[Mplayer-cvslog] CVS: main/etc codecs.conf,1.274,1.275

[D Richard Felker III]

On Mon, May 12, 2003 at 12:51:17PM +0200, Diego Biurrun wrote:
> Arpi writes:
>  > > Michael Niedermayer CVS writes:
>  > >  > Modified Files:
>  > >  > 	codecs.conf 
>  > >  > Log Message:
>  > >  > ffsvq3
>  > > 
>  > > You guys are incredible.  I'm really looking forward to somebody
>  > > optimizing this stuff, on my K6-III 500 I can play everything without
>  > > framedrop, except for some svq3 mov files.
>  > > 
>  > > Here are some benchmarks:
>  > > 
>  > > mplayer -vfm ffmpeg -benchmark -vo null -nosound /tmp/trailer_final_1000_dl.mov
>  > > BENCHMARKs: VC: 206.455s VO:   0.036s A:   0.000s Sys:   5.624s =  212.115s
>  > > BENCHMARK%: VC: 97.3315% VO:  0.0169% A:  0.0000% Sys:  2.6516% = 100.0000%
>  > > 
>  > > mplayer -benchmark -vo null -nosound /tmp/trailer_final_1000_dl.mov
>  > > BENCHMARKs: VC: 206.473s VO:   0.041s A:   0.000s Sys:   6.799s =  213.313s
>  > > BENCHMARK%: VC: 96.7935% VO:  0.0191% A:  0.0000% Sys:  3.1874% = 100.0000%
>  > > 
>  > > 
>  > > 
>  > > mplayer -vfm ffmpeg -benchmark -vo null -nosound /tmp/reloaded_teaser_1_640.mov
>  > > BENCHMARKs: VC:  35.542s VO:   0.017s A:   0.000s Sys:   2.446s =   38.005s
>  > > BENCHMARK%: VC: 93.5191% VO:  0.0439% A:  0.0000% Sys:  6.4370% = 100.0000%
>  > > 
>  > > mplayer -benchmark -vo null -nosound /tmp/reloaded_teaser_1_640.mov
>  > > BENCHMARKs: VC:  35.604s VO:   0.017s A:   0.000s Sys:   3.930s =   39.551s
>  > > BENCHMARK%: VC: 90.0195% VO:  0.0431% A:  0.0000% Sys:  9.9374% = 100.0000%
>  > > 
>  > > I made a few more runs, the results are similar, the FFmpeg decoder is
>  > > always slightly ahead of QT.
>  > 
>  > did you notice that ffsvq3 is the default, so -vfm ffmpeg changes nothing? :)
> 
> OK, pass the Cola bottle :-)
> 
> Here are the benchmarks again:
> 
> mplayer -vfm ffmpeg -benchmark -vo null -nosound reloaded_teaser_1_640.mov
> BENCHMARKs: VC:  35.678s VO:   0.019s A:   0.000s Sys:   2.420s =   38.117s
> BENCHMARK%: VC: 93.6029% VO:  0.0488% A:  0.0000% Sys:  6.3483% = 100.0000%
> 
> 
> mplayer -vfm qtvideo -benchmark -vo null -nosound reloaded_teaser_1_640.mov
> BENCHMARKs: VC:  35.782s VO:   0.022s A:   0.000s Sys:   3.434s =   39.238s
> BENCHMARK%: VC: 91.1911% VO:  0.0568% A:  0.0000% Sys:  8.7521% = 100.0000%
> 
> 
> 
> mplayer -vfm ffmpeg -benchmark -vo null -nosound trailer_final_1000_dl.mov
> BENCHMARKs: VC: 206.784s VO:   0.035s A:   0.000s Sys:   8.391s =  215.210s
> BENCHMARK%: VC: 96.0846% VO:  0.0164% A:  0.0000% Sys:  3.8989% = 100.0000%
> 
> 
> mplayer -vfm qtvideo -benchmark -vo null -nosound trailer_final_1000_dl.mov
> BENCHMARKs: VC: 211.267s VO:   0.040s A:   0.000s Sys:   7.929s =  219.236s
> BENCHMARK%: VC: 96.3654% VO:  0.0182% A:  0.0000% Sys:  3.6165% = 100.0000%
> 
> 
> Results are essentially the same, with FFmpeg slightly ahead of QT.

Considering that this is unoptimized, freshly-written code, I think
Apple should be extremely embarassed. Then again, proprietary crapware
houses never want to make their code perform well....especially when
they're also selling hardware.

Rich