What is the current state of real-world GPGPU programming? I have a relatively simple but extremely processor-intensive problem that would vastly benefit from the kind of simple parallelism I understand GPGPU offers. Are there currently any cross-platform solutions for this?
Name:
Anonymous2008-07-08 14:54
It should be pretty usable. Have a look at CUDA and Close-to-metal... And also http://en.wikipedia.org/wiki/BrookGPU , which I believe supports both popular brands of GPUs, runs on linux and windows, and is used in Folding@home.