Using 1000+ GPUs and 10000+ CPUs for Sedimentary Basin Simulations

In cutting-edge CPU/GPU hybrid clusters, such as Tianhe-1A, the aggregate CPU computing capability may amount to up to 1/3 of the aggregate GPU computing capability. It thus goes without saying that the CPUs and GPUs should jointly carry out the computational work. However, to effectively and simult...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Mei Wen, Huayou Su, Wenjie Wei, Nan Wu, Xing Cai, Chunyuan Zhang
Format: Tagungsbericht
Sprache:eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page 35
container_issue
container_start_page 27
container_title
container_volume
creator Mei Wen
Huayou Su
Wenjie Wei
Nan Wu
Xing Cai
Chunyuan Zhang
description In cutting-edge CPU/GPU hybrid clusters, such as Tianhe-1A, the aggregate CPU computing capability may amount to up to 1/3 of the aggregate GPU computing capability. It thus goes without saying that the CPUs and GPUs should jointly carry out the computational work. However, to effectively and simultaneously use both the hardware components requires great care when developing the parallel implementations. The challenges include (1) finding a balanced division of the workload between the CPU and GPU sides, and (2) hiding various overheads by overlapping computations with CPU-GPU data transfers and/or MPI communications. We study these issues in the context of real-world sedimentary basin simulations. Numerical experiments show that an appropriately devised CPU-GPU hybrid implementation is able to handle a global mesh resolution of 131,072*131,072, and a double-precision rate of 62 TFlops is achieved by using 1024 GPUs and 12288 CPU cores on Tianhe-1A. Such an extreme computing capability will be of great importance for carrying out high-resolution and continental-scale stratigraphic simulations in future.
doi_str_mv 10.1109/CLUSTER.2012.37
format Conference Proceeding
fullrecord <record><control><sourceid>ieee_6IE</sourceid><recordid>TN_cdi_ieee_primary_6337853</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><ieee_id>6337853</ieee_id><sourcerecordid>6337853</sourcerecordid><originalsourceid>FETCH-LOGICAL-i175t-5cdd4370ed114d100d6eb1e55d64028fa6610d9f5b76d3269395a79294b8ea9f3</originalsourceid><addsrcrecordid>eNotjktPwzAQhM1LIpScOXDJHSV4ba8fRxqVghQJRJpz5WAHGTUpisuBf495zGX0jTS7Q8gV0AqAmtu66drN6qViFFjF1RG5oEoaFJoqPCYZA6lLw5CfkNwoDUIqzgRj5pRkgMhKZEKckzzGd5qkQVMjM7LsYpjeCkjRTbF-7mJhJ_eLiesfHvZz0XoXRj8d7PxVLG1qFG0YP3f2EPZTvCRng91Fn__7gnT3q039UDZP68f6rikDKDyU-Oqc4Ip6ByBc-uCk78EjOiko04OVEqgzA_ZKOs6k4QatMsyIXntrBr4g1393g_d--zGHMc3ZSs6VRs6_ASfFSxc</addsrcrecordid><sourcetype>Publisher</sourcetype><iscdi>true</iscdi><recordtype>conference_proceeding</recordtype></control><display><type>conference_proceeding</type><title>Using 1000+ GPUs and 10000+ CPUs for Sedimentary Basin Simulations</title><source>IEEE Electronic Library (IEL) Conference Proceedings</source><creator>Mei Wen ; Huayou Su ; Wenjie Wei ; Nan Wu ; Xing Cai ; Chunyuan Zhang</creator><creatorcontrib>Mei Wen ; Huayou Su ; Wenjie Wei ; Nan Wu ; Xing Cai ; Chunyuan Zhang</creatorcontrib><description>In cutting-edge CPU/GPU hybrid clusters, such as Tianhe-1A, the aggregate CPU computing capability may amount to up to 1/3 of the aggregate GPU computing capability. It thus goes without saying that the CPUs and GPUs should jointly carry out the computational work. However, to effectively and simultaneously use both the hardware components requires great care when developing the parallel implementations. The challenges include (1) finding a balanced division of the workload between the CPU and GPU sides, and (2) hiding various overheads by overlapping computations with CPU-GPU data transfers and/or MPI communications. We study these issues in the context of real-world sedimentary basin simulations. Numerical experiments show that an appropriately devised CPU-GPU hybrid implementation is able to handle a global mesh resolution of 131,072*131,072, and a double-precision rate of 62 TFlops is achieved by using 1024 GPUs and 12288 CPU cores on Tianhe-1A. Such an extreme computing capability will be of great importance for carrying out high-resolution and continental-scale stratigraphic simulations in future.</description><identifier>ISSN: 1552-5244</identifier><identifier>ISBN: 9781467324229</identifier><identifier>ISBN: 1467324221</identifier><identifier>EISSN: 2168-9253</identifier><identifier>EISBN: 0769548075</identifier><identifier>EISBN: 9780769548074</identifier><identifier>DOI: 10.1109/CLUSTER.2012.37</identifier><identifier>CODEN: IEEPAD</identifier><language>eng</language><publisher>IEEE</publisher><subject>Computational modeling ; CPU-GPU hybrid computing ; Dual-lithology sedimentary basin simulation ; Graphics processing unit ; Hardware ; Instruction sets ; Kernel ; Mathematical model ; MPI/OpenMP/CUDA ; Multicore processing ; Tianhe-1A Hunan Solution</subject><ispartof>2012 IEEE International Conference on Cluster Computing, 2012, p.27-35</ispartof><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><linktohtml>$$Uhttps://ieeexplore.ieee.org/document/6337853$$EHTML$$P50$$Gieee$$H</linktohtml><link.rule.ids>309,310,780,784,789,790,2058,27925,54920</link.rule.ids><linktorsrc>$$Uhttps://ieeexplore.ieee.org/document/6337853$$EView_record_in_IEEE$$FView_record_in_$$GIEEE</linktorsrc></links><search><creatorcontrib>Mei Wen</creatorcontrib><creatorcontrib>Huayou Su</creatorcontrib><creatorcontrib>Wenjie Wei</creatorcontrib><creatorcontrib>Nan Wu</creatorcontrib><creatorcontrib>Xing Cai</creatorcontrib><creatorcontrib>Chunyuan Zhang</creatorcontrib><title>Using 1000+ GPUs and 10000+ CPUs for Sedimentary Basin Simulations</title><title>2012 IEEE International Conference on Cluster Computing</title><addtitle>CLUSTR</addtitle><description>In cutting-edge CPU/GPU hybrid clusters, such as Tianhe-1A, the aggregate CPU computing capability may amount to up to 1/3 of the aggregate GPU computing capability. It thus goes without saying that the CPUs and GPUs should jointly carry out the computational work. However, to effectively and simultaneously use both the hardware components requires great care when developing the parallel implementations. The challenges include (1) finding a balanced division of the workload between the CPU and GPU sides, and (2) hiding various overheads by overlapping computations with CPU-GPU data transfers and/or MPI communications. We study these issues in the context of real-world sedimentary basin simulations. Numerical experiments show that an appropriately devised CPU-GPU hybrid implementation is able to handle a global mesh resolution of 131,072*131,072, and a double-precision rate of 62 TFlops is achieved by using 1024 GPUs and 12288 CPU cores on Tianhe-1A. Such an extreme computing capability will be of great importance for carrying out high-resolution and continental-scale stratigraphic simulations in future.</description><subject>Computational modeling</subject><subject>CPU-GPU hybrid computing</subject><subject>Dual-lithology sedimentary basin simulation</subject><subject>Graphics processing unit</subject><subject>Hardware</subject><subject>Instruction sets</subject><subject>Kernel</subject><subject>Mathematical model</subject><subject>MPI/OpenMP/CUDA</subject><subject>Multicore processing</subject><subject>Tianhe-1A Hunan Solution</subject><issn>1552-5244</issn><issn>2168-9253</issn><isbn>9781467324229</isbn><isbn>1467324221</isbn><isbn>0769548075</isbn><isbn>9780769548074</isbn><fulltext>true</fulltext><rsrctype>conference_proceeding</rsrctype><creationdate>2012</creationdate><recordtype>conference_proceeding</recordtype><sourceid>6IE</sourceid><sourceid>RIE</sourceid><recordid>eNotjktPwzAQhM1LIpScOXDJHSV4ba8fRxqVghQJRJpz5WAHGTUpisuBf495zGX0jTS7Q8gV0AqAmtu66drN6qViFFjF1RG5oEoaFJoqPCYZA6lLw5CfkNwoDUIqzgRj5pRkgMhKZEKckzzGd5qkQVMjM7LsYpjeCkjRTbF-7mJhJ_eLiesfHvZz0XoXRj8d7PxVLG1qFG0YP3f2EPZTvCRng91Fn__7gnT3q039UDZP68f6rikDKDyU-Oqc4Ip6ByBc-uCk78EjOiko04OVEqgzA_ZKOs6k4QatMsyIXntrBr4g1393g_d--zGHMc3ZSs6VRs6_ASfFSxc</recordid><startdate>201209</startdate><enddate>201209</enddate><creator>Mei Wen</creator><creator>Huayou Su</creator><creator>Wenjie Wei</creator><creator>Nan Wu</creator><creator>Xing Cai</creator><creator>Chunyuan Zhang</creator><general>IEEE</general><scope>6IE</scope><scope>6IL</scope><scope>CBEJK</scope><scope>RIE</scope><scope>RIL</scope></search><sort><creationdate>201209</creationdate><title>Using 1000+ GPUs and 10000+ CPUs for Sedimentary Basin Simulations</title><author>Mei Wen ; Huayou Su ; Wenjie Wei ; Nan Wu ; Xing Cai ; Chunyuan Zhang</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-i175t-5cdd4370ed114d100d6eb1e55d64028fa6610d9f5b76d3269395a79294b8ea9f3</frbrgroupid><rsrctype>conference_proceedings</rsrctype><prefilter>conference_proceedings</prefilter><language>eng</language><creationdate>2012</creationdate><topic>Computational modeling</topic><topic>CPU-GPU hybrid computing</topic><topic>Dual-lithology sedimentary basin simulation</topic><topic>Graphics processing unit</topic><topic>Hardware</topic><topic>Instruction sets</topic><topic>Kernel</topic><topic>Mathematical model</topic><topic>MPI/OpenMP/CUDA</topic><topic>Multicore processing</topic><topic>Tianhe-1A Hunan Solution</topic><toplevel>online_resources</toplevel><creatorcontrib>Mei Wen</creatorcontrib><creatorcontrib>Huayou Su</creatorcontrib><creatorcontrib>Wenjie Wei</creatorcontrib><creatorcontrib>Nan Wu</creatorcontrib><creatorcontrib>Xing Cai</creatorcontrib><creatorcontrib>Chunyuan Zhang</creatorcontrib><collection>IEEE Electronic Library (IEL) Conference Proceedings</collection><collection>IEEE Proceedings Order Plan All Online (POP All Online) 1998-present by volume</collection><collection>IEEE Xplore All Conference Proceedings</collection><collection>IEEE Electronic Library (IEL)</collection><collection>IEEE Proceedings Order Plans (POP All) 1998-Present</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Mei Wen</au><au>Huayou Su</au><au>Wenjie Wei</au><au>Nan Wu</au><au>Xing Cai</au><au>Chunyuan Zhang</au><format>book</format><genre>proceeding</genre><ristype>CONF</ristype><atitle>Using 1000+ GPUs and 10000+ CPUs for Sedimentary Basin Simulations</atitle><btitle>2012 IEEE International Conference on Cluster Computing</btitle><stitle>CLUSTR</stitle><date>2012-09</date><risdate>2012</risdate><spage>27</spage><epage>35</epage><pages>27-35</pages><issn>1552-5244</issn><eissn>2168-9253</eissn><isbn>9781467324229</isbn><isbn>1467324221</isbn><eisbn>0769548075</eisbn><eisbn>9780769548074</eisbn><coden>IEEPAD</coden><abstract>In cutting-edge CPU/GPU hybrid clusters, such as Tianhe-1A, the aggregate CPU computing capability may amount to up to 1/3 of the aggregate GPU computing capability. It thus goes without saying that the CPUs and GPUs should jointly carry out the computational work. However, to effectively and simultaneously use both the hardware components requires great care when developing the parallel implementations. The challenges include (1) finding a balanced division of the workload between the CPU and GPU sides, and (2) hiding various overheads by overlapping computations with CPU-GPU data transfers and/or MPI communications. We study these issues in the context of real-world sedimentary basin simulations. Numerical experiments show that an appropriately devised CPU-GPU hybrid implementation is able to handle a global mesh resolution of 131,072*131,072, and a double-precision rate of 62 TFlops is achieved by using 1024 GPUs and 12288 CPU cores on Tianhe-1A. Such an extreme computing capability will be of great importance for carrying out high-resolution and continental-scale stratigraphic simulations in future.</abstract><pub>IEEE</pub><doi>10.1109/CLUSTER.2012.37</doi><tpages>9</tpages></addata></record>
fulltext fulltext_linktorsrc
identifier ISSN: 1552-5244
ispartof 2012 IEEE International Conference on Cluster Computing, 2012, p.27-35
issn 1552-5244
2168-9253
language eng
recordid cdi_ieee_primary_6337853
source IEEE Electronic Library (IEL) Conference Proceedings
subjects Computational modeling
CPU-GPU hybrid computing
Dual-lithology sedimentary basin simulation
Graphics processing unit
Hardware
Instruction sets
Kernel
Mathematical model
MPI/OpenMP/CUDA
Multicore processing
Tianhe-1A Hunan Solution
title Using 1000+ GPUs and 10000+ CPUs for Sedimentary Basin Simulations
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2024-12-25T01%3A48%3A17IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-ieee_6IE&rft_val_fmt=info:ofi/fmt:kev:mtx:book&rft.genre=proceeding&rft.atitle=Using%201000+%20GPUs%20and%2010000+%20CPUs%20for%20Sedimentary%20Basin%20Simulations&rft.btitle=2012%20IEEE%20International%20Conference%20on%20Cluster%20Computing&rft.au=Mei%20Wen&rft.date=2012-09&rft.spage=27&rft.epage=35&rft.pages=27-35&rft.issn=1552-5244&rft.eissn=2168-9253&rft.isbn=9781467324229&rft.isbn_list=1467324221&rft.coden=IEEPAD&rft_id=info:doi/10.1109/CLUSTER.2012.37&rft_dat=%3Cieee_6IE%3E6337853%3C/ieee_6IE%3E%3Curl%3E%3C/url%3E&rft.eisbn=0769548075&rft.eisbn_list=9780769548074&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_id=info:pmid/&rft_ieee_id=6337853&rfr_iscdi=true