Welcome to the HORIZON cluster

  • A quick start for dummies: QuickStart
  • A short description on getting started with the cluster: HOWTO
  • A short description of Job submission and some script examples.
  • Available softwares
  • IMPORTANT: note that this cluster compute nodes have large amounts of CPU cores (up to 64 cores) and of shared memory (up to 1TB RAM). The prime purpose of Horizon cluster is therefore for parallel jobs which make use of this large amount of shared memory, typically to post-process snapshots of simulations which ran on super computers (see Resource use policy).
  • IMPORTANT: note that on this cluster, the user files stored on /scratch* temporary disk space will be automatically erased after 4 (four) weeks,so please migrate your outputs to /data*/<user>/ (this storage space is NOT backed up). In contrast, your $HOME directory (limited by quota to 20 GB) is backed up.