What is Mjolnir?
Mjolnir is the Globe Institute high-performance computing (HPC) cluster at the University of Copenhagen. It gives research groups access to far more CPU, memory, and GPU capacity than a laptop or desktop, together with shared scientific software and reference databases.
What Mjolnir is for#
Use Mjolnir when your work needs more than a personal machine can provide:
- Long-running analyses that would occupy a laptop for days
- Jobs needing large amounts of memory, many CPU cores, or a GPU
- Pipelines run repeatedly over many samples or parameter sets
- Analyses that depend on the shared reference databases hosted on the cluster
How it works, in one paragraph#
You do not run your analysis directly when you log in. Mjolnir is a shared system, so work is coordinated by a scheduler called Slurm. You write a short script describing what you want to run and what resources it needs, submit it to Slurm, and Slurm runs it on a compute node as soon as suitable resources are free. This is the single most important difference from working on your own machine. How Mjolnir works shows the whole path in one picture, and First 15 minutes on Mjolnir walks you through it.
The shape of the cluster#
Mjolnir currently provides 20 compute nodes, ranging from 64 to 512 CPU cores each and from roughly 450 GB to 2 TB of memory, including three nodes with NVIDIA A100 GPUs. Work is distributed across several partitions — see Partitions, QoS, and fairshare for the current list and limits.
Where to start#
- Check whether you are eligible — Who can use Mjolnir?
- Request an account — Request access
- Connect and run something — First 15 minutes on Mjolnir
If you get stuck at any point, Getting help explains the supported route.