Loading…
Interconnect support for energy efficient and high bandwidth memory access in CMPs
The increasing number of on-chip cores and a limited number of memory controllers pose a critical problem for off-chip memory bandwidth. In this work, we propose an adaptive hybrid switching strategy with a dual crossbar router to provide low latency paths between cache and memory controllers. The p...
Saved in:
Published in: | Sustainable computing informatics and systems 2022-04, Vol.34, p.100720, Article 100720 |
---|---|
Main Authors: | , , , , , |
Format: | Article |
Language: | English |
Subjects: | |
Citations: | Items that this one cites |
Online Access: | Get full text |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
cited_by | |
---|---|
cites | cdi_FETCH-LOGICAL-c255t-f275f6e26cab1444b29bce56043aa8dc17b390dd17228bbe436666860d0597763 |
container_end_page | |
container_issue | |
container_start_page | 100720 |
container_title | Sustainable computing informatics and systems |
container_volume | 34 |
creator | Mondal, Hemanta Kumar Konar, Sarnava Hore, Poulomi Patra, Ramapati Sarkar, Pradipta Deb, Sujay |
description | The increasing number of on-chip cores and a limited number of memory controllers pose a critical problem for off-chip memory bandwidth. In this work, we propose an adaptive hybrid switching strategy with a dual crossbar router to provide low latency paths between cache and memory controllers. The performance is further improved by finding the optimal number and placement of memory controllers using machine learning approach with low overheads. The proposed architecture improves the average bandwidth of the network by 21.03% and reduces the network energy and memory access latency by 12.60% and 20.45%, respectively as compared to the traditional NoC architecture.
•A hybrid switching strategy with dual crossbar routers that allow simultaneous use of both circuit and packet-switched paths.•An optimal number and placement of memory controllers in the network using machine learning approach.•An energy-efficient router architecture using power-efficient drowsy and power-gating techniques.•New experimental evaluations with various application-specific traffic scenarios in terms of throughput, latency, improvements.•A detailed analysis of scalability and a summary of proposed and existing works |
doi_str_mv | 10.1016/j.suscom.2022.100720 |
format | article |
fullrecord | <record><control><sourceid>elsevier_cross</sourceid><recordid>TN_cdi_crossref_primary_10_1016_j_suscom_2022_100720</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><els_id>S2210537922000579</els_id><sourcerecordid>S2210537922000579</sourcerecordid><originalsourceid>FETCH-LOGICAL-c255t-f275f6e26cab1444b29bce56043aa8dc17b390dd17228bbe436666860d0597763</originalsourceid><addsrcrecordid>eNp9kM1KAzEUhbNQsNS-gYu8wNQkk0lmNoIUfwoVRXQdMslNm-IkJZkqfXtTxrVncy8XzuGeD6EbSpaUUHG7X-ZjNnFYMsJYORHJyAWaMUZJ1dSyu0KLnPekqBG0q_kMva_DCMnEEMCMOB8Ph5hG7GLCECBtTxic88ZDGLEOFu_8dof7sv14O-7wAENMJ6yNgZyxD3j18pav0aXTXxkWf3OOPh8fPlbP1eb1ab2631SGNc1YOSYbJ4AJo3vKOe9Z1xtoBOG11q01VPZ1R6ylkrG274HXoqgVxJKmk1LUc8SnXJNizgmcOiQ_6HRSlKgzDrVXEw51xqEmHMV2N9mg_PbtIal87mfA-lQYKBv9_wG_GWdsvQ</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>Interconnect support for energy efficient and high bandwidth memory access in CMPs</title><source>ScienceDirect Freedom Collection</source><creator>Mondal, Hemanta Kumar ; Konar, Sarnava ; Hore, Poulomi ; Patra, Ramapati ; Sarkar, Pradipta ; Deb, Sujay</creator><creatorcontrib>Mondal, Hemanta Kumar ; Konar, Sarnava ; Hore, Poulomi ; Patra, Ramapati ; Sarkar, Pradipta ; Deb, Sujay</creatorcontrib><description>The increasing number of on-chip cores and a limited number of memory controllers pose a critical problem for off-chip memory bandwidth. In this work, we propose an adaptive hybrid switching strategy with a dual crossbar router to provide low latency paths between cache and memory controllers. The performance is further improved by finding the optimal number and placement of memory controllers using machine learning approach with low overheads. The proposed architecture improves the average bandwidth of the network by 21.03% and reduces the network energy and memory access latency by 12.60% and 20.45%, respectively as compared to the traditional NoC architecture.
•A hybrid switching strategy with dual crossbar routers that allow simultaneous use of both circuit and packet-switched paths.•An optimal number and placement of memory controllers in the network using machine learning approach.•An energy-efficient router architecture using power-efficient drowsy and power-gating techniques.•New experimental evaluations with various application-specific traffic scenarios in terms of throughput, latency, improvements.•A detailed analysis of scalability and a summary of proposed and existing works</description><identifier>ISSN: 2210-5379</identifier><identifier>DOI: 10.1016/j.suscom.2022.100720</identifier><language>eng</language><publisher>Elsevier Inc</publisher><subject>Drowsy circuit and Power gating ; Hybrid switching ; Machine learning ; Network-on-Chip ; Optimal placement</subject><ispartof>Sustainable computing informatics and systems, 2022-04, Vol.34, p.100720, Article 100720</ispartof><rights>2022 Elsevier Inc.</rights><woscitedreferencessubscribed>false</woscitedreferencessubscribed><cites>FETCH-LOGICAL-c255t-f275f6e26cab1444b29bce56043aa8dc17b390dd17228bbe436666860d0597763</cites></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>314,776,780,27903,27904</link.rule.ids></links><search><creatorcontrib>Mondal, Hemanta Kumar</creatorcontrib><creatorcontrib>Konar, Sarnava</creatorcontrib><creatorcontrib>Hore, Poulomi</creatorcontrib><creatorcontrib>Patra, Ramapati</creatorcontrib><creatorcontrib>Sarkar, Pradipta</creatorcontrib><creatorcontrib>Deb, Sujay</creatorcontrib><title>Interconnect support for energy efficient and high bandwidth memory access in CMPs</title><title>Sustainable computing informatics and systems</title><description>The increasing number of on-chip cores and a limited number of memory controllers pose a critical problem for off-chip memory bandwidth. In this work, we propose an adaptive hybrid switching strategy with a dual crossbar router to provide low latency paths between cache and memory controllers. The performance is further improved by finding the optimal number and placement of memory controllers using machine learning approach with low overheads. The proposed architecture improves the average bandwidth of the network by 21.03% and reduces the network energy and memory access latency by 12.60% and 20.45%, respectively as compared to the traditional NoC architecture.
•A hybrid switching strategy with dual crossbar routers that allow simultaneous use of both circuit and packet-switched paths.•An optimal number and placement of memory controllers in the network using machine learning approach.•An energy-efficient router architecture using power-efficient drowsy and power-gating techniques.•New experimental evaluations with various application-specific traffic scenarios in terms of throughput, latency, improvements.•A detailed analysis of scalability and a summary of proposed and existing works</description><subject>Drowsy circuit and Power gating</subject><subject>Hybrid switching</subject><subject>Machine learning</subject><subject>Network-on-Chip</subject><subject>Optimal placement</subject><issn>2210-5379</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2022</creationdate><recordtype>article</recordtype><recordid>eNp9kM1KAzEUhbNQsNS-gYu8wNQkk0lmNoIUfwoVRXQdMslNm-IkJZkqfXtTxrVncy8XzuGeD6EbSpaUUHG7X-ZjNnFYMsJYORHJyAWaMUZJ1dSyu0KLnPekqBG0q_kMva_DCMnEEMCMOB8Ph5hG7GLCECBtTxic88ZDGLEOFu_8dof7sv14O-7wAENMJ6yNgZyxD3j18pav0aXTXxkWf3OOPh8fPlbP1eb1ab2631SGNc1YOSYbJ4AJo3vKOe9Z1xtoBOG11q01VPZ1R6ylkrG274HXoqgVxJKmk1LUc8SnXJNizgmcOiQ_6HRSlKgzDrVXEw51xqEmHMV2N9mg_PbtIal87mfA-lQYKBv9_wG_GWdsvQ</recordid><startdate>202204</startdate><enddate>202204</enddate><creator>Mondal, Hemanta Kumar</creator><creator>Konar, Sarnava</creator><creator>Hore, Poulomi</creator><creator>Patra, Ramapati</creator><creator>Sarkar, Pradipta</creator><creator>Deb, Sujay</creator><general>Elsevier Inc</general><scope>AAYXX</scope><scope>CITATION</scope></search><sort><creationdate>202204</creationdate><title>Interconnect support for energy efficient and high bandwidth memory access in CMPs</title><author>Mondal, Hemanta Kumar ; Konar, Sarnava ; Hore, Poulomi ; Patra, Ramapati ; Sarkar, Pradipta ; Deb, Sujay</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-c255t-f275f6e26cab1444b29bce56043aa8dc17b390dd17228bbe436666860d0597763</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2022</creationdate><topic>Drowsy circuit and Power gating</topic><topic>Hybrid switching</topic><topic>Machine learning</topic><topic>Network-on-Chip</topic><topic>Optimal placement</topic><toplevel>online_resources</toplevel><creatorcontrib>Mondal, Hemanta Kumar</creatorcontrib><creatorcontrib>Konar, Sarnava</creatorcontrib><creatorcontrib>Hore, Poulomi</creatorcontrib><creatorcontrib>Patra, Ramapati</creatorcontrib><creatorcontrib>Sarkar, Pradipta</creatorcontrib><creatorcontrib>Deb, Sujay</creatorcontrib><collection>CrossRef</collection><jtitle>Sustainable computing informatics and systems</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Mondal, Hemanta Kumar</au><au>Konar, Sarnava</au><au>Hore, Poulomi</au><au>Patra, Ramapati</au><au>Sarkar, Pradipta</au><au>Deb, Sujay</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Interconnect support for energy efficient and high bandwidth memory access in CMPs</atitle><jtitle>Sustainable computing informatics and systems</jtitle><date>2022-04</date><risdate>2022</risdate><volume>34</volume><spage>100720</spage><pages>100720-</pages><artnum>100720</artnum><issn>2210-5379</issn><abstract>The increasing number of on-chip cores and a limited number of memory controllers pose a critical problem for off-chip memory bandwidth. In this work, we propose an adaptive hybrid switching strategy with a dual crossbar router to provide low latency paths between cache and memory controllers. The performance is further improved by finding the optimal number and placement of memory controllers using machine learning approach with low overheads. The proposed architecture improves the average bandwidth of the network by 21.03% and reduces the network energy and memory access latency by 12.60% and 20.45%, respectively as compared to the traditional NoC architecture.
•A hybrid switching strategy with dual crossbar routers that allow simultaneous use of both circuit and packet-switched paths.•An optimal number and placement of memory controllers in the network using machine learning approach.•An energy-efficient router architecture using power-efficient drowsy and power-gating techniques.•New experimental evaluations with various application-specific traffic scenarios in terms of throughput, latency, improvements.•A detailed analysis of scalability and a summary of proposed and existing works</abstract><pub>Elsevier Inc</pub><doi>10.1016/j.suscom.2022.100720</doi></addata></record> |
fulltext | fulltext |
identifier | ISSN: 2210-5379 |
ispartof | Sustainable computing informatics and systems, 2022-04, Vol.34, p.100720, Article 100720 |
issn | 2210-5379 |
language | eng |
recordid | cdi_crossref_primary_10_1016_j_suscom_2022_100720 |
source | ScienceDirect Freedom Collection |
subjects | Drowsy circuit and Power gating Hybrid switching Machine learning Network-on-Chip Optimal placement |
title | Interconnect support for energy efficient and high bandwidth memory access in CMPs |
url | http://sfxeu10.hosted.exlibrisgroup.com/loughborough?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-24T03%3A27%3A52IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-elsevier_cross&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Interconnect%20support%20for%20energy%20efficient%20and%20high%20bandwidth%20memory%20access%20in%20CMPs&rft.jtitle=Sustainable%20computing%20informatics%20and%20systems&rft.au=Mondal,%20Hemanta%20Kumar&rft.date=2022-04&rft.volume=34&rft.spage=100720&rft.pages=100720-&rft.artnum=100720&rft.issn=2210-5379&rft_id=info:doi/10.1016/j.suscom.2022.100720&rft_dat=%3Celsevier_cross%3ES2210537922000579%3C/elsevier_cross%3E%3Cgrp_id%3Ecdi_FETCH-LOGICAL-c255t-f275f6e26cab1444b29bce56043aa8dc17b390dd17228bbe436666860d0597763%3C/grp_id%3E%3Coa%3E%3C/oa%3E%3Curl%3E%3C/url%3E&rft_id=info:oai/&rft_id=info:pmid/&rfr_iscdi=true |