Delta Send-Recv for Dynamic Pipelining in MPI Programs
Pipelining is necessary for efficient do-across parallelism but the use is difficult to automate because it requires send-receive analysis and loop blocking in both sender and receiver code. The blocking factor is statically chosen. This paper presents a new interface called delta send-recv. Through...
Gespeichert in:
| Hauptverfasser: | , , , |
|---|---|
| Format: | Tagungsbericht |
| Sprache: | eng |
| Schlagworte: | |
| Online-Zugang: | Volltext bestellen |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| container_end_page | 392 |
|---|---|
| container_issue | |
| container_start_page | 384 |
| container_title | |
| container_volume | |
| creator | Bao, Bin Ding, Chen Gao, Yaoqing Archambault, Roch |
| description | Pipelining is necessary for efficient do-across parallelism but the use is difficult to automate because it requires send-receive analysis and loop blocking in both sender and receiver code. The blocking factor is statically chosen. This paper presents a new interface called delta send-recv. Through compiler and run-time support, it enables dynamic pipelining. In program code, the interface is used to mark the related computation and communication. There is no need to restructure the computation code or compose multiple messages. At run time, the message size is dynamically determined, and multiple pipelines are chained among all tasks that participate in the delta communication. The new system is tested on kernel and reduced NAS benchmarks to show that it simplifies message-passing programming and improves program performance. |
| doi_str_mv | 10.1109/CCGrid.2012.113 |
| format | Conference Proceeding |
| fullrecord | <record><control><sourceid>acm_6IE</sourceid><recordid>TN_cdi_ieee_primary_6217445</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><ieee_id>6217445</ieee_id><sourcerecordid>acm_books_10_1109_CCGrid_2012_113</sourcerecordid><originalsourceid>FETCH-LOGICAL-a2033-2d06200515d8c45513b7c47223ecf3bba748ddc9c191f8807d38e7345615bacc3</originalsourceid><addsrcrecordid>eNqNkLlLxEAYxUdEUNetLWymtEn85k5Kyeq6sGLwqIe5sgzmWCYi7H9vllhYWj3ewSt-CF0TyAmB8q6q1in6nAKhU8BO0CUoWQouS1Ke_jXnaDmO0QKVSgol4QLJVWi_DH4Lvc9eg_vGzZDw6tCbLjpcx31oYx_7HY49fq43uE7DLpluvEJnjWnHsPzVBfp4fHivnrLty3pT3W8zQ4GxjHqQFEAQ4QvHhSDMKscVpSy4hllrFC-8d6UjJWmKApRnRVCMC0mENc6xBbqZf2MIQe9T7Ew6aEmJ4lxM7e3cGtdpOwyfoyagj0j0jEQfkUwBm6b5P6faphga9gNKF18M</addsrcrecordid><sourcetype>Publisher</sourcetype><iscdi>true</iscdi><recordtype>conference_proceeding</recordtype></control><display><type>conference_proceeding</type><title>Delta Send-Recv for Dynamic Pipelining in MPI Programs</title><source>IEEE Electronic Library (IEL) Conference Proceedings</source><creator>Bao, Bin ; Ding, Chen ; Gao, Yaoqing ; Archambault, Roch</creator><creatorcontrib>Bao, Bin ; Ding, Chen ; Gao, Yaoqing ; Archambault, Roch</creatorcontrib><description>Pipelining is necessary for efficient do-across parallelism but the use is difficult to automate because it requires send-receive analysis and loop blocking in both sender and receiver code. The blocking factor is statically chosen. This paper presents a new interface called delta send-recv. Through compiler and run-time support, it enables dynamic pipelining. In program code, the interface is used to mark the related computation and communication. There is no need to restructure the computation code or compose multiple messages. At run time, the message size is dynamically determined, and multiple pipelines are chained among all tasks that participate in the delta communication. The new system is tested on kernel and reduced NAS benchmarks to show that it simplifies message-passing programming and improves program performance.</description><identifier>ISBN: 0769546919</identifier><identifier>ISBN: 9780769546919</identifier><identifier>ISBN: 1467313955</identifier><identifier>ISBN: 9781467313957</identifier><identifier>EISBN: 0769546919</identifier><identifier>EISBN: 9780769546919</identifier><identifier>DOI: 10.1109/CCGrid.2012.113</identifier><language>eng</language><publisher>Washington, DC, USA: IEEE Computer Society</publisher><subject>Arrays ; communication-computation overlapping ; dynamic pipelining ; General and reference ; General and reference -- Cross-computing tools and techniques ; General and reference -- Cross-computing tools and techniques -- Performance ; Kernel ; Libraries ; Monitoring ; MPI ; Pipeline processing ; Receivers ; Social and professional topics ; Social and professional topics -- Professional topics ; Social and professional topics -- Professional topics -- Computing profession ; Social and professional topics -- Professional topics -- Computing profession -- Testing, certification and licensing</subject><ispartof>2012 12th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing (ccgrid 2012), 2012, p.384-392</ispartof><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><linktohtml>$$Uhttps://ieeexplore.ieee.org/document/6217445$$EHTML$$P50$$Gieee$$H</linktohtml><link.rule.ids>309,310,780,784,789,790,2058,27925,54920</link.rule.ids><linktorsrc>$$Uhttps://ieeexplore.ieee.org/document/6217445$$EView_record_in_IEEE$$FView_record_in_$$GIEEE</linktorsrc></links><search><creatorcontrib>Bao, Bin</creatorcontrib><creatorcontrib>Ding, Chen</creatorcontrib><creatorcontrib>Gao, Yaoqing</creatorcontrib><creatorcontrib>Archambault, Roch</creatorcontrib><title>Delta Send-Recv for Dynamic Pipelining in MPI Programs</title><title>2012 12th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing (ccgrid 2012)</title><addtitle>ccgrid</addtitle><description>Pipelining is necessary for efficient do-across parallelism but the use is difficult to automate because it requires send-receive analysis and loop blocking in both sender and receiver code. The blocking factor is statically chosen. This paper presents a new interface called delta send-recv. Through compiler and run-time support, it enables dynamic pipelining. In program code, the interface is used to mark the related computation and communication. There is no need to restructure the computation code or compose multiple messages. At run time, the message size is dynamically determined, and multiple pipelines are chained among all tasks that participate in the delta communication. The new system is tested on kernel and reduced NAS benchmarks to show that it simplifies message-passing programming and improves program performance.</description><subject>Arrays</subject><subject>communication-computation overlapping</subject><subject>dynamic pipelining</subject><subject>General and reference</subject><subject>General and reference -- Cross-computing tools and techniques</subject><subject>General and reference -- Cross-computing tools and techniques -- Performance</subject><subject>Kernel</subject><subject>Libraries</subject><subject>Monitoring</subject><subject>MPI</subject><subject>Pipeline processing</subject><subject>Receivers</subject><subject>Social and professional topics</subject><subject>Social and professional topics -- Professional topics</subject><subject>Social and professional topics -- Professional topics -- Computing profession</subject><subject>Social and professional topics -- Professional topics -- Computing profession -- Testing, certification and licensing</subject><isbn>0769546919</isbn><isbn>9780769546919</isbn><isbn>1467313955</isbn><isbn>9781467313957</isbn><isbn>0769546919</isbn><isbn>9780769546919</isbn><fulltext>true</fulltext><rsrctype>conference_proceeding</rsrctype><creationdate>2012</creationdate><recordtype>conference_proceeding</recordtype><sourceid>6IE</sourceid><sourceid>RIE</sourceid><recordid>eNqNkLlLxEAYxUdEUNetLWymtEn85k5Kyeq6sGLwqIe5sgzmWCYi7H9vllhYWj3ewSt-CF0TyAmB8q6q1in6nAKhU8BO0CUoWQouS1Ke_jXnaDmO0QKVSgol4QLJVWi_DH4Lvc9eg_vGzZDw6tCbLjpcx31oYx_7HY49fq43uE7DLpluvEJnjWnHsPzVBfp4fHivnrLty3pT3W8zQ4GxjHqQFEAQ4QvHhSDMKscVpSy4hllrFC-8d6UjJWmKApRnRVCMC0mENc6xBbqZf2MIQe9T7Ew6aEmJ4lxM7e3cGtdpOwyfoyagj0j0jEQfkUwBm6b5P6faphga9gNKF18M</recordid><startdate>20120513</startdate><enddate>20120513</enddate><creator>Bao, Bin</creator><creator>Ding, Chen</creator><creator>Gao, Yaoqing</creator><creator>Archambault, Roch</creator><general>IEEE Computer Society</general><general>IEEE</general><scope>6IE</scope><scope>6IL</scope><scope>CBEJK</scope><scope>RIE</scope><scope>RIL</scope></search><sort><creationdate>20120513</creationdate><title>Delta Send-Recv for Dynamic Pipelining in MPI Programs</title><author>Bao, Bin ; Ding, Chen ; Gao, Yaoqing ; Archambault, Roch</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-a2033-2d06200515d8c45513b7c47223ecf3bba748ddc9c191f8807d38e7345615bacc3</frbrgroupid><rsrctype>conference_proceedings</rsrctype><prefilter>conference_proceedings</prefilter><language>eng</language><creationdate>2012</creationdate><topic>Arrays</topic><topic>communication-computation overlapping</topic><topic>dynamic pipelining</topic><topic>General and reference</topic><topic>General and reference -- Cross-computing tools and techniques</topic><topic>General and reference -- Cross-computing tools and techniques -- Performance</topic><topic>Kernel</topic><topic>Libraries</topic><topic>Monitoring</topic><topic>MPI</topic><topic>Pipeline processing</topic><topic>Receivers</topic><topic>Social and professional topics</topic><topic>Social and professional topics -- Professional topics</topic><topic>Social and professional topics -- Professional topics -- Computing profession</topic><topic>Social and professional topics -- Professional topics -- Computing profession -- Testing, certification and licensing</topic><toplevel>online_resources</toplevel><creatorcontrib>Bao, Bin</creatorcontrib><creatorcontrib>Ding, Chen</creatorcontrib><creatorcontrib>Gao, Yaoqing</creatorcontrib><creatorcontrib>Archambault, Roch</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>Bao, Bin</au><au>Ding, Chen</au><au>Gao, Yaoqing</au><au>Archambault, Roch</au><format>book</format><genre>proceeding</genre><ristype>CONF</ristype><atitle>Delta Send-Recv for Dynamic Pipelining in MPI Programs</atitle><btitle>2012 12th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing (ccgrid 2012)</btitle><stitle>ccgrid</stitle><date>2012-05-13</date><risdate>2012</risdate><spage>384</spage><epage>392</epage><pages>384-392</pages><isbn>0769546919</isbn><isbn>9780769546919</isbn><isbn>1467313955</isbn><isbn>9781467313957</isbn><eisbn>0769546919</eisbn><eisbn>9780769546919</eisbn><abstract>Pipelining is necessary for efficient do-across parallelism but the use is difficult to automate because it requires send-receive analysis and loop blocking in both sender and receiver code. The blocking factor is statically chosen. This paper presents a new interface called delta send-recv. Through compiler and run-time support, it enables dynamic pipelining. In program code, the interface is used to mark the related computation and communication. There is no need to restructure the computation code or compose multiple messages. At run time, the message size is dynamically determined, and multiple pipelines are chained among all tasks that participate in the delta communication. The new system is tested on kernel and reduced NAS benchmarks to show that it simplifies message-passing programming and improves program performance.</abstract><cop>Washington, DC, USA</cop><pub>IEEE Computer Society</pub><doi>10.1109/CCGrid.2012.113</doi><tpages>9</tpages><oa>free_for_read</oa></addata></record> |
| fulltext | fulltext_linktorsrc |
| identifier | ISBN: 0769546919 |
| ispartof | 2012 12th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing (ccgrid 2012), 2012, p.384-392 |
| issn | |
| language | eng |
| recordid | cdi_ieee_primary_6217445 |
| source | IEEE Electronic Library (IEL) Conference Proceedings |
| subjects | Arrays communication-computation overlapping dynamic pipelining General and reference General and reference -- Cross-computing tools and techniques General and reference -- Cross-computing tools and techniques -- Performance Kernel Libraries Monitoring MPI Pipeline processing Receivers Social and professional topics Social and professional topics -- Professional topics Social and professional topics -- Professional topics -- Computing profession Social and professional topics -- Professional topics -- Computing profession -- Testing, certification and licensing |
| title | Delta Send-Recv for Dynamic Pipelining in MPI Programs |
| url | https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2024-12-22T10%3A44%3A31IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-acm_6IE&rft_val_fmt=info:ofi/fmt:kev:mtx:book&rft.genre=proceeding&rft.atitle=Delta%20Send-Recv%20for%20Dynamic%20Pipelining%20in%20MPI%20Programs&rft.btitle=2012%2012th%20IEEE/ACM%20International%20Symposium%20on%20Cluster,%20Cloud%20and%20Grid%20Computing%20(ccgrid%202012)&rft.au=Bao,%20Bin&rft.date=2012-05-13&rft.spage=384&rft.epage=392&rft.pages=384-392&rft.isbn=0769546919&rft.isbn_list=9780769546919&rft.isbn_list=1467313955&rft.isbn_list=9781467313957&rft_id=info:doi/10.1109/CCGrid.2012.113&rft_dat=%3Cacm_6IE%3Eacm_books_10_1109_CCGrid_2012_113%3C/acm_6IE%3E%3Curl%3E%3C/url%3E&rft.eisbn=0769546919&rft.eisbn_list=9780769546919&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_id=info:pmid/&rft_ieee_id=6217445&rfr_iscdi=true |