Portable compilers for OpenMP

Seung Jai Min, Seon Wook Kim, Michael Voss, Sang Ik Lee, Rudolf Eigenmann

Research output: Chapter in Book/Report/Conference proceedingConference contribution

12 Citations (Scopus)


The recent parallel language standard for shared memory multiprocessor (SMP) machines, OpenMP, promises a simple and portable interface for programmers who wish to exploit parallelism explicitly. In this paper, we present our effort to develop portable compilers for the OpenMP parallel directive language. Our compiler consists of two parts. Part one is an OpenMP parallelizer, which transforms sequential languages into OpenMP. Part two transforms programs written in OpenMP into thread-based form and links with our runtime library. Both compilers are built on the Polaris compiler infrastructure. We present performance measurements showing that our compiler yields results comparable to those of commercial OpenMP compilers. Our infrastructure is freely available with the intent to enable research projects on OpenMP-related language development and compiler techniques.

Original languageEnglish
Title of host publicationOpenMP Shared Memory Parallel Programming - International Workshop on OpenMP Applications and Tools,WOMPAT 2001, Proceedings
EditorsRudolf Eigenmann, Michael J. Voss
PublisherSpringer Verlag
Number of pages9
ISBN (Print)9783540445876
Publication statusPublished - 2001
Externally publishedYes
EventInternational Workshop on OpenMP Applications and Tools, WOMPAT 2001 - West Lafayette, United States
Duration: 2001 Jul 302001 Jul 31

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349


OtherInternational Workshop on OpenMP Applications and Tools, WOMPAT 2001
Country/TerritoryUnited States
CityWest Lafayette

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)


Dive into the research topics of 'Portable compilers for OpenMP'. Together they form a unique fingerprint.

Cite this