Bryan Del Rizzo: We announced we were going to acquire back in February of 2008. We closed the acquisition a few weeks later, then we started moving the PhysX libraries and SDK over to our GPUs. I don't know how familiar you are with the Ageia SDK but it was developed in . So we have something called which is our architecture for allowing application developers to harness the power of the GPU cores for nontraditional tasks other than, say, 3D rendering. CUDA's been a part of our GPUs for about three years now. It's actually a hardware part. We've actually got transistors inside the chip that do the parallel processing as opposed to just traditional 3D rendering. Our GPUs have two modes, the GPU mode and then the CUDA mode for parallel processing, and the CUDA architecture is based on C. Any developer that knows C can write a program for CUDA.

GO: How do I know if my NVIDIA card supports PhysX or not?

BDR: Anything 8800 GTX or above supports CUDA, and therefore would support PhysX.

GO: Is there a way to tell what sort of performance metrics you can expect by enabling PhysX in a game that supports it?

BDR: Having PhysX run on the GPU is fundamentally days and nights ahead of what the original Ageia physics processor could do, just because of the number of cores we have in our GPU. At the low end, even in terms of our motherboard GPUs and the integrated space, those have a minimum of 16 cores. A GTX 280, just by comparison, has 240 cores. Compare that against even a Core 2 Quad or Core 2 Duo on the CPU side. You have up to four cores there, versus up to 240 on one of our GPUs currently. That's the whole basis of this CUDA stuff, that you can take advantage of all these extras cores and use them for parallel processing.