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

by

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

More Related Contents:

  1. How is CUDA memory managed?
  2. How to measure the inner kernel time in NVIDIA CUDA?
  3. How do CUDA blocks/warps/threads map onto CUDA cores?
  4. Understanding CUDA grid dimensions, block dimensions and threads organization (simple explanation) [closed]
  5. What can I do against ‘CUDA driver version is insufficient for CUDA runtime version’?
  6. What is a bank conflict? (Doing Cuda/OpenCL programming)
  7. nvidia-smi Volatile GPU-Utilization explanation?
  8. How do I use Nvidia Multi-process Service (MPS) to run multiple non-MPI CUDA applications?
  9. How can I make tensorflow run on a GPU with capability 2.x?
  10. CUDA determining threads per block, blocks per grid
  11. How does CUDA assign device IDs to GPUs?
  12. What is the canonical way to check for errors using the CUDA runtime API?
  13. How do I choose grid and block dimensions for CUDA kernels?
  14. Unspecified launch failure on Memcpy
  15. Different CUDA versions shown by nvcc and NVIDIA-smi
  16. What is the correct version of CUDA for my nvidia driver?
  17. GPU Emulator for CUDA programming without the hardware [closed]
  18. How to get the CUDA version?
  19. What kind of variables consume registers in CUDA?
  20. In CUDA, what is memory coalescing, and how is it achieved?
  21. CUDA apps time out & fail after several seconds – how to work around this?
  22. Why is NVIDIA Pascal GPUs slow on running CUDA Kernels when using cudaMallocManaged
  23. How to use 2D Arrays in CUDA?
  24. How can I check the progress of matrix multiplication?
  25. why do we need cudaDeviceSynchronize(); in kernels with device-printf?
  26. CUBLAS: Incorrect inversion for matrix with zero pivot
  27. Using maximum shared memory in Cuda
  28. Cuda Mutex, why deadlock?
  29. CUDA function pointers
  30. Using an array of device function pointers

Leave a Comment