I'm writing some prototypes for running parallel processes on my laptop that has GeForce 8700M and would like to get desktop. I wonder which option will run faster one Telsa C1060 or a pair of GeForce GTX295? Either first or second option allowed from my purchasing dept.
migrated from stackoverflow.com Jul 20 '09 at 20:49
The NVIDIA Tesla C1060 is not a Graphics Card. It is a pluggable processing unit. You can plug anywhere from 1 - 4 of them into your motherboards PCI x16 2.0 slots and then you write NVIDIA Cuda code to offload processing to these cards from your CPU.
They are quite effective when used in this fashion and perform close to a small supercomputer.
I can't find a single review, benchmark or price for the Tesla C1060 so I suggest you go with the pair of GTX295s since they're true and tried. Being a first adopter seldom works out well.
UPDATE: The price point for the Tesla C1060 seems to be around 1700$.
UPDATE2: Very hard to find any kind of benchmarks for the card, this is the closest I found, indicating that it's (in this benchmark at least) very close to a GTX295, which might indicate that the benchmark isn't very parallellizable.