On Thu, 2003-10-16 at 19:00, Keith Whitwell wrote: > Felix Kühling wrote: > > IIRC, the radeon driver still uses AGP writeback. This doesn't work > > reliably on my system and I disabled it in my local tree. Some people > > (including myself) have been thinking about an efficient algorithm to > > detect unreliable writeback, but AFAICT nobody came up with anything yet > > (at least no implementation). Should we just disable it for good? > > How often do we actually look at the read pointer? It's only really useful > for detecting an empty ring, and we have other registers for that. The other > time it's examined is for a *full* ring. > > If we read it lazily, and only when the ring appears to fill, that actually > wouldn't be that many reads.
True. OTOH, once we're using the recommended memory layout (I'm almost done with the transition), we could use plain PCI transfers for the writeback, which should be more reliable. Also, I have code for clients to read the values directly from the writeback page, which could save some context switches. -- Earthling Michel Dänzer \ Debian (powerpc), XFree86 and DRI developer Software libre enthusiast \ http://svcs.affero.net/rm.php?r=daenzer ------------------------------------------------------- This SF.net email is sponsored by: SF.net Giveback Program. SourceForge.net hosts over 70,000 Open Source Projects. See the people who have HELPED US provide better services: Click here: http://sourceforge.net/supporters.php _______________________________________________ Dri-devel mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/dri-devel