Compiling Scientific Programs for Scalable Parallel Systems

Abstract

This report details research into compiler technology to support machine-independent data parallel programming for scientific application. The investigation focused on focused on design and development of dHPF, an advanced prototype compiler for High Performance FORTRAN (HPF). The research performed in this project included new techniques for recognizing implicit parallelism in sequential programs, a powerful and precise set-based framework for analysis and transformation of data-parallel programs, support for sophisticated data and computation partitioning, effective support for generation and optimization of parallel code for message-passing and distribution shared-memory systems; techniques for supporting sophisticated data-parallel applications with irregular structure, and an integrated compiled effort that included a multi-level memory hierarchy simulator and techniques for improving locality in irregular computations.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Feb 01, 2001
Accession Number
ADA389314

Entities

People

  • Guohua Jin
  • John Mellor-crummey
  • Ken Kennedy
  • Robert J. Fowler
  • Vikram Adve

Organizations

  • Rice University

Tags

Communities of Interest

  • Energy and Power Technologies
  • Human Systems
  • Weapons Technologies

DTIC Thesaurus Topics

  • Artificial Intelligence
  • Computational Fluid Dynamics
  • Computational Science
  • Computer Programming
  • Computer Programs
  • Computer Science
  • Computers
  • Fluid Dynamics
  • High Performance Computing
  • Military Research
  • Molecular Dynamics
  • Parallel Computing
  • Parallel Processing
  • Simulations
  • Simulators
  • Spatial Partitioning
  • Three Dimensional

Fields of Study

  • Computer science

Readers

  • Computational Linguistics
  • Parallel and Distributed Computing.
  • Technical Research and Report Writing.