[vorbis] Windows XP and MP3
Segher Boessenkool
segher at chello.nl
Fri Apr 13 10:02:31 PDT 2001
fungus wrote:
>
> Michael Smith wrote:
> >
> > There's been a lot of optimisation work, and the current decoder is better
> > than the 'bad' mp3 decoders as far as speed goes. I'd be surprised if it
> > was still more than twice as slow.
It still is, and probably will be for some time. Sorry about that.
I won't be doing (much) more floor0 optimizations, as floor1 will be a lot
faster (and better) anyway. For MDCT I will first make a nice altivec
optimized implementation, as I have a nice'n'shiny G4 now. G3 implementation
is almost ready. 3dnow! will follow. SSE will only be done (by me) if someone
donates me a P3 machine.
> Is that in the beta 4 or is it still in the unofficial
> "nightly CVS" versions?
My latest opts are not in the nightly yet. People have been busy lately.
> > It's useful to note that there are at least two (that I know of) games
> > currently available (i.e. actual commercial releases) that use vorbis for
> > their in-game audio.
>
That's a nice thing to hear. Note that for such a constrained
environment (control over both the decoder _and_ the streams
it will play), a lot of optimizations are possible, which are
_not_ possible with the generic decoder. You don't have to be
(backwards) compatible, for example; you can throw out support
for every feature you don't use.
> I'm working on one now where I can have up to six audio
> streams going at the same time[1], along with ov_seek_pcm()s
> happening in each channel.
>
> I probably have to come up with a work-around for the seek
> slowness, but the basic decoder is still very slow for a
You can just lseek() to the position in the stream you need,
and start decoding there. No need to ov_seek_pcm(). You make
the streams, so you can know exactly at what byte position
in the stream your sounds start. No need to seek to a time
offset. Just precalculate all of this, and store it somewhere
along with the streams.
> game. FWIW, Winamp currently registers aout 5% CPU usage
> with .ogg files, and 1% with .mp3 files.
Wow. winamp 1% with mp3? That's an _extremely_ fast processor
you have, or your method of measurement is not all that
exact.
> Just like another poster in this thread, I don't really
> have time to start digging around in the libs and learning
> what's going on (I wish I did!), I have to rely on the
> work of the vorbis team.
>
> Is speed being worked on for the final release, or is
> Version 1.0 just going to be the beta 4 code but with
> the bugs fixed.
Added features and speed increase is planned for 1.0.
Dagdag,
Segher
--- >8 ----
List archives: http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/
To unsubscribe from this list, send a message to 'vorbis-request at xiph.org'
containing only the word 'unsubscribe' in the body. No subject is needed.
Unsubscribe messages sent to the list will be ignored/filtered.
More information about the Vorbis
mailing list