A task parallel implementation of fast multipole methods

Handle URI:
http://hdl.handle.net/10754/564624
Title:
A task parallel implementation of fast multipole methods
Authors:
Taura, Kenjiro; Nakashima, Jun; Yokota, Rio ( 0000-0001-7573-7873 ) ; Maruyama, Naoya
Abstract:
This paper describes a task parallel implementation of ExaFMM, an open source implementation of fast multipole methods (FMM), using a lightweight task parallel library MassiveThreads. Although there have been many attempts on parallelizing FMM, experiences have almost exclusively been limited to formulation based on flat homogeneous parallel loops. FMM in fact contains operations that cannot be readily expressed in such conventional but restrictive models. We show that task parallelism, or parallel recursions in particular, allows us to parallelize all operations of FMM naturally and scalably. Moreover it allows us to parallelize a ''mutual interaction'' for force/potential evaluation, which is roughly twice as efficient as a more conventional, unidirectional force/potential evaluation. The net result is an open source FMM that is clearly among the fastest single node implementations, including those on GPUs; with a million particles on a 32 cores Sandy Bridge 2.20GHz node, it completes a single time step including tree construction and force/potential evaluation in 65 milliseconds. The study clearly showcases both programmability and performance benefits of flexible parallel constructs over more monolithic parallel loops. © 2012 IEEE.
KAUST Department:
Extreme Computing Research Center
Publisher:
Institute of Electrical and Electronics Engineers (IEEE)
Journal:
2012 SC Companion: High Performance Computing, Networking Storage and Analysis
Conference/Event name:
2012 SC Companion: High Performance Computing, Networking Storage and Analysis, SCC 2012
Issue Date:
Nov-2012
DOI:
10.1109/SC.Companion.2012.86
Type:
Conference Paper
ISBN:
9780769549569
Appears in Collections:
Conference Papers; Extreme Computing Research Center

Full metadata record

DC FieldValue Language
dc.contributor.authorTaura, Kenjiroen
dc.contributor.authorNakashima, Junen
dc.contributor.authorYokota, Rioen
dc.contributor.authorMaruyama, Naoyaen
dc.date.accessioned2015-08-04T07:05:23Zen
dc.date.available2015-08-04T07:05:23Zen
dc.date.issued2012-11en
dc.identifier.isbn9780769549569en
dc.identifier.doi10.1109/SC.Companion.2012.86en
dc.identifier.urihttp://hdl.handle.net/10754/564624en
dc.description.abstractThis paper describes a task parallel implementation of ExaFMM, an open source implementation of fast multipole methods (FMM), using a lightweight task parallel library MassiveThreads. Although there have been many attempts on parallelizing FMM, experiences have almost exclusively been limited to formulation based on flat homogeneous parallel loops. FMM in fact contains operations that cannot be readily expressed in such conventional but restrictive models. We show that task parallelism, or parallel recursions in particular, allows us to parallelize all operations of FMM naturally and scalably. Moreover it allows us to parallelize a ''mutual interaction'' for force/potential evaluation, which is roughly twice as efficient as a more conventional, unidirectional force/potential evaluation. The net result is an open source FMM that is clearly among the fastest single node implementations, including those on GPUs; with a million particles on a 32 cores Sandy Bridge 2.20GHz node, it completes a single time step including tree construction and force/potential evaluation in 65 milliseconds. The study clearly showcases both programmability and performance benefits of flexible parallel constructs over more monolithic parallel loops. © 2012 IEEE.en
dc.publisherInstitute of Electrical and Electronics Engineers (IEEE)en
dc.subjectdivide and conqueren
dc.subjectExaFMMen
dc.subjectfast multipole methodsen
dc.subjectFMMen
dc.subjectMassiveThreadsen
dc.subjecttask parallelismen
dc.titleA task parallel implementation of fast multipole methodsen
dc.typeConference Paperen
dc.contributor.departmentExtreme Computing Research Centeren
dc.identifier.journal2012 SC Companion: High Performance Computing, Networking Storage and Analysisen
dc.conference.date10 November 2012 through 16 November 2012en
dc.conference.name2012 SC Companion: High Performance Computing, Networking Storage and Analysis, SCC 2012en
dc.conference.locationSalt Lake City, UTen
dc.contributor.institutionGraduate School of Information Science and Technology, University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japanen
dc.contributor.institutionRIKEN Advanced Institute for Computational Science, 7-1-26 Minatojima-minami-machi, Chuo-ku Kobe, Hyogo 650-0047, Japanen
kaust.authorYokota, Rioen
All Items in KAUST are protected by copyright, with all rights reserved, unless otherwise indicated.