[MPlayer-dev-eng] Re: Vorbis decoding performance

[Rich Felker]

On Wed, Aug 02, 2006 at 10:38:39PM +0200, Balatoni Denes wrote:
> Hi!
> 
> Thanks for the benchmarks, looks good!
> IMHO the reason why ffvorbis was so slow on the k6-III is the
> non-SSE floating point mdct (k6-III has no SSE, right?).

Yes.

> BTW if there was a fixed-point mdct in ffmpeg (xiph's can be stolen, but 
> that's not too creative, is it) a blazing fast ffvorbis could be made, 

Yes. :))))
Tremor is not at all efficient and it's still the fastest on my cpu.
I'm fairly confident that an optimized decoder could be roughly the
same performance as mp3 decoding, which would make a big difference to
usability when it comes to high-res or h264 videos with vorbis audio.

> assuming that 32 bit is enough for mdct coefficients - the vorbis spec 
> explicitly warns against this, and "mandates" 48 bits (and in fact tremor 
> uses some tricky floating-point for this reason), but 32 bit is easily enough 
> for all current vorbis streams/encoders.

Yes, 32 bits is perfectly fine. If anyone really cares the decoder
could detect nonsense streams that need more precision and switch to a
slower algorithm.

Rich