Skip to main navigation Skip to search Skip to main content

Simplifying heterogeneous embedded systems programming based on OpenMP

  • Lei Huang
  • , Barbara Chapman
  • University of Houston

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Modern computer architectures have increased their paces to move into heterogeneity by integrating specialized processing elements such as GPUs, FPGAs, DSPs into a single system together with general processors. The integration allows developers to maximize their applications performance by mapping the different modules to different processing elements based on the characteristics. However, the current programming models are not comprehensive, portable or expressive enough to support developers' needs. In this paper, we are exploring the current programming approach on heterogeneous embedded systems, as well as studying the popular shared memory programming model OpenMP for how to adapt it to heterogeneous systems. We present our initial design of OpenMP extensions to support work distribution and data communication between different processing elements.

Original languageEnglish
Title of host publicationParallel Computing
Subtitle of host publicationFrom Multicores and GPU's to Petascale
PublisherIOS Press BV
Pages728-735
Number of pages8
ISBN (Print)9781607505297
DOIs
StatePublished - 2010

Publication series

NameAdvances in Parallel Computing
Volume19
ISSN (Print)0927-5452

Keywords

  • Heterogeneous Systems
  • High Performance Computing
  • OpenMP

Fingerprint

Dive into the research topics of 'Simplifying heterogeneous embedded systems programming based on OpenMP'. Together they form a unique fingerprint.

Cite this