See Gamespot last benchmark: http://www.gamespot.com/features/6177688/p-6.html

Interesting to see the HD 2900XT win the performance crown, beating the 8800GTX by approximately 10%.

What is much more interesting though is how the X1950Pro trumps its direct competitor, the 7900GS. I've always thought that the X1950Pro had more long-term potential, well, this 40% victory is proving it.

Note, however, that the HD 2900XT is currently doing much worse in Windows Vista. The wide discrepancy between XP and Vista results suggest that Vista performance could be improved in a subsequent driver release. NVidia's drivers look much more mature in this benchmark.

Note to those who've been recommending the 640MB version of the 8800GTS "for high resolutions": it performs significantly worse than an overclocked 320MB at 2048x1536...

Some of the weirdest benchmarks I've seen lately. Thoughts?

People on neowin.net told me this is a "The way it's meant to be played" nvidia-sponsored game.

HAHAHAHA!!!!

ok... so the 8800 GTX out performs 2900 XT only in Vista.... Don't need to act like fanboy about it

It is because the lower mb versions of video cards use better ram. 4 fps isnt a big dif, where do you see ati whipping nvidia by 10%??

That is pretty amazing, and one of the reasons I quit going to game spot years ago. If you go to a more reputable gaming source like firingsquad

http://www.firingsquad.com/hardwar [...] /page5.asp

Isn't it amazing how the ATI card doesn't come in faster than any of the Nvidia cards but the old 7k series?

I wonder what kind of a financial deal ATI reached with gamespot on that one.

My question is, why would you have a top of the line graphics card if your not running dx10? DX10 numbers are all that matter to me. All the other benchmarks Ive seen have nvidias cards winning too.

The FiringSquad tests are under Windows Vista, while Gamespot are under XP. And Gamespot show that the HD 2900XT does a LOT worse in Vista than in XP. That wide discrepancy suggest a driver update would help in Vista.

Bioshock DX9 and DX10 modes show only negligible differences, both visually and in the framerate, as FiringSquad and Gamespot show.

Trust gamespot for all the lastest gaming reviews and previews. But not for hardware I recommend you use this site "TomsHardware.com" It's really good in fact you are posting on it right now

Uh, the tests were done under XP with Gamespot and under Vista Ultimate on Firingsquad. Big difference. It pays to actually read the small print before you start throwing baseless accusations around. I'll give you that Firingsquad is a little bit more reputable when it comes to hardware, but that doesn't mean that writers at gamespot are getting paid off by anyone either. That being said, I wish they would have benched with a quad core to see if it offers any advantage, though I seriously doubt it, I don't think the unreal engine supports more than 2 at this point.

Since most people are using XP at this point, I think this is a rather impressive showing for the 2900XT, although it is only one game and could be a fluke. We'll see when Crysis and some of the other big fall titles start making their appearances.

The HD2900XT also beat the 8800GTX in Rainbow Six Vegas, which uses UE3.

I call this bull simply based on the fact that it's from Gamespot. They gave the game a 9.0/10, which is the same rating they gave Perfect Dark Zero.

ROFL Yea what's up with that ?

I saw a review about DX10 on this site that showed ATI doing pretty good during DX10 games. DX10 is sooo young still.

"Bioshock strongly favors the ATI Radeon"

4-5 fps is not what i call "strongly" especially when you look at dx10 scores and when the ultra is missing.