Overlapping communication with computation using OpenMP tasks on the GTS magnetic fusion code

Robert Preissl; Alice Koniges; Stephan Ethier; Weixing Wang; Nathan Wichmann

doi:10.3233/SPR-2010-0311

Overlapping communication with computation using OpenMP tasks on the GTS magnetic fusion code

Robert Preissl
, Alice Koniges
, Stephan Ethier
, Weixing Wang
, Nathan Wichmann

PPPL Computational Science

Princeton University

Research output: Contribution to journal › Article › peer-review

Abstract

Application codes in a variety of areas are being updated for performance on the latest architectures. In this paper we examine an application, which comes from magnetic fusion for performance acceleration with a particular emphasis on methods that are applicable for many/multicore and future architectural designs. We take an important magnetic fusion particle code that already includes several levels of parallelism including hybrid MPI combined with OpenMP. We study how to include new advanced hybrid models, which extend the applicability of OpenMP tasks and exploit multi-threaded MPI support to overlap communication and computation. Experiments carried out on Cray XT4 and XT5 machines resulting in a speed-up of up to 35% of the investigated GTS particle shifter kernel show the benefits and applicability of this approach.

Original language	American English
Pages (from-to)	139-151
Number of pages	13
Journal	Scientific Programming
Volume	18
Issue number	3-4
DOIs	https://doi.org/10.3233/SPR-2010-0311
State	Published - 2010

ASJC Scopus subject areas

Software
Computer Science Applications

Keywords

GTS
OpenMP task
communication overlap
hybrid MPI and OpenMP computing

Access to Document

10.3233/SPR-2010-0311

Cite this

@article{9814637d0749472e877d754a7d9313fc,

title = "Overlapping communication with computation using OpenMP tasks on the GTS magnetic fusion code",

abstract = "Application codes in a variety of areas are being updated for performance on the latest architectures. In this paper we examine an application, which comes from magnetic fusion for performance acceleration with a particular emphasis on methods that are applicable for many/multicore and future architectural designs. We take an important magnetic fusion particle code that already includes several levels of parallelism including hybrid MPI combined with OpenMP. We study how to include new advanced hybrid models, which extend the applicability of OpenMP tasks and exploit multi-threaded MPI support to overlap communication and computation. Experiments carried out on Cray XT4 and XT5 machines resulting in a speed-up of up to 35\% of the investigated GTS particle shifter kernel show the benefits and applicability of this approach.",

keywords = "GTS, OpenMP task, communication overlap, hybrid MPI and OpenMP computing",

author = "Robert Preissl and Alice Koniges and Stephan Ethier and Weixing Wang and Nathan Wichmann",

year = "2010",

doi = "10.3233/SPR-2010-0311",

language = "American English",

volume = "18",

pages = "139--151",

journal = "Scientific Programming",

issn = "1058-9244",

publisher = "John Wiley and Sons Inc",

number = "3-4",

}

TY - JOUR

T1 - Overlapping communication with computation using OpenMP tasks on the GTS magnetic fusion code

AU - Preissl, Robert

AU - Koniges, Alice

AU - Ethier, Stephan

AU - Wang, Weixing

AU - Wichmann, Nathan

PY - 2010

Y1 - 2010

N2 - Application codes in a variety of areas are being updated for performance on the latest architectures. In this paper we examine an application, which comes from magnetic fusion for performance acceleration with a particular emphasis on methods that are applicable for many/multicore and future architectural designs. We take an important magnetic fusion particle code that already includes several levels of parallelism including hybrid MPI combined with OpenMP. We study how to include new advanced hybrid models, which extend the applicability of OpenMP tasks and exploit multi-threaded MPI support to overlap communication and computation. Experiments carried out on Cray XT4 and XT5 machines resulting in a speed-up of up to 35% of the investigated GTS particle shifter kernel show the benefits and applicability of this approach.

AB - Application codes in a variety of areas are being updated for performance on the latest architectures. In this paper we examine an application, which comes from magnetic fusion for performance acceleration with a particular emphasis on methods that are applicable for many/multicore and future architectural designs. We take an important magnetic fusion particle code that already includes several levels of parallelism including hybrid MPI combined with OpenMP. We study how to include new advanced hybrid models, which extend the applicability of OpenMP tasks and exploit multi-threaded MPI support to overlap communication and computation. Experiments carried out on Cray XT4 and XT5 machines resulting in a speed-up of up to 35% of the investigated GTS particle shifter kernel show the benefits and applicability of this approach.

KW - GTS

KW - OpenMP task

KW - communication overlap

KW - hybrid MPI and OpenMP computing

UR - https://www.scopus.com/pages/publications/79551697181

UR - https://www.scopus.com/pages/publications/79551697181#tab=citedBy

U2 - 10.3233/SPR-2010-0311

DO - 10.3233/SPR-2010-0311

M3 - Article

SN - 1058-9244

VL - 18

SP - 139

EP - 151

JO - Scientific Programming

JF - Scientific Programming

IS - 3-4

ER -

Overlapping communication with computation using OpenMP tasks on the GTS magnetic fusion code

Abstract

ASJC Scopus subject areas

Keywords

Access to Document

Other files and links

Fingerprint

Cite this