Turning PC into SuperComputer with OpenGL/DirectX

[bozo]

I have heard that is possible to get exceptional performance in processing speed using Sound and Graphics card CPU's
rather than the main one which we are all familiar with, main reason being the different architecture.
Apparently onboard cpu for these peripheral cards can be much faster with mathematical operations.

Out of curiosity I would like to learn about this concept, does anyone here have any links to source code or tutorials which
show how this can be done?

thanks.

forgot to mention, all i know so far is that you use callback functions from OpenGL/DirectX libraries, but i have no experience
in this area at all.

[Mark_Larson]

  Their are quite a few resources in this area.  I wouldn't call it a "super computer", though it can be faster than the processor.  The main webpage you should check out is www.gpgpu.org.  You can find links to other resources from there.  I also found this one on slashdot a year or more ago.  It's for a C library that supports generic non-graphic orientated APIs.  I've seen people doing FFT or other cool things in a GPU.


http://graphics.stanford.edu/projects/brookgpu/index.html

[bozo]

thanks Mark!, I'll check it out.