User story: As a user of HPC, I'd benefit from a template project that wraps an existing project, so that I can associate HPC files with a project without adding mess to the project itself.
Typically an HPC user will be deploying an existing codebase for use on an HPC system. The codebase is liable to run across multiple systems, including different HPC systems, locally and potentially cloud systems such as Azure or AWS.
The HPC system has a bunch of requirements — files, scripts and containers specifically — that will be needed to allow it to run on an HPC system and which are liable to be specific to that HPC system.
It can be helpful to keep these in a repository, but at the same time it's typically undesirable to store these with the code in the original repository, because they're specific to the HPC system in use.
This task is therefore to design, develop and document a template repository that ensures a healthy HPC project structure.
At its heart, this should involve allowing a repository to be created with HPC-specific files (e.g. SLURM sbatch scripts) but with the original code as a submodule.
User story: As a user of HPC, I'd benefit from a template project that wraps an existing project, so that I can associate HPC files with a project without adding mess to the project itself.
Typically an HPC user will be deploying an existing codebase for use on an HPC system. The codebase is liable to run across multiple systems, including different HPC systems, locally and potentially cloud systems such as Azure or AWS.
The HPC system has a bunch of requirements — files, scripts and containers specifically — that will be needed to allow it to run on an HPC system and which are liable to be specific to that HPC system.
It can be helpful to keep these in a repository, but at the same time it's typically undesirable to store these with the code in the original repository, because they're specific to the HPC system in use.
This task is therefore to design, develop and document a template repository that ensures a healthy HPC project structure.
At its heart, this should involve allowing a repository to be created with HPC-specific files (e.g. SLURM
sbatchscripts) but with the original code as a submodule.