Actually concerning these benchmarks and the conclusions drawn: I had some questions.
Something notable in the results was that it appeared that games designed for older hardware (earlier versions or using older engines) ran far better on Catalyst than the the open source drivers. Newer games ran closer to the same or sometimes faster. Is this because newer games are forced to use more unoptimized fallbacks for that old hardware narrowing the difference?
I know every time benchmarks are done people chirp in and say why didn't you do this or that - I'm just curious because I have rose glass memories of being able to play NWN with maxed out settings on my T60, and I haven't been able to do that without severe slowdowns on the gallium drivers as of a couple months ago.
Also obviously maxing out settings that arn't supported like AA is cheating.