Hello, I have a few partitions in the HPC of my work, and some of them have nodes with multiple GPUs. Since this is a shared space I would like to not block a whole node when I can use just 1 or 2 GPUs, but deepconsensus uses all of the GPUs available even if I limit the number of requested GPUs in SLURM. There is a way to avoid that deepconsensus uses all the available resources?
Hello, I have a few partitions in the HPC of my work, and some of them have nodes with multiple GPUs. Since this is a shared space I would like to not block a whole node when I can use just 1 or 2 GPUs, but deepconsensus uses all of the GPUs available even if I limit the number of requested GPUs in SLURM. There is a way to avoid that deepconsensus uses all the available resources?