Optimizing Loop-Level Parallelism in Cray XMT Applications

S-2487-14 - Dec 2009

Provides guidance and techniques for improving parallelization for Cray XMT programs.

Links

Available formats

Adobe Acrobat Reader S-2487-14.pdf
Valid HTML 4.01 html-S-2487-14

Table of contents

Introduction
Overview of the Cray XMT System
    2.1  The Cray XMT Programming Environment and Tools
    2.2  Language Extensions
    2.3  Loop-level Parallelism and Canal
Identifying Parallelism
    3.1  Conditions for Safe Parallelization
Compiler Transformations
    4.1  Scalar Expansion
    4.2  Reductions
    4.3  Linear Recurrences
    4.4  Nested Parallelism and Loop Collapse
Pragmas
Implementing Parallelism
    6.1  Parallel Regions
    6.2  Styles of Parallelism
    6.3  Loop Scheduling
An Example Loop
Summary
Acknowledgments
10  About the Authors
11  Selected Bibliography

Software Releases this book supports

Product Version Sub Product Release Date

Other versions of this book

Publication Number Release Date Supported Software Releases
S-2487-10 Oct 2009 unsupported