On Die, 2011-10-04 at 15:52 -0700, Aaron Plattner wrote: > This caused a performance regression because previously, miTrapezoids > would render its mask image into a scratch pixmap and then perform the > final composite using a possibly-accelerated RENDER Composite call. > Now, fbTrapezoids does the final composite in software, leading to, for > example, http://www.nvnews.net/vbulletin/showthread.php?t=166698 > > Of course, I could simply copy the old miTrapezoids implementation into > our driver, but it seems hard to believe that our driver is the only one > that benefited from the "rasterize in software, composite in hardware" > behavior of the old code. Should I re-add miTrapezoids as a helper that > drivers can plug in when they want the old behavior, just revert this > change, or go the "only nvidia gets to go fast" copy & paste route?
FWIW, EXA already had a copy of miTrapezoids before this change to avoid GPU<->CPU ping-pong (GPU clear, CPU rasterization, GPU composition). -- Earthling Michel Dänzer | http://www.amd.com Libre software enthusiast | Debian, X and DRI developer _______________________________________________ [email protected]: X.Org development Archives: http://lists.x.org/archives/xorg-devel Info: http://lists.x.org/mailman/listinfo/xorg-devel
