Hi,
we're excited to get started with this package and are trying to run the basic example right now. We keep having an issue with the line >>> cuda.select_device(rank), and get the error message
numba.cuda.cudadrv.error.CudaSupportError: Error at driver init:
[803] Call to cuInit results in UNKNOWN_CUDA_ERROR:
We suspect that this is because the package is trying to run the program using MPI, but we only have a single gpu.
Is there a flag to run it without MPI?
Best, Tobias and Tuan