GPU performance question: is there any way in a fragment shader (or equivalent) to if() around a multisampled texture read based on whether or not the texture actually had multiple samples at that location? For high MSAA (16x) render targets, the blending takes a long time...
-
-
It seems like this would save a lot of work on blocks of pixels that do not have an edge passing through them, since the entire warp would skip the slow path... but maybe I'm wrong.
Show this threadThanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Wait, don't shader branches execute in lockstep on all GPU cores as fast as the slowest branch? Thus if only 1 pixel in the needs blending all the others would just stall?
-
Yes, but since most pixels aren't multisampled, you still have a high chance of getting all zeroes in the warp and skipping the high-cost code.
End of conversation
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.