[MPlayer-dev-eng] [PATCH] remove mp3lib

[Ivan Kalvachev]

On 9/20/11, Reimar Döffinger <Reimar.Doeffinger at gmx.de> wrote:
> Hello,
> I am still interested in seeing performance numbers and statistics from
> ffmp3 vs. mp3lib when ffmp3 is slower, but I guess it's time to start
> some more specific moves to get rid of it.
>

I've done some benchmarks on my old Athlon TBird 1GHz.
(no sse of any kind, only mmx1/2, 3dnow1/2).

$ mplayer -loop 3 -ao pcm:fast:file=/dev/null -benchmark -quiet -ac
mp3 stream.dump
MPlayer SVN-r34123-4.5.2 (C) 2000-2011 MPlayer Team
Selected audio codec: [mp3] afm: mp3lib (mp3lib MPEG layer-2, layer-3)
AO: [pcm] 48000Hz 2ch s16le (2 bytes per sample)
BENCHMARKs: VC:   0.000s VO:   0.000s A:  49.342s Sys:   0.416s =   49.758s
BENCHMARKs: VC:   0.000s VO:   0.000s A:  49.436s Sys:   0.436s =   49.872s
BENCHMARKs: VC:   0.000s VO:   0.000s A:  49.516s Sys:   0.409s =   49.924s

$ mplayer -loop 3 -ao pcm:fast:file=/dev/null -benchmark -quiet -ac
ffmp3 stream.dump
Selected audio codec: [ffmp3] afm: ffmpeg (FFmpeg MPEG layer-3 audio)
AO: [pcm] 48000Hz 2ch s16le (2 bytes per sample)
BENCHMARKs: VC:   0.000s VO:   0.000s A: 123.222s Sys:   0.406s =  123.627s
BENCHMARKs: VC:   0.000s VO:   0.000s A: 123.053s Sys:   0.415s =  123.467s
BENCHMARKs: VC:   0.000s VO:   0.000s A: 123.167s Sys:   0.444s =  123.611s

$ mplayer -loop 3 -ao pcm:fast:file=/dev/null -benchmark -quiet -ac
ffmp3float stream.dump
Selected audio codec: [ffmp3float] afm: ffmpeg (FFmpeg MPEG layer-3 audio)
AO: [pcm] 48000Hz 2ch floatle (4 bytes per sample)
BENCHMARKs: VC:   0.000s VO:   0.000s A:  56.801s Sys:   0.760s =   57.562s
BENCHMARKs: VC:   0.000s VO:   0.000s A:  56.831s Sys:   0.775s =   57.606s
BENCHMARKs: VC:   0.000s VO:   0.000s A:  56.863s Sys:   0.766s =   57.629s

$ mplayer -loop 3 -ao pcm:fast:file=/dev/null -benchmark -quiet -ac
ffmp3float -af format=s16le stream.dump
Selected audio codec: [ffmp3float] afm: ffmpeg (FFmpeg MPEG layer-3 audio)
AO: [pcm] 48000Hz 2ch s16le (2 bytes per sample)
BENCHMARKs: VC:   0.000s VO:   0.000s A:  62.408s Sys:   0.635s =   63.042s
BENCHMARKs: VC:   0.000s VO:   0.000s A:  62.791s Sys:   0.668s =   63.459s
BENCHMARKs: VC:   0.000s VO:   0.000s A:  62.795s Sys:   0.633s =   63.428s

=====
Even if we assume the extra 6 seconds are because of the ugly float
conversion and clipping i've done in af_format, even then mp3lib is
still around 16% faster than the float only case.

Do you want oprofile from my cpu too?