Overloads
Overloads
- Subject: Overloads
- From: Glenn Maynard <email@hidden>
- Date: Sat, 22 May 2004 00:37:33 -0400
- Mail-copies-to: nobody
We're having reports of overloads on older machines, especially during
data loading.
This is hard for us to track--there's only one of us on a Mac, he's on
a new machine that doesn't have the problem, and we don't have access
(even remote) to any that do.
The code is fairly standard, as far as the IOProc is concerned: grab
data from some lockless FIFOs, mix them together, passing the results
through an AudioConverter. The mixing isn't heavily optimized, but
during loads, only a BGM is playing, so that isn't being stressed.
All decoding happens in a separate thread, which is priority boosted
(or the IOProc was starving during loads) but not realtime.
For reference, if anyone's bored enough to actually read code:
http://cvs.sf.net/viewcvs.py/stepmania/stepmania/src/arch/Sound/RageSoundDriver_CA.cpp?view=markup
and an example log:
http://www.stolaf.edu/people/voss/files/info.txt
I'm hoping somebody can point out something obviously wrong that I'm
overlooking. If I had access to a machine that can reproduce the problem,
I could track it down further, but currently all I can do is add logs,
wait until the next testing snapshot is released (1-2 weeks), and wait for
bug reports to arrive to see the results--a slow turnaround.
(I ran some basic stats before releasing a snapshot recently, tracking
average time--under 1ms per IOProc buffer--but I should have tracked
worst-case times, or perhaps keep track of the duration of the last
IOProc, and log it in the overload handler, which I'll do for the
next snapshot.)
--
Glenn Maynard
_______________________________________________
coreaudio-api mailing list | email@hidden
Help/Unsubscribe/Archives:
http://www.lists.apple.com/mailman/listinfo/coreaudio-api
Do not post admin requests to the list. They will be ignored.