Mercurial > libavcodec.hg
annotate synth_filter.c @ 11352:6e0af2cfdcfe libavcodec
Do MC and IDCT in coding (hilbert) order
This increases the slice size to 64 pixels, due to having to decode an
entire chroma superblock row per slice.
This can be up to 6% slower depending on clip and CPU, but is necessary
for future optimizations that gain significantly more than was lost.
author | conrad |
---|---|
date | Wed, 03 Mar 2010 23:27:40 +0000 |
parents | 90daf9b3083d |
children | 4b3da727d832 |
rev | line source |
---|---|
10467 | 1 /* |
2 * copyright (c) 2008 Michael Niedermayer <michaelni@gmx.at> | |
3 * | |
4 * This file is part of FFmpeg. | |
5 * | |
6 * FFmpeg is free software; you can redistribute it and/or | |
7 * modify it under the terms of the GNU Lesser General Public | |
8 * License as published by the Free Software Foundation; either | |
9 * version 2.1 of the License, or (at your option) any later version. | |
10 * | |
11 * FFmpeg is distributed in the hope that it will be useful, | |
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU | |
14 * Lesser General Public License for more details. | |
15 * | |
16 * You should have received a copy of the GNU Lesser General Public | |
17 * License along with FFmpeg; if not, write to the Free Software | |
18 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA | |
19 */ | |
20 | |
21 #include "synth_filter.h" | |
22 | |
23 void ff_synth_filter_float(FFTContext *imdct, | |
10479 | 24 float *synth_buf_ptr, int *synth_buf_offset, |
25 float synth_buf2[32], const float window[512], | |
26 float out[32], const float in[32], float scale, float bias) | |
10467 | 27 { |
28 float *synth_buf= synth_buf_ptr + *synth_buf_offset; | |
29 int i, j; | |
30 | |
31 ff_imdct_half(imdct, synth_buf, in); | |
32 | |
33 for (i = 0; i < 16; i++){ | |
10482
90daf9b3083d
Change whitespace placement a little to improve readabiliy slightly.
michael
parents:
10479
diff
changeset
|
34 float a= synth_buf2[i ]; |
90daf9b3083d
Change whitespace placement a little to improve readabiliy slightly.
michael
parents:
10479
diff
changeset
|
35 float b= synth_buf2[i + 16]; |
10467 | 36 float c= 0; |
37 float d= 0; | |
38 for (j = 0; j < 512 - *synth_buf_offset; j += 64){ | |
10482
90daf9b3083d
Change whitespace placement a little to improve readabiliy slightly.
michael
parents:
10479
diff
changeset
|
39 a += window[i + j ]*(-synth_buf[15 - i + j ]); |
90daf9b3083d
Change whitespace placement a little to improve readabiliy slightly.
michael
parents:
10479
diff
changeset
|
40 b += window[i + j + 16]*( synth_buf[ i + j ]); |
90daf9b3083d
Change whitespace placement a little to improve readabiliy slightly.
michael
parents:
10479
diff
changeset
|
41 c += window[i + j + 32]*( synth_buf[16 + i + j ]); |
90daf9b3083d
Change whitespace placement a little to improve readabiliy slightly.
michael
parents:
10479
diff
changeset
|
42 d += window[i + j + 48]*( synth_buf[31 - i + j ]); |
10467 | 43 } |
44 for ( ; j < 512; j += 64){ | |
10482
90daf9b3083d
Change whitespace placement a little to improve readabiliy slightly.
michael
parents:
10479
diff
changeset
|
45 a += window[i + j ]*(-synth_buf[15 - i + j - 512]); |
90daf9b3083d
Change whitespace placement a little to improve readabiliy slightly.
michael
parents:
10479
diff
changeset
|
46 b += window[i + j + 16]*( synth_buf[ i + j - 512]); |
90daf9b3083d
Change whitespace placement a little to improve readabiliy slightly.
michael
parents:
10479
diff
changeset
|
47 c += window[i + j + 32]*( synth_buf[16 + i + j - 512]); |
90daf9b3083d
Change whitespace placement a little to improve readabiliy slightly.
michael
parents:
10479
diff
changeset
|
48 d += window[i + j + 48]*( synth_buf[31 - i + j - 512]); |
10467 | 49 } |
10482
90daf9b3083d
Change whitespace placement a little to improve readabiliy slightly.
michael
parents:
10479
diff
changeset
|
50 out[i ] = a*scale + bias; |
90daf9b3083d
Change whitespace placement a little to improve readabiliy slightly.
michael
parents:
10479
diff
changeset
|
51 out[i + 16] = b*scale + bias; |
90daf9b3083d
Change whitespace placement a little to improve readabiliy slightly.
michael
parents:
10479
diff
changeset
|
52 synth_buf2[i ] = c; |
90daf9b3083d
Change whitespace placement a little to improve readabiliy slightly.
michael
parents:
10479
diff
changeset
|
53 synth_buf2[i + 16] = d; |
10467 | 54 } |
10482
90daf9b3083d
Change whitespace placement a little to improve readabiliy slightly.
michael
parents:
10479
diff
changeset
|
55 *synth_buf_offset= (*synth_buf_offset - 32)&511; |
10467 | 56 } |