How can I use Singularity containers to install and run software on my institution’s HPC cluster?
Here is a tutorial for using Singularity on the CRC computing cluster at the University of Notre Dame, which uses the SGE job scheduling system. It serves as a general introduction to containers and may be applicable to other computing clusters as well. The tutorial walks you through the process of building a container and running a GPU-accelerated PyTorch program on a GPU node. It also has tips for managing Python packages efficiently.
Singularity is available on compute nodes on Odyssey and user can using SBATCH or an interactive session in SLURM to their workflow using singularity. We also have documentation on how you can convert docker or other OCI containers to a singularity image and launch them. https://www.rc.fas.harvard.edu/resources/documentation/software/singularity-on-odyssey/