remove DECLARE_ALIGNED_{8,16} macros
mru
parents: 11368
diff changeset
697 # define LOCAL_ALIGNED_8(t, v, s, ...) DECLARE_ALIGNED(8, t, v) s __VA_ARGS__
11194
7b3f6955462b Add LOCAL_ALIGNED() macro for declaring aligned local arrays
mru
parents: 10961
diff changeset
698 #else
7b3f6955462b Add LOCAL_ALIGNED() macro for declaring aligned local arrays
mru
parents: 10961
diff changeset
699 # define LOCAL_ALIGNED_8(t, v, s, ...) LOCAL_ALIGNED(8, t, v, s, __VA_ARGS__)
7b3f6955462b Add LOCAL_ALIGNED() macro for declaring aligned local arrays
mru
parents: 10961
diff changeset
700 #endif
7b3f6955462b Add LOCAL_ALIGNED() macro for declaring aligned local arrays
mru
parents: 10961
diff changeset
701
7b3f6955462b Add LOCAL_ALIGNED() macro for declaring aligned local arrays
mru
parents: 10961
diff changeset
702 #if HAVE_LOCAL_ALIGNED_16
11369
98970e51365a Remove DECLARE_ALIGNED_{8,16} macros
mru
parents: 11368
diff changeset
703 # define LOCAL_ALIGNED_16(t, v, s, ...) DECLARE_ALIGNED(16, t, v) s __VA_ARGS__
11194
7b3f6955462b Add LOCAL_ALIGNED() macro for declaring aligned local arrays
mru
parents: 10961
diff changeset
704 #else
7b3f6955462b Add LOCAL_ALIGNED() macro for declaring aligned local arrays
mru
parents: 10961
diff changeset
705 # define LOCAL_ALIGNED_16(t, v, s, ...) LOCAL_ALIGNED(16, t, v, s, __VA_ARGS__)
7b3f6955462b Add LOCAL_ALIGNED() macro for declaring aligned local arrays
mru
parents: 10961
diff changeset
706 #endif
7b3f6955462b Add LOCAL_ALIGNED() macro for declaring aligned local arrays
mru
parents: 10961
diff changeset
707
252
ddb1a0e94cf4 - Added PSNR feature to libavcodec and ffmpeg. By now just Y PSNR until I'm
pulento
parents: 214
diff changeset
708 /* PSNR */
1064
b32afefe7d33 * UINTX -> uintx_t INTX -> intx_t
kabi
parents: 1040
diff changeset
709 void get_psnr(uint8_t *orig_image[3], uint8_t *coded_image[3],
252
ddb1a0e94cf4 - Added PSNR feature to libavcodec and ffmpeg. By now just Y PSNR until I'm
pulento
parents: 214
diff changeset
710 int orig_linesize[3], int coded_linesize,
ddb1a0e94cf4 - Added PSNR feature to libavcodec and ffmpeg. By now just Y PSNR until I'm
pulento
parents: 214
diff changeset
711 AVCodecContext *avctx);
781
6f5e87957bcb new generic FFT/MDCT code for audio codecs
bellard
parents: 764
diff changeset
712
6056
558c1fd0ee72 Fix typo in macro name: WARPER8_16_SQ --> WRAPPER8_16_SQ.
diego
parents: 5997
diff changeset
713 #define WRAPPER8_16(name8, name16)\
1708
dea5b2946999 interlaced motion estimation
michael
parents: 1653
diff changeset
714 static int name16(void /*MpegEncContext*/ *s, uint8_t *dst, uint8_t *src, int stride, int h){\
dea5b2946999 interlaced motion estimation
michael
parents: 1653
diff changeset
715 return name8(s, dst , src , stride, h)\
dea5b2946999 interlaced motion estimation
michael
parents: 1653
diff changeset
716 +name8(s, dst+8 , src+8 , stride, h);\
dea5b2946999 interlaced motion estimation
michael
parents: 1653
diff changeset
717 }
dea5b2946999 interlaced motion estimation
michael
parents: 1653
diff changeset
718
6056
558c1fd0ee72 Fix typo in macro name: WARPER8_16_SQ --> WRAPPER8_16_SQ.
diego
parents: 5997
diff changeset
719 #define WRAPPER8_16_SQ(name8, name16)\
1708
dea5b2946999 interlaced motion estimation
michael
parents: 1653
diff changeset
720 static int name16(void /*MpegEncContext*/ *s, uint8_t *dst, uint8_t *src, int stride, int h){\
dea5b2946999 interlaced motion estimation
michael
parents: 1653
diff changeset
721 int score=0;\
dea5b2946999 interlaced motion estimation
michael
parents: 1653
diff changeset
722 score +=name8(s, dst , src , stride, 8);\
dea5b2946999 interlaced motion estimation
michael
parents: 1653
diff changeset
723 score +=name8(s, dst+8 , src+8 , stride, 8);\
dea5b2946999 interlaced motion estimation
michael
parents: 1653
diff changeset
724 if(h==16){\
dea5b2946999 interlaced motion estimation
michael
parents: 1653
diff changeset
725 dst += 8*stride;\
dea5b2946999 interlaced motion estimation
michael
parents: 1653
diff changeset
726 src += 8*stride;\
dea5b2946999 interlaced motion estimation
michael
parents: 1653
diff changeset
727 score +=name8(s, dst , src , stride, 8);\
dea5b2946999 interlaced motion estimation
michael
parents: 1653
diff changeset
728 score +=name8(s, dst+8 , src+8 , stride, 8);\
dea5b2946999 interlaced motion estimation
michael
parents: 1653
diff changeset
729 }\
dea5b2946999 interlaced motion estimation
michael
parents: 1653
diff changeset
730 return score;\
936
caa77cd960c0 qpel encoding
michaelni
parents: 894
diff changeset
731 }
caa77cd960c0 qpel encoding
michaelni
parents: 894
diff changeset
732
4240
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
733
9436
dfed6a243bab Mark src constant in copy_block*().
alexc
parents: 8976
diff changeset
734 static inline void copy_block2(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride, int h)
4240
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
735 {
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
736 int i;
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
737 for(i=0; i<h; i++)
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
738 {
5520
c16a59ef6a86 * renaming (ST|LD)(16|32|64) -> AV_(R|W)N(16|32|64)
romansh
parents: 5291
diff changeset
739 AV_WN16(dst , AV_RN16(src ));
4240
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
740 dst+=dstStride;
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
741 src+=srcStride;
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
742 }
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
743 }
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
744
9436
dfed6a243bab Mark src constant in copy_block*().
alexc
parents: 8976
diff changeset
745 static inline void copy_block4(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride, int h)
4240
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
746 {
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
747 int i;
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
748 for(i=0; i<h; i++)
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
749 {
5520
c16a59ef6a86 * renaming (ST|LD)(16|32|64) -> AV_(R|W)N(16|32|64)
romansh
parents: 5291
diff changeset
750 AV_WN32(dst , AV_RN32(src ));
4240
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
751 dst+=dstStride;
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
752 src+=srcStride;
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
753 }
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
754 }
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
755
9436
dfed6a243bab Mark src constant in copy_block*().
alexc
parents: 8976
diff changeset
756 static inline void copy_block8(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride, int h)
4240
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
757 {
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
758 int i;
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
759 for(i=0; i<h; i++)
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
760 {
5520
c16a59ef6a86 * renaming (ST|LD)(16|32|64) -> AV_(R|W)N(16|32|64)
romansh
parents: 5291
diff changeset
761 AV_WN32(dst , AV_RN32(src ));
c16a59ef6a86 * renaming (ST|LD)(16|32|64) -> AV_(R|W)N(16|32|64)
romansh
parents: 5291
diff changeset
762 AV_WN32(dst+4 , AV_RN32(src+4 ));
4240
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
763 dst+=dstStride;
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
764 src+=srcStride;
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
765 }
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
766 }
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
767
9436
dfed6a243bab Mark src constant in copy_block*().
alexc
parents: 8976
diff changeset
768 static inline void copy_block9(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride, int h)
4240
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
769 {
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
770 int i;
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
771 for(i=0; i<h; i++)
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
772 {
5520
c16a59ef6a86 * renaming (ST|LD)(16|32|64) -> AV_(R|W)N(16|32|64)
romansh
parents: 5291
diff changeset
773 AV_WN32(dst , AV_RN32(src ));
c16a59ef6a86 * renaming (ST|LD)(16|32|64) -> AV_(R|W)N(16|32|64)
romansh
parents: 5291
diff changeset
774 AV_WN32(dst+4 , AV_RN32(src+4 ));
4240
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
775 dst[8]= src[8];
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
776 dst+=dstStride;
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
777 src+=srcStride;
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
778 }
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
779 }
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
780
9436
dfed6a243bab Mark src constant in copy_block*().
alexc
parents: 8976
diff changeset
781 static inline void copy_block16(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride, int h)
4240
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
782 {
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
783 int i;
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
784 for(i=0; i<h; i++)
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
785 {
5520
c16a59ef6a86 * renaming (ST|LD)(16|32|64) -> AV_(R|W)N(16|32|64)
romansh
parents: 5291
diff changeset
786 AV_WN32(dst , AV_RN32(src ));
c16a59ef6a86 * renaming (ST|LD)(16|32|64) -> AV_(R|W)N(16|32|64)
romansh
parents: 5291
diff changeset
787 AV_WN32(dst+4 , AV_RN32(src+4 ));
c16a59ef6a86 * renaming (ST|LD)(16|32|64) -> AV_(R|W)N(16|32|64)
romansh
parents: 5291
diff changeset
788 AV_WN32(dst+8 , AV_RN32(src+8 ));
c16a59ef6a86 * renaming (ST|LD)(16|32|64) -> AV_(R|W)N(16|32|64)
romansh
parents: 5291
diff changeset
789 AV_WN32(dst+12, AV_RN32(src+12));
4240
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
790 dst+=dstStride;
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
791 src+=srcStride;
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
792 }
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
793 }
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
794
9436
dfed6a243bab Mark src constant in copy_block*().
alexc
parents: 8976
diff changeset
795 static inline void copy_block17(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride, int h)
4240
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
796 {
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
797 int i;
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
798 for(i=0; i<h; i++)
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
799 {
5520
c16a59ef6a86 * renaming (ST|LD)(16|32|64) -> AV_(R|W)N(16|32|64)
romansh
parents: 5291
diff changeset
800 AV_WN32(dst , AV_RN32(src ));
c16a59ef6a86 * renaming (ST|LD)(16|32|64) -> AV_(R|W)N(16|32|64)
romansh
parents: 5291
diff changeset
801 AV_WN32(dst+4 , AV_RN32(src+4 ));
c16a59ef6a86 * renaming (ST|LD)(16|32|64) -> AV_(R|W)N(16|32|64)
romansh
parents: 5291
diff changeset
802 AV_WN32(dst+8 , AV_RN32(src+8 ));
c16a59ef6a86 * renaming (ST|LD)(16|32|64) -> AV_(R|W)N(16|32|64)
romansh
parents: 5291
diff changeset
803 AV_WN32(dst+12, AV_RN32(src+12));
4240
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
804 dst[16]= src[16];
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
805 dst+=dstStride;
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
806 src+=srcStride;
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
807 }
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
808 }
e880555ad2c6 Move copy_block* functions to dsp dsputil.h
gpoirier
parents: 4239
diff changeset
809
7760
c4a4495715dd Globally rename the header inclusion guard names.
stefano
parents: 7687
diff changeset
810 #endif /* AVCODEC_DSPUTIL_H */