Because what this forum and most other communities on the internet doesn't understand is that the end-user gaming market is less than 10% of nvidia's profits, and they don't give a shit to persuade people like you or me into waiting.
The real profit margins comes from the workstation market with Quadro and the server market with Tesla. Even their fucking chipset market with Apple and ION is bigger than the GeForce market.
The thing above generate millions from OEMs and large companies. The GeForce market moves pennies compared to that. Yet the only people that are interested in performance numbers are the GeForce users that are impatient nerds or fanboys.
Why would nvidia jeopardize millions of dollars with an early architecture launch to satisfy you? Why would they care about releasing early benchmarks to satisfy you?
There are gigantic companies waiting to purchase the new Quadro and Tesla cards regardless of when they come out simply because they will perform better than the old architecture, they will save time, and time = money. Nvidia has no other competitor anyways, ATI does not exist in that market.
So just wait, be patient and hope that their GPGPU and CUDA engineered moster can handle graphics just as well as crunches numbers. And remember that personal computers are nothing but a small fraction of their business.
If you need a card now, buy ATI. It's simple. You don't play the waiting game if you NEED something. I don't think this is can be classified as a daily indispensable item like food or shelter, so you should be able to wait FFS.
/end rant, dont take it personal
Well, it's not quite like that, but pretty close.
The revenue figures from GPUs are a significant proportion of the overall NV business, while the GPGPU revenue is tiny, but the discrete chip margin is much much lower than that of GPGPU and its ilk.
Discrete is very important still to NV, but in terms of margins it's not as good as other areas.
Bear in mind that NV is going to basically lose all of its chipset business in the not too distant future and they are let with 2 sources of profit: GPGPU and discrete cards.
Now if you consider the discrete market, the very high end isn't a huge proportion of it, but revenues are still high, maybe 25% of total is somewhere above >$300 cards, by revenue, not volume, which means it's still a useful market segment, but it's by no means the be all and end all for NV, like you say, and their bread and butter is still going to be the more mainstream segments and (they hope) the GPGPU area in the future.
For NV this card doesn't even really matter as a totally functional product.
It will do two things: Give them bragging rights in discrete GPUs and allow them to show what they can do in the GPGPU arena going forward.
Fermi isn't going to be (IMO) the massive seller in the GPGPU market, but it is positioned as a breathrough product to say "hey, this is what we can do, you should start considering this route", and then future designs can be positioned to really win big business and replace the chipset losses.
Anyone who is looking at Fermi for GPGPU solutions and doesn't already use them will likely wait for a Rev2 product anyway, and the fact that Fermi is broken might not make a huge difference (you have to plan and budget these things).
Facebook
Twitter