The cluster is specialised to run SMP and MPI jobs but is also suitable to run HTC jobs (many single core jobs).
The Vega cluster is composed of 44 nodes each featuring 4 CPUs for a total of 64 cores and 256 GB of RAM. Compute nodes are inter-connected via an Infiniband QDR network. There is a GPFS storage of ~70 TB with reasonable IO performances. Two compute nodes have 1x an NVIDIA GPGPU K20 accelerator.
The cluster was installed in 2013 and hasn't changed since then. The workload on the cluster is generally high. The maximum walltime is of 14 days.
Vega is freely accessible by the members of all the Wallonie-Bruxelles universities (ULB, UCL, ULg, UNamur & UMons). You need to create a CECI account to access the clusters. See the CECI documentation for more details. We impose limitations on the number and type of jobs that can be executed since this is a shared environment and therefore everyone need a chance to get jobs started.
|Number of running jobs||Maximum 500 jobs|
|Number of concurrent usable cores||Maximum 1024|
|Number of jobs in the queue per user||Maximum 2000 jobs|
|Memory per job||Maximum is 256 GB per core/node.|
|Home & work space||Quota (soft and hard) set to 6 TB. Home and work (scratch) space are the same.|
|Walltime for jobs||Maximum allowed is 336 hours|