Asynchronous OpenCL/MPI Numerical Simulations of Conservation Laws

Asynchronous OpenCL/MPI Numerical Simulations of Conservation Laws

Hyperbolic conservation laws are important mathematical models for describing many phenomena in physics or engineering. The Finite Volume (FV) method and the Discontinuous Galerkin (DG) method are two popular methods for solving conservation laws on computers. In this paper, we present several FV an...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Helluy, Philippe, Strub, Thomas, Massaro, Michel, Roberts, Malcolm
Format: Buchkapitel
Sprache:eng
Schlagworte:
Analysis of PDEs
Discontinuous Galerkin
Gauss Point
Interface Zone
Mathematics
Task Graph
Volume Zone
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page 565
container_issue
container_start_page 547
container_title
container_volume
creator Helluy, Philippe
Strub, Thomas
Massaro, Michel
Roberts, Malcolm
description Hyperbolic conservation laws are important mathematical models for describing many phenomena in physics or engineering. The Finite Volume (FV) method and the Discontinuous Galerkin (DG) method are two popular methods for solving conservation laws on computers. In this paper, we present several FV and DG numerical simulations that we have realized with the OpenCL and MPI paradigms. First, we compare two optimized implementations of the FV method on a regular grid: an OpenCL implementation and a more traditional OpenMP implementation. We compare the efficiency of the approach on several CPU and GPU architectures of different brands. Then we present how we have implemented the DG method in the OpenCL/MPI framework in order to achieve high efficiency. The implementation relies on a splitting of the DG mesh into subdomains and subzones. Different kernels are compiled according to the zone properties. In addition, we rely on the OpenCL asynchronous task graph in order to overlap OpenCL computations, memory transfers and MPI communications.
doi_str_mv 10.1007/978-3-319-40528-5_25
format Book Chapter
fullrecord <record><control><sourceid>hal_sprin</sourceid><recordid>TN_cdi_hal_primary_oai_HAL_hal_01134222v2</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>oai_HAL_hal_01134222v2</sourcerecordid><originalsourceid>FETCH-LOGICAL-h278t-d60b845ce509fd25f13b17fcf8f6174f3f25748c4cc409791ff15be0d0bddb783</originalsourceid><addsrcrecordid>eNo1kEtPAjEQgOsrEZF_4GGvHiozfdD2iBsVklVM1HPTfVQ2LrtkKxj-vQX0NMk3XyaZj5AbhDsEUGOjNOWUo6ECJNNUWiZPyBWP5ADEKRkwNIqqqJ-RUfT_dxM8JwMU3FDFpb4koxDqHIBpQMVxQO6nYdcWy75ru01IFuuqTbPx8-s8edmsqr4uXJO81atN477rrg1J55M0zqrfHkCSuZ9wTS68a0I1-ptD8vH48J7OaLZ4mqfTjC6Z0t-0nECuhSwqCcaXTHrkOSpfeO0nqITnnkkldCGKQoBRBr1HmVdQQl6WefxoSG6Pd5euseu-Xrl-ZztX29k0s3sGiFwwxrYsuuzohii2n1Vv8677ChbB7ovaWMhyGxvZQ0C7L8p_AQXfY6A</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>book_chapter</recordtype></control><display><type>book_chapter</type><title>Asynchronous OpenCL/MPI Numerical Simulations of Conservation Laws</title><source>Springer Books</source><creator>Helluy, Philippe ; Strub, Thomas ; Massaro, Michel ; Roberts, Malcolm</creator><contributor>Bungartz, Hans-Joachim ; Nagel, Wolfgang E. ; Neumann, Philipp</contributor><creatorcontrib>Helluy, Philippe ; Strub, Thomas ; Massaro, Michel ; Roberts, Malcolm ; Bungartz, Hans-Joachim ; Nagel, Wolfgang E. ; Neumann, Philipp</creatorcontrib><description>Hyperbolic conservation laws are important mathematical models for describing many phenomena in physics or engineering. The Finite Volume (FV) method and the Discontinuous Galerkin (DG) method are two popular methods for solving conservation laws on computers. In this paper, we present several FV and DG numerical simulations that we have realized with the OpenCL and MPI paradigms. First, we compare two optimized implementations of the FV method on a regular grid: an OpenCL implementation and a more traditional OpenMP implementation. We compare the efficiency of the approach on several CPU and GPU architectures of different brands. Then we present how we have implemented the DG method in the OpenCL/MPI framework in order to achieve high efficiency. The implementation relies on a splitting of the DG mesh into subdomains and subzones. Different kernels are compiled according to the zone properties. In addition, we rely on the OpenCL asynchronous task graph in order to overlap OpenCL computations, memory transfers and MPI communications.</description><identifier>ISSN: 1439-7358</identifier><identifier>ISBN: 9783319405261</identifier><identifier>ISBN: 3319405268</identifier><identifier>ISBN: 3319405284</identifier><identifier>ISBN: 9783319405285</identifier><identifier>EISSN: 2197-7100</identifier><identifier>EISBN: 3319405284</identifier><identifier>EISBN: 9783319405285</identifier><identifier>DOI: 10.1007/978-3-319-40528-5_25</identifier><language>eng</language><publisher>Cham: Springer International Publishing</publisher><subject>Analysis of PDEs ; Discontinuous Galerkin ; Gauss Point ; Interface Zone ; Mathematics ; Task Graph ; Volume Zone</subject><ispartof>Software for Exascale Computing - SPPEXA 2013-2015, 2016, p.547-565</ispartof><rights>Springer International Publishing Switzerland 2016</rights><rights>Distributed under a Creative Commons Attribution 4.0 International License</rights><lds50>peer_reviewed</lds50><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed><orcidid>0000-0002-1077-1867</orcidid><relation>Lecture Notes in Computational Science and Engineering</relation></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><linktopdf>$$Uhttps://link.springer.com/content/pdf/10.1007/978-3-319-40528-5_25$$EPDF$$P50$$Gspringer$$H</linktopdf><linktohtml>$$Uhttps://link.springer.com/10.1007/978-3-319-40528-5_25$$EHTML$$P50$$Gspringer$$H</linktohtml><link.rule.ids>230,775,776,780,789,881,27902,38232,41418,42487</link.rule.ids><backlink>$$Uhttps://hal.science/hal-01134222$$DView record in HAL$$Hfree_for_read</backlink></links><search><contributor>Bungartz, Hans-Joachim</contributor><contributor>Nagel, Wolfgang E.</contributor><contributor>Neumann, Philipp</contributor><creatorcontrib>Helluy, Philippe</creatorcontrib><creatorcontrib>Strub, Thomas</creatorcontrib><creatorcontrib>Massaro, Michel</creatorcontrib><creatorcontrib>Roberts, Malcolm</creatorcontrib><title>Asynchronous OpenCL/MPI Numerical Simulations of Conservation Laws</title><title>Software for Exascale Computing - SPPEXA 2013-2015</title><description>Hyperbolic conservation laws are important mathematical models for describing many phenomena in physics or engineering. The Finite Volume (FV) method and the Discontinuous Galerkin (DG) method are two popular methods for solving conservation laws on computers. In this paper, we present several FV and DG numerical simulations that we have realized with the OpenCL and MPI paradigms. First, we compare two optimized implementations of the FV method on a regular grid: an OpenCL implementation and a more traditional OpenMP implementation. We compare the efficiency of the approach on several CPU and GPU architectures of different brands. Then we present how we have implemented the DG method in the OpenCL/MPI framework in order to achieve high efficiency. The implementation relies on a splitting of the DG mesh into subdomains and subzones. Different kernels are compiled according to the zone properties. In addition, we rely on the OpenCL asynchronous task graph in order to overlap OpenCL computations, memory transfers and MPI communications.</description><subject>Analysis of PDEs</subject><subject>Discontinuous Galerkin</subject><subject>Gauss Point</subject><subject>Interface Zone</subject><subject>Mathematics</subject><subject>Task Graph</subject><subject>Volume Zone</subject><issn>1439-7358</issn><issn>2197-7100</issn><isbn>9783319405261</isbn><isbn>3319405268</isbn><isbn>3319405284</isbn><isbn>9783319405285</isbn><isbn>3319405284</isbn><isbn>9783319405285</isbn><fulltext>true</fulltext><rsrctype>book_chapter</rsrctype><creationdate>2016</creationdate><recordtype>book_chapter</recordtype><recordid>eNo1kEtPAjEQgOsrEZF_4GGvHiozfdD2iBsVklVM1HPTfVQ2LrtkKxj-vQX0NMk3XyaZj5AbhDsEUGOjNOWUo6ECJNNUWiZPyBWP5ADEKRkwNIqqqJ-RUfT_dxM8JwMU3FDFpb4koxDqHIBpQMVxQO6nYdcWy75ru01IFuuqTbPx8-s8edmsqr4uXJO81atN477rrg1J55M0zqrfHkCSuZ9wTS68a0I1-ptD8vH48J7OaLZ4mqfTjC6Z0t-0nECuhSwqCcaXTHrkOSpfeO0nqITnnkkldCGKQoBRBr1HmVdQQl6WefxoSG6Pd5euseu-Xrl-ZztX29k0s3sGiFwwxrYsuuzohii2n1Vv8677ChbB7ovaWMhyGxvZQ0C7L8p_AQXfY6A</recordid><startdate>2016</startdate><enddate>2016</enddate><creator>Helluy, Philippe</creator><creator>Strub, Thomas</creator><creator>Massaro, Michel</creator><creator>Roberts, Malcolm</creator><general>Springer International Publishing</general><scope>1XC</scope><scope>VOOES</scope><orcidid>https://orcid.org/0000-0002-1077-1867</orcidid></search><sort><creationdate>2016</creationdate><title>Asynchronous OpenCL/MPI Numerical Simulations of Conservation Laws</title><author>Helluy, Philippe ; Strub, Thomas ; Massaro, Michel ; Roberts, Malcolm</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-h278t-d60b845ce509fd25f13b17fcf8f6174f3f25748c4cc409791ff15be0d0bddb783</frbrgroupid><rsrctype>book_chapters</rsrctype><prefilter>book_chapters</prefilter><language>eng</language><creationdate>2016</creationdate><topic>Analysis of PDEs</topic><topic>Discontinuous Galerkin</topic><topic>Gauss Point</topic><topic>Interface Zone</topic><topic>Mathematics</topic><topic>Task Graph</topic><topic>Volume Zone</topic><toplevel>peer_reviewed</toplevel><toplevel>online_resources</toplevel><creatorcontrib>Helluy, Philippe</creatorcontrib><creatorcontrib>Strub, Thomas</creatorcontrib><creatorcontrib>Massaro, Michel</creatorcontrib><creatorcontrib>Roberts, Malcolm</creatorcontrib><collection>Hyper Article en Ligne (HAL)</collection><collection>Hyper Article en Ligne (HAL) (Open Access)</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Helluy, Philippe</au><au>Strub, Thomas</au><au>Massaro, Michel</au><au>Roberts, Malcolm</au><au>Bungartz, Hans-Joachim</au><au>Nagel, Wolfgang E.</au><au>Neumann, Philipp</au><format>book</format><genre>bookitem</genre><ristype>CHAP</ristype><atitle>Asynchronous OpenCL/MPI Numerical Simulations of Conservation Laws</atitle><btitle>Software for Exascale Computing - SPPEXA 2013-2015</btitle><seriestitle>Lecture Notes in Computational Science and Engineering</seriestitle><date>2016</date><risdate>2016</risdate><spage>547</spage><epage>565</epage><pages>547-565</pages><issn>1439-7358</issn><eissn>2197-7100</eissn><isbn>9783319405261</isbn><isbn>3319405268</isbn><isbn>3319405284</isbn><isbn>9783319405285</isbn><eisbn>3319405284</eisbn><eisbn>9783319405285</eisbn><abstract>Hyperbolic conservation laws are important mathematical models for describing many phenomena in physics or engineering. The Finite Volume (FV) method and the Discontinuous Galerkin (DG) method are two popular methods for solving conservation laws on computers. In this paper, we present several FV and DG numerical simulations that we have realized with the OpenCL and MPI paradigms. First, we compare two optimized implementations of the FV method on a regular grid: an OpenCL implementation and a more traditional OpenMP implementation. We compare the efficiency of the approach on several CPU and GPU architectures of different brands. Then we present how we have implemented the DG method in the OpenCL/MPI framework in order to achieve high efficiency. The implementation relies on a splitting of the DG mesh into subdomains and subzones. Different kernels are compiled according to the zone properties. In addition, we rely on the OpenCL asynchronous task graph in order to overlap OpenCL computations, memory transfers and MPI communications.</abstract><cop>Cham</cop><pub>Springer International Publishing</pub><doi>10.1007/978-3-319-40528-5_25</doi><tpages>19</tpages><orcidid>https://orcid.org/0000-0002-1077-1867</orcidid><oa>free_for_read</oa></addata></record>
fulltext fulltext
identifier ISSN: 1439-7358
ispartof Software for Exascale Computing - SPPEXA 2013-2015, 2016, p.547-565
issn 1439-7358
2197-7100
language eng
recordid cdi_hal_primary_oai_HAL_hal_01134222v2
source Springer Books
subjects Analysis of PDEs
Discontinuous Galerkin
Gauss Point
Interface Zone
Mathematics
Task Graph
Volume Zone
title Asynchronous OpenCL/MPI Numerical Simulations of Conservation Laws
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-02-07T00%3A53%3A21IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-hal_sprin&rft_val_fmt=info:ofi/fmt:kev:mtx:book&rft.genre=bookitem&rft.atitle=Asynchronous%20OpenCL/MPI%20Numerical%20Simulations%20of%20Conservation%20Laws&rft.btitle=Software%20for%20Exascale%20Computing%20-%20SPPEXA%202013-2015&rft.au=Helluy,%20Philippe&rft.date=2016&rft.spage=547&rft.epage=565&rft.pages=547-565&rft.issn=1439-7358&rft.eissn=2197-7100&rft.isbn=9783319405261&rft.isbn_list=3319405268&rft.isbn_list=3319405284&rft.isbn_list=9783319405285&rft_id=info:doi/10.1007/978-3-319-40528-5_25&rft_dat=%3Chal_sprin%3Eoai_HAL_hal_01134222v2%3C/hal_sprin%3E%3Curl%3E%3C/url%3E&rft.eisbn=3319405284&rft.eisbn_list=9783319405285&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_id=info:pmid/&rfr_iscdi=true