Dissecting the Software-Based Measurement of CPU Energy Consumption: A Comparative Analysis
Information and Communications Technologies (ICT) are an increasingly important contributor to the environmental crisis. Computer scientists need tools for measuring the footprint of the code they produce and for optimizing it. Running Average Power Limit (RAPL) is a low-level interface designed by...
Gespeichert in:
| Veröffentlicht in: | IEEE transactions on parallel and distributed systems Jg. 36; H. 1; S. 96 - 107 |
|---|---|
| Hauptverfasser: | , |
| Format: | Journal Article |
| Sprache: | Englisch |
| Veröffentlicht: |
IEEE
01.01.2025
Institute of Electrical and Electronics Engineers |
| Schlagworte: | |
| ISSN: | 1045-9219, 1558-2183 |
| Online-Zugang: | Volltext |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| Abstract | Information and Communications Technologies (ICT) are an increasingly important contributor to the environmental crisis. Computer scientists need tools for measuring the footprint of the code they produce and for optimizing it. Running Average Power Limit (RAPL) is a low-level interface designed by Intel that provides a measure of the energy consumption of a CPU (and more) without the need for additional hardware. Since 2017, it is available on most x86 processors, including AMD processors. More and more people are using RAPL for energy measurement, mostly like a black box without deep knowledge of its behavior. Unfortunately, this causes mistakes when implementing measurement tools. In this article, we propose to come back to the basic mechanisms that allow to use RAPL measurements and present a critical analysis of their operations. In addition to long-established mechanisms, we explore the suitability of the recent eBPF technology (formerly and abbreviation for extended Berkeley Packet Filter) for working with RAPL. We release an implementation in Rust that avoids the pitfalls we detected in existing tools, improving correctness, timing accuracy and performance, with desirable properties for monitoring and profiling parallel applications. We provide an experimental study with multiple benchmarks and processor models to evaluate the efficiency of the various mechanisms and their impact on parallel software. We show that no mechanism provides a significant performance advantage over the others. However, they differ significantly in terms of ease-of-use and resiliency. We believe that this work will help the community to develop correct, resilient and lightweight measurement tools. |
|---|---|
| AbstractList | Every day, we experience the effects of the global warming: extreme weather events, major forest fires, storms, global warming, etc.The scientific community acknowledges that this crisis is a consequence of human activities where Information and Communications Technologies (ICT) are an increasingly important contributor.Computer scientists need tools for measuring the footprint of the code they produce and for optimizing it. Running Average Power Limit (RAPL) is a low-level interface designed by Intel that provides a measure of the energy consumption of a CPU (and more) without the need for additional hardware. Since 2017, it is available on most computing devices, including non-Intel devices such as AMD processors.More and more people are using RAPL for energy measurement, mostly like a black box without deep knowledge of its behavior.Unfortunately, this causes mistakes when implementing measurement tools.In this paper, we propose to come back to the basic mechanisms that allow to use RAPL measurements and present a critical analysis of their operations. In addition to long-established mechanisms, we explore the suitability of the recent eBPF technology (formerly and abbreviation for extended Berkeley Packet Filter) for working with RAPL.For each mechanism, we release an implementation in Rust that avoids the pitfalls we detected in existing tools, improving correctness, timing accuracy and performance. These new implementations have desirable properties for monitoring and profiling parallel applications.We also provide an experimental study with multiple benchmarks and processor models (Intel and AMD) in order to evaluate the efficiency of the various mechanisms and their impact on parallel software.These experiments show that no mechanism provides a significant performance advantage over the others. However, they differ significantly in terms of ease-of-use and resiliency.We believe that this work will help the community to develop correct, resilient and lightweight measurement tools. Information and Communications Technologies (ICT) are an increasingly important contributor to the environmental crisis. Computer scientists need tools for measuring the footprint of the code they produce and for optimizing it. Running Average Power Limit (RAPL) is a low-level interface designed by Intel that provides a measure of the energy consumption of a CPU (and more) without the need for additional hardware. Since 2017, it is available on most x86 processors, including AMD processors. More and more people are using RAPL for energy measurement, mostly like a black box without deep knowledge of its behavior. Unfortunately, this causes mistakes when implementing measurement tools. In this article, we propose to come back to the basic mechanisms that allow to use RAPL measurements and present a critical analysis of their operations. In addition to long-established mechanisms, we explore the suitability of the recent eBPF technology (formerly and abbreviation for extended Berkeley Packet Filter) for working with RAPL. We release an implementation in Rust that avoids the pitfalls we detected in existing tools, improving correctness, timing accuracy and performance, with desirable properties for monitoring and profiling parallel applications. We provide an experimental study with multiple benchmarks and processor models to evaluate the efficiency of the various mechanisms and their impact on parallel software. We show that no mechanism provides a significant performance advantage over the others. However, they differ significantly in terms of ease-of-use and resiliency. We believe that this work will help the community to develop correct, resilient and lightweight measurement tools. |
| Author | Raffin, Guillaume Trystram, Denis |
| Author_xml | – sequence: 1 givenname: Guillaume orcidid: 0009-0009-9156-121X surname: Raffin fullname: Raffin, Guillaume email: guillaume.raffin@univ-grenoble-alpes.fr organization: CNRS, Inria, Grenoble INP, LIG, Univ. Grenoble Alpes, Grenoble, France – sequence: 2 givenname: Denis orcidid: 0000-0002-2623-6922 surname: Trystram fullname: Trystram, Denis email: denis.trystram@univ-grenoble-alpes.fr organization: CNRS, Inria, Grenoble INP, LIG, Univ. Grenoble Alpes, Grenoble, France |
| BackLink | https://hal.science/hal-04420527$$DView record in HAL |
| BookMark | eNp9kE1LAzEQhoNUsFV_gOAhVw9b87nZeKtt_YCKgu3JQ8jGWY20uyWJlf57d2kF8eBc5oP3nWGeAerVTQ0InVEypJToy_nT5HnICBNDLjTjPD9AfSplkTFa8F5bEyEzzag-QoMYPwihQhLRRy8THyO45Os3nN4BPzdV-rIBsmsb4RU_gI2fAVZQJ9xUePy0wNMawtsWj5s6fq7WyTf1FR617Wptg01-A3hU2-U2-niCDiu7jHC6z8docTOdj--y2ePt_Xg0yxxXJGWFFEC4lK5SbeS5cloJpYHy0mrNiqpkzlUuLxUBaYuS6QKUKywFlpdMaH6MLnZ73-3SrINf2bA1jfXmbjQz3YwIwYhkasNbrdppXWhiDFAZ55PtvkjB-qWhxHQ8TcfTdDzNnmfrpH-cP6f-85zvPB4AfumVyLkg_BunxYJs |
| CODEN | ITDSEO |
| CitedBy_id | crossref_primary_10_3390_fi17050203 crossref_primary_10_2298_CSIS240722023K |
| Cites_doi | 10.1145/2425248.2425252 10.1016/j.jpdc.2019.07.007 10.1109/ICPPW.2010.38 10.2172/983318 10.1007/978-3-319-09228-7_1 10.1145/3401335.3401716 10.1109/ITC30.2018.00039 10.1109/CLUSTER.2017.66 10.1145/2834800.2834807 10.23919/DATE51398.2021.9474110 10.1145/2989081.2989088 10.1109/IPDPSW.2015.70 10.1145/3177754 10.1126/science.1259855 10.1145/3432261.3432263 10.1145/3433210.3437517 10.1109/Cluster48925.2021.00087 10.1109/ISPASS.2013.6557170 10.1109/CCGrid49817.2020.00-45 |
| ContentType | Journal Article |
| Copyright | Attribution |
| Copyright_xml | – notice: Attribution |
| DBID | 97E RIA RIE AAYXX CITATION 1XC VOOES |
| DOI | 10.1109/TPDS.2024.3492336 |
| DatabaseName | IEEE All-Society Periodicals Package (ASPP) 2005–Present IEEE All-Society Periodicals Package (ASPP) 1998–Present IEEE/IET Electronic Library CrossRef Hyper Article en Ligne (HAL) Hyper Article en Ligne (HAL) (Open Access) |
| DatabaseTitle | CrossRef |
| DatabaseTitleList | |
| Database_xml | – sequence: 1 dbid: RIE name: IEEE url: https://ieeexplore.ieee.org/ sourceTypes: Publisher |
| DeliveryMethod | fulltext_linktorsrc |
| Discipline | Engineering Computer Science |
| EISSN | 1558-2183 |
| EndPage | 107 |
| ExternalDocumentID | oai:HAL:hal-04420527v3 10_1109_TPDS_2024_3492336 10746340 |
| Genre | orig-research |
| GrantInformation_xml | – fundername: MIAI@Grenoble Alpes grantid: ANR-19-P3IA-0003 |
| GroupedDBID | --Z -~X .DC 0R~ 29I 4.4 5GY 5VS 6IK 97E AAJGR AARMG AASAJ AAWTH ABAZT ABFSI ABQJQ ABVLG ACGFO ACIWK AENEX AETIX AGQYO AGSQL AHBIQ AI. AIBXA AKJIK AKQYR ALLEH ALMA_UNASSIGNED_HOLDINGS ASUFR ATWAV BEFXN BFFAM BGNUA BKEBE BPEOZ CS3 DU5 E.L EBS EJD HZ~ H~9 ICLAB IEDLZ IFIPE IFJZH IPLJI JAVBF LAI M43 MS~ O9- OCL P2P PQQKQ RIA RIE RNI RNS RZB TN5 TWZ UHB VH1 AAYXX CITATION 1XC VOOES |
| ID | FETCH-LOGICAL-c370t-854e0355cf7777667c97479e13ba9928fb2ccfc6b70e5a8b298e7c8a1e26b2493 |
| IEDL.DBID | RIE |
| ISICitedReferencesCount | 3 |
| ISICitedReferencesURI | http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=001367278200002&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D |
| ISSN | 1045-9219 |
| IngestDate | Tue Oct 28 06:37:11 EDT 2025 Sat Nov 29 06:06:51 EST 2025 Tue Nov 18 22:24:37 EST 2025 Wed Aug 27 02:32:55 EDT 2025 |
| IsDoiOpenAccess | true |
| IsOpenAccess | true |
| IsPeerReviewed | true |
| IsScholarly | true |
| Issue | 1 |
| Keywords | software measurement RAPL library (Running Average Power Limit) energy consumption performance analysis energy efficiency |
| Language | English |
| License | https://ieeexplore.ieee.org/Xplorehelp/downloads/license-information/IEEE.html https://doi.org/10.15223/policy-029 https://doi.org/10.15223/policy-037 Attribution: http://creativecommons.org/licenses/by |
| LinkModel | DirectLink |
| MergedId | FETCHMERGED-LOGICAL-c370t-854e0355cf7777667c97479e13ba9928fb2ccfc6b70e5a8b298e7c8a1e26b2493 |
| ORCID | 0000-0002-2623-6922 0009-0009-9156-121X |
| OpenAccessLink | https://hal.science/hal-04420527 |
| PageCount | 12 |
| ParticipantIDs | ieee_primary_10746340 crossref_primary_10_1109_TPDS_2024_3492336 hal_primary_oai_HAL_hal_04420527v3 crossref_citationtrail_10_1109_TPDS_2024_3492336 |
| PublicationCentury | 2000 |
| PublicationDate | 2025-01-01 |
| PublicationDateYYYYMMDD | 2025-01-01 |
| PublicationDate_xml | – month: 01 year: 2025 text: 2025-01-01 day: 01 |
| PublicationDecade | 2020 |
| PublicationTitle | IEEE transactions on parallel and distributed systems |
| PublicationTitleAbbrev | TPDS |
| PublicationYear | 2025 |
| Publisher | IEEE Institute of Electrical and Electronics Engineers |
| Publisher_xml | – name: IEEE – name: Institute of Electrical and Electronics Engineers |
| References | ref13 ref35 Freitag (ref6) 2021; 2 Raffin (ref32) 2023 ref37 ref14 Bouza Heguerte (ref30) 2023; 5 ref11 Courty (ref19) 2024 Belkhir (ref3) 2018; 177 ref17 ref18 Hilty (ref7) 2006; 21 (ref36) 2019 ref24 ref23 Eranian (ref38) 2022 (ref16) 2016; 3B ref26 ref48 Andrae (ref2) 2019; 8 (ref5) 2023 ref25 ref41 ref22 Rasoldier (ref10) Bourdon (ref20) 2013; 92 DeLoss (ref9) 2023 ref28 ref27 ref49 Gruber (ref34) 2015 ref8 ref4 Petit (ref21) 2020 Jay (ref29) Malmodin (ref1) 2018; 10 |
| References_xml | – volume: 177 start-page: 448 volume-title: J. Cleaner Prod. year: 2018 ident: ref3 article-title: Assessing ICT global emissions footprint: Trends to 2040 & recommendations – volume: 92 start-page: 43 year: 2013 ident: ref20 article-title: PowerAPI: A software library to monitor the energy consumed at the process-level publication-title: ERCIM News – start-page: 1 volume-title: Proc. 23rd IEEE/ACM Int. Symp. Cluster Cloud Internet Comput. ident: ref29 article-title: An experimental comparison of software-based power meters: Focus on CPU and GPU – ident: ref35 doi: 10.1145/2425248.2425252 – year: 2023 ident: ref5 article-title: Environmental impacts of digital technology: 5-year trends and 5G governance – ident: ref27 doi: 10.1016/j.jpdc.2019.07.007 – ident: ref37 doi: 10.1109/ICPPW.2010.38 – volume-title: Proc. 8th Workshop Comput. Limits ident: ref10 article-title: How realistic are claims about the benefits of using digital technologies for GHG emissions mitigation? – year: 2015 ident: ref34 article-title: Review overflow handling for RAPL counters – ident: ref49 doi: 10.2172/983318 – ident: ref8 doi: 10.1007/978-3-319-09228-7_1 – volume: 2 issue: 9 volume-title: Patterns year: 2021 ident: ref6 article-title: The real climate and transformative impact of ICT: A critique of estimates, trends, and regulations – ident: ref11 doi: 10.1145/3401335.3401716 – ident: ref48 doi: 10.1109/ITC30.2018.00039 – ident: ref17 doi: 10.1109/CLUSTER.2017.66 – ident: ref24 doi: 10.1145/2834800.2834807 – year: 2023 ident: ref9 article-title: AI helping to unravel complexity of climate, weather and land use, find solutions to climate change – ident: ref4 doi: 10.23919/DATE51398.2021.9474110 – ident: ref25 doi: 10.1145/2989081.2989088 – year: 2019 ident: ref36 article-title: PyJoules – volume: 8 start-page: 27 year: 2019 ident: ref2 article-title: Prediction studies of electricity use of global computing in 2030 publication-title: Int. J. Sci. Eng. Investigations – ident: ref23 doi: 10.1109/IPDPSW.2015.70 – volume: 3B year: 2016 ident: ref16 publication-title: Syst. Programm. Guide, Part 2 – ident: ref26 doi: 10.1145/3177754 – year: 2022 ident: ref38 article-title: perf/x86/rapl: Fix amd event handling (commit 0036fb0) – year: 2020 ident: ref21 article-title: Scaphandre – year: 2023 ident: ref32 article-title: Detect and correct overflows of the RAPL microjoule counter – volume: 5 volume-title: Environ. Res. Commun. year: 2023 ident: ref30 article-title: How to estimate carbon footprint when training deep learning models? A guide and review – ident: ref13 doi: 10.1126/science.1259855 – ident: ref14 doi: 10.1145/3432261.3432263 – volume: 21 start-page: 1618 issue: 11 volume-title: Environ. Modelling Softw. year: 2006 ident: ref7 article-title: The relevance of information and communication technologies for environmental sustainability–A prospective simulation study – ident: ref41 doi: 10.1145/3433210.3437517 – year: 2024 ident: ref19 article-title: mlco2/codecarbon: v2.4.1 publication-title: Zenodo – ident: ref28 doi: 10.1109/Cluster48925.2021.00087 – ident: ref22 doi: 10.1109/ISPASS.2013.6557170 – ident: ref18 doi: 10.1109/CCGrid49817.2020.00-45 – volume: 10 issue: 9 volume-title: Sustainability year: 2018 ident: ref1 article-title: The energy and carbon footprint of the global ICT and E&M sectors 2010–2015 |
| SSID | ssj0014504 |
| Score | 2.4715698 |
| Snippet | Information and Communications Technologies (ICT) are an increasingly important contributor to the environmental crisis. Computer scientists need tools for... Every day, we experience the effects of the global warming: extreme weather events, major forest fires, storms, global warming, etc.The scientific community... |
| SourceID | hal crossref ieee |
| SourceType | Open Access Repository Enrichment Source Index Database Publisher |
| StartPage | 96 |
| SubjectTerms | Central Processing Unit Climate change Closed box Computer Science Distributed, Parallel, and Cluster Computing Energy consumption Energy efficiency Environmental monitoring Global warming Performance analysis RAPL library (Running Average Power Limit) Software measurement Weather forecasting |
| Title | Dissecting the Software-Based Measurement of CPU Energy Consumption: A Comparative Analysis |
| URI | https://ieeexplore.ieee.org/document/10746340 https://hal.science/hal-04420527 |
| Volume | 36 |
| WOSCitedRecordID | wos001367278200002&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D |
| hasFullText | 1 |
| inHoldings | 1 |
| isFullTextHit | |
| isPrint | |
| journalDatabaseRights | – providerCode: PRVIEE databaseName: IEEE customDbUrl: eissn: 1558-2183 dateEnd: 99991231 omitProxy: false ssIdentifier: ssj0014504 issn: 1045-9219 databaseCode: RIE dateStart: 19900101 isFulltext: true titleUrlDefault: https://ieeexplore.ieee.org/ providerName: IEEE |
| link | http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV3NT8IwFG-EeNCDKGLErzTGk8mgdN26ekOEcFBCAiQkHpa1e4smBgwC_vu-lvFx0cSdlqZblr227_3ex-8RcpcJXLWQIjrJVOSJRGdeFBrmSQWJDFBfZcBcswnZ60XjsernxequFgYAXPIZ1Oyti-WnU7OwrrK6TR4MfYEIvSBluCrW2oQMROB6BSK8CDyF-zAPYTaYqg_7TwOEglzULBef7-iYt0qo8GZTIHd6qzjV0in986OOyVFuQ9LmSugnZA8mZVJa92eg-XYtk8MdssFT8vpkQ-_GpjlTtProAA_g72QG3iMqspS-bJ2FdJrRVn9E264ukLZclaY7Wh5ok7a2fOF0TWlSIaNOe9jqenlrBc_4ks29KBDA0NQwmcQrDKWxuEJBw9eJUjzKNDcmM6GWDIIk0lxFIE2UNICHGhGbf0aKk-kEzglFe0JrE6AyNEqkGu0fH8-QlOtEQAAprxK2_texyXnHbfuLj9jhD6ZiK57YiifOxVMl95tHPlekG39NvkUBbuZZuuxu8zm2Y0wIzgIul36VVKzkdt62EtrFL-OX5IDbRr_O13JFivPZAq7JvlnO379mN27h_QAjXNO7 |
| linkProvider | IEEE |
| linkToHtml | http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV3dT9swED9tMGnsAVgB0QHDmvY0KeA6dhzvrRRQEaWqRCsh8RDFzkUgoRZ1bffv7-ymHy9DIk-R5URRzvbd7z5-B_CzlLRqsSB0Upo0krktozRxPNIGc61IX5XIQ7MJ3e2mDw-mVxWrh1oYRAzJZ3jmb0Msvxi5qXeVnfvkwSSWhNA3lSTgMy_XWgYNpArdAglgqMjQTqyCmA1uzvu9y3sCg0KeeTa-OBAyr9TQxyefBLnWXSUol-udd37WLmxXViRrzsX-FT7gsAY7iw4NrNqwNfiyRje4B4-XPvjufKIzI7uP3dMR_DcfY3RBqqxgdyt3IRuVrNUbsKtQGchaoU4zHC6_WZO1VozhbEFqsg-D66t-qx1VzRUiF2s-iVIlkZOx4UpNV5Jo55GFwUZsc2NEWlrhXOkSqzmqPLXCpKhdmjdQJJYwW3wAG8PREA-BkUVhrVOkDp2RhSULKKZTpBA2l6iwEHXgi3-duYp53DfAeMkCAuEm8-LJvHiySjx1-LV85HVOu_HW5B8kwOU8T5jdbnYyP8alFFwJPYvrsO8lt_a2udC-_Wf8FD63-3edrHPTvT2CLeHb_gbPyzFsTMZTPIFPbjZ5_jP-HhbhP1GY1wI |
| openUrl | ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Ajournal&rft.genre=article&rft.atitle=Dissecting+the+Software-Based+Measurement+of+CPU+Energy+Consumption%3A+A+Comparative+Analysis&rft.jtitle=IEEE+transactions+on+parallel+and+distributed+systems&rft.au=Raffin%2C+Guillaume&rft.au=Trystram%2C+Denis&rft.date=2025-01-01&rft.pub=IEEE&rft.issn=1045-9219&rft.volume=36&rft.issue=1&rft.spage=96&rft.epage=107&rft_id=info:doi/10.1109%2FTPDS.2024.3492336&rft.externalDocID=10746340 |
| thumbnail_l | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=1045-9219&client=summon |
| thumbnail_m | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=1045-9219&client=summon |
| thumbnail_s | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=1045-9219&client=summon |