On Wed, 2016-08-24 at 12:27 +0900, Michel Dänzer wrote: > This change results in write-combining for the PBO CPU mapping. > Apparently, fbPutXYImage ends up either reading from the PBO, or at > least writing to it in a WC-unfriendly manner, causing a big slowdown.
It does read from the destination, and the implementation is pretty much the least cache-friendly thing one could imagine. fbPutXYImage works by walking plane-by-plane over the source image and combining each plane with the destination according to the rop mode, so even if the equivalent ZPixmap operation wouldn't read the destination (say, pm of ~0 and rop of GXcopy), the XYPixmap version will. - ajax _______________________________________________ xorg-devel@lists.x.org: X.Org development Archives: http://lists.x.org/archives/xorg-devel Info: https://lists.x.org/mailman/listinfo/xorg-devel