-
Notifications
You must be signed in to change notification settings - Fork 64
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
CL_OUT_OF_RESOURCES on Nvidia #73
Comments
I also get follwing errors in kernel log:
I can run nheqminer on CUDA without problems. |
Currently there is no support for CUDA/Nvidia devices. I hope someone to will implement CUDA support soon. |
Shouldn't OpenCL also run on Nvidia devices? |
Yes NVIDIA support OpenCL. You should be able to mine with NVIDIA GPU |
Thus my report. |
Hey Kubu! Good to hear from you. So from my minimal understanding of hardware programming, I think the problem from Nvidia is partly coming from the choice in our kernel to use wavefronts (amd hardware optmized) instead of warps (nvida hardware optimized). It has to do with the way the gpu schedules the thousands of threads it can use to do the equihash work. A warp likes to schedule 32 threads at a time while an AMD wavefront likes to schedule 64. Here you can see we're scheduling the kernel local workgroup for the optimized AMD wavefront. https://github.com/nginnever/zogminer/blob/master/src/libzogminer/cl_zogminer.cpp#L403 Good news is the kernel developer I believe has mentioned plans to support warps. |
I know that Nvidia was not target of this miner, but there are not many CUDA miners right now and I wanted to try it out on my 1070.
System info: up-to-date Arch, i7-6800k and 1070.
Trying to run the benchmark, I get following log:
and the backtrace in coredump is:
The text was updated successfully, but these errors were encountered: