I notice the single Tesla card is going to be retailed at 1,500 which is quite a bit more than the 8800 GTX which it seems it is comparable to. The difference I see is that the Tesla has 1.5gb of global memory and is conforms to the IEEE standard for the 32bit floating point precision yada yada.
Is there something that the Teslas will be able to do with CUDA that justifies the high cost other than the memory/IEEE standard?
Wait, are we a generation past the G80 already? I’d also be interested in seeing some comparisons with the current hardware.
I think that four-GPU server board looks pretty interesting–is there any price info for that yet?
From what I’ve read, the Tesla is a modified G80 (correct me if I’m wrong). I’m just wondering at why the projected price is so high. I suppose no one really knows, except for the engineers. Heh.
See this article for pricing. If you look around on google you will find the same prices.