How do I select which GPU to run a job on?

The problem was caused by not setting the CUDA_VISIBLE_DEVICES variable within the shell correctly.

To specify CUDA device 1 for example, you would set the CUDA_VISIBLE_DEVICES using

export CUDA_VISIBLE_DEVICES=1

or

CUDA_VISIBLE_DEVICES=1 ./cuda_executable

The former sets the variable for the life of the current shell, the latter only for the lifespan of that particular executable invocation.

If you want to specify more than one device, use

export CUDA_VISIBLE_DEVICES=0,1

or

CUDA_VISIBLE_DEVICES=0,1 ./cuda_executable

Related Posts:

  1. NVIDIA NVML Driver/library version mismatch
  2. Is it possible to run CUDA on AMD GPUs?
  3. How to get the CUDA version?
  4. How to get the CUDA version?
  5. Error Message : Cannot find or open the PDB file
  6. Cudamemcpy function usage
  7. CUDA driver version is insufficient for CUDA runtime version
  8. Error compiling CUDA from Command Prompt
  9. How to verify CuDNN installation?
  10. When to call cudaDeviceSynchronize?
  11. Does __syncthreads() synchronize all threads in the grid?
  12. ImportError: libcublas.so.9.0: cannot open shared object file
  13. Tensorflow not running on GPU
  14. Use of cudamalloc(). Why the double pointer?
  15. X Error of failed request: BadValue (integer parameter out of range for operation)

Leave a Comment

tech