Energy-efficient interconnect via Router Parking
The increase in on-chip core counts in Chip Multi Processors (CMPs) has led to the adoption of interconnects such as Mesh and Torus, which consume an increasing fraction of the chip power. Moreover, as technology and voltage continue to scale down, static power consumes a larger fraction of the tota...
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 | 519 |
---|---|
container_issue | |
container_start_page | 508 |
container_title | |
container_volume | |
creator | Samih, A. Ren Wang Krishna, A. Maciocco, C. Tai, C. Solihin, Y. |
description | The increase in on-chip core counts in Chip Multi Processors (CMPs) has led to the adoption of interconnects such as Mesh and Torus, which consume an increasing fraction of the chip power. Moreover, as technology and voltage continue to scale down, static power consumes a larger fraction of the total power; reducing it is increasingly important for energy proportional computing. Currently, processor designers strive to send under-utilized cores into deep sleep states in order to reduce idling power and improve overall energy efficiency. However, even in state-of-the-art CMP designs, when a core goes to sleep the router attached to it remains active in order to continue packet forwarding. In this paper, we propose Router Parking - selectively power-gating routers attached to parked cores. Router Parking ensures that network connectivity is maintained, and limits the average interconnect latency impact of packet detouring around parked routers. We present two Router Parking algorithms - an aggressive approach to park as many routers as possible, and a conservative approach that parks a limited set of routers in order to keep the impact on latency increase minimal. Further, we propose an adaptive policy to choose between the two algorithms at run-time. We evaluate our algorithms using both synthetic traffic as well as real workloads taken from SPEC CPU2006 and PARSEC 2.1 benchmark suites. Our evaluation results show that Router Parking can achieve significant savings in the total interconnect energy (average of 32%, 40% and 41% for the synthetic, SPEC CPU2006, and PARSEC 2.1 workloads, respectively). |
doi_str_mv | 10.1109/HPCA.2013.6522345 |
format | Conference Proceeding |
fullrecord | <record><control><sourceid>ieee_6IE</sourceid><recordid>TN_cdi_ieee_primary_6522345</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><ieee_id>6522345</ieee_id><sourcerecordid>6522345</sourcerecordid><originalsourceid>FETCH-LOGICAL-i175t-45e44daa1b9ed5542a63e7a9e0a919e89c72bbe40bc199cd4d90acd8427e5fa43</originalsourceid><addsrcrecordid>eNo1j9tKw0AURccbmNZ-gPiSH0g8czmZOY8ltFYoWETBtzKZnJTxMpUkCv17BevThrVgwRbiWkIpJdDtalPPSwVSlxUqpQ2eiBlZJ01lNaKz6lRkSltXKNAvZ2LyL1Cdi0yihgIc2UsxGYZXAFCEMhOwSNzvDgV3XQyR05jHNHIf9ilxGPPv6PPH_dcvyTe-f4tpdyUuOv8-8Oy4U_G8XDzVq2L9cHdfz9dFlBbHwiAb03ovG-IW0ShfabaeGDxJYkfBqqZhA02QRKE1LYEPrTPKMnbe6Km4-etGZt5-9vHD94ft8bn-ATfqSIw</addsrcrecordid><sourcetype>Publisher</sourcetype><iscdi>true</iscdi><recordtype>conference_proceeding</recordtype></control><display><type>conference_proceeding</type><title>Energy-efficient interconnect via Router Parking</title><source>IEEE Electronic Library (IEL) Conference Proceedings</source><creator>Samih, A. ; Ren Wang ; Krishna, A. ; Maciocco, C. ; Tai, C. ; Solihin, Y.</creator><creatorcontrib>Samih, A. ; Ren Wang ; Krishna, A. ; Maciocco, C. ; Tai, C. ; Solihin, Y.</creatorcontrib><description>The increase in on-chip core counts in Chip Multi Processors (CMPs) has led to the adoption of interconnects such as Mesh and Torus, which consume an increasing fraction of the chip power. Moreover, as technology and voltage continue to scale down, static power consumes a larger fraction of the total power; reducing it is increasingly important for energy proportional computing. Currently, processor designers strive to send under-utilized cores into deep sleep states in order to reduce idling power and improve overall energy efficiency. However, even in state-of-the-art CMP designs, when a core goes to sleep the router attached to it remains active in order to continue packet forwarding. In this paper, we propose Router Parking - selectively power-gating routers attached to parked cores. Router Parking ensures that network connectivity is maintained, and limits the average interconnect latency impact of packet detouring around parked routers. We present two Router Parking algorithms - an aggressive approach to park as many routers as possible, and a conservative approach that parks a limited set of routers in order to keep the impact on latency increase minimal. Further, we propose an adaptive policy to choose between the two algorithms at run-time. We evaluate our algorithms using both synthetic traffic as well as real workloads taken from SPEC CPU2006 and PARSEC 2.1 benchmark suites. Our evaluation results show that Router Parking can achieve significant savings in the total interconnect energy (average of 32%, 40% and 41% for the synthetic, SPEC CPU2006, and PARSEC 2.1 workloads, respectively).</description><identifier>ISSN: 1530-0897</identifier><identifier>ISBN: 1467355852</identifier><identifier>ISBN: 9781467355858</identifier><identifier>EISSN: 2378-203X</identifier><identifier>EISBN: 9781467355872</identifier><identifier>EISBN: 1467355860</identifier><identifier>EISBN: 1467355879</identifier><identifier>EISBN: 9781467355865</identifier><identifier>DOI: 10.1109/HPCA.2013.6522345</identifier><language>eng</language><publisher>IEEE</publisher><subject>Fabrics ; Frequency modulation ; Network topology ; Partitioning algorithms ; Routing ; System recovery ; Topology</subject><ispartof>2013 IEEE 19th International Symposium on High Performance Computer Architecture (HPCA), 2013, p.508-519</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/6522345$$EHTML$$P50$$Gieee$$H</linktohtml><link.rule.ids>309,310,780,784,789,790,2058,27925,54920</link.rule.ids><linktorsrc>$$Uhttps://ieeexplore.ieee.org/document/6522345$$EView_record_in_IEEE$$FView_record_in_$$GIEEE</linktorsrc></links><search><creatorcontrib>Samih, A.</creatorcontrib><creatorcontrib>Ren Wang</creatorcontrib><creatorcontrib>Krishna, A.</creatorcontrib><creatorcontrib>Maciocco, C.</creatorcontrib><creatorcontrib>Tai, C.</creatorcontrib><creatorcontrib>Solihin, Y.</creatorcontrib><title>Energy-efficient interconnect via Router Parking</title><title>2013 IEEE 19th International Symposium on High Performance Computer Architecture (HPCA)</title><addtitle>HPCA</addtitle><description>The increase in on-chip core counts in Chip Multi Processors (CMPs) has led to the adoption of interconnects such as Mesh and Torus, which consume an increasing fraction of the chip power. Moreover, as technology and voltage continue to scale down, static power consumes a larger fraction of the total power; reducing it is increasingly important for energy proportional computing. Currently, processor designers strive to send under-utilized cores into deep sleep states in order to reduce idling power and improve overall energy efficiency. However, even in state-of-the-art CMP designs, when a core goes to sleep the router attached to it remains active in order to continue packet forwarding. In this paper, we propose Router Parking - selectively power-gating routers attached to parked cores. Router Parking ensures that network connectivity is maintained, and limits the average interconnect latency impact of packet detouring around parked routers. We present two Router Parking algorithms - an aggressive approach to park as many routers as possible, and a conservative approach that parks a limited set of routers in order to keep the impact on latency increase minimal. Further, we propose an adaptive policy to choose between the two algorithms at run-time. We evaluate our algorithms using both synthetic traffic as well as real workloads taken from SPEC CPU2006 and PARSEC 2.1 benchmark suites. Our evaluation results show that Router Parking can achieve significant savings in the total interconnect energy (average of 32%, 40% and 41% for the synthetic, SPEC CPU2006, and PARSEC 2.1 workloads, respectively).</description><subject>Fabrics</subject><subject>Frequency modulation</subject><subject>Network topology</subject><subject>Partitioning algorithms</subject><subject>Routing</subject><subject>System recovery</subject><subject>Topology</subject><issn>1530-0897</issn><issn>2378-203X</issn><isbn>1467355852</isbn><isbn>9781467355858</isbn><isbn>9781467355872</isbn><isbn>1467355860</isbn><isbn>1467355879</isbn><isbn>9781467355865</isbn><fulltext>true</fulltext><rsrctype>conference_proceeding</rsrctype><creationdate>2013</creationdate><recordtype>conference_proceeding</recordtype><sourceid>6IE</sourceid><sourceid>RIE</sourceid><recordid>eNo1j9tKw0AURccbmNZ-gPiSH0g8czmZOY8ltFYoWETBtzKZnJTxMpUkCv17BevThrVgwRbiWkIpJdDtalPPSwVSlxUqpQ2eiBlZJ01lNaKz6lRkSltXKNAvZ2LyL1Cdi0yihgIc2UsxGYZXAFCEMhOwSNzvDgV3XQyR05jHNHIf9ilxGPPv6PPH_dcvyTe-f4tpdyUuOv8-8Oy4U_G8XDzVq2L9cHdfz9dFlBbHwiAb03ovG-IW0ShfabaeGDxJYkfBqqZhA02QRKE1LYEPrTPKMnbe6Km4-etGZt5-9vHD94ft8bn-ATfqSIw</recordid><startdate>201302</startdate><enddate>201302</enddate><creator>Samih, A.</creator><creator>Ren Wang</creator><creator>Krishna, A.</creator><creator>Maciocco, C.</creator><creator>Tai, C.</creator><creator>Solihin, Y.</creator><general>IEEE</general><scope>6IE</scope><scope>6IL</scope><scope>CBEJK</scope><scope>RIE</scope><scope>RIL</scope></search><sort><creationdate>201302</creationdate><title>Energy-efficient interconnect via Router Parking</title><author>Samih, A. ; Ren Wang ; Krishna, A. ; Maciocco, C. ; Tai, C. ; Solihin, Y.</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-i175t-45e44daa1b9ed5542a63e7a9e0a919e89c72bbe40bc199cd4d90acd8427e5fa43</frbrgroupid><rsrctype>conference_proceedings</rsrctype><prefilter>conference_proceedings</prefilter><language>eng</language><creationdate>2013</creationdate><topic>Fabrics</topic><topic>Frequency modulation</topic><topic>Network topology</topic><topic>Partitioning algorithms</topic><topic>Routing</topic><topic>System recovery</topic><topic>Topology</topic><toplevel>online_resources</toplevel><creatorcontrib>Samih, A.</creatorcontrib><creatorcontrib>Ren Wang</creatorcontrib><creatorcontrib>Krishna, A.</creatorcontrib><creatorcontrib>Maciocco, C.</creatorcontrib><creatorcontrib>Tai, C.</creatorcontrib><creatorcontrib>Solihin, Y.</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 Explore</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>Samih, A.</au><au>Ren Wang</au><au>Krishna, A.</au><au>Maciocco, C.</au><au>Tai, C.</au><au>Solihin, Y.</au><format>book</format><genre>proceeding</genre><ristype>CONF</ristype><atitle>Energy-efficient interconnect via Router Parking</atitle><btitle>2013 IEEE 19th International Symposium on High Performance Computer Architecture (HPCA)</btitle><stitle>HPCA</stitle><date>2013-02</date><risdate>2013</risdate><spage>508</spage><epage>519</epage><pages>508-519</pages><issn>1530-0897</issn><eissn>2378-203X</eissn><isbn>1467355852</isbn><isbn>9781467355858</isbn><eisbn>9781467355872</eisbn><eisbn>1467355860</eisbn><eisbn>1467355879</eisbn><eisbn>9781467355865</eisbn><abstract>The increase in on-chip core counts in Chip Multi Processors (CMPs) has led to the adoption of interconnects such as Mesh and Torus, which consume an increasing fraction of the chip power. Moreover, as technology and voltage continue to scale down, static power consumes a larger fraction of the total power; reducing it is increasingly important for energy proportional computing. Currently, processor designers strive to send under-utilized cores into deep sleep states in order to reduce idling power and improve overall energy efficiency. However, even in state-of-the-art CMP designs, when a core goes to sleep the router attached to it remains active in order to continue packet forwarding. In this paper, we propose Router Parking - selectively power-gating routers attached to parked cores. Router Parking ensures that network connectivity is maintained, and limits the average interconnect latency impact of packet detouring around parked routers. We present two Router Parking algorithms - an aggressive approach to park as many routers as possible, and a conservative approach that parks a limited set of routers in order to keep the impact on latency increase minimal. Further, we propose an adaptive policy to choose between the two algorithms at run-time. We evaluate our algorithms using both synthetic traffic as well as real workloads taken from SPEC CPU2006 and PARSEC 2.1 benchmark suites. Our evaluation results show that Router Parking can achieve significant savings in the total interconnect energy (average of 32%, 40% and 41% for the synthetic, SPEC CPU2006, and PARSEC 2.1 workloads, respectively).</abstract><pub>IEEE</pub><doi>10.1109/HPCA.2013.6522345</doi><tpages>12</tpages></addata></record> |
fulltext | fulltext_linktorsrc |
identifier | ISSN: 1530-0897 |
ispartof | 2013 IEEE 19th International Symposium on High Performance Computer Architecture (HPCA), 2013, p.508-519 |
issn | 1530-0897 2378-203X |
language | eng |
recordid | cdi_ieee_primary_6522345 |
source | IEEE Electronic Library (IEL) Conference Proceedings |
subjects | Fabrics Frequency modulation Network topology Partitioning algorithms Routing System recovery Topology |
title | Energy-efficient interconnect via Router Parking |
url | https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-03T17%3A38%3A44IST&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=Energy-efficient%20interconnect%20via%20Router%20Parking&rft.btitle=2013%20IEEE%2019th%20International%20Symposium%20on%20High%20Performance%20Computer%20Architecture%20(HPCA)&rft.au=Samih,%20A.&rft.date=2013-02&rft.spage=508&rft.epage=519&rft.pages=508-519&rft.issn=1530-0897&rft.eissn=2378-203X&rft.isbn=1467355852&rft.isbn_list=9781467355858&rft_id=info:doi/10.1109/HPCA.2013.6522345&rft_dat=%3Cieee_6IE%3E6522345%3C/ieee_6IE%3E%3Curl%3E%3C/url%3E&rft.eisbn=9781467355872&rft.eisbn_list=1467355860&rft.eisbn_list=1467355879&rft.eisbn_list=9781467355865&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_id=info:pmid/&rft_ieee_id=6522345&rfr_iscdi=true |