Hello Nick and everyone,
I noticed that, too: same hardware setup, same video card, but different performance in the two reviews. However, there is one difference between them: the driver versions.
- The 9800GTX review uses ForceWare 169.44 beta (GeForce 8 series) / 174.53 beta (9800 GTX) and Catalyst 8.3 (Radeon cards).
- The GTX 260/280 review uses ForceWare 175.16 (GeForce 8 and 9 series) and Catalyst 8.5 (Radeon cards).
One might think this shouldn't make a big difference. But it does. With each new driver update, NVIDIA and AMD (usually) try to improve performance by optimizing their drivers for the various titles out there (e.g. Crysis, World In Conflict etc.)
But it seems it can also happen the other way round, and TomsHardware noticed this themselves. In a Nov'07 review, they experienced a 7% performance loss of NVIDIA cards when they updated the drivers from ForceWare 169.01 to 169.09. (See
this page for the perplexing results.)
I'm not saying NVIDIA does this on purpose to make you buy new video cards. And just for the record,
I've also seen these little variations of performance in other reviews, also in established, trustworthy magazines. Anyway, you shouldn't consider benchmark frame rates as set in stone. While we get ever-changing drivers, they are more of a snapshot, giving you an idea of the performance of a video card, also in comparison to its "rivals". Nothing more, and nothing less.
Hope this helps. Kind regards.