https://github.com/cscfi/csc-env-julia

Julia language environment including MPI.jl, CUDA.jl and AMDGPU.jl preferences for HPC clusters at CSC.
https://github.com/cscfi/csc-env-julia

amdgpu ansible cuda hpc julia julia-language mpi

Last synced: 24 days ago
JSON representation

Julia language environment including MPI.jl, CUDA.jl and AMDGPU.jl preferences for HPC clusters at CSC.

• Host: GitHub
• URL: https://github.com/cscfi/csc-env-julia
• Owner: CSCfi
• License: mit
• Created: 2023-11-02T11:47:39.000Z (over 1 year ago)
• Default Branch: main
• Last Pushed: 2025-01-09T07:45:04.000Z (about 1 month ago)
• Last Synced: 2025-01-22T02:36:54.729Z (24 days ago)
• Topics: amdgpu, ansible, cuda, hpc, julia, julia-language, mpi
• Language: Shell
• Homepage:
• Size: 371 KB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE.txt

Awesome Lists containing this project

README

        
# csc-env-julia

Ansible playbooks for installing a shared environment for the [Julia language](https://julialang.org/) on [Puhti](https://docs.csc.fi/computing/systems-puhti/), [Mahti](https://docs.csc.fi/computing/systems-mahti/), and [LUMI](https://docs.lumi-supercomputer.eu/) high-performance computing clusters.

The environments include plain Julia using official binaries and global [preferences](https://github.com/JuliaPackaging/Preferences.jl) for using system installation binaries for [MPI.jl](https://github.com/JuliaParallel/MPI.jl), [CUDA.jl](https://github.com/JuliaGPU/CUDA.jl), and [AMDGPU.jl](https://github.com/JuliaGPU/AMDGPU.jl).

The clusters use [Lmod](https://lmod.readthedocs.io/en/latest/) for environment modules and [Slurm](https://slurm.schedmd.com/) for managing workloads.

Documentation is available for [using the Julia environment](https://docs.csc.fi/apps/julia/) and [running Julia batch jobs on the cluster](https://docs.csc.fi/support/tutorials/julia/).

We aim to follow the conventions described on the [Julia on HPC clusters](https://juliahpc.github.io/) page.

The `Argcfile.sh` wrapper script for installing different environments depends on [argc](https://github.com/sigoden/argc) and [ansible](https://github.com/ansible/ansible).

We can use it as follows:

```bash

# 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`.