How to get the CUDA version?

As Jared mentions in a comment, from the command line:

nvcc --version

(or /usr/local/cuda/bin/nvcc --version) gives the CUDA compiler version (which matches the toolkit version).

From application code, you can query the runtime API version with

cudaRuntimeGetVersion()

or the driver API version with

cudaDriverGetVersion()

As Daniel points out, deviceQuery is an SDK sample app that queries the above, along with device capabilities.

As others note, you can also check the contents of the version.txt using (e.g., on Mac or Linux)

cat /usr/local/cuda/version.txt

However, if there is another version of the CUDA toolkit installed other than the one symlinked from /usr/local/cuda, this may report an inaccurate version if another version is earlier in your PATH than the above, so use with caution.

Related Posts:

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

Leave a Comment