ATI has announced its commitment to what it calls Stream Computing, and is partnering with Stanford University to release a GPU-enabled Folding @ Home
Stream Computing is ATI's term for general purpose GPU use. The GPU has multiple piplines, as many as 48 right now, and this makes it a massively parallel processor. Rather than just using it to render graphics, why not use it to solve massively parallel data problems, too?
Speaking to us in San Francisco, ATI's Dave Orton explained that certain types of data problems map well to this kind of computation. One of these is scientific analysis of data, just as Folding @ Home does.
To this end, Stanford has released a new, GPU-enabled Folding client which users can download. Using the massively parallel computational power, Stanford testers have seen a 15x increase in performance over a standard CPU working alone.
Folding teams powered with ATI cards should be able to get work units done rather quicker than their competitors.
If you fancy having a go, you can join the bit-tech team
Let us know your thoughts on general purpose GPU usage over in the forums.