DIMCA: An Area-Efficient Digital In-Memory Computing Macro Featuring Approximate Arithmetic Hardware in 28 nm
Recent SRAM-based in-memory computing (IMC) hardware demonstrates high energy efficiency and throughput for matrix-vector multiplication (MVM), the dominant kernel for deep neural networks (DNNs). Earlier IMC macros have employed analog-mixed-signal (AMS) arithmetic hardware. However, those so-calle...
Gespeichert in:
| Veröffentlicht in: | IEEE journal of solid-state circuits Jg. 59; H. 3; S. 960 - 971 |
|---|---|
| Hauptverfasser: | , , , , , , |
| Format: | Journal Article |
| Sprache: | Englisch |
| Veröffentlicht: |
New York
IEEE
01.03.2024
The Institute of Electrical and Electronics Engineers, Inc. (IEEE) |
| Schlagworte: | |
| ISSN: | 0018-9200, 1558-173X |
| Online-Zugang: | Volltext |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| Abstract | Recent SRAM-based in-memory computing (IMC) hardware demonstrates high energy efficiency and throughput for matrix-vector multiplication (MVM), the dominant kernel for deep neural networks (DNNs). Earlier IMC macros have employed analog-mixed-signal (AMS) arithmetic hardware. However, those so-called AIMCs suffer from process, voltage, and temperature (PVT) variations. Digital IMC (DIMC) macros, on the other hand, exhibit better robustness against PVT variations, but they tend to require more silicon area. This article proposes novel DIMC hardware featuring approximate arithmetic (DIMCA) to improve area efficiency without hurting compute density (CD). We also propose an approximation-aware training model and a customized number format to compensate for the accuracy degradation caused by the approximation hardware. We prototyped the test chip in 28-nm CMOS. It contains two versions: the DIMCA with single-approximate hardware (DIMCA1) and DIMCA with double-approximate hardware (DIMCA2). The measurement results show that DIMCA1 supports a 4 b-activation and 1 b-weight (4 b/1 b) CNN model, achieving 327 kb/mm2, 458-990 TOPS/W (normalized to 1 b/1 b), 8.27-392 TOPS/mm2 (normalized to 1 b/1 b), and 90.41% accuracy for CIFAR-10. DIMCA2 supports a 1 b/1 b CNN model, achieving 485 kb/mm2, 932-2219 TOPS/W, 14.4-607 TOPS/mm2, and 86.96% accuracy for CIFAR-10. |
|---|---|
| AbstractList | Recent SRAM-based in-memory computing (IMC) hardware demonstrates high energy efficiency and throughput for matrix–vector multiplication (MVM), the dominant kernel for deep neural networks (DNNs). Earlier IMC macros have employed analog-mixed-signal (AMS) arithmetic hardware. However, those so-called AIMCs suffer from process, voltage, and temperature (PVT) variations. Digital IMC (DIMC) macros, on the other hand, exhibit better robustness against PVT variations, but they tend to require more silicon area. This article proposes novel DIMC hardware featuring approximate arithmetic (DIMCA) to improve area efficiency without hurting compute density (CD). We also propose an approximation-aware training model and a customized number format to compensate for the accuracy degradation caused by the approximation hardware. We prototyped the test chip in 28-nm CMOS. It contains two versions: the DIMCA with single-approximate hardware (DIMCA1) and DIMCA with double-approximate hardware (DIMCA2). The measurement results show that DIMCA1 supports a 4 b-activation and 1 b-weight (4 b/1 b) CNN model, achieving 327 kb/mm2, 458–990 TOPS/W (normalized to 1 b/1 b), 8.27–392 TOPS/mm2 (normalized to 1 b/1 b), and 90.41% accuracy for CIFAR-10. DIMCA2 supports a 1 b/1 b CNN model, achieving 485 kb/mm2, 932–2219 TOPS/W, 14.4–607 TOPS/mm2, and 86.96% accuracy for CIFAR-10. |
| Author | Lin, Chuan-Tung Krishnamurthy, Ram Kumar Seok, Mingoo Zhang, Bo Chen, Gregory K. Wang, Dewei Knag, Phil C. |
| Author_xml | – sequence: 1 givenname: Chuan-Tung orcidid: 0000-0002-7345-1916 surname: Lin fullname: Lin, Chuan-Tung organization: Department of Electrical Engineering, Columbia University, New York, NY, USA – sequence: 2 givenname: Dewei orcidid: 0009-0006-5970-8144 surname: Wang fullname: Wang, Dewei organization: Department of Electrical Engineering, Columbia University, New York, NY, USA – sequence: 3 givenname: Bo orcidid: 0000-0003-3603-6016 surname: Zhang fullname: Zhang, Bo organization: Department of Electrical Engineering, Columbia University, New York, NY, USA – sequence: 4 givenname: Gregory K. orcidid: 0000-0002-4813-3844 surname: Chen fullname: Chen, Gregory K. organization: Circuit Research Laboratory, Intel Corporation, Hillsboro, OR, USA – sequence: 5 givenname: Phil C. orcidid: 0000-0001-6794-8806 surname: Knag fullname: Knag, Phil C. organization: Circuit Research Laboratory, Intel Corporation, Hillsboro, OR, USA – sequence: 6 givenname: Ram Kumar orcidid: 0000-0002-2428-7099 surname: Krishnamurthy fullname: Krishnamurthy, Ram Kumar organization: Circuit Research Laboratory, Intel Corporation, Hillsboro, OR, USA – sequence: 7 givenname: Mingoo orcidid: 0000-0002-9722-0979 surname: Seok fullname: Seok, Mingoo email: ms4415@columbia.edu organization: Department of Electrical Engineering, Columbia University, New York, NY, USA |
| BookMark | eNp9UE1Lw0AUXETBtvoDBA8LnlP3M9l4C6kfFYsHFbyF1_SlrjSbutmi_nu3tAfx4Okxj5l5b2ZIDl3nkJAzzsacs_zy_umpHAsm5FhKLjXPD8iAa20SnsnXQzJgjJskF4wdk2Hfv0eolOED0k6ms7K4ooWjhUdIrpvG1hZdoBO7tAFWdOqSGbad_6Zl1643wbolnUHtO3qDEDZ-i4v12ndftoWA0caGtxaDrekd-MUneKTWUWGoa0_IUQOrHk_3c0Rebq6fy7vk4fF2WhYPSS1yFRJUKgMOtV4IjpkQDXItEIzRWc5AZRoWDW-EgXnaMBM3BoArOcdMgQCj5Yhc7HzjWx8b7EP13m28iycrkUuhU5WpNLKyHSuG6XuPTVXHxMF2Lniwq4qzatttte222nZb7buNSv5HufYxvf_-V3O-01hE_MUXaSqFkT-h04Xh |
| CODEN | IJSCBC |
| CitedBy_id | crossref_primary_10_1109_LSSC_2024_3369058 crossref_primary_10_1109_TVLSI_2025_3576889 crossref_primary_10_1109_JSSC_2024_3362699 crossref_primary_10_1109_JSSC_2024_3463691 crossref_primary_10_1109_TCSI_2024_3497187 crossref_primary_10_1016_j_vlsi_2025_102525 |
| Cites_doi | 10.1109/isscc42614.2022.9731762 10.1109/JSSC.2020.2992886 10.1109/isscc42614.2022.9731659 10.1109/ICCV48922.2021.00060 10.1109/MSSC.2019.2922889 10.1145/2742060.2743760 10.1109/isscc42613.2021.9365766 10.1109/isscc42613.2021.9365928 10.1109/ISOCC.2015.7401667 10.1109/ISSCC42615.2023.10067720 10.1109/JSSC.2018.2867275 10.1109/JSSC.2019.2939682 10.1109/JSSC.2021.3108344 10.1109/JSSC.2021.3061508 10.1109/isscc42613.2021.9365791 10.1109/isscc42615.2023.10067305 10.1109/MSSC.2022.3182935 10.1109/isscc19947.2020.9063111 10.1007/978-3-030-58452-8_13 10.1109/ISSCC42614.2022.9731754 10.1038/s41565-020-0655-z 10.1109/JSSC.2019.2952773 10.1109/JSSC.2019.2963616 10.1109/CVPR42600.2020.01079 10.1109/JSSC.2016.2616357 10.1109/isscc42614.2022.9731545 10.1109/isscc42614.2022.9731757 10.1109/JSSC.2020.3029586 10.1109/JSSC.2021.3119018 10.1109/CICC53496.2022.9772781 10.1109/JSSC.2019.2899730 10.1109/3.748832 10.1109/VLSITechnologyandCir46769.2022.9830438 10.1109/TCAD.2012.2217962 10.1109/ICCV48922.2021.00986 10.1109/JSSC.2018.2880918 |
| ContentType | Journal Article |
| Copyright | Copyright The Institute of Electrical and Electronics Engineers, Inc. (IEEE) 2024 |
| Copyright_xml | – notice: Copyright The Institute of Electrical and Electronics Engineers, Inc. (IEEE) 2024 |
| DBID | 97E RIA RIE AAYXX CITATION 7SP 8FD L7M |
| DOI | 10.1109/JSSC.2023.3313519 |
| DatabaseName | IEEE All-Society Periodicals Package (ASPP) 2005–Present IEEE All-Society Periodicals Package (ASPP) 1998–Present IEEE/IET Electronic Library (IEL) (UW System Shared) CrossRef Electronics & Communications Abstracts Technology Research Database Advanced Technologies Database with Aerospace |
| DatabaseTitle | CrossRef Technology Research Database Advanced Technologies Database with Aerospace Electronics & Communications Abstracts |
| DatabaseTitleList | Technology Research Database |
| Database_xml | – sequence: 1 dbid: RIE name: IEEE Electronic Library (IEL) url: https://ieeexplore.ieee.org/ sourceTypes: Publisher |
| DeliveryMethod | fulltext_linktorsrc |
| Discipline | Engineering |
| EISSN | 1558-173X |
| EndPage | 971 |
| ExternalDocumentID | 10_1109_JSSC_2023_3313519 10266328 |
| Genre | orig-research |
| GrantInformation_xml | – fundername: NSF grantid: 1919147 funderid: 10.13039/100000001 – fundername: Semiconductor Research Corporation (SRC) grantid: Task 2810.034 funderid: 10.13039/100000028 |
| GroupedDBID | -~X .DC 0R~ 29I 3EH 4.4 41~ 5GY 5VS 6IK 97E AAJGR AARMG AASAJ AAWTH ABAZT ABQJQ ABVLG ACGFS ACIWK ACNCT AENEX AETIX AGQYO AGSQL AHBIQ AI. AIBXA AKJIK AKQYR ALLEH ALMA_UNASSIGNED_HOLDINGS ATWAV BEFXN BFFAM BGNUA BKEBE BPEOZ CS3 DU5 EBS EJD F5P HZ~ H~9 IAAWW IBMZZ ICLAB IFIPE IFJZH IPLJI JAVBF LAI M43 O9- OCL P2P PZZ RIA RIE RNS TAE TN5 UKR VH1 AAYXX CITATION 7SP 8FD L7M |
| ID | FETCH-LOGICAL-c294t-e447a1ac5d21e722fe152ea885790a475adf1f28ab6f080a48aa143be74a2a853 |
| IEDL.DBID | RIE |
| ISICitedReferencesCount | 19 |
| ISICitedReferencesURI | http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=001078947700001&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D |
| ISSN | 0018-9200 |
| IngestDate | Sun Nov 09 08:17:37 EST 2025 Sat Nov 29 02:11:55 EST 2025 Tue Nov 18 22:16:04 EST 2025 Wed Aug 27 01:58:40 EDT 2025 |
| IsPeerReviewed | true |
| IsScholarly | true |
| Issue | 3 |
| 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 |
| LinkModel | DirectLink |
| MergedId | FETCHMERGED-LOGICAL-c294t-e447a1ac5d21e722fe152ea885790a475adf1f28ab6f080a48aa143be74a2a853 |
| Notes | ObjectType-Article-1 SourceType-Scholarly Journals-1 ObjectType-Feature-2 content type line 14 |
| ORCID | 0000-0001-6794-8806 0000-0002-4813-3844 0000-0002-2428-7099 0000-0003-3603-6016 0000-0002-9722-0979 0000-0002-7345-1916 0009-0006-5970-8144 |
| PQID | 2932564746 |
| PQPubID | 85482 |
| PageCount | 12 |
| ParticipantIDs | proquest_journals_2932564746 crossref_citationtrail_10_1109_JSSC_2023_3313519 ieee_primary_10266328 crossref_primary_10_1109_JSSC_2023_3313519 |
| PublicationCentury | 2000 |
| PublicationDate | 2024-03-01 |
| PublicationDateYYYYMMDD | 2024-03-01 |
| PublicationDate_xml | – month: 03 year: 2024 text: 2024-03-01 day: 01 |
| PublicationDecade | 2020 |
| PublicationPlace | New York |
| PublicationPlace_xml | – name: New York |
| PublicationTitle | IEEE journal of solid-state circuits |
| PublicationTitleAbbrev | JSSC |
| PublicationYear | 2024 |
| Publisher | IEEE The Institute of Electrical and Electronics Engineers, Inc. (IEEE) |
| Publisher_xml | – name: IEEE – name: The Institute of Electrical and Electronics Engineers, Inc. (IEEE) |
| References | ref13 ref35 ref12 ref34 ref15 ref37 ref14 ref36 ref31 ref30 ref11 ref33 ref10 ref32 ref2 ref17 ref16 ref19 ref18 ref24 ref23 ref26 ref25 ref20 ref22 ref21 ref28 ref27 ref29 ref8 ref7 ref9 ref4 ref3 ref6 ref5 Brown (ref1); 33 |
| References_xml | – ident: ref28 doi: 10.1109/isscc42614.2022.9731762 – ident: ref12 doi: 10.1109/JSSC.2020.2992886 – ident: ref37 doi: 10.1109/isscc42614.2022.9731659 – ident: ref4 doi: 10.1109/ICCV48922.2021.00060 – ident: ref14 doi: 10.1109/MSSC.2019.2922889 – ident: ref34 doi: 10.1145/2742060.2743760 – ident: ref25 doi: 10.1109/isscc42613.2021.9365766 – ident: ref8 doi: 10.1109/isscc42613.2021.9365928 – ident: ref32 doi: 10.1109/ISOCC.2015.7401667 – volume: 33 start-page: 1877 volume-title: Proc. Adv. Neural Inf. Process. Syst. ident: ref1 article-title: Language models are few-shot learners – ident: ref31 doi: 10.1109/ISSCC42615.2023.10067720 – ident: ref35 doi: 10.1109/JSSC.2018.2867275 – ident: ref23 doi: 10.1109/JSSC.2019.2939682 – ident: ref18 doi: 10.1109/JSSC.2021.3108344 – ident: ref24 doi: 10.1109/JSSC.2021.3061508 – ident: ref9 doi: 10.1109/isscc42613.2021.9365791 – ident: ref30 doi: 10.1109/isscc42615.2023.10067305 – ident: ref19 doi: 10.1109/MSSC.2022.3182935 – ident: ref7 doi: 10.1109/isscc19947.2020.9063111 – ident: ref3 doi: 10.1007/978-3-030-58452-8_13 – ident: ref26 doi: 10.1109/ISSCC42614.2022.9731754 – ident: ref16 doi: 10.1038/s41565-020-0655-z – ident: ref21 doi: 10.1109/JSSC.2019.2952773 – ident: ref11 doi: 10.1109/JSSC.2019.2963616 – ident: ref2 doi: 10.1109/CVPR42600.2020.01079 – ident: ref6 doi: 10.1109/JSSC.2016.2616357 – ident: ref29 doi: 10.1109/isscc42614.2022.9731545 – ident: ref10 doi: 10.1109/isscc42614.2022.9731757 – ident: ref17 doi: 10.1109/JSSC.2020.3029586 – ident: ref15 doi: 10.1109/JSSC.2021.3119018 – ident: ref20 doi: 10.1109/CICC53496.2022.9772781 – ident: ref22 doi: 10.1109/JSSC.2019.2899730 – ident: ref36 doi: 10.1109/3.748832 – ident: ref27 doi: 10.1109/VLSITechnologyandCir46769.2022.9830438 – ident: ref33 doi: 10.1109/TCAD.2012.2217962 – ident: ref5 doi: 10.1109/ICCV48922.2021.00986 – ident: ref13 doi: 10.1109/JSSC.2018.2880918 |
| SSID | ssj0014481 |
| Score | 2.5273652 |
| Snippet | Recent SRAM-based in-memory computing (IMC) hardware demonstrates high energy efficiency and throughput for matrix-vector multiplication (MVM), the dominant... Recent SRAM-based in-memory computing (IMC) hardware demonstrates high energy efficiency and throughput for matrix–vector multiplication (MVM), the dominant... |
| SourceID | proquest crossref ieee |
| SourceType | Aggregation Database Enrichment Source Index Database Publisher |
| StartPage | 960 |
| SubjectTerms | Accuracy Approximate computing Approximation approximation-aware training Arithmetic Artificial neural networks Compressors Computation Deep learning Hardware In-memory computing in-memory computing (IMC) Integrated circuit modeling Mathematical analysis Multiplication neural network accelerators Static random access memory |
| Title | DIMCA: An Area-Efficient Digital In-Memory Computing Macro Featuring Approximate Arithmetic Hardware in 28 nm |
| URI | https://ieeexplore.ieee.org/document/10266328 https://www.proquest.com/docview/2932564746 |
| Volume | 59 |
| WOSCitedRecordID | wos001078947700001&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 Electronic Library (IEL) customDbUrl: eissn: 1558-173X dateEnd: 99991231 omitProxy: false ssIdentifier: ssj0014481 issn: 0018-9200 databaseCode: RIE dateStart: 19660101 isFulltext: true titleUrlDefault: https://ieeexplore.ieee.org/ providerName: IEEE |
| link | http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV3PS-wwEA66ePAd_P1wdZUcPD3I2mazTeptWRUVVx6o4K1Mm4mv8LbKuv76752kVRRR8NbDTAj9ks6XTmY-xnZQG3RR5ITVMhdK2kRAoaXwKaVYRi7u1S3zT_XZmbm6Sv82xeqhFgYRw-Uz7PrHkMu3N8W9_1VGO5zCSU-aWTardVIXa72lDOicUcvjxbSDCfsmhRlH6e7J-fmw63XCu71eUKT7EISCqsqnT3GIL4eLP5zZEltoiCQf1MgvsxmsVtivd-0FV9l4_3g0HOzxQUVmCOIgtIuggfh-ee3FQvhxJUb-qu0zr9UdyI2PgCbNPTUMFYx84LuOP5XEbJGGKaf_xr7ukfuU_yNMkJcVl4ZX4zV2eXhwMTwSjbyCKGSqpgKV0hBD0bcyRi2lQ4rlCMb0dRqB0n2wLnbSQJ444pWgDACxqxy1AgkU5n-zVnVT4TrjFoj4SU10pOgriFxqXA5JDs46a5yybRa9vu-saHqPewmM_1k4g0Rp5iHKPERZA1Gb_Xlzua0bb3xnvOYxeWdYw9FmnVdUs2Zv3mVEcIjnKa2SjS_cNtk8ja7qq2Yd1ppO7nGLzRUP0_Jush2W3QssftNv |
| linkProvider | IEEE |
| linkToHtml | http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV1Nb9QwEB1BQaIc-CxioYAPnJC8OF5n7XBbbVt1YXeF1CL1Fk3iMURi02q75ePfM3bSqgiBxC2HGcfKizPPGc88gNdkHQWlgvRWV9JoP5ZYWy1jSinTKmSjrmX-3C6X7uSk-NgXq6daGCJKh89oGC9TLt-f1hfxVxmvcA4nI-1uwq3cGK26cq2rpAHvNDqBvIzXMKPfJzEzVbx9f3Q0HUal8OFolDTpfgtDSVflj49xijAH9_9zbg_gXk8lxaTD_iHcoPYR3L3WYPAxrPZmi-nknZi0bEYo91PDCB5I7DWfo1yImLVyEQ_b_hSdvgO7iQXypEUkh6mGUUxi3_EfDXNb4mGazZdVrHwUMen_HdckmlZoJ9rVDnw62D-eHspeYEHWujAbScZYzLDOvc7Iah2Iozmhc7ktFBqbow9Z0A6rcWBmicYhMr-qyBrUyIH-CWy1py09BeGRqZ-2TEjq3KAKhQsVjisMPngXjB-AunzeZd13H48iGF_LtAtRRRkhKiNEZQ_RAN5cuZx1rTf-ZbwTMblm2MExgN1LVMt-dZ6XTHGY6Rlrxs_-4vYK7hweL-blfLb88By2-U6mO3i2C1ub9QW9gNv1t01zvn6ZXsFfVXDWtg |
| 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=DIMCA%3A+An+Area-Efficient+Digital+In-Memory+Computing+Macro+Featuring+Approximate+Arithmetic+Hardware+in+28+nm&rft.jtitle=IEEE+journal+of+solid-state+circuits&rft.au=Chuan-Tung%2C+Lin&rft.au=Wang%2C+Dewei&rft.au=Zhang%2C+Bo&rft.au=Chen%2C+Gregory+K&rft.date=2024-03-01&rft.pub=The+Institute+of+Electrical+and+Electronics+Engineers%2C+Inc.+%28IEEE%29&rft.issn=0018-9200&rft.eissn=1558-173X&rft.volume=59&rft.issue=3&rft.spage=960&rft_id=info:doi/10.1109%2FJSSC.2023.3313519&rft.externalDBID=NO_FULL_TEXT |
| thumbnail_l | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=0018-9200&client=summon |
| thumbnail_m | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=0018-9200&client=summon |
| thumbnail_s | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=0018-9200&client=summon |