For some reason you still refuse to acknowledge the fact that high-end dominance brings success to all product lines.
SexyK - Mindshare is notoriously hard to judge, so while I definitely agree that high-end dominance has some serious perks, I'm again dubious that it's the high-end dominance of the 8800-series that has driven nVidia's chipset and mobile GPU sales. The vast majority of these products are not directly bought directly by consumers but through OEMs who, one would think, would have the good sense not to buy a product simply because another product from the same vendor is tops on the performance list. The x1900 series topped the 7900 in a lot of benchmarks but it didn't make the x1600's any more appealing.
I want you to understand that I do appreciate your point in general, but I need more than just the assertion that the 8800-series dominance is what drives OEMs to buy nVidia chipsets and mobile GPUs. I'm not at all trying to say that it has no impact. I'm really just trying to say that I think it has its biggest impact on a very small subset of nVidia's market, and that it's easy to overstate it's importance. If you've done any research on the subject on the web where nVidia's OEM customers talk about how the presence of the 8800 has affected their purchasing decisions, I'd genuinely love to see it. I'm always willing to be educated about how these businesses work.
From what I've observed however, very, very few people actively buy their own GPUs per se--the GPUs come with the PCs themselves. Most often, the decision as to what GPU options to give the consumer has already been made by the manufacturer. You get a few options from Dell, for example, that most buyers judge by the price of the card and the size of the memory buffer. As an aside, many of my friends don't even know the names of nVidia or ATI. Even when they're purchasing a system and selecting a card from one of them it just goes in one ear and out the other. They buy a computer from a major manufacturer like Dell that does what they need and, when it no longer does what they need, they buy a new one. Dell is looking at the best volume deal in several specific price segments in order to determine what to offer. If Dell tells them that a certain GPU is kick-arse, they believe it and buy it. They don't browse around on AT reading benchmarks.
On some level, I'm still aghast when I go to manufacturers like Gateway and still see 'gaming' computers being sold with x1950xt's. I've seen the benchmarks and I wouldn't buy one in a million years. Yet people still buy them because that's what they're offered by those big manufacturers. 3 out of 4 of Dell's XPS 710s have the default option as a 7900gs. In fact, it's not until to you select the system that has a base price of $2K that you get the option for an 8800-series.
The reason I mention this is merely to point out that I sincerely doubt that Dell's decision to offer a 7900gs has anything to do with the 8800's performance (or the battle between the 7900 and the x1900 series, per se). It most likely has everything to do with the price they could get for the 7900gs relative to the performance Dell thinks their customers will demand. Dropping $2000 on a computer that has a 7900gs seems imbecilic to me, but I've gotten interested in following the GPU benchmarks. Most of Dell's customers haven't. They merrily buy a $2K gaming rig with only a 7900gs and go their way.
As to the timing issue, you may very, very well be right. I don't know. Just like everyone else I'm speculating. I don't know how long after seeing R600 it will take nVidia to get new parts in the channel. I really don't. If they've already got parts manufactured ready to ship to their partners for final assembly, then you're clearly right, they'll be able to have parts in the channel nearly instantaneously.
Do you think that nVidia already has manufactured G80 refresh parts, or does it just not take but 20-30 days to get parts in the channel from start to finish? That's an honest question, I really want to know this. I don't have a great grasp of either company's manufacturing capacity or how their product cycles work.
It would just strike me as a bad idea for nVidia to have too many G80-refreshes already made when they don't know how R600 is going to perform. If they're banking on R600 only besting G80 by 15% and they target 10% over that for the G80 refresh but R600 comes in with a 35-40% improvement on the G80, then they would have a refresh already manufactured that wouldn't retake the crown.
At any rate, right or wrong, there's no need to 'bow down' to anyone. That's more my point than anything else. I'm not really posting because I want or need to be right. I'm posting because it's an interesting exercise to try and judge how the market will play out. If you're right (and you really, really might be), then you're right. It will be a little bit of a downer simply because it will mean a slightly increased chance that AMD will abandon the really high end GPU market because nVidia has kicked the crap out of them so consistently that they don't even want to compete any more. Great for nVidia, not so great for consumers as the cost to entering this market is pretty high. That's not to say that some other major player couldn't emerge.
If nVidia out-times AMD/ATI again, however, they clearly deserve all the business props in the world.
I'll just make two final observations and be on my way.
Holding back a market-leading product hoping to somehow corner the DX10 market when your competitor already has already launched DX10 hardware at pricepoints from $250 on up is not smart business.
I'm not sure I completely agree with this. More because I don't think it's about 'cornering' the market for ATI. I think at this stage it's just about getting first dibs for once, even if it is ever so brief. Even if ATI had the best part out for the first 30 days of a major DX10 launch, that would be a step in the right direction for them. I guess I also think that having DX10 parts out right now is not really relevant. As someone else said, it's DX9 performance that is driving the 8800 sales. DX10 compatibility is really icing on the cake. I think both companies want to have the best DX10 parts available when people actually start upgrading their hardware because of DX10 games.
Everyone knows that 8800's are DX10 because they've been around for months. When a DX10 game comes out, less-sophisticaed consumers will say "Oh my friend has an 8800 and i know it's DX10, so I'll go pick up an 8600." They won't say "Oh I know 8800's are DX10, but I'll run out and pick up an X2600XT which hasn't been on the market until now and I know very little about."
Just as a last point, I think the less sophisticated consumers (as I indicated above) have their choices made for them by the major manufacturers. I think you may be really talking about the less sophisticated performance buyers who follow the GPU market somewhat. The really savvy performance buyers are going to look at both benchmarks and choose the best price/performance ratio. The big manufacturers are going to choose the product that provides the best price in relation to the performance expected by their uneducated consumers.
In the end, I think the 7600's success wasn't because the 7800 had come out long before the x1800/x1900, creating positive mind-share, nor was it because the 7900 kicked the crap out of the x1900 (it didn't), it was really because the 7600 kicked the crud out of anything ATI had in that price range. It was a great, profitable, mid range part. The price/performance ratio hooked both the savvy performance crowd and the major manufacturers. nVidia could offer the 7600 cheaper than ATI could offer a comparable part.
Cheers and thanks for your thoughts.