Hi, On Tue, Apr 04, 2006 at 03:49:56PM +0100, Mike Hearn wrote: > So in this case either the CPU time goes way high when the 3D scene > first appears, or maybe my 3D driver (nvidia) is not allowing > pre-emption enough.
nvidia... nvidia... Hrmmm... might this just be caused by the recent annoying sched_yield nvidia driver issue that also managed to char-coal Xgl completely? (jagged window moves, abysmal performance, ...)? Just a stab in the dark, but... > I guess I'm uneasy about this idea of keeping cpu time low, even for > threads we control, because how do we debug it? How does one monitor a > scheduler anyway? Maybe the problem is our audio thread is using too > much cpu time - ok, how do we measure that? How much is too much? What > threshold do we have to beat to get scheduling latency low enough to > work properly? How do we figure out if the scheduler is doing what we > think it's doing? What if it is a driver problem? This all seems very > opaque to me. Exactly my thoughts. Debugging it may be somewhat difficult, but if it finally works then a non-realtime solution (whatever the problem actually is) will be so much nicer from a clean design point of view. Andreas
