Loading…

RAPL in Action: Experiences in Using RAPL for Power Measurements

To improve energy efficiency and comply with the power budgets, it is important to be able to measure the power consumption of cloud computing servers. Intel’s Running Average Power Limit (RAPL) interface is a powerful tool for this purpose. RAPL provides power limiting features and accurate energy...

Full description

Saved in:
Bibliographic Details
Published in:ACM transactions on modeling and performance evaluation of computing systems 2018-06, Vol.3 (2), p.1-26
Main Authors: Khan, Kashif Nizam, Hirki, Mikael, Niemi, Tapio, Nurminen, Jukka K., Ou, Zhonghong
Format: Article
Language:English
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-c187t-1d11a5dd852034376b912449d354113bbf0b9181c407611c0e7b98ac5742be983
container_end_page 26
container_issue 2
container_start_page 1
container_title ACM transactions on modeling and performance evaluation of computing systems
container_volume 3
creator Khan, Kashif Nizam
Hirki, Mikael
Niemi, Tapio
Nurminen, Jukka K.
Ou, Zhonghong
description To improve energy efficiency and comply with the power budgets, it is important to be able to measure the power consumption of cloud computing servers. Intel’s Running Average Power Limit (RAPL) interface is a powerful tool for this purpose. RAPL provides power limiting features and accurate energy readings for CPUs and DRAM, which are easily accessible through different interfaces on large distributed computing systems. Since its introduction, RAPL has been used extensively in power measurement and modeling. However, the advantages and disadvantages of RAPL have not been well investigated yet. To fill this gap, we conduct a series of experiments to disclose the underlying strengths and weaknesses of the RAPL interface by using both customized microbenchmarks and three well-known application level benchmarks: Stream , Stress-ng , and ParFullCMS . Moreover, to make the analysis as realistic as possible, we leverage two production-level power measurement datasets from the Taito , a supercomputing cluster of the Finnish Center of Scientific Computing and also replicate our experiments on Amazon EC2. Our results illustrate different aspects of RAPL and document the findings through comprehensive analysis. Our observations reveal that RAPL readings are highly correlated with plug power, promisingly accurate enough, and have negligible performance overhead. Experimental results suggest RAPL can be a very useful tool to measure and monitor the energy consumption of servers without deploying any complex power meters. We also show that there are still some open issues, such as driver support, non-atomicity of register updates, and unpredictable timings that might weaken the usability of RAPL in certain scenarios. For such scenarios, we pinpoint solutions and workarounds.
doi_str_mv 10.1145/3177754
format article
fullrecord <record><control><sourceid>crossref</sourceid><recordid>TN_cdi_crossref_primary_10_1145_3177754</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>10_1145_3177754</sourcerecordid><originalsourceid>FETCH-LOGICAL-c187t-1d11a5dd852034376b912449d354113bbf0b9181c407611c0e7b98ac5742be983</originalsourceid><addsrcrecordid>eNo9j0FrwkAQRhdpQbGpfyG3ntLOZGYzu8cgVguBirTnkN0kEFFTsl767xtp8PQ9vsODp9QK4RWR9RuhiGieqUVKkiWUsTzcmexcRSEcAQAzEsO0UM-HfF_E3SXO_bXrL0_qsa1OoYmmXarv983XepcUn9uPdV4kHo1cE6wRK13XRqdAPNqdxZTZ1qQZkZxrYXwMegbJED004qypvBZOXWMNLdXLv9cPfQhD05Y_Q3euht8SobyVlFMJ_QE05TXu</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>RAPL in Action: Experiences in Using RAPL for Power Measurements</title><source>Association for Computing Machinery:Jisc Collections:ACM OPEN Journals 2023-2025 (reading list)</source><creator>Khan, Kashif Nizam ; Hirki, Mikael ; Niemi, Tapio ; Nurminen, Jukka K. ; Ou, Zhonghong</creator><creatorcontrib>Khan, Kashif Nizam ; Hirki, Mikael ; Niemi, Tapio ; Nurminen, Jukka K. ; Ou, Zhonghong</creatorcontrib><description>To improve energy efficiency and comply with the power budgets, it is important to be able to measure the power consumption of cloud computing servers. Intel’s Running Average Power Limit (RAPL) interface is a powerful tool for this purpose. RAPL provides power limiting features and accurate energy readings for CPUs and DRAM, which are easily accessible through different interfaces on large distributed computing systems. Since its introduction, RAPL has been used extensively in power measurement and modeling. However, the advantages and disadvantages of RAPL have not been well investigated yet. To fill this gap, we conduct a series of experiments to disclose the underlying strengths and weaknesses of the RAPL interface by using both customized microbenchmarks and three well-known application level benchmarks: Stream , Stress-ng , and ParFullCMS . Moreover, to make the analysis as realistic as possible, we leverage two production-level power measurement datasets from the Taito , a supercomputing cluster of the Finnish Center of Scientific Computing and also replicate our experiments on Amazon EC2. Our results illustrate different aspects of RAPL and document the findings through comprehensive analysis. Our observations reveal that RAPL readings are highly correlated with plug power, promisingly accurate enough, and have negligible performance overhead. Experimental results suggest RAPL can be a very useful tool to measure and monitor the energy consumption of servers without deploying any complex power meters. We also show that there are still some open issues, such as driver support, non-atomicity of register updates, and unpredictable timings that might weaken the usability of RAPL in certain scenarios. For such scenarios, we pinpoint solutions and workarounds.</description><identifier>ISSN: 2376-3639</identifier><identifier>EISSN: 2376-3647</identifier><identifier>DOI: 10.1145/3177754</identifier><language>eng</language><ispartof>ACM transactions on modeling and performance evaluation of computing systems, 2018-06, Vol.3 (2), p.1-26</ispartof><lds50>peer_reviewed</lds50><woscitedreferencessubscribed>false</woscitedreferencessubscribed><cites>FETCH-LOGICAL-c187t-1d11a5dd852034376b912449d354113bbf0b9181c407611c0e7b98ac5742be983</cites><orcidid>0000-0002-3209-5783</orcidid></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>314,780,784,27924,27925</link.rule.ids></links><search><creatorcontrib>Khan, Kashif Nizam</creatorcontrib><creatorcontrib>Hirki, Mikael</creatorcontrib><creatorcontrib>Niemi, Tapio</creatorcontrib><creatorcontrib>Nurminen, Jukka K.</creatorcontrib><creatorcontrib>Ou, Zhonghong</creatorcontrib><title>RAPL in Action: Experiences in Using RAPL for Power Measurements</title><title>ACM transactions on modeling and performance evaluation of computing systems</title><description>To improve energy efficiency and comply with the power budgets, it is important to be able to measure the power consumption of cloud computing servers. Intel’s Running Average Power Limit (RAPL) interface is a powerful tool for this purpose. RAPL provides power limiting features and accurate energy readings for CPUs and DRAM, which are easily accessible through different interfaces on large distributed computing systems. Since its introduction, RAPL has been used extensively in power measurement and modeling. However, the advantages and disadvantages of RAPL have not been well investigated yet. To fill this gap, we conduct a series of experiments to disclose the underlying strengths and weaknesses of the RAPL interface by using both customized microbenchmarks and three well-known application level benchmarks: Stream , Stress-ng , and ParFullCMS . Moreover, to make the analysis as realistic as possible, we leverage two production-level power measurement datasets from the Taito , a supercomputing cluster of the Finnish Center of Scientific Computing and also replicate our experiments on Amazon EC2. Our results illustrate different aspects of RAPL and document the findings through comprehensive analysis. Our observations reveal that RAPL readings are highly correlated with plug power, promisingly accurate enough, and have negligible performance overhead. Experimental results suggest RAPL can be a very useful tool to measure and monitor the energy consumption of servers without deploying any complex power meters. We also show that there are still some open issues, such as driver support, non-atomicity of register updates, and unpredictable timings that might weaken the usability of RAPL in certain scenarios. For such scenarios, we pinpoint solutions and workarounds.</description><issn>2376-3639</issn><issn>2376-3647</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2018</creationdate><recordtype>article</recordtype><recordid>eNo9j0FrwkAQRhdpQbGpfyG3ntLOZGYzu8cgVguBirTnkN0kEFFTsl767xtp8PQ9vsODp9QK4RWR9RuhiGieqUVKkiWUsTzcmexcRSEcAQAzEsO0UM-HfF_E3SXO_bXrL0_qsa1OoYmmXarv983XepcUn9uPdV4kHo1cE6wRK13XRqdAPNqdxZTZ1qQZkZxrYXwMegbJED004qypvBZOXWMNLdXLv9cPfQhD05Y_Q3euht8SobyVlFMJ_QE05TXu</recordid><startdate>20180630</startdate><enddate>20180630</enddate><creator>Khan, Kashif Nizam</creator><creator>Hirki, Mikael</creator><creator>Niemi, Tapio</creator><creator>Nurminen, Jukka K.</creator><creator>Ou, Zhonghong</creator><scope>AAYXX</scope><scope>CITATION</scope><orcidid>https://orcid.org/0000-0002-3209-5783</orcidid></search><sort><creationdate>20180630</creationdate><title>RAPL in Action</title><author>Khan, Kashif Nizam ; Hirki, Mikael ; Niemi, Tapio ; Nurminen, Jukka K. ; Ou, Zhonghong</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-c187t-1d11a5dd852034376b912449d354113bbf0b9181c407611c0e7b98ac5742be983</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2018</creationdate><toplevel>peer_reviewed</toplevel><toplevel>online_resources</toplevel><creatorcontrib>Khan, Kashif Nizam</creatorcontrib><creatorcontrib>Hirki, Mikael</creatorcontrib><creatorcontrib>Niemi, Tapio</creatorcontrib><creatorcontrib>Nurminen, Jukka K.</creatorcontrib><creatorcontrib>Ou, Zhonghong</creatorcontrib><collection>CrossRef</collection><jtitle>ACM transactions on modeling and performance evaluation of computing systems</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Khan, Kashif Nizam</au><au>Hirki, Mikael</au><au>Niemi, Tapio</au><au>Nurminen, Jukka K.</au><au>Ou, Zhonghong</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>RAPL in Action: Experiences in Using RAPL for Power Measurements</atitle><jtitle>ACM transactions on modeling and performance evaluation of computing systems</jtitle><date>2018-06-30</date><risdate>2018</risdate><volume>3</volume><issue>2</issue><spage>1</spage><epage>26</epage><pages>1-26</pages><issn>2376-3639</issn><eissn>2376-3647</eissn><abstract>To improve energy efficiency and comply with the power budgets, it is important to be able to measure the power consumption of cloud computing servers. Intel’s Running Average Power Limit (RAPL) interface is a powerful tool for this purpose. RAPL provides power limiting features and accurate energy readings for CPUs and DRAM, which are easily accessible through different interfaces on large distributed computing systems. Since its introduction, RAPL has been used extensively in power measurement and modeling. However, the advantages and disadvantages of RAPL have not been well investigated yet. To fill this gap, we conduct a series of experiments to disclose the underlying strengths and weaknesses of the RAPL interface by using both customized microbenchmarks and three well-known application level benchmarks: Stream , Stress-ng , and ParFullCMS . Moreover, to make the analysis as realistic as possible, we leverage two production-level power measurement datasets from the Taito , a supercomputing cluster of the Finnish Center of Scientific Computing and also replicate our experiments on Amazon EC2. Our results illustrate different aspects of RAPL and document the findings through comprehensive analysis. Our observations reveal that RAPL readings are highly correlated with plug power, promisingly accurate enough, and have negligible performance overhead. Experimental results suggest RAPL can be a very useful tool to measure and monitor the energy consumption of servers without deploying any complex power meters. We also show that there are still some open issues, such as driver support, non-atomicity of register updates, and unpredictable timings that might weaken the usability of RAPL in certain scenarios. For such scenarios, we pinpoint solutions and workarounds.</abstract><doi>10.1145/3177754</doi><tpages>26</tpages><orcidid>https://orcid.org/0000-0002-3209-5783</orcidid></addata></record>
fulltext fulltext
identifier ISSN: 2376-3639
ispartof ACM transactions on modeling and performance evaluation of computing systems, 2018-06, Vol.3 (2), p.1-26
issn 2376-3639
2376-3647
language eng
recordid cdi_crossref_primary_10_1145_3177754
source Association for Computing Machinery:Jisc Collections:ACM OPEN Journals 2023-2025 (reading list)
title RAPL in Action: Experiences in Using RAPL for Power Measurements
url http://sfxeu10.hosted.exlibrisgroup.com/loughborough?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2024-12-23T00%3A32%3A37IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-crossref&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=RAPL%20in%20Action:%20Experiences%20in%20Using%20RAPL%20for%20Power%20Measurements&rft.jtitle=ACM%20transactions%20on%20modeling%20and%20performance%20evaluation%20of%20computing%20systems&rft.au=Khan,%20Kashif%20Nizam&rft.date=2018-06-30&rft.volume=3&rft.issue=2&rft.spage=1&rft.epage=26&rft.pages=1-26&rft.issn=2376-3639&rft.eissn=2376-3647&rft_id=info:doi/10.1145/3177754&rft_dat=%3Ccrossref%3E10_1145_3177754%3C/crossref%3E%3Cgrp_id%3Ecdi_FETCH-LOGICAL-c187t-1d11a5dd852034376b912449d354113bbf0b9181c407611c0e7b98ac5742be983%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