Mercurial > mplayer.hg

comparison libswscale/yuv2rgb_altivec.c @ 27155:d5d7ecbbf0d8

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Place license header at the top of the file for consistency.
author diego
date Fri, 04 Jul 2008 13:12:47 +0000
parents e69f41b2c8c0
children 23f1738030fc
comparison
equal deleted inserted replaced
27154:01526c8e2d75 27155:d5d7ecbbf0d8
1 /* 1 /*
2 marc.hoffman@analog.com March 8, 2004 2 * AltiVec acceleration for colorspace conversion
3 3 *
4 AltiVec acceleration for colorspace conversion revision 0.2 4 * copyright (C) 2004 Marc Hoffman <marc.hoffman@analog.com>
5 5 *
6 convert I420 YV12 to RGB in various formats,
7 it rejects images that are not in 420 formats
8 it rejects images that don't have widths of multiples of 16
9 it rejects images that don't have heights of multiples of 2
10 reject defers to C simulation codes.
11
12 lots of optimizations to be done here
13
14 1. need to fix saturation code, I just couldn't get it to fly with packs and adds.
15 so we currently use max min to clip
16
17 2. the inefficient use of chroma loading needs a bit of brushing up
18
19 3. analysis of pipeline stalls needs to be done, use shark to identify pipeline stalls
20
21
22 MODIFIED to calculate coeffs from currently selected color space.
23 MODIFIED core to be a macro which you spec the output format.
24 ADDED UYVY conversion which is never called due to some thing in SWSCALE.
25 CORRECTED algorithim selection to be strict on input formats.
26 ADDED runtime detection of altivec.
27
28 ADDED altivec_yuv2packedX vertical scl + RGB converter
29
30 March 27,2004
31 PERFORMANCE ANALYSIS
32
33 The C version use 25% of the processor or ~250Mips for D1 video rawvideo used as test
34 The ALTIVEC version uses 10% of the processor or ~100Mips for D1 video same sequence
35
36 720*480*30 ~10MPS
37
38 so we have roughly 10clocks per pixel this is too high something has to be wrong.
39
40 OPTIMIZED clip codes to utilize vec_max and vec_packs removing the need for vec_min.
41
42 OPTIMIZED DST OUTPUT cache/dma controls. we are pretty much
43 guaranteed to have the input video frame it was just decompressed so
44 it probably resides in L1 caches. However we are creating the
45 output video stream this needs to use the DSTST instruction to
46 optimize for the cache. We couple this with the fact that we are
47 not going to be visiting the input buffer again so we mark it Least
48 Recently Used. This shaves 25% of the processor cycles off.
49
50 Now MEMCPY is the largest mips consumer in the system, probably due
51 to the inefficient X11 stuff.
52
53 GL libraries seem to be very slow on this machine 1.33Ghz PB running
54 Jaguar, this is not the case for my 1Ghz PB. I thought it might be
55 a versioning issues, however I have libGL.1.2.dylib for both
56 machines. ((We need to figure this out now))
57
58 GL2 libraries work now with patch for RGB32
59
60 NOTE quartz vo driver ARGB32_to_RGB24 consumes 30% of the processor
61
62 Integrated luma prescaling adjustment for saturation/contrast/brightness adjustment.
63 */
64
65 /*
66 * This file is part of FFmpeg. 6 * This file is part of FFmpeg.
67 * 7 *
68 * FFmpeg is free software; you can redistribute it and/or modify 8 * FFmpeg is free software; you can redistribute it and/or modify
69 * it under the terms of the GNU General Public License as published by 9 * it under the terms of the GNU General Public License as published by
70 * the Free Software Foundation; either version 2 of the License, or 10 * the Free Software Foundation; either version 2 of the License, or
77 * 17 *
78 * You should have received a copy of the GNU General Public License 18 * You should have received a copy of the GNU General Public License
79 * along with FFmpeg; if not, write to the Free Software 19 * along with FFmpeg; if not, write to the Free Software
80 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA 20 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
81 */ 21 */
22
23 /*
24 convert I420 YV12 to RGB in various formats,
25 it rejects images that are not in 420 formats
26 it rejects images that don't have widths of multiples of 16
27 it rejects images that don't have heights of multiples of 2
28 reject defers to C simulation codes.
29
30 lots of optimizations to be done here
31
32 1. need to fix saturation code, I just couldn't get it to fly with packs and adds.
33 so we currently use max min to clip
34
35 2. the inefficient use of chroma loading needs a bit of brushing up
36
37 3. analysis of pipeline stalls needs to be done, use shark to identify pipeline stalls
38
39
40 MODIFIED to calculate coeffs from currently selected color space.
41 MODIFIED core to be a macro which you spec the output format.
42 ADDED UYVY conversion which is never called due to some thing in SWSCALE.
43 CORRECTED algorithim selection to be strict on input formats.
44 ADDED runtime detection of altivec.
45
46 ADDED altivec_yuv2packedX vertical scl + RGB converter
47
48 March 27,2004
49 PERFORMANCE ANALYSIS
50
51 The C version use 25% of the processor or ~250Mips for D1 video rawvideo used as test
52 The ALTIVEC version uses 10% of the processor or ~100Mips for D1 video same sequence
53
54 720*480*30 ~10MPS
55
56 so we have roughly 10clocks per pixel this is too high something has to be wrong.
57
58 OPTIMIZED clip codes to utilize vec_max and vec_packs removing the need for vec_min.
59
60 OPTIMIZED DST OUTPUT cache/dma controls. we are pretty much
61 guaranteed to have the input video frame it was just decompressed so
62 it probably resides in L1 caches. However we are creating the
63 output video stream this needs to use the DSTST instruction to
64 optimize for the cache. We couple this with the fact that we are
65 not going to be visiting the input buffer again so we mark it Least
66 Recently Used. This shaves 25% of the processor cycles off.
67
68 Now MEMCPY is the largest mips consumer in the system, probably due
69 to the inefficient X11 stuff.
70
71 GL libraries seem to be very slow on this machine 1.33Ghz PB running
72 Jaguar, this is not the case for my 1Ghz PB. I thought it might be
73 a versioning issues, however I have libGL.1.2.dylib for both
74 machines. ((We need to figure this out now))
75
76 GL2 libraries work now with patch for RGB32
77
78 NOTE quartz vo driver ARGB32_to_RGB24 consumes 30% of the processor
79
80 Integrated luma prescaling adjustment for saturation/contrast/brightness adjustment.
81 */
82 82
83 #include <stdio.h> 83 #include <stdio.h>
84 #include <stdlib.h> 84 #include <stdlib.h>
85 #include <string.h> 85 #include <string.h>
86 #include <inttypes.h> 86 #include <inttypes.h>