Using OSCAR to Win the Cluster Challenge

Author

Greidanus, Paul ; Klok, Gordon Willem

Author_Institution

Center of Excellence in Integrated Nanotools, Univ. of Alberta, Edmonton, AB

fYear

2008

fDate

9-11 June 2008

Firstpage

47

Lastpage

51

Abstract

This paper discusses using OSCAR (open source cluster administration resource) to compete and win, in the first ever cluster challenge competition at supercomputing 2007. The cluster challenge invited teams of undergraduate students to assemble a cluster and compete for 44 hours on the conference floor. The teams competed in two main areas; (i) achieving the best results on the high performance computing challenge benchmark and (ii) who could complete the most of a selection of data sets for several common scientific applications. In the this document we explain why Team Alberta chose OSCAR, how we feel it contributed to our success and our experiences using it.

Keywords

open systems; public domain software; workstation clusters; cluster challenge; high performance computing; open source cluster administration resource; Application software; Assembly; Chemistry; Distributed computing; Hardware; High performance computing; Oceans; Open source software; Packaging; Software packages; Cluster; HPC; OSCAR; System Administration;

fLanguage

English

Publisher

ieee

Conference_Titel

High Performance Computing Systems and Applications, 2008. HPCS 2008. 22nd International Symposium on

Conference_Location

Quebec City, Que.

ISSN

1550-5243

Print_ISBN

978-0-7695-3250-9

Type

conf

DOI

10.1109/HPCS.2008.22

Filename

4556072