Exploring Fully Offloaded GPU Stream-Aware Message Passing
Modern heterogeneous supercomputing systems are comprised of CPUs, GPUs, and high-speed network interconnects. Communication libraries supporting efficient data transfers involving memory buffers from the GPU memory typically require the CPU to orchestrate the data transfer operations. A new offload...
Gespeichert in:
Veröffentlicht in: | arXiv.org 2023-06 |
---|---|
Hauptverfasser: | , , , , |
Format: | Artikel |
Sprache: | eng |
Schlagworte: | |
Online-Zugang: | Volltext |
Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
container_end_page | |
---|---|
container_issue | |
container_start_page | |
container_title | arXiv.org |
container_volume | |
creator | Namashivayam, Naveen Kandalla, Krishna White, James B Kaplan, Larry Pagel, Mark |
description | Modern heterogeneous supercomputing systems are comprised of CPUs, GPUs, and high-speed network interconnects. Communication libraries supporting efficient data transfers involving memory buffers from the GPU memory typically require the CPU to orchestrate the data transfer operations. A new offload-friendly communication strategy, stream-triggered (ST) communication, was explored to allow offloading the synchronization and data movement operations from the CPU to the GPU. A Message Passing Interface (MPI) one-sided active target synchronization based implementation was used as an exemplar to illustrate the proposed strategy. A latency-sensitive nearest neighbor microbenchmark was used to explore the various performance aspects of the implementation. The offloaded implementation shows significant on-node performance advantages over standard MPI active RMA (36%) and point-to-point (61%) communication. The current multi-node improvement is less (23% faster than standard active RMA but 11% slower than point-to-point), but plans are in progress to purse further improvements. |
format | Article |
fullrecord | <record><control><sourceid>proquest</sourceid><recordid>TN_cdi_proquest_journals_2831118379</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2831118379</sourcerecordid><originalsourceid>FETCH-proquest_journals_28311183793</originalsourceid><addsrcrecordid>eNpjYuA0MjY21LUwMTLiYOAtLs4yMDAwMjM3MjU15mSwcq0oyMkvysxLV3ArzcmpVPBPS8vJT0xJTVFwDwhVCC4pSk3M1XUsTyxKVfBNLS5OTE9VCEgsLgZq4GFgTUvMKU7lhdLcDMpuriHOHroFRfmFpanFJfFZ-aVFeUCpeCMLY0NDQwtjc0tj4lQBAHQFNiw</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2831118379</pqid></control><display><type>article</type><title>Exploring Fully Offloaded GPU Stream-Aware Message Passing</title><source>Free E- Journals</source><creator>Namashivayam, Naveen ; Kandalla, Krishna ; White, James B ; Kaplan, Larry ; Pagel, Mark</creator><creatorcontrib>Namashivayam, Naveen ; Kandalla, Krishna ; White, James B ; Kaplan, Larry ; Pagel, Mark</creatorcontrib><description>Modern heterogeneous supercomputing systems are comprised of CPUs, GPUs, and high-speed network interconnects. Communication libraries supporting efficient data transfers involving memory buffers from the GPU memory typically require the CPU to orchestrate the data transfer operations. A new offload-friendly communication strategy, stream-triggered (ST) communication, was explored to allow offloading the synchronization and data movement operations from the CPU to the GPU. A Message Passing Interface (MPI) one-sided active target synchronization based implementation was used as an exemplar to illustrate the proposed strategy. A latency-sensitive nearest neighbor microbenchmark was used to explore the various performance aspects of the implementation. The offloaded implementation shows significant on-node performance advantages over standard MPI active RMA (36%) and point-to-point (61%) communication. The current multi-node improvement is less (23% faster than standard active RMA but 11% slower than point-to-point), but plans are in progress to purse further improvements.</description><identifier>EISSN: 2331-8422</identifier><language>eng</language><publisher>Ithaca: Cornell University Library, arXiv.org</publisher><subject>Data transfer (computers) ; Message passing ; Synchronism</subject><ispartof>arXiv.org, 2023-06</ispartof><rights>2023. This work is published under http://arxiv.org/licenses/nonexclusive-distrib/1.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>780,784</link.rule.ids></links><search><creatorcontrib>Namashivayam, Naveen</creatorcontrib><creatorcontrib>Kandalla, Krishna</creatorcontrib><creatorcontrib>White, James B</creatorcontrib><creatorcontrib>Kaplan, Larry</creatorcontrib><creatorcontrib>Pagel, Mark</creatorcontrib><title>Exploring Fully Offloaded GPU Stream-Aware Message Passing</title><title>arXiv.org</title><description>Modern heterogeneous supercomputing systems are comprised of CPUs, GPUs, and high-speed network interconnects. Communication libraries supporting efficient data transfers involving memory buffers from the GPU memory typically require the CPU to orchestrate the data transfer operations. A new offload-friendly communication strategy, stream-triggered (ST) communication, was explored to allow offloading the synchronization and data movement operations from the CPU to the GPU. A Message Passing Interface (MPI) one-sided active target synchronization based implementation was used as an exemplar to illustrate the proposed strategy. A latency-sensitive nearest neighbor microbenchmark was used to explore the various performance aspects of the implementation. The offloaded implementation shows significant on-node performance advantages over standard MPI active RMA (36%) and point-to-point (61%) communication. The current multi-node improvement is less (23% faster than standard active RMA but 11% slower than point-to-point), but plans are in progress to purse further improvements.</description><subject>Data transfer (computers)</subject><subject>Message passing</subject><subject>Synchronism</subject><issn>2331-8422</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2023</creationdate><recordtype>article</recordtype><sourceid>ABUWG</sourceid><sourceid>AFKRA</sourceid><sourceid>AZQEC</sourceid><sourceid>BENPR</sourceid><sourceid>CCPQU</sourceid><sourceid>DWQXO</sourceid><recordid>eNpjYuA0MjY21LUwMTLiYOAtLs4yMDAwMjM3MjU15mSwcq0oyMkvysxLV3ArzcmpVPBPS8vJT0xJTVFwDwhVCC4pSk3M1XUsTyxKVfBNLS5OTE9VCEgsLgZq4GFgTUvMKU7lhdLcDMpuriHOHroFRfmFpanFJfFZ-aVFeUCpeCMLY0NDQwtjc0tj4lQBAHQFNiw</recordid><startdate>20230627</startdate><enddate>20230627</enddate><creator>Namashivayam, Naveen</creator><creator>Kandalla, Krishna</creator><creator>White, James B</creator><creator>Kaplan, Larry</creator><creator>Pagel, Mark</creator><general>Cornell University Library, arXiv.org</general><scope>8FE</scope><scope>8FG</scope><scope>ABJCF</scope><scope>ABUWG</scope><scope>AFKRA</scope><scope>AZQEC</scope><scope>BENPR</scope><scope>BGLVJ</scope><scope>CCPQU</scope><scope>DWQXO</scope><scope>HCIFZ</scope><scope>L6V</scope><scope>M7S</scope><scope>PIMPY</scope><scope>PQEST</scope><scope>PQQKQ</scope><scope>PQUKI</scope><scope>PRINS</scope><scope>PTHSS</scope></search><sort><creationdate>20230627</creationdate><title>Exploring Fully Offloaded GPU Stream-Aware Message Passing</title><author>Namashivayam, Naveen ; Kandalla, Krishna ; White, James B ; Kaplan, Larry ; Pagel, Mark</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-proquest_journals_28311183793</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2023</creationdate><topic>Data transfer (computers)</topic><topic>Message passing</topic><topic>Synchronism</topic><toplevel>online_resources</toplevel><creatorcontrib>Namashivayam, Naveen</creatorcontrib><creatorcontrib>Kandalla, Krishna</creatorcontrib><creatorcontrib>White, James B</creatorcontrib><creatorcontrib>Kaplan, Larry</creatorcontrib><creatorcontrib>Pagel, Mark</creatorcontrib><collection>ProQuest SciTech Collection</collection><collection>ProQuest Technology Collection</collection><collection>Materials Science & Engineering Collection</collection><collection>ProQuest Central (Alumni Edition)</collection><collection>ProQuest Central UK/Ireland</collection><collection>ProQuest Central Essentials</collection><collection>ProQuest Central</collection><collection>Technology Collection</collection><collection>ProQuest One Community College</collection><collection>ProQuest Central Korea</collection><collection>SciTech Premium Collection</collection><collection>ProQuest Engineering Collection</collection><collection>Engineering Database</collection><collection>Publicly Available Content Database</collection><collection>ProQuest One Academic Eastern Edition (DO NOT USE)</collection><collection>ProQuest One Academic</collection><collection>ProQuest One Academic UKI Edition</collection><collection>ProQuest Central China</collection><collection>Engineering Collection</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Namashivayam, Naveen</au><au>Kandalla, Krishna</au><au>White, James B</au><au>Kaplan, Larry</au><au>Pagel, Mark</au><format>book</format><genre>document</genre><ristype>GEN</ristype><atitle>Exploring Fully Offloaded GPU Stream-Aware Message Passing</atitle><jtitle>arXiv.org</jtitle><date>2023-06-27</date><risdate>2023</risdate><eissn>2331-8422</eissn><abstract>Modern heterogeneous supercomputing systems are comprised of CPUs, GPUs, and high-speed network interconnects. Communication libraries supporting efficient data transfers involving memory buffers from the GPU memory typically require the CPU to orchestrate the data transfer operations. A new offload-friendly communication strategy, stream-triggered (ST) communication, was explored to allow offloading the synchronization and data movement operations from the CPU to the GPU. A Message Passing Interface (MPI) one-sided active target synchronization based implementation was used as an exemplar to illustrate the proposed strategy. A latency-sensitive nearest neighbor microbenchmark was used to explore the various performance aspects of the implementation. The offloaded implementation shows significant on-node performance advantages over standard MPI active RMA (36%) and point-to-point (61%) communication. The current multi-node improvement is less (23% faster than standard active RMA but 11% slower than point-to-point), but plans are in progress to purse further improvements.</abstract><cop>Ithaca</cop><pub>Cornell University Library, arXiv.org</pub><oa>free_for_read</oa></addata></record> |
fulltext | fulltext |
identifier | EISSN: 2331-8422 |
ispartof | arXiv.org, 2023-06 |
issn | 2331-8422 |
language | eng |
recordid | cdi_proquest_journals_2831118379 |
source | Free E- Journals |
subjects | Data transfer (computers) Message passing Synchronism |
title | Exploring Fully Offloaded GPU Stream-Aware Message Passing |
url | https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2024-12-30T08%3A42%3A52IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest&rft_val_fmt=info:ofi/fmt:kev:mtx:book&rft.genre=document&rft.atitle=Exploring%20Fully%20Offloaded%20GPU%20Stream-Aware%20Message%20Passing&rft.jtitle=arXiv.org&rft.au=Namashivayam,%20Naveen&rft.date=2023-06-27&rft.eissn=2331-8422&rft_id=info:doi/&rft_dat=%3Cproquest%3E2831118379%3C/proquest%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_pqid=2831118379&rft_id=info:pmid/&rfr_iscdi=true |