High Performance Computing clusters at CERN
Since end 2017, two new HPC cluster have been setup here at CERN by the IT department for users of MPI applications.
Access
ABP members who need access to the cluster can request it by writing to abp.computing[AT]cern.ch in order to be added to the relevant e-group (service-hpc-be).
Please note that these are dedicated facilities for parallel computing applications, i.e. applications running with MPI across more than 32 cores for each simulation (hence requiring multiple nodes). Simulations requiring less than 32 cores should be submitted to the CERN batch service.
Documentation and other resources
Documentation on the cluster and its usage can be found at:
https://batchdocs.web.cern.ch/linuxhpc/index.html
Support and further info can be found in the Service Portal:
https://cern.service-now.com/service-portal?id=service_element&name=High-Performance-Computing
A dedicated workshop in two sessions took place in 2020. The corresponding indico pages can be found at:
https://indico.cern.ch/event/867459/
https://indico.cern.ch/event/916903/
Regular meetings are held by IT to share information on the facility development:
https://indico.cern.ch/category/7950/
Some further information can be found in dedicated presentations in past ABP-CWG meetings: