It looks like you're new here. If you want to get involved, click one of these buttons!
Some people here have expressed hope that, while Nvidia's top GPU chip for this generation, GK100, was cancelled, the next generation GK110 should come somewhat soonish. Well, no.
"The GK110 won't reach the market until the fourth quarter of 2012, and multiple folks from Nvidia forthrightly admitted to us that those chips are already sold out through the end of 2012. All of those sales are to supercomputing clusters and the like, where each chip commands a higher price than it would aboard a video card."
And that's assuming everything goes well. Based on Nvidia's recent history, it probably won't. With more than double the transistors of GK104, GK110 will likely be Nvidia's largest chip ever by die size. The last huge die that Nvidia was able to bring to market without massive delays was G80. For those keeping score, that was in 2006.
And then once Nvidia does have dies, what will they do with them? Tesla cards at $4000 each? Quadro cards at $4000 each? GeForce cards at $800 each? The last option gets nothing (or perhaps token quantities for a paper launch) until the first two are satisfied.
The good news is that GK110 will be heavily based on the established Kepler architecture, and built on what will by then be a mature process node. So two of the major factors that fouled the launches of GK10* and GF10* will be absent. But 600+ mm^2 dies are still hard to make if you want them to work properly, which is why AMD declines to even try.
Still, once Nvidia gets the cards working, GK110-based GeForce cards will be the undisputed king of graphical performance. Ever since AMD went to the small die strategy four generations ago, their plan was to cede the top end to Nvidia, while building the best $100 and $200 and $300 cards that they could. The problem for Nvidia is that they haven't been able to fill that top end properly since G80. When your die is 70% bigger than your competitor's, you really need to win in performance by more than 20% or so.
Now that Nvidia finally has a basic architecture competitive with AMD's, they have a chance at having a top end card that is by far the best. A single GK110-based card has a serious shot at being better than two of AMD's top end cards in CrossFire, at least if you ignore average frame rates and consider actual gaming experience, where CrossFire and SLI aren't nearly as good as you'd expect from the average frame rates.
But that's only if Nvidia can deliver. Can they? AMD has had their entire Southern Islands lineup available for more than two months now, while Nvidia has had nothing but paper launches of Kepler until some GeForce GTX 670s started showing up recently. GK110 will probably be up against the top end GPU from AMD's next generation, not Tahiti, but with it unlikely that AMD will either go with an enormous die or find miraculous RV770-style efficiency improvements, GK110 will probably win the top end, and probably by a lot. Delays won't change that unless GK110 barely beats AMD's 20/22 nm GPUs to market, and that's probably not coming until 2014.