csc-env-julia

Ansible playbooks for installing a shared environment for the Julia language on Puhti, Mahti, and LUMI high-performance computing clusters. The environments include plain Julia using official binaries and global preferences for using system installation binaries for MPI.jl, CUDA.jl, and AMDGPU.jl. The clusters use Lmod for environment modules and Slurm for managing workloads. Documentation is available for using the Julia environment and running Julia batch jobs on the cluster. We aim to follow the conventions described on the Julia on HPC clusters page.

The Argcfile.sh wrapper script for installing different environments depends on argc and ansible. We can use it as follows:

# Install Julia binaries
argc install-julia-linux-x86-64 --version 1.10.7 --system puhti

# Install MPI.jl preferences
argc install-mpi --version 0.20.0 --system puhti

# Install CUDA.jl preferences
argc install-cuda --version 5.2.0 --system puhti

# Install AMDGPU.jl preferences
argc install-amdgpu --version 1.1.3 --system lumi

# Install JupyterLab and notebook for IJulia.jl
argc install-jupyter ---version 4.3.4 --system puhti

For more information, use argc --help.