Interesting. A bit disappointing.
I read somewhere that the X1700XT was going to be around X1800XT performance. With those specs, it's not going to be anywhere close. As a matter of fact, I would expect the X1700XT to have to run in excess of 550 MHz if it is to perform at least as well as the 7600GT. Fortunately, at 80nm, 550+ MHz shouldn't be a problem, but it doesn't look any better than the 7600GT on paper.
You'd think one of these companies would try to "go for the b@lls" so to speak, but they seem to be perfectly content to simply match one another (a 12/36 pipeline configuration for the X1700XT would be a much more agressive way to compete with the 7600GT, and since it would still use 128-bit RAM, it would leave enough of a gap at high resolutions to not cannibalize sales of the top cards).
The X1900GTO should be a solid card (probably matching the X1800XT) but where does that put the still-unreleased X1800GTO? At the $250 price point?
Right now, Nvidia's strategy makes more sense to me: the 7600GT and 7900GT are well designed cards that need minimal cooling and perform well. ATI's top X1700 should be a good card, but they will have two "cut down" cards competing at attractive pricepoints: the X1800GTO (~$250) and X1900GTO (~$300). And these cards, depending on whether they are entirely new cores or simply modified X1800XT's and X1900XT's, may cost more to manufacture than Nvidia's cards. Meaning prices can't drop as quickly, and profit margins are lower.
The only thing I could see making the GTO's hot sellers is if they can unlock pipelines to become XT's, like the previous generation (X800GTO2). Otherwise, IMO, the 7900GT is a more attractive card atm. Although I guess we won't know actual final performance until ATI releases these cards.