Originally posted by: AnAndAustin
AunixP35, Developing the technology andf building the plants is expensive but the production costs of 0.13mu are MUCH smaller than 0.18mu and as such it will be much cheaper for nVidia to produce their 0.13mu NV30 than for ATI to produce their 0.15mu Rad9700, in any case ATI had all those costs when switching from 0.18mu Rad8500 to 0.15mu Rad9700 (IIRC) but 0.15mu isn't as good or as cheap to produce as 0.13mu. This would make things much better for nVidia, certainly in the long term, however as of NOW it has enabled ATI to get their product on the market first (if they hurry the F up).
Couple of things to remember...
1) Neither nVIDIA or ATI has any fabs. They outsource their production to fabs such as TSMC or UMC. They focus on their core competency which is chip design and leave production to folks who are better at it.
2) .13 is a more economically viable alternative ONLY when yields are good. You produce more chips per platter which allows you to leverage economies of scale. The last number I heard was that TSMC's .13 yields were around 15%. If TSMC can't get the ball rolling nVIDIA will be in a position where they will not be able to meet demand. We, the consumer, end up getting screwed and paying $500 for a NV30 based card. The .15 process the R300 is being produced on is quite mature at this point and I would imagine from a cost standpoint, they will most likely be cheaper to manufacture for a short time.
3) One other thing some folks don't seem to realize... nVIDIA and ATI pay for the final product that meets their requirements... up front manfacturing costs are carried by the fabs.
As a few other folks have said... trying to figure out which card to buy and speculating on a product that won't be released for at least 3 more months is an exercise in futility, especially considering the fact that the NV30 specs have not officially been announced. I imagine both cards will be solid products, the question boils down to need and cost. Of course the NV30 is probably going to be faster... it will most likely run @ a faster clock speed with faster memory and be released later than the R300. That's the way the graphics market works.
If you need a faster card soon, get the R300. If you can wait.. wait until the NV30 is out and read a couple of reviews and buy the one that you feel best about.
I doubt the performance delta will be great enough to make a difference with current machines, especially considering that these two products will no longer be limiting performance... it will be the rest of your system. When a 3.5+ GHZ CPU hits the streets with PCI X, Hyperthreading, DDRII, SATA... then it might be time to upgrade again.
It will be interesting to see what Anand has to say about the NV30 after he clarifies his little blurb on the main page. His bias towards nVIDIA is clearly evident and has been for a couple years.