Cluster description
Global Description#

Some important definitions
- Login node: a node to which a user logs in and through which they submit their jobs to the cluster
- Compute node: a node on which user-submitted jobs are executed
- Compute node: a node on which user-submitted jobs are executed
Hardware#
Compute nodes#
CPU compute nodes#
⚠️ The values below can change. Check the current situation:
sinfo -Ne --format "%.15N %.4c %.7z %.7m" -S c,m,N | uniq
| Nbr nodes | CPU | RAM (GB ) |
|---|---|---|
| 16 | 255 | 1500 |
| 8 | 51 | 1000 |
| TOTAL | 4488 | 32000 |
| (Last update: 22/07/2026) |
GPU nodes#
| Nbr nodes | GPU | CPU | RAM (GB ) | Type | Disk /tmp | Disk /bank |
|---|---|---|---|---|---|---|
| 2 | 2 | 80 | 512 | NVIDIA l40s | ~64 GB | 3.5 TB |
| (Last update: 22/07/2026) |
GPU Instance Profiles#
⚠️ The values below can change. To check the current situation:
sinfo -Ne -p gpu --format "%.15N %.4c %.7m %G"
| Profile Name | GPU Memory | Number of Instances Available |
|---|---|---|
| l40s | 24GB | 4 |
Storage#
@TODO
DELL PowerScale:
- 2.5 PB of project storage
- 7 TB of scratch
Software layer#
-
The cluster is managed by Slurm.
-
Scientific software and tools are available through Environment Modules; they are mainly based on Conda environements or Apptainer images.
-
Deployment and configuration are powered by DevOps-style orchestration.