Keith Packard <[email protected]> writes: > Eric Anholt <[email protected]> writes: > >> This possibly is a minor hit for immediate mode renderers (no >> difference on copypixin100 on my hsw, n=12), but it gives important >> information about drawing bounds to a deferred renderer (3.1x >> improvement in copypixwin100 on vc4). > > If you're going to set the scissor to the destination rectangle every > time, you might as well also switch this to draw a single triangle > instead of a quad?
Maybe? That would be a separate change, at least. And I'd rather see us just fix the fact that we ever draw to the front buffer in general. > Also, we use scissors all over the place; is this the only one which > gains significantly on your hardware? No, scissors are always a good idea for me, and the rest of the code I've seen didn't try to optimize out turning on a scissor.
signature.asc
Description: PGP signature
_______________________________________________ [email protected]: X.Org development Archives: http://lists.x.org/archives/xorg-devel Info: http://lists.x.org/mailman/listinfo/xorg-devel
