Cargando…

Performance of space charge simulations using High Performance Computing (HPC) cluster

In 2016 a collaboration agreement between CERN and Istituto Nazionale di Fisica Nucleare (INFN) through its Centro Nazionale Analisi Fotogrammi (CNAF, Bologna) was signed [1], which foresaw the purchase and installation of a cluster of 20 nodes with 32 cores each, connected with InfiniBand, at CNAF...

Descripción completa

Detalles Bibliográficos
Autores principales: Bartosik, Hannes, Rumolo, Giovanni
Lenguaje:eng
Publicado: 2017
Materias:
Acceso en línea:http://cds.cern.ch/record/2276464
_version_ 1780955201783463936
author Bartosik, Hannes
Rumolo, Giovanni
author_facet Bartosik, Hannes
Rumolo, Giovanni
author_sort Bartosik, Hannes
collection CERN
description In 2016 a collaboration agreement between CERN and Istituto Nazionale di Fisica Nucleare (INFN) through its Centro Nazionale Analisi Fotogrammi (CNAF, Bologna) was signed [1], which foresaw the purchase and installation of a cluster of 20 nodes with 32 cores each, connected with InfiniBand, at CNAF for the use of CERN members to develop parallelized codes as well as conduct massive simulation campaigns with the already available parallelized tools. As outlined in [1], after the installation and the set up of the first 12 nodes, the green light to proceed with the procurement and installation of the next 8 nodes can be given only after successfully passing an acceptance test based on two specific benchmark runs. This condition is necessary to consider the first batch of the cluster operational and complying with the desired performance specifications. In this brief note, we report the results of the above mentioned acceptance test.
id cern-2276464
institution Organización Europea para la Investigación Nuclear
language eng
publishDate 2017
record_format invenio
spelling cern-22764642019-09-30T06:29:59Zhttp://cds.cern.ch/record/2276464engBartosik, HannesRumolo, GiovanniPerformance of space charge simulations using High Performance Computing (HPC) clusterComputing and ComputersAccelerators and Storage RingsIn 2016 a collaboration agreement between CERN and Istituto Nazionale di Fisica Nucleare (INFN) through its Centro Nazionale Analisi Fotogrammi (CNAF, Bologna) was signed [1], which foresaw the purchase and installation of a cluster of 20 nodes with 32 cores each, connected with InfiniBand, at CNAF for the use of CERN members to develop parallelized codes as well as conduct massive simulation campaigns with the already available parallelized tools. As outlined in [1], after the installation and the set up of the first 12 nodes, the green light to proceed with the procurement and installation of the next 8 nodes can be given only after successfully passing an acceptance test based on two specific benchmark runs. This condition is necessary to consider the first batch of the cluster operational and complying with the desired performance specifications. In this brief note, we report the results of the above mentioned acceptance test.CERN-ACC-NOTE-2017-0048oai:cds.cern.ch:22764642017-08-01
spellingShingle Computing and Computers
Accelerators and Storage Rings
Bartosik, Hannes
Rumolo, Giovanni
Performance of space charge simulations using High Performance Computing (HPC) cluster
title Performance of space charge simulations using High Performance Computing (HPC) cluster
title_full Performance of space charge simulations using High Performance Computing (HPC) cluster
title_fullStr Performance of space charge simulations using High Performance Computing (HPC) cluster
title_full_unstemmed Performance of space charge simulations using High Performance Computing (HPC) cluster
title_short Performance of space charge simulations using High Performance Computing (HPC) cluster
title_sort performance of space charge simulations using high performance computing (hpc) cluster
topic Computing and Computers
Accelerators and Storage Rings
url http://cds.cern.ch/record/2276464
work_keys_str_mv AT bartosikhannes performanceofspacechargesimulationsusinghighperformancecomputinghpccluster
AT rumologiovanni performanceofspacechargesimulationsusinghighperformancecomputinghpccluster