Hmm...is it just me, or is anyone else suprised the the FX5900 is only about 20% slower then the 9800 in Halo? Considering that Halo is supposed to make decent use of PS 2.0 effects, I expected the margin to be much bigger (refering to the 45.xx dets, btw). Nothing like difference we've seen in TR:AOD and HL2. And actually, it seems the det 50s don't really improve speed that much. Actually it's very odd that they make a bigger difference at lower res then at higher res (the difference between the 50.xx and 45.xx dets in Halo and TR at 1600x1200 is less then 1 FPS). Hell, I'd expect a bigger boost from the cheats alone, let alone any real optimisations they made. Though it's still very odd...why is it only certain backgrounds are darker? What is causing the backgrounds to be darker like that? Obviously simply making the background darker won't impact performence, but what effect is it that's being cut out that's causing it?
Actually, even TR:AOD isn't as bad as some of the benchies I saw. Every other benchmark I saw put it at around less then half the FPS of the 9800, but in this benchmark it's only around 40% slower. Sure, it's a huge margin, but the 41 FPS the 5900 pumps out is a lot more playable then the 25 FPS I saw before. I'd like to see what happens with HL2 once the public benchmark is out.