I am a physics student working on some modeling and simulation tasks. As I am on a budget, I was thinking the 8800GTS with 640MB would be reasonable (newEgg has an EVGA for $360 after rebate). The 320MB is $300. I have never been disappointed with additional memory, so I thought that doubling the memory for $60 would be a good thing. Correct? I think my PC may be happier with a few less watts being drained so the GTS seemed a better choice than the GTX.
Does anyone know what the new Quadro FX 5600 can do with CUDA? The release notes only state the 8 Series work , but the press release on the 5600 state CUDA capable.
Yes, more memory can only be better than less in general External Image, but really how much will be enough for you depends on your application and working set size, so I can’t give you any hard advice here.
The release notes only mention the GeForce 8 Series simply because they were posted before the Quadro FX 4600 and 5600 were announced. Both boards fully support CUDA.
Thanks for the note. Is there an estimate of the GFlops possible on the Quadro FX 5600? Is it significantly better thean the 80GTX? - which I think was listed around 330 on the graoh on page 1 CUDA Programming Guide ver 0.8.
The Quadro FX 5600 has the same basic hardware specifications as the GeForce 8800 (same number of multiprocessors etc.). The clock speeds are slightly higher, I believe.
