Build with CUDA#
In order to compile CUDA, you will need to ensure nvcc is available.
The CUDA Toolkit is available in the nvhpc module:
module load nvhpc
To test that the module has loaded properly:
srun --pty -t 00:00:30 --partition=gpu --gres=gpu:h100:1 -N 1 -n 1 /bin/bash -l
# Once the session on the compute node has begun:
nvcc --version
echo "nvcc --version" > check-cuda.sh
chmod u+x check-cuda.sh
sbatch -t 00:10:00 --partition=gpu --gres=gpu:h100:1 -N 1 -n 8 check-cuda.sh
Once completed, slurm-<jobid>.out should contain the output of nvcc --version