Closed drdivine closed 4 years ago
Currently, multi GPU support is not implemented, but you can split the reads and call Chiron multiple times, in each call, use the following command to choose the GPU, for example, choose the first GPU:
export CUDA_VISIBLE_DEVICES=0
and then run chiron using chiron call
Describe the requested feature It would be great if I have multiple GPUs that the evaluation i.e.
chiron call
would work on mulitple GPUs. Any advice for making this work? Additional context Add any other context about the problem here.