Folding Forum

Hello, i'm relatively new to folding and have completed several WU's using the GPU client with my Geforce GTX560TI . I was curious if it is possible to use both my Core I5 2500 and my GPU to work together on the same WU? If so how? Sorry if this is silly question. Thanks in advance.

Yes.

The Nvidia GPU driver uses so little CPU that you can get full use of your CPUs. (AMD Drivers use more CPU, and it is often best to leave 1 CPU free to handle that load)

The v7 Client makes it easy to run CPU and GPU clients.

Hi there quak3man, welcome to the Folding@home support forum!

While it's possible to process separate WUs, each using different hardware under traditional v6 clients or the new v7 client, currently there is no way to have them process the same Work Unit. Such technology is under development: http://folding.typepad.com/news/2012/02 ... n-fah.html

Sorry, NO. FAH does not currently have a client that can use the processing power of both the CPU and GPU on the SAME work unit.

However, as noted above, you can use both types of hardware to increase your contibutions to folding, just with separate WUs on each type. Currently, that's more productive anyway. .

Thanks everyone for the quick and helpful replies. In that case i will be sure to download the appropriate client for use my my CPU as well and run that as well as my GPU client.

I have a follow-up question on this subject. I have just upgraded my GPU (to an XFX-HD-6950), and I want to find the best balance between my CPU client and the GPU client (I'm running V7 of the beta). Would I get the best overall results (i.e., greatest number of WU's each day) if I set the CPU/SMP to "0" and let the client manage everything, or should I set the CPU/SMP to a specific number of cores (2, or 4?)?

Hello Crestwood, welcome to the forum.

We could only guess, since you didn't tell us what kind of CPU you have. Even then, it would still be a good guess. Run the GPU by itself and note the PPD. Run SMP by itself, and note the PPD. Then run both on all cores. Then run both with SMP using one less core. The one with the best total score wins.

Crestwood wrote:I have a follow-up question on this subject. I have just upgraded my GPU (to an XFX-HD-6950), and I want to find the best balance between my CPU client and the GPU client (I'm running V7 of the beta). Would I get the best overall results (i.e., greatest number of WU's each day) if I set the CPU/SMP to "0" and let the client manage everything, or should I set the CPU/SMP to a specific number of cores (2, or 4?)?

The V7 client will try to use all CPU cores by default. AMD GPUs typically consume one CPU core while folding, which would slow down the SMP folding. Likewise the SMP folding may impact GPU folding, but that's a guess. Like 7im said, try out some combinations and see. Maybe you'll get better performance by taking the SMP down a core.

Thanks for the info. I'll experiment a bit to see what combo yields the best results. I should have described my system a bit more fully. It's a CyberPower build, which has been upgraded since it was first assembled. Current configuration includes: Gigabyte GA-P55A-UD3; Intel i7-870 CPU with CoolerMaster V8 cooler; XFX HD-6950 GPU; 2x4GB of 1600 mHz RAM; Corsair TX750 watt PSU.

With an i7 it's usually recommended to drop the SMP client to -smp 6 if you need to free up a core due to GPU folding. Certain projects cannot run on 7 threads (or other prime numbers >5) and will fail instantly; getting the client to use 6 threads may decrease performance a bit but will ensure the every project will run properly (not counting an unstable overclock, the odd bad WU, etc.)

Zagen30 wrote:getting the client to use 6 threads may decrease performance a bit but will ensure . . .

It will decrease SMP performance compared using 8 threads, but only very slightly. It will most definitely increase performance on a [SMP+ATI GPU] system.