Hi all. I’m trying to get CUDA up on my machine to use with MXNET in python. I had installed CUDA, and MXNET, but was getting an error from python telling me no GPU was detected.
After doing some digging, I think that my CUDA install might not be working correctly.
I’ve downloaded CUDA 9.2 and installed and patched it. I’m building the samples library using Visual Studio 2017.
When I rebuild the samples, I get the following message:
========== Rebuild All: 41 succeeded, 122 failed, 0 skipped ==========
I’m not sure that those should all be succeeding but that seems like it possibly indicates a problem?
When I run the deviceQuery.exe sample, I get the following output:
deviceQuery.exe Starting…
CUDA Device Query (Runtime API) version (CUDART static linking)
cudaGetDeviceCount returned 30
→ unknown error
Result = FAIL
Obviously this isn’t very helpful. Does anyone know what I can try in order to get this working?
Specs / Other info:
Windows 8.1 PC, Intel CPU
Geforce GTX 760 GPU
Fresh install (today) of both NVIDA drivers and CUDA 9.2
Have restarted computer multiple times.