Prevent kernel timeout on second GPU device

Dear all,

I know that the OS does prevent too long kernel execution times on GPU’s by canceling an execution with a timeout-failure. I also know that—on a desktop machine—one should get a second GPU if possible as then the first may still be used uninterrupted by the Window manager of the OS. However, I now do have a second GPU (second 8800GT in a Mac Pro) but this one still has it’s kernelExecTimeout enabled.

How can I disable the kernel exec timeout under OS X and use my second 8800GT solely for computation purposes? Ideally, I would configure the same behaviour for both CUDA and OpenCL.



This was posted on the Apple OpenCL forum. An Apple employee stated there that rescinding the timeout for the second GPU was an interesting idea they would consider for the next release.

Thanks, coleb, for the reply. I guess the Apple OpenCL forum is only available to ADC members?

Also: do you know how the desired behavior might be achieved using CUDA? I guess this should be a driver configuration issue on Apple’s side, too.



If I recall you may be able to get an account without buying the full fledged membership. Though I have the membership so I’m not positive. There isn’t as much activity on that board as this one, though I do like it because the Apple employees are very responsive.

I know nothing about CUDA, we stayed away from GPU computing until there was an open standard.