Data-Pattern-Based Predictive On-Chip Power Meter in DNN Accelerator
Advanced power management techniques, such as voltage drop mitigation and fast power management, can greatly enhance energy efficiency in contemporary hardware design. Nevertheless, the implementation of these innovative techniques necessitates accurate and fine-grained power modeling, as well as ti...
Saved in:
Published in | IEEE transactions on computer-aided design of integrated circuits and systems p. 1 |
---|---|
Main Authors | , , , , , , |
Format | Journal Article |
Language | English |
Published |
IEEE
10.06.2024
|
Subjects | |
Online Access | Get full text |
Cover
Loading…
Abstract | Advanced power management techniques, such as voltage drop mitigation and fast power management, can greatly enhance energy efficiency in contemporary hardware design. Nevertheless, the implementation of these innovative techniques necessitates accurate and fine-grained power modeling, as well as timely responses for effective coordination with the power management unit. Additionally, existing performance-counter-based and RTL-based on-chip power meters have difficulty in providing sufficient response time for fast power and voltage management scenarios. In this paper, we propose PROPHET, a data-pattern-based power modeling method for multiply-accumulate-based (MACC) DNN accelerators. Our proposed power model extracts the pre-defined data patterns during memory access and then a pre-trained power model can predict the dynamic power of the DNN accelerators. Thus, PROPHET can predict dynamic power and provide sufficient responding time for power management units. In the experiments, we evaluate our predictive power model in four DNN accelerators with different dataflows and data types. In power model training and verification, our proposed data-patterns-based power model can realize the 2-cycle temporal resolution with R2>0.9, NMAE<7%, and the area and power overhead lower than 4.5%. |
---|---|
AbstractList | Advanced power management techniques, such as voltage drop mitigation and fast power management, can greatly enhance energy efficiency in contemporary hardware design. Nevertheless, the implementation of these innovative techniques necessitates accurate and fine-grained power modeling, as well as timely responses for effective coordination with the power management unit. Additionally, existing performance-counter-based and RTL-based on-chip power meters have difficulty in providing sufficient response time for fast power and voltage management scenarios. In this paper, we propose PROPHET, a data-pattern-based power modeling method for multiply-accumulate-based (MACC) DNN accelerators. Our proposed power model extracts the pre-defined data patterns during memory access and then a pre-trained power model can predict the dynamic power of the DNN accelerators. Thus, PROPHET can predict dynamic power and provide sufficient responding time for power management units. In the experiments, we evaluate our predictive power model in four DNN accelerators with different dataflows and data types. In power model training and verification, our proposed data-patterns-based power model can realize the 2-cycle temporal resolution with R2>0.9, NMAE<7%, and the area and power overhead lower than 4.5%. |
Author | Xie, Zhiyao Zhang, Yipu Zhang, Wei Liang, Tingyuan Lin, Zhe Jiang, Jingbo Peng, Jian |
Author_xml | – sequence: 1 givenname: Jian orcidid: 0000-0001-7335-3198 surname: Peng fullname: Peng, Jian organization: Dept. of Electronic and Computer Engineering, Hong Kong University of Science and Technology, Hong Kong, Hong Kong – sequence: 2 givenname: Tingyuan orcidid: 0000-0002-0390-2320 surname: Liang fullname: Liang, Tingyuan organization: Huawei, China – sequence: 3 givenname: Jingbo orcidid: 0000-0003-0268-9844 surname: Jiang fullname: Jiang, Jingbo organization: Dept. of Electronic and Computer Engineering, Hong Kong University of Science and Technology, Hong Kong, Hong Kong – sequence: 4 givenname: Yipu surname: Zhang fullname: Zhang, Yipu organization: Dept. of Electronic and Computer Engineering, Hong Kong University of Science and Technology, Hong Kong, Hong Kong – sequence: 5 givenname: Zhe surname: Lin fullname: Lin, Zhe organization: School of Integrated Circuit, Sun Yat-sen University, China – sequence: 6 givenname: Zhiyao orcidid: 0000-0002-4442-592X surname: Xie fullname: Xie, Zhiyao organization: Dept. of Electronic and Computer Engineering, Hong Kong University of Science and Technology, Hong Kong, Hong Kong – sequence: 7 givenname: Wei orcidid: 0000-0002-7622-6714 surname: Zhang fullname: Zhang, Wei organization: Dept. of Electronic and Computer Engineering, Hong Kong University of Science and Technology, Hong Kong, Hong Kong |
BookMark | eNpNkM1OwkAUhScGEwF9ABMX8wKD907np7PEVtQEoQv2zTDciTXYkmmj8e2FwMLV2Xzn5OSbsFHbtcTYPcIMEdzjppiXMwlSzTKF0tn8io3RZVYo1DhiY5A2FwAWbtik7z8BUGnpxqws_eBF5YeBUiuefE87XiXaNWFovomvW1F8NAdedT-U-DsdKd60vFyt-DwE2lPyQ5du2XX0-57uLjllm8XzpngVy_XLWzFfimCOTzCLMnhjt2rrATPlgzOIOebk0Hi1taS0owAxgtUYndxJjD6QssYGqyGbMjzPhtT1faJYH1Lz5dNvjVCfLNQnC_XJQn2xcOw8nDsNEf3jtVbG2OwPupNZBQ |
CODEN | ITCSDI |
ContentType | Journal Article |
DBID | 97E RIA RIE AAYXX CITATION |
DOI | 10.1109/TCAD.2024.3412978 |
DatabaseName | IEEE All-Society Periodicals Package (ASPP) 2005-present IEEE All-Society Periodicals Package (ASPP) 1998–Present IEEE Xplore CrossRef |
DatabaseTitle | CrossRef |
DatabaseTitleList | |
Database_xml | – sequence: 1 dbid: RIE name: IEEE Xplore url: https://proxy.k.utb.cz/login?url=https://ieeexplore.ieee.org/ sourceTypes: Publisher |
DeliveryMethod | fulltext_linktorsrc |
Discipline | Engineering |
EISSN | 1937-4151 |
EndPage | 1 |
ExternalDocumentID | 10_1109_TCAD_2024_3412978 10554667 |
Genre | orig-research |
GroupedDBID | --Z -~X 0R~ 29I 4.4 5GY 6IK 97E AAJGR AASAJ ABQJQ ABVLG ACGFS ACIWK ACNCT AENEX AKJIK ALMA_UNASSIGNED_HOLDINGS ASUFR ATWAV BEFXN BFFAM BGNUA BKEBE BPEOZ CS3 DU5 EBS HZ~ IFIPE IPLJI JAVBF LAI M43 O9- OCL P2P PZZ RIA RIE RNS TN5 5VS AAYXX AETIX AI. AIBXA ALLEH CITATION EJD H~9 IBMZZ ICLAB IFJZH RIG VH1 VJK |
ID | FETCH-LOGICAL-c637-13f2ca67b4ba0134ac9611818e916a4b7e459ec0ff0751f92d21face4767c7503 |
IEDL.DBID | RIE |
ISSN | 0278-0070 |
IngestDate | Thu Aug 22 11:27:04 EDT 2024 Wed Jul 03 05:40:34 EDT 2024 |
IsPeerReviewed | true |
IsScholarly | true |
Language | English |
LinkModel | DirectLink |
MergedId | FETCHMERGED-LOGICAL-c637-13f2ca67b4ba0134ac9611818e916a4b7e459ec0ff0751f92d21face4767c7503 |
ORCID | 0000-0002-7622-6714 0000-0002-0390-2320 0000-0003-0268-9844 0000-0001-7335-3198 0000-0002-4442-592X |
PageCount | 1 |
ParticipantIDs | ieee_primary_10554667 crossref_primary_10_1109_TCAD_2024_3412978 |
PublicationCentury | 2000 |
PublicationDate | 20240610 |
PublicationDateYYYYMMDD | 2024-06-10 |
PublicationDate_xml | – month: 6 year: 2024 text: 20240610 day: 10 |
PublicationDecade | 2020 |
PublicationTitle | IEEE transactions on computer-aided design of integrated circuits and systems |
PublicationTitleAbbrev | TCAD |
PublicationYear | 2024 |
Publisher | IEEE |
Publisher_xml | – name: IEEE |
SSID | ssj0014529 |
Score | 2.4673824 |
Snippet | Advanced power management techniques, such as voltage drop mitigation and fast power management, can greatly enhance energy efficiency in contemporary hardware... |
SourceID | crossref ieee |
SourceType | Aggregation Database Publisher |
StartPage | 1 |
SubjectTerms | Arrays Data models data patterns DNN accelerator Energy efficiency Meters on-chip power meter Power modeling power prediction Power system management System-on-chip Voltage measurement |
Title | Data-Pattern-Based Predictive On-Chip Power Meter in DNN Accelerator |
URI | https://ieeexplore.ieee.org/document/10554667 |
hasFullText | 1 |
inHoldings | 1 |
isFullTextHit | |
isPrint | |
link | http://utb.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwjV1La8MwDDZrT9thz451L3zYaeA2SV07OXbtShks66GD3oKt2KwM0lLSy379LCcdYTAYuZiQgC0JS_b3SSLkgYNF-oJkEZiE8ZgPmFYKWGBBGvdoUbF8UzF75y_L4bJOVve5MMYYTz4zPRx6LD9fww6vyvrYzJELIVukFQdRlaz1AxkggugvVLBkrDPkGsIMg6S_cKtyR8GI99yeHfmWag0n1Oiq4p3K9ISk--lUXJLP3q7UPfj6Vanx3_M9Jcd1eElHlT2ckQNTnJOjRtHBCzKZqFKxua-rWbAn58VyOt8iXoM7H30r2PhjtaFzbJ9GX5EtQ1cFnaQpHQE4J-Vx-Q5ZTJ8X4xmreykwEAPJwoGNQAmpuVYu6OMKEoEpp7Fx4aHiWho-TAwE1roQIrRJlEehVWC4FBIQ6rwk7WJdmCtCg5yH1lrQudB8AEmiIZdxruI8DNQwVF3yuJdttqkqZmT-pBEkGSoiQ0VktSK6pINia3xYSez6j_c35BB_R65WGNySdrndmTsXFZT63lvDNwfcsno |
link.rule.ids | 315,786,790,802,27955,27956,55107 |
linkProvider | IEEE |
linkToHtml | http://utb.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwjV1LSwMxEA5aD-rBt1ifOXgS0u5u06R7rK2larv2UKG3JZlNsAjbUtqLv95MditFEGQvYVmWZGbITPJ9M0PIPQeL9AXJIjAx4y3eYFopYIEFadyjRcHyTUT_nb9MmpMyWd3nwhhjPPnM1HDosfxsBiu8KqtjM0cuhNwmO87RB7JI1_oBDRBD9FcqWDTWmXIJYoZBXB-7dbnDYMRrbteOfFO1DTe00VfFu5XeIUnWEyrYJJ-11VLX4OtXrcZ_z_iIHJQBJm0XFnFMtkx-QvY3yg6ekm5XLRUb-cqaOXt0fiyjowUiNrj30becdT6mczrCBmp0iHwZOs1pN0loG8C5KY_Mn5Fx72nc6bOymwID0ZAsbNgIlJCaa-XCPq4gFph02jIuQFRcS8ObsYHAWhdEhDaOsii0CgyXQgKCneekks9yc0FokPHQWgs6E5o3II41ZLKVqVYWBqoZqip5WMs2nRc1M1J_1gjiFBWRoiLSUhFVcoZi2_iwkNjlH-_vyG5_PBykg-fk9Yrs4a-QuRUG16SyXKzMjYsRlvrWW8Y3oua1zg |
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=Data-Pattern-Based+Predictive+On-Chip+Power+Meter+in+DNN+Accelerator&rft.jtitle=IEEE+transactions+on+computer-aided+design+of+integrated+circuits+and+systems&rft.au=Peng%2C+Jian&rft.au=Liang%2C+Tingyuan&rft.au=Jiang%2C+Jingbo&rft.au=Zhang%2C+Yipu&rft.date=2024-06-10&rft.issn=0278-0070&rft.eissn=1937-4151&rft.spage=1&rft.epage=1&rft_id=info:doi/10.1109%2FTCAD.2024.3412978&rft.externalDBID=n%2Fa&rft.externalDocID=10_1109_TCAD_2024_3412978 |
thumbnail_l | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=0278-0070&client=summon |
thumbnail_m | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=0278-0070&client=summon |
thumbnail_s | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=0278-0070&client=summon |