Re: [Plib-users] Video Lockups

[Steve B. <sjb...@ai...>]

Christian Mayer wrote:

> It might be a hardware problem. Eg. increased power demand for the
> graphics accelerator during OpenGL drops the voltage and thus it gets
> too low.

Yes - I wondered that - which is why I asked whether other OpenGL programs
had the problem.

GeForce cards use a LOT of power - we've blown up PC power supplies at work
where the PC was stuffed with lots of cards/RAM/disks and had a cheap power
supply.

Also, there have been cases where certain motherboards had very thin circuit
board traces carrying the power to the AGP slot - those have been known to
fail when the graphics card is forced to do LOTS of work.

> This should happen with other OpenGL (or Direct3D) applications as well.
> And it should be extremely bad with benchmarks. Tried those?

I agree - that's an important test.
 
> But on the other hand: the GeForce2 MX isn't known to require that much
> power.

I strongly disagree - I don't know specifically about the 2/MX but GeForce
cards in general seem to be especially power hungry and we've had a TON of
problems relating to that.

> Note: A big power supply (i.e. many Watts) doesn't necesary help. It's
> important that it can "support" enough Amperes for every voltage.

Yep.
 
> >  * If WinAMP keeps running, so should other applications that don't
> >    access the graphics.  You could write a script that sleeps for
> >    (say) five minutes and then kills the PLIB application.  You should
> >    be able to use the resulting debug stuff (a 'core' file or something?)
> >    to see where PLIB is stuck.
> 
> Does Winamp play only the current file or does it also load the next one
> from the playlist and goes on? (Perhaps you are only listening to the
> buffer and not to winamp any more)

That's a good theory...he *said* it kept playing until he hit reset - but
if the sound card has a large enough buffer - and if WinAMP uses it, that
could be a long wait.
 
> PS: All PLIB versions (+ FGFS) run very well on my W2K + GeForce2MX.

That's what I thought.  It's encouraging to hear definite confirmation
though.

If I had to bet though, my money would be on the board-vendor-driver
issue.

----------------------------- Steve Baker -------------------------------
Mail : <sjb...@ai...>   WorkMail: <sj...@li...>
URLs : http://www.sjbaker.org
       http://plib.sf.net http://tuxaqfh.sf.net http://tuxkart.sf.net
       http://prettypoly.sf.net http://freeglut.sf.net
       http://toobular.sf.net   http://lodestone.sf.net