Printable View
The newer graphics cards (GeForce FX & Radeon 9800) can use floating point operations and should in theory be much faster in crunching numbers than your ordinairy processor. Has anybody tried writing a client that uses a DX9 graphics card for processing?
Just curious :)
http://gpgpu.org/
check this out.
every DC project must have GeForce clients =)))))))
Does a Geforce client exist for the Distibuted Folding Project? I've looked but couldn't find one...
Check out the threads under the Science listing. DF is not big on floating point math. Most of the processing is simple indexing, so a fast FPU is unnecessary.
I'm not sure if it would be good idea to use GPU as the second CPU in the system for help in crunching distributed computing projects, or if it could be implemented at all. If it could it would definitely be worth a try - my R9700Pro is sitting idle when DF is running, and has probably more computing power than AXP3200+ it is working with...
I'd like to see someone writing the Pi calculation program for GPU and compare the results with CPUs available today.