Interpretive Performance Prediction for Parallel Application Development

Manish Parashar, Salim Hariri

Research output: Contribution to journalArticlepeer-review

11 Scopus citations

Abstract

Application software development for high-performance parallel computing (HPC) is a non trivial process; its complexity can be primarily attributed to the increased degrees of freedom that have to be resolved and tuned in such an environment. Performance prediction tools enable a developer to evaluate available design alternatives and can assist in HPC application software development. In this paper we first present a novel "interpretive" approach for accurate and cost-effective performance prediction. The approach has been used to develop an interpretive HPF/Fortran 90D application performance prediction framework. The accuracy and usability of the performance prediction framework are experimentally validated. We then outline the stages typically encountered during application software development for HPC and highlight the significance and requirements of a performance prediction tool at relevant stages. Numerical results using benchmarking kernels and application codes are presented to demonstrate the application of the interpretive performance prediction framework at different stages of the HPC application software development process.

Original languageEnglish (US)
Pages (from-to)17-47
Number of pages31
JournalJournal of Parallel and Distributed Computing
Volume60
Issue number1
DOIs
StatePublished - Jan 2000

Keywords

  • HPC application software development
  • HPF/Fortran 90D application development
  • Interpretive performance prediction

ASJC Scopus subject areas

  • Software
  • Theoretical Computer Science
  • Hardware and Architecture
  • Computer Networks and Communications
  • Artificial Intelligence

Fingerprint Dive into the research topics of 'Interpretive Performance Prediction for Parallel Application Development'. Together they form a unique fingerprint.

Cite this