The GPU codes were developed in CUDA 11, which is the version that should be loaded (e.g. via module command on an HPC cluster). Some cluster is default to version 12 now. Thus, this information should be noted somewhere, but preferably do a version check and raise error.
Not really sure where to do this check -- I guess it would have to be done in the script itself that runs the sim (because it would have to run after the slurm job has been created).
The GPU codes were developed in CUDA 11, which is the version that should be loaded (e.g. via
module
command on an HPC cluster). Some cluster is default to version 12 now. Thus, this information should be noted somewhere, but preferably do a version check and raise error.