I only have 1 nvidia GPU with CUDA installed - how to I ensure only 1 GPU used?
RuntimeError: CUDA error: invalid device ordinal
CUDA kernel errors might be asynchronously reported at some other API call,so the stacktrace below might be incorrect.
For debugging consider passing CUDA_LAUNCH_BLOCKING=1.
I only have 1 nvidia GPU with CUDA installed - how to I ensure only 1 GPU used?
RuntimeError: CUDA error: invalid device ordinal CUDA kernel errors might be asynchronously reported at some other API call,so the stacktrace below might be incorrect. For debugging consider passing CUDA_LAUNCH_BLOCKING=1.