I/O in parallel applications: the weakest link

Rajeev Thakur, Ewing Lusk, William Gropp

Research output: Contribution to journalArticlepeer-review

Abstract

Parallel computers are increasingly being used to run large-scale applications that also have huge input/output (I/O) requirements. However, many applications obtain poor I/O performance on modem parallel machines. This two-part special issue of the Internationa/Journal of High Performance Computing Applications contains papers that describe the I/O requirements and the techniques used to perform I/O in real parallel applications. The authors first explain how the I/O application program interface (API) plays a critical role in enabling such applications to achieve high I/O performance. They describe how the commonly used UNIX I/O interface is inappropriate for parallel I/O and how an explicitly parallel API with support for collective I/O can help the underlying I/O hardware and software perform I/O efficiently. They then describe MPI-IO, a recently defined, standard, portable API specifically designed for high performance parallel I/O. They conclude with an overview of the papers in Part 1 and Part 2 of this special issue.

Original languageEnglish (US)
Pages (from-to)389-395
Number of pages7
JournalInternational Journal of High Performance Computing Applications
Volume12
Issue number4
DOIs
StatePublished - 1998
Externally publishedYes

ASJC Scopus subject areas

  • Software
  • Theoretical Computer Science
  • Hardware and Architecture

Fingerprint

Dive into the research topics of 'I/O in parallel applications: the weakest link'. Together they form a unique fingerprint.

Cite this