Types of scientific computing resources available to the researchers:
Recherche
Liens UNIL
Infrastructures
Wally
Wally is a 1536 cores Intel Xeon based HPC cluster offering the following features:
Job category supported --> - Any suitable
- interactive job
- tightly coupled parallel job
- embarrassingly parallel job
- sequential long running jobs (< 10 days )
- A maximum of 1000 jobs per user in the queuing system
Access --> via ssh to wally-front1.unil.ch
OS --> CentOS Linux release 7.6.1810 (Core)
Access node --> 1x dual socket Intel(R) Xeon Gold 6148 CPU @ 2.40GHz CPU (total 40 cores), 786 GB RAM
Compute Nodes --> 91 x dual socket Intel(R) Xeon CPU E5-2660@ 2.20GHz (total 16 cores), 64 GB RAM
usable for batch processing or interactive access
5 x dual socket Intel(R) Xeon CPU E5-2660@ 2.20GHz (total 16 cores), 96 GB RAM
usable for batch processing or interactive access
Storage --> - 5 GB of shared home storage available on the /users/<unil_username> path
please, DO NOT USE the shared home space for computation
- 346 TB of NOT BACKED UP shared scratch available on the /scratch/wally path
- local node scratch available on the /scratch/local/ path
/scratch/local/daily/ <-- any file older than 1 day is erased
/scratch/local/weekly/ <-- any file older than 10 days is erased
Interconnect --> Infiniband 40 Gb/s
Network --> 1 Gb/s Ethernet
Parallel env. --> - SMP (on a single 16 cores node)
- MPI : OpenMPI
Batch scheduler --> SLURM https://slurm.schedmd.com
Submission queues --> - normal (the default one)
- max 512 monothreaded compute slots
- max 24 hours of Wall Clock Time per slot
- long
- max 96 monothreaded compute slots
- max 240 hours of Wall Clock Time per slot
A "First Steps with SLURM" tutorial. Please read this tutorial before to start to use the HPC cluster
Axiom
Axiom will be a future extension to the Wally cluster including specialized compute nodes.
It will be available in a few weeks.
The content of this page will evolve during the next months.
Stay tuned !
