This Section describes how to access the SLURM Linux HPC resource.
Please bear in mind that Linux HPC is reserved for workloads that need multiple nodes per job. If you only need single-node parallelism, please use HTCondor.
Accessing Linux HPC
Access to Linux HPC is only supported via ssh through hpc-batch.cern.ch. You may connect from CERN directly, or from LXPLUS if accessing from outside of CERN's networks.
The HPC batch cluster using SLURM is dedicated for users of MPI applications in the Accelerator and Technology sector at CERN or the Theory Lattice QCD team.
Access to the cluster is granted via e-groups. In general users should only be granted access by their local HPC experts and supervisors.
Regarding approval of user requests, please refer to: KB0004975
Other users are kindly asked to use the regular batch computing service on HTCondor. It shall be noted that there is far more capacity on HTCondor than on the SLURM MPI cluster.