High-performance computing: Clusters, constellations, MPPs, and future directions

Jack Dongarra, Thomas Sterling, Horst Simon, Erich Strohmaier

Research output: Contribution to journalReview articlepeer-review

54 Scopus citations

Abstract

A view of the past, present and future of high-performance computing is given. The paper implies that every parallel system employing replicated resources is a cluster. The objective of the paper is to restore the strength and value of the term 'cluster' by degeneralizing its applicability to a restrict subset of parallel computers. This class is further considered in conjugation with its complemeting term constellation, beowolf class, and massively parallel processing systems (MPP), based on the classification used by the Top 500 list.

Original languageEnglish
Pages (from-to)51-59
Number of pages9
JournalComputing in Science and Engineering
Volume7
Issue number2
DOIs
StatePublished - Mar 2005

Fingerprint

Dive into the research topics of 'High-performance computing: Clusters, constellations, MPPs, and future directions'. Together they form a unique fingerprint.

Cite this