10-30-2010, 09:09 AM
Quote:Radeons eat dirt when using SSAA
I'de like to see you prove that. They both have the same performance hit. 1/n where n is the number of samples since all you are doing is raising the internal resolution.
Let's make p = pixel fillrate, r = resolution per frame, f = framerate, n = number of samples.
Normally:
p/r = f
with ssaa it should therefore be:
p/(r x n) = f
Of course this is just for the max framerate that the gpu is capable of. You also have to factor in external bottlenecks. Since supersampling increases EVERYTHING equally it is impossible to design a gpu that has a different performance hit than another. You are probably thinking of msaa. Since radeon gpus used to have terrible msaa performance due to crappy rops.