On Mon, 7 Mar 2005, David Dawes wrote: > On Mon, Mar 07, 2005 at 03:12:55AM -0500, Greg Stark wrote: > >Mark Vojkovich <mvojkovi@xxxxxxxxxxx> writes: > > > >> It's conceivable that overlay fifo underflow in the chip could cause the > >> overlay to stop functioning. > > > >Well I don't see any errors logged about "fifo underflow". > > > >This is why I'm not a hardware guy. When you have no visibility to what's > >going on how are you supposed to debug anything? > > I don't know how related this is, but ... > > I've just tried Xv with a G400, and I don't see the symptoms you've > described. What I do see, at 24-bit and screen resolutions > 1280x1024, > and with a small Xv window is visual problems with the image on the right > half of the window. If I expand the Xv window to near full screen these > disappear. I see this with mplayer, and with Mark's 'suzi' demo. > > I don't see these problems at 16-bit, or at lower screen resolutions. That's a memory bandwidth issue. Scaling it up makes it go away because the overlay needs to fetch overlay pixels less often (relative to framebuffer scanout by the DAC). OverclockMem may help. Remember that when the overlay is bilinear scaling, it needs to fetch four overlay pixels for every RGB pixel read out from the framebuffer. Caches internal to the chip are the only thing that makes this possible. Caching works better when upscaling because you use the same pixels for a longer time. Mark. _______________________________________________ XFree86 mailing list XFree86@xxxxxxxxxxx http://XFree86.Org/mailman/listinfo/xfree86
