Also note their benchmark of a desktop with a Radeon 9600 Pro. Well, hate to break it to ya, but the 9600 Pro is a laptop-only card. I smell fish.
No mention anywhere of what processor the "desktop benchmarks" use either. No mention of which machines are HT-capeable, and no mention of which of those machines have had HT enabled. No mention of which graphics cards in the test may have been underclocked (generally for heat reasons).
Software-wise, no mention of version of drivers, minimal mention of which settings were used in the test software (no mention of texture detail, dynamic lighting, dynamic shadows, or any other video-card-intensive settings).
Yeah, I smell lots of fish.
Alienware seems to have a marketing agreement with nVidia (go to their site - look at all the huge nVidia graphics, nVidia cards come standard on all machines recently released). They have big marketing releases about nVida support (but not ATI support).
I agree that it would be nice to see the two cards compared in the same machine running the same software. And, obviously, don't let ATI, nVidia, or any close partner of one of those companies set up the test.
Which is my problem with testing in an Alienware system - to my mind they are potentially biased by their partnership with nVidia. I mean, when Alienware publically says nVida cards are preferred on their performance-oriented systems, wouldn't it be embarrasing if an ATI card performed better?