[MPlayer-dev-eng] Re: Re: PATCH [0/12] CoreAVCDecoder support
Bernd Ernesti
mplayer-dev-eng at lists.veego.de
Sun Feb 18 09:39:44 CET 2007
On Sun, Feb 18, 2007 at 03:06:07AM -0500, Alexander Strange wrote:
>
> On Feb 18, 2007, at 2:48 AM, Bernd Ernesti wrote:
>
> >On Sun, Feb 18, 2007 at 01:42:02AM +0100, Michael Niedermayer wrote:
> >[..]
> >
> >>maybe someone should try benchmarking h264.c compiled with -Os ?
> >
> >Same file used for benchmarking (harry_potter_goblet-tlr2_h720p.mov):
> >
> >-Os (h264idct.c and h264.c, all other files were -04):
>
> Just to make sure your tests don't get too complex, let me point out
> that -O4 doesn't exist. All levels above -O3 are -O3.
I know, but thats what the mplayer Makefile uses.
> You could also try gccs 2 and 4 if you've got them, but the inlining
> code was only rewritten to be good very recently (and hasn't made it
> to a release) so there probably isn't that much difference.
That is not so easy right now.
I can only offer a faster system with gcc version 4.1.2 20061021 prerelease.
-O4:
BENCHMARKs: VC: 27.561s VO: 0.017s A: 0.000s Sys: 1.105s = 28.682s
BENCHMARK%: VC: 96.0887% VO: 0.0594% A: 0.0000% Sys: 3.8519% = 100.0000%
or
BENCHMARKs: VC: 27.539s VO: 0.017s A: 0.000s Sys: 1.105s = 28.662s
BENCHMARK%: VC: 96.0832% VO: 0.0601% A: 0.0000% Sys: 3.8567% = 100.0000%
-Os:
BENCHMARKs: VC: 29.739s VO: 0.017s A: 0.000s Sys: 1.097s = 30.853s
BENCHMARK%: VC: 96.3879% VO: 0.0553% A: 0.0000% Sys: 3.5568% = 100.0000%
or
BENCHMARKs: VC: 29.730s VO: 0.017s A: 0.000s Sys: 1.100s = 30.846s
BENCHMARK%: VC: 96.3799% VO: 0.0555% A: 0.0000% Sys: 3.5646% = 100.0000%
Ups, thats the other way around.
vendor_id : AuthenticAMD
cpu family : 15
model : 3
model name : AMD Unknown K7 (Athlon) (686-class)
stepping : 2
cpu MHz : 2411.86
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush 20 acpi mmx fxsr sse sse2 ht tm ia64 31
This is with only one CPU used (the other is not attached).
Bernd
More information about the MPlayer-dev-eng
mailing list