A distributed computing approach to improve the performance of the Parallel Ocean Program (v2.1)

B.J.C. van Werkhoven; M. Kliphuis; H.A. Dijkstra; S.A. Brunnabend; M.A.J. van Meersbergen; F.J. Seinstra; H.E. Bal

doi:10.5194/gmd-7-267-2014

A distributed computing approach to improve the performance of the Parallel Ocean Program (v2.1)

B.J.C. van Werkhoven, M. Kliphuis, H.A. Dijkstra, S.A. Brunnabend, M.A.J. van Meersbergen, F.J. Seinstra, H.E. Bal

Research output: Contribution to Journal › Article › Academic › peer-review

Abstract

The Parallel Ocean Program (POP) is used in many strongly eddying ocean circulation simulations. Ideally it would be desirable to be able to do thousand-yearlong simulations, but the current performance of POP prohibits these types of simulations. In this work, using a new distributed computing approach, two methods to improve the performance of POP are presented. The first is a blockpartitioning scheme for the optimization of the load balancing of POP such that it can be run efficiently in a multiplatform setting. The second is the implementation of part of the POP model code on graphics processing units (GPUs). We show that the combination of both innovations also leads to a substantial performance increase when running POP simultaneously over multiple computational platforms.

Original language	English
Pages (from-to)	267-281
Journal	Geoscientific Model Development Discussions
Volume	7
DOIs	https://doi.org/10.5194/gmd-7-267-2014
Publication status	Published - 2014

Access to Document

10.5194/gmd-7-267-2014

Fingerprint Dive into the research topics of 'A distributed computing approach to improve the performance of the Parallel Ocean Program (v2.1)'. Together they form a unique fingerprint.

Cite this

@article{2bddb0440e4448aea1a9a0102bcd55f8,

title = "A distributed computing approach to improve the performance of the Parallel Ocean Program (v2.1)",

abstract = "The Parallel Ocean Program (POP) is used in many strongly eddying ocean circulation simulations. Ideally it would be desirable to be able to do thousand-yearlong simulations, but the current performance of POP prohibits these types of simulations. In this work, using a new distributed computing approach, two methods to improve the performance of POP are presented. The first is a blockpartitioning scheme for the optimization of the load balancing of POP such that it can be run efficiently in a multiplatform setting. The second is the implementation of part of the POP model code on graphics processing units (GPUs). We show that the combination of both innovations also leads to a substantial performance increase when running POP simultaneously over multiple computational platforms.",

author = "{van Werkhoven}, B.J.C. and M. Kliphuis and H.A. Dijkstra and S.A. Brunnabend and {van Meersbergen}, M.A.J. and F.J. Seinstra and H.E. Bal",

year = "2014",

doi = "10.5194/gmd-7-267-2014",

language = "English",

volume = "7",

pages = "267--281",

journal = "Geoscientific Model Development Discussions",

issn = "1991-9611",

publisher = "Copernicus Publications",

}

A distributed computing approach to improve the performance of the Parallel Ocean Program (v2.1). / van Werkhoven, B.J.C.; Kliphuis, M.; Dijkstra, H.A.; Brunnabend, S.A.; van Meersbergen, M.A.J.; Seinstra, F.J.; Bal, H.E.

In: Geoscientific Model Development Discussions, Vol. 7, 2014, p. 267-281.

Research output: Contribution to Journal › Article › Academic › peer-review

TY - JOUR

T1 - A distributed computing approach to improve the performance of the Parallel Ocean Program (v2.1)

AU - van Werkhoven, B.J.C.

AU - Kliphuis, M.

AU - Dijkstra, H.A.

AU - Brunnabend, S.A.

AU - van Meersbergen, M.A.J.

AU - Seinstra, F.J.

AU - Bal, H.E.

PY - 2014

Y1 - 2014

N2 - The Parallel Ocean Program (POP) is used in many strongly eddying ocean circulation simulations. Ideally it would be desirable to be able to do thousand-yearlong simulations, but the current performance of POP prohibits these types of simulations. In this work, using a new distributed computing approach, two methods to improve the performance of POP are presented. The first is a blockpartitioning scheme for the optimization of the load balancing of POP such that it can be run efficiently in a multiplatform setting. The second is the implementation of part of the POP model code on graphics processing units (GPUs). We show that the combination of both innovations also leads to a substantial performance increase when running POP simultaneously over multiple computational platforms.

AB - The Parallel Ocean Program (POP) is used in many strongly eddying ocean circulation simulations. Ideally it would be desirable to be able to do thousand-yearlong simulations, but the current performance of POP prohibits these types of simulations. In this work, using a new distributed computing approach, two methods to improve the performance of POP are presented. The first is a blockpartitioning scheme for the optimization of the load balancing of POP such that it can be run efficiently in a multiplatform setting. The second is the implementation of part of the POP model code on graphics processing units (GPUs). We show that the combination of both innovations also leads to a substantial performance increase when running POP simultaneously over multiple computational platforms.

U2 - 10.5194/gmd-7-267-2014

DO - 10.5194/gmd-7-267-2014

M3 - Article

VL - 7

SP - 267

EP - 281

JO - Geoscientific Model Development Discussions

JF - Geoscientific Model Development Discussions

SN - 1991-9611

ER -