[MPlayer-dev-eng] vo_xv broken

Alex Beregszaszi alex at fsn.hu
Thu May 15 20:09:49 CEST 2003


Hi,

looked further, I found out that still lot of vo's depend on the
swscaler :(

mga_common.c:#include "../postproc/rgb2rgb.h"
mga_common.c:   sws_rgb2rgb_init(get_sws_cpuflags());
However it only calls that, but doesn't uses the rgb2rgb stuffs.

vo_directfb.c:#include "../postproc/rgb2rgb.h"
This uses heavily the yuv2rgb funcs. But here's vo_directfb2 which works
better, and the vo_directfb is abandoned.

vo_fbdev.c:#include "../postproc/rgb2rgb.h"
And this also -- this needs to be fixed! Any volunteers?

vo_gl.c:#include "../postproc/rgb2rgb.h"
vo_gl2.c:#include "../postproc/rgb2rgb.h"
Dunno about these, never used them.

vo_x11.c:#include "../postproc/rgb2rgb.h"
This is ok. But I would like to see it without these hacks.

vo_yuv4mpeg.c: * Based on the pgm output plugin, the rgb2rgb postproc
filter, divxdec,
vo_yuv4mpeg.c:#include "../postproc/rgb2rgb.h"
This needs to be fixed.

Also attached a patch against vo_xv.c which removes the BGR24 support.

Btw, I found this bug becouse I was cleaning up vd_lcl.c to get rid of
the
csp converters written in C. So here are some benchmarks with the
original
vd_lcl and with a bit modified vd_lcl without the use of those
converters (still not optimal):

before:
BENCHMARKs: VC:   3.834s VO:   1.905s A:   0.000s Sys:   0.247s =   
5.986s
BENCHMARK%: VC: 64.0445% VO: 31.8222% A:  0.0000% Sys:  4.1332% =
100.0000%

after:
BENCHMARKs: VC:   2.804s VO:   1.816s A:   0.000s Sys:   0.248s =   
4.868s
BENCHMARK%: VC: 57.5931% VO: 37.3070% A:  0.0000% Sys:  5.0999% =
100.0000%

-- 
Alex Beregszaszi <alex at fsn.hu>
(MPlayer Core Developer -- http://www.mplayerhq.hu/)
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: vo_xv.diff
URL: <http://lists.mplayerhq.hu/pipermail/mplayer-dev-eng/attachments/20030515/2626b3d3/attachment.txt>


More information about the MPlayer-dev-eng mailing list