On Mon, 14 May 2001, Matt wrote:

> On Sun, 13 May 2001, Matt wrote:
>
> > On 13 May 2001 [EMAIL PROTECTED] wrote:
> >
> > >   src/sound      : soundserver_sdl.c
> > >
> > > Log message:
> > > * Fixed Win32 specific code in main.c
> > > * (a) Added #defineable update debug code to operations.c (more to come)
> > > * (b) Fixed soundserver_sdl to lock the mutex before waiting on it
> >
> > in SDL_SemWatiTimeout, it spins on the WaitForSingleObject() seemingly
> > forever. (I've only let it run for 10 minutes or so.)
>
> That appeared to fix that problem; now there's another one:
>
> CPU usage is fine and music plays fine in LSL3, until I enter the scene
> with Larry's wife and her new girlfriend. CPU usage goes back to using as
> much as possible, fighting with the main thread, and the music becomes
> very slow.

In HQ1, the slowness happens no matter what. The sound thread is always
fighting with the main thread for CPU time.

Looking more and more like sched_yield wa a better alternative..


--
http://www.clock.org/~matt


Reply via email to