Re: real-time priority, CoreAudio, etc.
Re: real-time priority, CoreAudio, etc.
- Subject: Re: real-time priority, CoreAudio, etc.
- From: "David A. Gatwood" <email@hidden>
- Date: Fri, 24 Aug 2001 15:17:18 -0700 (PDT)
On Fri, 24 Aug 2001, James Farwell wrote:
>
Another problem mpg123 has is that there are no ppc assembly routines for
>
decoding, so it just uses generic C routines. This is why mpg123 takes a
>
lot more cpu than slower intel machines. For example, when playing the same
>
mp3, mpeg123 took 4% cpu on my G3 500 and 0.1% on my k6/233 linux box.
>
Granted the k6 will have a little advantage because of mmx, but the sheer
>
speed difference there should easily make up for that. It would be really
>
nice if someone wrote some streamlined ppc assembley decoding routines for
>
decoding mp3s that we could add to mpg123.
If we were using PPC 8xx machines (or was it 4xx), there'd be a microcode
patch to do it, but I assume that's not quite what you meant. ;-)
Anyway, I can duplicate this problem (stalls while moving Mac OS X windows
around) just as easily with a raw .wav file and esdplay, so the problem
probably has little to do with the cpu horsepower of the decode. Best
guess is that either everyone's (mpg123/esdplay/ogg123 and esd) priority
is too low or there are long stalls in the socket communication between
the player and esd... or it could be be an issue in communication with
CoreAudio. Any thoughts?
Later,
David
---------------------------------------------------------------------
All your source are belong to us....