We're definitely not going to do like some drivers and try to write to the GPU when the GPU can't accept writes, thereby throttling the bus and introducing huge latencies.
Instead, we'll opt to do things RIGHT and have all the necessary hardware in place so you can let interrupts tell you when you can do stuff. Tasks SLEEP when they have to wait, rather than busy-waiting; DMA bursts aren't overly long, causing huge latencies; etc. On Wed, 2 Feb 2005 21:31:40 +0100, Nicolas Boulay <[EMAIL PROTECTED]> wrote: > For information, i have heard a lot from real time computer programer that the > graphicals driver could be a nightmare. They often take too much cpu slice > time. > > I think i have heard about using 100 ms of latency in some closed prioretary > video driver (locking PCI bus for this time). That's very udge for a 2.6 > system that is supposed to react in 1 ms. > _______________________________________________ Open-graphics mailing list [email protected] http://lists.duskglow.com/mailman/listinfo/open-graphics List service provided by Duskglow Consulting, LLC (www.duskglow.com)
