Originally Posted by Croft
About every test goes to the 9800 GTX+; and in the ones it looses, it is only behind by 1 or 2 FPS. I know everyone is jumping on the ATi bandwagon, but that doesn't mean you have to ignore what your eyes tell you, way to fail.
The ATI leads in minimum frame-rates in quite a few tests. I'd rather have a game average 35 FPS and not drop below 20 than have a game average 50 FPS and not drop below 10. It's also worth noting that most games show the HD4850 surpassing the G92 architecture in DirectX 10 (or at least performing closer percentage-wise), which indicates to me that ATI's closer adherence to Microsoft's DirectX API has payed off. The HD4850 also consumes slightly less power - it may run hotter thanks to the single-slot cooler, but that's nothing that can't (and shouldn't) be fixed with a decent aftermarket cooler, or even some proper fan settings.
To all the people that wanted to see Crossfire tests, that's just begging for more inaccuracies. Once you start running different motherboards in a test bed, you're going to get even more variance than you would on a single board. Since the review was based around the 9800GTX+ it's to be expected that SLI is the only configuration they were able to test. It would have been nice to see Crossfire tests, but I would have rather them drop the 9800GTX+ altogether, and test some other cards at the same price point like the 9800GTX or even 8800GT.