Conclusion
Nvidia maintains that we’ll see GF100-based hardware in Q1 of this year—up to two months from now and as many as six months after AMD began shipping its Radeon HD 5870. Typically, that’d be a debilitating deficit to overcome. But if a single GF100 demonstrates the chops in today’s games to do battle against ATI’s flagship (which, by the way, now starts at $650 and spans up to $720), then we can comfortably posit that shipping DirectX 11-capable hardware six months late means little to Nvidia’s future, even if it’s eating up the company’s earnings today.
More concerning, perhaps, is that this three billion transistor chip will likely struggle to find its way into an affordable price segment. At least until Nvidia starts talking about derivatives, GF100-based boards will remain exclusive to the folks able to afford Radeon HD 5800-series cards.
For those who are in the market for high-end graphics, however, it’d seem that good things are on the way. It would have been difficult to walk away from the specifications Nvidia presented and the preliminary numbers it offered without being impressed. Seeing a more-than-doubling of performance in some of today’s games versus GeForce GTX 285 and incredible potential in tomorrow’s (thanks to an architecture optimized for geometric complexity and GPU-based compute capabilities) sets GF100 up to be one of Nvidia’s most game-changing designs.
Of course, we can’t let Nvidia off the hook quite that easily. All of the benchmark numbers we’ve seen come from the company’s own boxes using cards with undisclosed clocks. What we have is little more than a preview of hardware to come. That hardware is expected to be expensive, power-hungry, and hot. We don’t know exactly when it’ll drop, how many models Nvidia will build on the GF100 GPU, or how much they’ll cost.
Meanwhile, AMD is shipping DirectX 11 hardware from $99 to $649 (though you’ll need to spend at least $150 to get playable DirectX 11 performance). It’s offering Eyefinity across the board, which, contrary to the Doubting Thomas’ out there, isn’t a gimmick and is in fact viable for both gaming and productivity. And it maintains its appeal in home theater environments, too.
The Inevitable “We’ll See”
So even as AMD looks to maintain a couple of its most notable advantages over Nvidia’s graphics card lineup, the green team has the bump and a long, floating set for what we’re expecting to be a fairly spectacular spike when GF100-based cards start shipping. More than likely, it’ll compound its own list of advantages, adding a leg up in gaming and compute performance to its PhysX, CUDA, and GeForce 3D Vision support. GF100 is an ambitious effort—we’re sure of that. But it’ll take a card in the lab to demonstrate how Nvidia’s latest effort fares beyond its academic virtues.
