(Geo) Well, for someone who remembers the excitement that was generated by Dave Orton piffling that R300 t'weren't nuttin', it was R400 that he was excited about. . . and knowing that all the roadmap changes and reshufflings in the interim mean that R600 finally brings that (no doubt improved in the interim) to fruition in the PC space. . . it's a bit disappointing to see that it's not a killer enthusiast card at this point. The price/performance might be very good, we'll have to see what NV does and how AMD responds. Tho AMD does have a smaller die and, presumably, lesser memory costs vs even GTS 640MB to compete on price/performance with.
We all felt that G80 was a killer piece of hardware last fall and had surprised AMD with just how true that was. Today's results I think confirm that unquestionably.
Sometimes in a long competitive career you just have to tip your cap to the other fellow, say "well played", and move on to the next round with a gleam in your eye to get some payback.
From the engineering side, I can see where the NV30 comparisons have at least some validity. But it's on the marketing side where the comparison falls down so far, and where NV30 went from disappointment to scandal/flamewars for the ages. By pricing HD 2900 XT where they have, hopefully AMD will have removed that aspect of the equation. It was in insisting that 5800 Ultra was actually faster than 9700 Pro and priced the same MSRP (and I think maybe even a bit higher at street) that the tragedy really kicked into gear with follow on consequences of strained logic out of marketing, and driver engineers pressured into "aggressively optimized" drivers. Realistic pricing and market positioning should (hopefully --and I will add that I always know where my pitchfork is when needed) make it unnecessary for that aspect of the NV30 = R600 story to materialize.
Personally, I don't know what to make of CFAA just yet. Maybe more work there will pay some dividends, and Rys in his examples seemed to find some spots where it would pay off. Whether it gets to be you can find a setting for a specific game and leave it there all game is not really clear to me yet. It would be awesome if AMD found a way to make that a plug-in so community types could play with their own algos and distribute them to other users. Dunno if that's practical even a little bit.
Of course, having the absolute performance leader is a nice thing. It's called "the halo effect", and every IHV would rather have it than not. However, it's also been proven that it's not an absolute requirement to be reasonably successful in the market. Certainly NV made a good bid in the GF6 era by having a price/performance leader in GF6800GT at the top, and a kick-ass midrange part in GF6600GT. So it's doable. And of course AMD has bundling opportunities galore now with its own suite of parts from cpu, to chipset, to gpu that pretty much ensures decent sales there to OEMs at least in the low and middle part of the market. Given those facts, particularly the relationship with the parent company now, I think that the fate of 2400 and 2600 is probably much more important to AMD than is 2900. As an enthusiast, that makes me a little sad, but it's probably true nonetheless.
Having said that, this being the first new generation flagship released under the AMD banner, and it also being the first ATI flagship since R8500 that can't at release make at least a defensible argument to being performance king, I can also imagine a lot of people clucking their tongue and wanting to blame AMD for pulling back on the ATI reigns. Some folks will want to see this as evidence of AMD slowly abandoning the competition for all out performance champ at the top. I don't think that's what happened here, so I don't think it's a fair rap to lay at AMD's door.
Facebook
Twitter