Skip to main navigation Skip to search Skip to main content

Towards Efficient Remote OpenMP Offloading

  • Wenbin Lu
  • , Baodi Shan
  • , Eric Raut
  • , Jie Meng
  • , Mauricio Araya-Polo
  • , Johannes Doerfert
  • , Abid M. Malik
  • , Barbara Chapman
  • Stony Brook University
  • Total S.A.
  • Argonne National Laboratory

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

11 Scopus citations

Abstract

On modern heterogeneous HPC systems, the most popular way to realize distributed computation is the hybrid programming model of MPI+X (X being OpenMP/CUDA/etc.), as it has been proven to perform well with various scientific applications. However, application developers prefer to use a single coherent programming model over a hybrid model, as maintainability and portability decrease per additional model. Recent work [14] has shown that the OpenMP device offloading model could be used to program distributed accelerator-based HPC systems with minimal changes to the application. In this paper, we improve the performance of OpenMP remote offloading through various runtime optimizations, guided by a detailed overhead analysis. Evaluation of our work is conducted using an industrial-level seismic modeling code, Minimod, as well as two proxy-apps, XSBench and RSBench. Results show that, compared to the baseline version, our optimizations can reduce offloading latencies by up to 92%, and raise application parallel efficiency by at least 25.2% when running with 16 GPUs. We then point out why strong scaling is still difficult with OpenMP remote offloading, and propose further improvements to the runtime to increase scalability.

Original languageEnglish
Title of host publicationOpenMP in a Modern World
Subtitle of host publicationFrom Multi-device Support to Meta Programming - 18th International Workshop on OpenMP, IWOMP 2022, Proceedings
EditorsMichael Klemm, Bronis R. de Supinski, Jannis Klinkenberg, Brandon Neth
PublisherSpringer Science and Business Media Deutschland GmbH
Pages17-31
Number of pages15
ISBN (Print)9783031159213
DOIs
StatePublished - 2022
Event18th International Workshop on OpenMP, IWOMP 2022 - Chattanooga, United States
Duration: Sep 27 2022Sep 30 2022

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume13527 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference18th International Workshop on OpenMP, IWOMP 2022
Country/TerritoryUnited States
CityChattanooga
Period09/27/2209/30/22

Keywords

  • distributed computing
  • GPGPU
  • OpenMP

Fingerprint

Dive into the research topics of 'Towards Efficient Remote OpenMP Offloading'. Together they form a unique fingerprint.

Cite this