I submitted a job and it runs fine. However, it is slow and I found that it is not utilizing GPUs.
#!/bin/bash -l #SBATCH -o out #SBATCH -e err conda activate tensorflow-gpu python train.py
After the the job was submitted, I logged in to the compute node by running:
run --pty --jobid 12345 /bin/bash
and then run nvidia-smi
.
nvidia-smi
to return information about GPU usage.nvidia-smi
returns no devices were found