This cluster has been decommissioned in July 2020.
Hosted at UCLouvain, the HMEM cluster consisted mainly of 12 fat nodes, each with
48 cores (four 12-core AMD Opteron 6174 @ 2.2 GHz).
Memory configurations included:
- 2 nodes with 512 GB RAM
- 7 nodes with 256 GB RAM
- 3 nodes with 128 GB RAM
All nodes were interconnected through a fast Infiniband QDR fabric and provided 1.7 TB of local RAID scratch.
Additionally, local disks across nodes were aggregated into a 12 TB global BeeGFS filesystem.
Suitable for:
Large shared-memory jobs requiring very large RAM (100+ GB) and many cores (24+).
Max wall time: 15 days.
Resources
- Home directory (50 GB per user)
- Global working directory
/globalfs($GLOBALSCRATCH) - Local working directory
/scratch($LOCALSCRATCH) - Internet access available on compute nodes
- Queues*:
- Low, Middle, High queues — Max 15 days, 40 running jobs/user
- Fast queue — 24 hours, no access to
$GLOBALSCRATCH
Access / Support
This cluster has been decommissioned and is no longer accessible.
