Sparseloop: An Analytical Approach To Sparse Tensor Accelerator Modeling

In recent years, many accelerators have been proposed to efficiently process sparse tensor algebra applications (e.g., sparse neural networks). However, these proposals are single points in a large and diverse design space. The lack of systematic description and modeling support for these sparse ten...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:2022 55th IEEE/ACM International Symposium on Microarchitecture (MICRO) S. 1377 - 1395
Hauptverfasser: Wu, Yannan Nellie, Tsai, Po-An, Parashar, Angshuman, Sze, Vivienne, Emer, Joel S.
Format: Tagungsbericht
Sprache:Englisch
Veröffentlicht: IEEE 01.10.2022
Schlagworte:
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Abstract In recent years, many accelerators have been proposed to efficiently process sparse tensor algebra applications (e.g., sparse neural networks). However, these proposals are single points in a large and diverse design space. The lack of systematic description and modeling support for these sparse tensor accelerators impedes hardware designers from efficient and effective design space exploration. This paper first presents a unified taxonomy to systematically describe the diverse sparse tensor accelerator design space. Based on the proposed taxonomy, it then introduces Sparseloop, the first fast, accurate, and flexible analytical modeling framework to enable early-stage evaluation and exploration of sparse tensor accelerators. Sparseloop comprehends a large set of architecture specifications, including various dataflows and sparse acceleration features (e.g., elimination of zero-based compute). Using these specifications, Sparseloop evaluates a design's processing speed and energy efficiency while accounting for data movement and compute incurred by the employed dataflow, including the savings and overhead introduced by the sparse acceleration features using stochastic density models. Across representative accelerator designs and workloads, Sparseloop achieves over 2000× faster modeling speed than cycle-level simulations, maintains relative performance trends, and achieves 0.1% to 8% average error. The paper also presents example use cases of Sparseloop in different accelerator design flows to reveal important design insights.
AbstractList In recent years, many accelerators have been proposed to efficiently process sparse tensor algebra applications (e.g., sparse neural networks). However, these proposals are single points in a large and diverse design space. The lack of systematic description and modeling support for these sparse tensor accelerators impedes hardware designers from efficient and effective design space exploration. This paper first presents a unified taxonomy to systematically describe the diverse sparse tensor accelerator design space. Based on the proposed taxonomy, it then introduces Sparseloop, the first fast, accurate, and flexible analytical modeling framework to enable early-stage evaluation and exploration of sparse tensor accelerators. Sparseloop comprehends a large set of architecture specifications, including various dataflows and sparse acceleration features (e.g., elimination of zero-based compute). Using these specifications, Sparseloop evaluates a design's processing speed and energy efficiency while accounting for data movement and compute incurred by the employed dataflow, including the savings and overhead introduced by the sparse acceleration features using stochastic density models. Across representative accelerator designs and workloads, Sparseloop achieves over 2000× faster modeling speed than cycle-level simulations, maintains relative performance trends, and achieves 0.1% to 8% average error. The paper also presents example use cases of Sparseloop in different accelerator design flows to reveal important design insights.
Author Tsai, Po-An
Parashar, Angshuman
Sze, Vivienne
Wu, Yannan Nellie
Emer, Joel S.
Author_xml – sequence: 1
  givenname: Yannan Nellie
  surname: Wu
  fullname: Wu, Yannan Nellie
  email: nelliewu@mit.edu
  organization: MIT,Cambridge,US
– sequence: 2
  givenname: Po-An
  surname: Tsai
  fullname: Tsai, Po-An
  email: poant@nvidia.com
  organization: NVIDIA,Westford,US
– sequence: 3
  givenname: Angshuman
  surname: Parashar
  fullname: Parashar, Angshuman
  email: aparashar@nvidia.com
  organization: NVIDIA,Westford,US
– sequence: 4
  givenname: Vivienne
  surname: Sze
  fullname: Sze, Vivienne
  email: sze@mit.edu
  organization: MIT,Cambridge,US
– sequence: 5
  givenname: Joel S.
  surname: Emer
  fullname: Emer, Joel S.
  email: jsemer@mit.edu
  organization: MIT / NVIDIA,Cambridge,US
BookMark eNotjF1LwzAYRiMoqLO_QIT8gdV8f3hXirrBxkDr9UiTN1qITUl7s3_vYMID51wcnnt0PeYREHqipKaU2Of9tv04SMWEqRlhrCaEWHWFKqsNVUoKxTTjt6ia56EnkkttmJR3aPM5uTJDynl6wc14nkunZfAu4WaaSnb-B3cZXyrcwTjnghvvIUFxy9n3OUAaxu8HdBNdmqH65wp9vb127Wa9O7xv22a3dszoZU2ZFD6AdZRy46gPXnqhbOyDAsJ5jNQI6fpAg3YqisiJNFxCpFIYHwjhK_R4-R0A4DiV4deV09Faxg3R_A8SDU3-
CODEN IEEPAD
ContentType Conference Proceeding
DBID 6IE
6IL
CBEJK
RIE
RIL
DOI 10.1109/MICRO56248.2022.00096
DatabaseName IEEE Electronic Library (IEL) Conference Proceedings
IEEE Proceedings Order Plan All Online (POP All Online) 1998-present by volume
IEEE Xplore All Conference Proceedings
IEEE Electronic Library (IEL)
IEEE Proceedings Order Plans (POP All) 1998-Present
DatabaseTitleList
Database_xml – sequence: 1
  dbid: RIE
  name: IEEE Electronic Library (IEL)
  url: https://ieeexplore.ieee.org/
  sourceTypes: Publisher
DeliveryMethod fulltext_linktorsrc
EISBN 9781665462723
1665462728
EndPage 1395
ExternalDocumentID 9923807
Genre orig-research
GroupedDBID 6IE
6IL
ACM
ALMA_UNASSIGNED_HOLDINGS
APO
CBEJK
LHSKQ
RIE
RIL
ID FETCH-LOGICAL-a287t-1254cde9a1138a1cdc5c469fbd6e033ff1845abd1d7a6f4f305835ef1548cd003
IEDL.DBID RIE
ISICitedReferencesCount 47
ISICitedReferencesURI http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=000886530600080&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
IngestDate Wed Aug 27 02:51:45 EDT 2025
IsPeerReviewed false
IsScholarly true
Language English
LinkModel DirectLink
MergedId FETCHMERGED-LOGICAL-a287t-1254cde9a1138a1cdc5c469fbd6e033ff1845abd1d7a6f4f305835ef1548cd003
PageCount 19
ParticipantIDs ieee_primary_9923807
PublicationCentury 2000
PublicationDate 2022-Oct.
PublicationDateYYYYMMDD 2022-10-01
PublicationDate_xml – month: 10
  year: 2022
  text: 2022-Oct.
PublicationDecade 2020
PublicationTitle 2022 55th IEEE/ACM International Symposium on Microarchitecture (MICRO)
PublicationTitleAbbrev MICRO
PublicationYear 2022
Publisher IEEE
Publisher_xml – name: IEEE
SSID ssib053578255
Score 2.4622266
Snippet In recent years, many accelerators have been proposed to efficiently process sparse tensor algebra applications (e.g., sparse neural networks). However, these...
SourceID ieee
SourceType Publisher
StartPage 1377
SubjectTerms Analytical modeling
Analytical models
Computational modeling
Hardware Accelerator
Neural networks
Stochastic processes
Systematics
Taxonomy
Tensor computation
Tensors
Title Sparseloop: An Analytical Approach To Sparse Tensor Accelerator Modeling
URI https://ieeexplore.ieee.org/document/9923807
WOSCitedRecordID wos000886530600080&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
link http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV09T8MwELVKxcAEqEV8ywMjpkltJw5bhag6lQqK1K2Kz2cJCSVVafn9nJ3QLixsUWIpyjnWu2ffe8fYHYDKweVSOF9YoVKLwvihFQYJELBQ1oGJzSby6dQsFsWsw-53WhhEjMVn-BAu41m-q2EbtsoGBWUjJkjHD_I8a7Rav_-ODq4tlB63Ip00KQb0Ga8vBO-xgmsYfTmjNf--iUrEkPHx_95-wvp7MR6f7WDmlHWw6rHJ24oYKX7W9eqRjyoevUXitjQftS7hfF7zZhSfE1et13wEQCATz9V56IEWlOh99j5-nj9NRNsUQZREbjaCEhIFDosyTaUpU3CggSiuty7DRErvibLp0rrU5WXmlaf1TEkW-kBNwNEaPmPdqq7wnPGMhmkl0etE0kNplfYlJi4xPtNEiy5YL0RhuWp8L5ZtAC7_vn3FjkKYm0K3a9bdrLd4ww7he_Pxtb6Nk_UDHu-WzA
linkProvider IEEE
linkToHtml http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV3PS8MwFA5DBT2pbOJvc_BoXLskbeptiGPinEMr7Daa5AUEacfc_Pt9Set28eKttIHSl4bvfcn7vkfItTEiNTblzLpMMxFrYMr1NFOAgACZ0Nao0GwiHY_VdJpNWuRmrYUBgFB8Brf-Mpzl28qs_FZZN8NsRHnp-LYUohfVaq3fv0d63xZMkBuZThxlXfyQ1xcE-FDD1QvOnMGcf9NGJaDIYP9_7z8gnY0cj07WQHNIWlC2yfBtjpwUPqtqfkf7JQ3uImFjmvYbn3CaV7QeRXNkq9WC9o1BmAkn69R3QfNa9A55Hzzk90PWtEVgBdKbJcOURBgLWRHHXBWxsUYaJLlO2wQizp1D0iYLbWObFokTDlc0plngPDkxFlfxEdkqqxKOCU1wmBQcnIw4PuRaSFdAZCPlEonE6IS0fRRm89r5YtYE4PTv21dkd5g_j2ajx_HTGdnzIa_L3s7J1nKxgguyY76XH1-LyzBxPyjsmhM
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%3Abook&rft.genre=proceeding&rft.title=2022+55th+IEEE%2FACM+International+Symposium+on+Microarchitecture+%28MICRO%29&rft.atitle=Sparseloop%3A+An+Analytical+Approach+To+Sparse+Tensor+Accelerator+Modeling&rft.au=Wu%2C+Yannan+Nellie&rft.au=Tsai%2C+Po-An&rft.au=Parashar%2C+Angshuman&rft.au=Sze%2C+Vivienne&rft.date=2022-10-01&rft.pub=IEEE&rft.spage=1377&rft.epage=1395&rft_id=info:doi/10.1109%2FMICRO56248.2022.00096&rft.externalDocID=9923807