Wondering what NVIDIA’s plans are for the PhysX tech it acquired recently? The Tech Report has confirmation from NVIDIA CEO Jen-Hsun Huang that the company plans to incorporate support for that hardware-accelerated physics middleware into its GeForce 8 range of GPUs.
But it won’t be doing it in hardware. It doesn’t need to. The massively parallel nature of the GPU is already designed to handle the type of processing required for hardware physics. All it takes is a bit of software magic, and PhysX will run happily on GPUs that support CUDA.
What’s CUDA? It stands for Compute Unified Device Architecture, and put simply, it’s a tech included in the GeForce 8 range that allows the GPU to run general purpose code. According to the TR article, NVIDIA will port the PhysX middleware to a variant of the C programming language that can run on GPUs. Then it’s just a matter of executing the code on the video card and voila – hardware PhysX support without the need for a PhysX card.
If anything’s going to help the adoption of the PhysX middleware, it’s knowing you only need one piece of hardware to take advantage of it. Note that PhysX is free, while Havok is not, so the latter will undoubtedly be keeping a close eye on this development.
Huang didn’t put a date on support, but I suspect it’ll come in the form of a driver update in the not-too-distant future.
GeForce 8 graphics processors to gain PhysX support [Tech Report, via Blue’s News]