Efficient GPU Offloading with OpenMP for a Hyperbolic Finite Volume Solver on Dynamically Adaptive Meshes

Mario Wille, Tobias Weinzierl, Gonzalo Brito Gadeschi, Michael Bader

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

2 Scopus citations

Abstract

We identify and show how to overcome an OpenMP bottleneck in the administration of GPU memory. It arises for a wave equation solver on dynamically adaptive block-structured Cartesian meshes, which keeps all CPU threads busy and allows all of them to offload sets of patches to the GPU. Our studies show that multithreaded, concurrent, non-deterministic access to the GPU leads to performance breakdowns, since the GPU memory bookkeeping as offered through OpenMP’s map clause, i.e., the allocation and freeing, becomes another runtime challenge besides expensive data transfer and actual computation. We, therefore, propose to retain the memory management responsibility on the host: A caching mechanism acquires memory on the accelerator for all CPU threads, keeps hold of this memory and hands it out to the offloading threads upon demand. We show that this user-managed, CPU-based memory administration helps us to overcome the GPU memory bookkeeping bottleneck and speeds up the time-to-solution of Finite Volume kernels by more than an order of magnitude.

Original languageEnglish
Title of host publicationHigh Performance Computing - 38th International Conference, ISC High Performance 2023, Proceedings
EditorsAbhinav Bhatele, Jeff Hammond, Marc Baboulin, Carola Kruse
PublisherSpringer Science and Business Media Deutschland GmbH
Pages65-85
Number of pages21
ISBN (Print)9783031320408
DOIs
StatePublished - 2023
Event38th International Conference on High Performance Computing, ISC High Performance 2023 - Hamburg, Germany
Duration: 21 May 202325 May 2023

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume13948 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference38th International Conference on High Performance Computing, ISC High Performance 2023
Country/TerritoryGermany
CityHamburg
Period21/05/2325/05/23

Keywords

  • Dynamically adaptive mesh refinement
  • GPU offloading
  • Multithreading
  • OpenMP

Fingerprint

Dive into the research topics of 'Efficient GPU Offloading with OpenMP for a Hyperbolic Finite Volume Solver on Dynamically Adaptive Meshes'. Together they form a unique fingerprint.

Cite this