The paper describes the results of some benchmarking tests aimed to verify and validate all the solutions implemented during the deployment of a HPC heterogeneous resource acquired by the data center of the University of Naples "Federico II" thanks to the funds of the IBiSCo (Infrastructure for Big data and Scientific COmputing) Italian National Project. The first set of benchmarks evaluates how the network interconnection technologies affect the inter- and intra-node communications of GP-GPU workloads. The second set evaluates the performance of the Lustre parallel file system to ensure an efficient environment for data-intensive applications. The tests, especially those that analyze the lower level of the middleware (micro-benchmarks), seem to confirm the ability of the resource to guarantee the expected performance.

Benchmarking a High Performance Computing Heterogeneous Cluster

Carracciuolo Luisa
Primo
;
2023

Abstract

The paper describes the results of some benchmarking tests aimed to verify and validate all the solutions implemented during the deployment of a HPC heterogeneous resource acquired by the data center of the University of Naples "Federico II" thanks to the funds of the IBiSCo (Infrastructure for Big data and Scientific COmputing) Italian National Project. The first set of benchmarks evaluates how the network interconnection technologies affect the inter- and intra-node communications of GP-GPU workloads. The second set evaluates the performance of the Lustre parallel file system to ensure an efficient environment for data-intensive applications. The tests, especially those that analyze the lower level of the middleware (micro-benchmarks), seem to confirm the ability of the resource to guarantee the expected performance.
2023
Istituto per i Polimeri, Compositi e Biomateriali - IPCB
978-3-031-30445-3
Benchmarking
High Performance Computing
Heterogeneous Computing
GP-GPU
InfiniBand
NVLink
Lustre
CUDA
RDMA
UCX
MPI
File in questo prodotto:
File Dimensione Formato  
prod_481000-doc_197671.pdf

solo utenti autorizzati

Descrizione: Benchmarking a High Performance Computing Heterogeneous Cluster
Tipologia: Versione Editoriale (PDF)
Licenza: NON PUBBLICO - Accesso privato/ristretto
Dimensione 1.24 MB
Formato Adobe PDF
1.24 MB Adobe PDF   Visualizza/Apri   Richiedi una copia
BenchmarkingPaper.pdf

accesso aperto

Descrizione: Benchmarking a High Performance Computing Heterogeneous Cluster
Tipologia: Documento in Post-print
Licenza: Creative commons
Dimensione 1.25 MB
Formato Adobe PDF
1.25 MB Adobe PDF Visualizza/Apri

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/20.500.14243/435479
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 2
  • ???jsp.display-item.citation.isi??? ND
social impact