On the Impact of OpenMP Task Granularity

Thierry Gautier 1 Christian Pérez 1 Jérôme Richard 1
1 AVALON - Algorithms and Software Architectures for Distributed and HPC Platforms
Inria Grenoble - Rhône-Alpes, LIP - Laboratoire de l'Informatique du Parallélisme
Abstract : Tasks are a good support for composition. During the development of a high-level component model for HPC, we have experimented to manage parallelism from components using OpenMP tasks. Since version 4-0, the standard proposes a model with dependent tasks that seems very attractive because it enables the description of dependencies between tasks generated by different components without breaking maintainability constraints such as separation of concerns. The paper presents our feedback on using OpenMP in our context. We discover that our main issues are a too coarse task granularity for our expected performance on classical OpenMP runtimes, and a harmful task throttling heuristic counter-productive for our applications. We present a completion time breakdown of task management in the Intel OpenMP runtime and propose extensions evaluated on a testbed application coming from the Gysela application in plasma physics.
Complete list of metadatas

Cited literature [7 references]  Display  Hide  Download

https://hal.inria.fr/hal-01901806
Contributor : Christian Perez <>
Submitted on : Wednesday, November 14, 2018 - 4:07:33 PM
Last modification on : Thursday, November 15, 2018 - 9:26:59 AM
Long-term archiving on : Friday, February 15, 2019 - 3:58:50 PM

File

iwomp_hal.pdf
Files produced by the author(s)

Identifiers

Collections

Citation

Thierry Gautier, Christian Pérez, Jérôme Richard. On the Impact of OpenMP Task Granularity. IWOMP 2018 - 14th International Workshop on OpenMP for Evolving Architectures, Sep 2018, Barcelone, Spain. pp.205-221, ⟨10.1007/978-3-319-98521-3_14⟩. ⟨hal-01901806⟩

Share

Metrics

Record views

66

Files downloads

232