Loading…

Algorithm to minimize MPI communications in the parallelized fast multipole method combined with molecular dynamics calculations

In the era of exascale supercomputers, large‐scale, and long‐time molecular dynamics (MD) calculations are expected to make breakthroughs in various fields of science and technology. Here, we propose a new algorithm to improve the parallelization performance of message passing interface (MPI)‐commun...

Full description

Saved in:
Bibliographic Details
Published in:Journal of computational chemistry 2021-06, Vol.42 (15), p.1073-1087
Main Authors: Andoh, Yoshimichi, Ichikawa, Shin‐ichi, Sakashita, Tatsuya, Yoshii, Noriyuki, Okazaki, Susumu
Format: Article
Language:English
Subjects:
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
cited_by cdi_FETCH-LOGICAL-c3974-40fb8aef3a17b457e7538887f7e5b8b8e706704020e851a454d8ba2ee401386f3
cites cdi_FETCH-LOGICAL-c3974-40fb8aef3a17b457e7538887f7e5b8b8e706704020e851a454d8ba2ee401386f3
container_end_page 1087
container_issue 15
container_start_page 1073
container_title Journal of computational chemistry
container_volume 42
creator Andoh, Yoshimichi
Ichikawa, Shin‐ichi
Sakashita, Tatsuya
Yoshii, Noriyuki
Okazaki, Susumu
description In the era of exascale supercomputers, large‐scale, and long‐time molecular dynamics (MD) calculations are expected to make breakthroughs in various fields of science and technology. Here, we propose a new algorithm to improve the parallelization performance of message passing interface (MPI)‐communication in the MPI‐parallelized fast multipole method (FMM) combined with MD calculations under three‐dimensional periodic boundary conditions. Our approach enables a drastic reduction in the amount of communication data, including the atomic coordinates and multipole coefficients, both of which are required to calculate the electrostatic interaction by using the FMM. In communications of multipole coefficients, the reduction rate of communication data in the new algorithm relative to the amount of data in the conventional one increases as both the number of FMM levels and the number of MPI processes increase. The aforementioned rate increase could exceed 50% as the number of MPI processes becomes larger for very large systems. The proposed algorithm, named the minimum‐transferred data (MTD) method, should enable large‐scale and long‐time MD calculations to be calculated efficiently, under the condition of massive MPI‐parallelization on exascale supercomputers. A new algorithm is proposed for message passing interface (MPI)‐communication suitable for the fast multipole method (FMM) combined with molecular dynamics (MD) calculations under three‐dimensional periodic boundary conditions. The method, named the minimum transferred‐data (MTD) method, can reduce the elapsed time of MPI communications in the MPI‐parallelized FMM drastically, enabling efficient large‐scale and long‐time MD calculations on exascale supercomputers.
doi_str_mv 10.1002/jcc.26524
format article
fullrecord <record><control><sourceid>proquest_cross</sourceid><recordid>TN_cdi_proquest_miscellaneous_2506519767</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2518638722</sourcerecordid><originalsourceid>FETCH-LOGICAL-c3974-40fb8aef3a17b457e7538887f7e5b8b8e706704020e851a454d8ba2ee401386f3</originalsourceid><addsrcrecordid>eNp1kU1r3DAQhkVoSLZpD_0DRdBLe3CiT0s-hqX5IiE5pNCbkeVxV4tkbSWbsD31p0ebTXso9DQw88zDMC9CHyg5pYSws7W1p6yWTBygBSVNXTVafX-DFoQ2rNK1pMfobc5rQgiXtThCx5wrXRbpAv0-9z9ictMq4Cni4EYX3C_Adw_X2MYQ5tFZM7k4ZuxGPK0Ab0wy3oMvVI8HkyccZj-5TfSAA0yr2O8WOzeW8VPx4lAmdvYm4X47muBsxtb4XefF-w4dDsZneP9aT9C3i6-Py6vq9v7yenl-W1neKFEJMnTawMANVZ2QCpTkWms1KJCd7jQoUisiCCOgJTVCil53hgEIQrmuB36CPu-9mxR_zpCnNrhswXszQpxzyyQpj2pUrQr66R90Hec0lusKRXXNtWKsUF_2lE0x5wRDu0kumLRtKWl3sbQllvYllsJ-fDXOXYD-L_knhwKc7YEn52H7f1N7s1zulc9H8pgA</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2518638722</pqid></control><display><type>article</type><title>Algorithm to minimize MPI communications in the parallelized fast multipole method combined with molecular dynamics calculations</title><source>Wiley-Blackwell Read &amp; Publish Collection</source><creator>Andoh, Yoshimichi ; Ichikawa, Shin‐ichi ; Sakashita, Tatsuya ; Yoshii, Noriyuki ; Okazaki, Susumu</creator><creatorcontrib>Andoh, Yoshimichi ; Ichikawa, Shin‐ichi ; Sakashita, Tatsuya ; Yoshii, Noriyuki ; Okazaki, Susumu</creatorcontrib><description>In the era of exascale supercomputers, large‐scale, and long‐time molecular dynamics (MD) calculations are expected to make breakthroughs in various fields of science and technology. Here, we propose a new algorithm to improve the parallelization performance of message passing interface (MPI)‐communication in the MPI‐parallelized fast multipole method (FMM) combined with MD calculations under three‐dimensional periodic boundary conditions. Our approach enables a drastic reduction in the amount of communication data, including the atomic coordinates and multipole coefficients, both of which are required to calculate the electrostatic interaction by using the FMM. In communications of multipole coefficients, the reduction rate of communication data in the new algorithm relative to the amount of data in the conventional one increases as both the number of FMM levels and the number of MPI processes increase. The aforementioned rate increase could exceed 50% as the number of MPI processes becomes larger for very large systems. The proposed algorithm, named the minimum‐transferred data (MTD) method, should enable large‐scale and long‐time MD calculations to be calculated efficiently, under the condition of massive MPI‐parallelization on exascale supercomputers. A new algorithm is proposed for message passing interface (MPI)‐communication suitable for the fast multipole method (FMM) combined with molecular dynamics (MD) calculations under three‐dimensional periodic boundary conditions. The method, named the minimum transferred‐data (MTD) method, can reduce the elapsed time of MPI communications in the MPI‐parallelized FMM drastically, enabling efficient large‐scale and long‐time MD calculations on exascale supercomputers.</description><identifier>ISSN: 0192-8651</identifier><identifier>EISSN: 1096-987X</identifier><identifier>DOI: 10.1002/jcc.26524</identifier><identifier>PMID: 33780021</identifier><language>eng</language><publisher>Hoboken, USA: John Wiley &amp; Sons, Inc</publisher><subject>Algorithms ; Boundary conditions ; Communication ; efficient MPI communication ; exascale supercomputers ; fast multipole method ; Mathematical analysis ; Message passing ; Molecular dynamics ; molecular dynamics calculation ; MPI parallelization ; Multipoles ; Parallel processing ; Reduction ; Supercomputers</subject><ispartof>Journal of computational chemistry, 2021-06, Vol.42 (15), p.1073-1087</ispartof><rights>2021 Wiley Periodicals LLC.</rights><lds50>peer_reviewed</lds50><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed><citedby>FETCH-LOGICAL-c3974-40fb8aef3a17b457e7538887f7e5b8b8e706704020e851a454d8ba2ee401386f3</citedby><cites>FETCH-LOGICAL-c3974-40fb8aef3a17b457e7538887f7e5b8b8e706704020e851a454d8ba2ee401386f3</cites><orcidid>0000-0002-9843-467X ; 0000-0001-5027-4120 ; 0000-0002-3361-2113</orcidid></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>314,780,784,27924,27925</link.rule.ids><backlink>$$Uhttps://www.ncbi.nlm.nih.gov/pubmed/33780021$$D View this record in MEDLINE/PubMed$$Hfree_for_read</backlink></links><search><creatorcontrib>Andoh, Yoshimichi</creatorcontrib><creatorcontrib>Ichikawa, Shin‐ichi</creatorcontrib><creatorcontrib>Sakashita, Tatsuya</creatorcontrib><creatorcontrib>Yoshii, Noriyuki</creatorcontrib><creatorcontrib>Okazaki, Susumu</creatorcontrib><title>Algorithm to minimize MPI communications in the parallelized fast multipole method combined with molecular dynamics calculations</title><title>Journal of computational chemistry</title><addtitle>J Comput Chem</addtitle><description>In the era of exascale supercomputers, large‐scale, and long‐time molecular dynamics (MD) calculations are expected to make breakthroughs in various fields of science and technology. Here, we propose a new algorithm to improve the parallelization performance of message passing interface (MPI)‐communication in the MPI‐parallelized fast multipole method (FMM) combined with MD calculations under three‐dimensional periodic boundary conditions. Our approach enables a drastic reduction in the amount of communication data, including the atomic coordinates and multipole coefficients, both of which are required to calculate the electrostatic interaction by using the FMM. In communications of multipole coefficients, the reduction rate of communication data in the new algorithm relative to the amount of data in the conventional one increases as both the number of FMM levels and the number of MPI processes increase. The aforementioned rate increase could exceed 50% as the number of MPI processes becomes larger for very large systems. The proposed algorithm, named the minimum‐transferred data (MTD) method, should enable large‐scale and long‐time MD calculations to be calculated efficiently, under the condition of massive MPI‐parallelization on exascale supercomputers. A new algorithm is proposed for message passing interface (MPI)‐communication suitable for the fast multipole method (FMM) combined with molecular dynamics (MD) calculations under three‐dimensional periodic boundary conditions. The method, named the minimum transferred‐data (MTD) method, can reduce the elapsed time of MPI communications in the MPI‐parallelized FMM drastically, enabling efficient large‐scale and long‐time MD calculations on exascale supercomputers.</description><subject>Algorithms</subject><subject>Boundary conditions</subject><subject>Communication</subject><subject>efficient MPI communication</subject><subject>exascale supercomputers</subject><subject>fast multipole method</subject><subject>Mathematical analysis</subject><subject>Message passing</subject><subject>Molecular dynamics</subject><subject>molecular dynamics calculation</subject><subject>MPI parallelization</subject><subject>Multipoles</subject><subject>Parallel processing</subject><subject>Reduction</subject><subject>Supercomputers</subject><issn>0192-8651</issn><issn>1096-987X</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2021</creationdate><recordtype>article</recordtype><recordid>eNp1kU1r3DAQhkVoSLZpD_0DRdBLe3CiT0s-hqX5IiE5pNCbkeVxV4tkbSWbsD31p0ebTXso9DQw88zDMC9CHyg5pYSws7W1p6yWTBygBSVNXTVafX-DFoQ2rNK1pMfobc5rQgiXtThCx5wrXRbpAv0-9z9ictMq4Cni4EYX3C_Adw_X2MYQ5tFZM7k4ZuxGPK0Ab0wy3oMvVI8HkyccZj-5TfSAA0yr2O8WOzeW8VPx4lAmdvYm4X47muBsxtb4XefF-w4dDsZneP9aT9C3i6-Py6vq9v7yenl-W1neKFEJMnTawMANVZ2QCpTkWms1KJCd7jQoUisiCCOgJTVCil53hgEIQrmuB36CPu-9mxR_zpCnNrhswXszQpxzyyQpj2pUrQr66R90Hec0lusKRXXNtWKsUF_2lE0x5wRDu0kumLRtKWl3sbQllvYllsJ-fDXOXYD-L_knhwKc7YEn52H7f1N7s1zulc9H8pgA</recordid><startdate>20210605</startdate><enddate>20210605</enddate><creator>Andoh, Yoshimichi</creator><creator>Ichikawa, Shin‐ichi</creator><creator>Sakashita, Tatsuya</creator><creator>Yoshii, Noriyuki</creator><creator>Okazaki, Susumu</creator><general>John Wiley &amp; Sons, Inc</general><general>Wiley Subscription Services, Inc</general><scope>NPM</scope><scope>AAYXX</scope><scope>CITATION</scope><scope>JQ2</scope><scope>7X8</scope><orcidid>https://orcid.org/0000-0002-9843-467X</orcidid><orcidid>https://orcid.org/0000-0001-5027-4120</orcidid><orcidid>https://orcid.org/0000-0002-3361-2113</orcidid></search><sort><creationdate>20210605</creationdate><title>Algorithm to minimize MPI communications in the parallelized fast multipole method combined with molecular dynamics calculations</title><author>Andoh, Yoshimichi ; Ichikawa, Shin‐ichi ; Sakashita, Tatsuya ; Yoshii, Noriyuki ; Okazaki, Susumu</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-c3974-40fb8aef3a17b457e7538887f7e5b8b8e706704020e851a454d8ba2ee401386f3</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2021</creationdate><topic>Algorithms</topic><topic>Boundary conditions</topic><topic>Communication</topic><topic>efficient MPI communication</topic><topic>exascale supercomputers</topic><topic>fast multipole method</topic><topic>Mathematical analysis</topic><topic>Message passing</topic><topic>Molecular dynamics</topic><topic>molecular dynamics calculation</topic><topic>MPI parallelization</topic><topic>Multipoles</topic><topic>Parallel processing</topic><topic>Reduction</topic><topic>Supercomputers</topic><toplevel>peer_reviewed</toplevel><toplevel>online_resources</toplevel><creatorcontrib>Andoh, Yoshimichi</creatorcontrib><creatorcontrib>Ichikawa, Shin‐ichi</creatorcontrib><creatorcontrib>Sakashita, Tatsuya</creatorcontrib><creatorcontrib>Yoshii, Noriyuki</creatorcontrib><creatorcontrib>Okazaki, Susumu</creatorcontrib><collection>PubMed</collection><collection>CrossRef</collection><collection>ProQuest Computer Science Collection</collection><collection>MEDLINE - Academic</collection><jtitle>Journal of computational chemistry</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Andoh, Yoshimichi</au><au>Ichikawa, Shin‐ichi</au><au>Sakashita, Tatsuya</au><au>Yoshii, Noriyuki</au><au>Okazaki, Susumu</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Algorithm to minimize MPI communications in the parallelized fast multipole method combined with molecular dynamics calculations</atitle><jtitle>Journal of computational chemistry</jtitle><addtitle>J Comput Chem</addtitle><date>2021-06-05</date><risdate>2021</risdate><volume>42</volume><issue>15</issue><spage>1073</spage><epage>1087</epage><pages>1073-1087</pages><issn>0192-8651</issn><eissn>1096-987X</eissn><abstract>In the era of exascale supercomputers, large‐scale, and long‐time molecular dynamics (MD) calculations are expected to make breakthroughs in various fields of science and technology. Here, we propose a new algorithm to improve the parallelization performance of message passing interface (MPI)‐communication in the MPI‐parallelized fast multipole method (FMM) combined with MD calculations under three‐dimensional periodic boundary conditions. Our approach enables a drastic reduction in the amount of communication data, including the atomic coordinates and multipole coefficients, both of which are required to calculate the electrostatic interaction by using the FMM. In communications of multipole coefficients, the reduction rate of communication data in the new algorithm relative to the amount of data in the conventional one increases as both the number of FMM levels and the number of MPI processes increase. The aforementioned rate increase could exceed 50% as the number of MPI processes becomes larger for very large systems. The proposed algorithm, named the minimum‐transferred data (MTD) method, should enable large‐scale and long‐time MD calculations to be calculated efficiently, under the condition of massive MPI‐parallelization on exascale supercomputers. A new algorithm is proposed for message passing interface (MPI)‐communication suitable for the fast multipole method (FMM) combined with molecular dynamics (MD) calculations under three‐dimensional periodic boundary conditions. The method, named the minimum transferred‐data (MTD) method, can reduce the elapsed time of MPI communications in the MPI‐parallelized FMM drastically, enabling efficient large‐scale and long‐time MD calculations on exascale supercomputers.</abstract><cop>Hoboken, USA</cop><pub>John Wiley &amp; Sons, Inc</pub><pmid>33780021</pmid><doi>10.1002/jcc.26524</doi><tpages>15</tpages><orcidid>https://orcid.org/0000-0002-9843-467X</orcidid><orcidid>https://orcid.org/0000-0001-5027-4120</orcidid><orcidid>https://orcid.org/0000-0002-3361-2113</orcidid><oa>free_for_read</oa></addata></record>
fulltext fulltext
identifier ISSN: 0192-8651
ispartof Journal of computational chemistry, 2021-06, Vol.42 (15), p.1073-1087
issn 0192-8651
1096-987X
language eng
recordid cdi_proquest_miscellaneous_2506519767
source Wiley-Blackwell Read & Publish Collection
subjects Algorithms
Boundary conditions
Communication
efficient MPI communication
exascale supercomputers
fast multipole method
Mathematical analysis
Message passing
Molecular dynamics
molecular dynamics calculation
MPI parallelization
Multipoles
Parallel processing
Reduction
Supercomputers
title Algorithm to minimize MPI communications in the parallelized fast multipole method combined with molecular dynamics calculations
url http://sfxeu10.hosted.exlibrisgroup.com/loughborough?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2024-12-27T09%3A39%3A32IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest_cross&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Algorithm%20to%20minimize%20MPI%20communications%20in%20the%20parallelized%20fast%20multipole%20method%20combined%20with%20molecular%20dynamics%20calculations&rft.jtitle=Journal%20of%20computational%20chemistry&rft.au=Andoh,%20Yoshimichi&rft.date=2021-06-05&rft.volume=42&rft.issue=15&rft.spage=1073&rft.epage=1087&rft.pages=1073-1087&rft.issn=0192-8651&rft.eissn=1096-987X&rft_id=info:doi/10.1002/jcc.26524&rft_dat=%3Cproquest_cross%3E2518638722%3C/proquest_cross%3E%3Cgrp_id%3Ecdi_FETCH-LOGICAL-c3974-40fb8aef3a17b457e7538887f7e5b8b8e706704020e851a454d8ba2ee401386f3%3C/grp_id%3E%3Coa%3E%3C/oa%3E%3Curl%3E%3C/url%3E&rft_id=info:oai/&rft_pqid=2518638722&rft_id=info:pmid/33780021&rfr_iscdi=true