Scalable high-performance architecture for convolutional ternary neural networks on FPGA

Thanks to their excellent performances on typical artificial intelligence problems, deep neural networks have drawn a lot of interest lately. However, this comes at the cost of large computational needs and high power consumption. Benefiting from high precision at acceptable hardware cost on these d...

Full description

Saved in:
Bibliographic Details
Published in:International Conference on Field-programmable Logic and Applications pp. 1 - 7
Main Authors: Prost-Boucle, Adrien, Bourge, Alban, Petrot, Frederic, Alemdar, Hande, Caldwell, Nicholas, Leroy, Vincent
Format: Conference Proceeding
Language:English
Published: Ghent University 01.09.2017
Subjects:
ISSN:1946-1488
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Abstract Thanks to their excellent performances on typical artificial intelligence problems, deep neural networks have drawn a lot of interest lately. However, this comes at the cost of large computational needs and high power consumption. Benefiting from high precision at acceptable hardware cost on these difficult problems is a challenge. To address it, we advocate the use of ternary neural networks (TNN) that, when properly trained, can reach results close to the state of the art using floatingpoint arithmetic. We present a highly versatile FPGA friendly architecture for TNN in which we can vary both the number of bits of the input data and the level of parallelism at synthesis time, allowing to trade throughput for hardware resources and power consumption. To demonstrate the efficiency of our proposal, we implement high-complexity convolutional neural networks on the Xilinx Virtex-7 VC709 FPGA board. While reaching a better accuracy than comparable designs, we can target either high throughput or low power. We measure a throughput up to 27 000 fps at ≈7W or up to 8.36 TMAC/s at ≈13 W.
AbstractList Thanks to their excellent performances on typical artificial intelligence problems, deep neural networks have drawn a lot of interest lately. However, this comes at the cost of large computational needs and high power consumption. Benefiting from high precision at acceptable hardware cost on these difficult problems is a challenge. To address it, we advocate the use of ternary neural networks (TNN) that, when properly trained, can reach results close to the state of the art using floatingpoint arithmetic. We present a highly versatile FPGA friendly architecture for TNN in which we can vary both the number of bits of the input data and the level of parallelism at synthesis time, allowing to trade throughput for hardware resources and power consumption. To demonstrate the efficiency of our proposal, we implement high-complexity convolutional neural networks on the Xilinx Virtex-7 VC709 FPGA board. While reaching a better accuracy than comparable designs, we can target either high throughput or low power. We measure a throughput up to 27 000 fps at ≈7W or up to 8.36 TMAC/s at ≈13 W.
Author Petrot, Frederic
Caldwell, Nicholas
Bourge, Alban
Alemdar, Hande
Leroy, Vincent
Prost-Boucle, Adrien
Author_xml – sequence: 1
  givenname: Adrien
  surname: Prost-Boucle
  fullname: Prost-Boucle, Adrien
  email: Adrien.Prost-Boucle@univ-grenoble-alpes.fr
  organization: Grenoble INP, Univ. Grenoble Alpes, Grenoble, France
– sequence: 2
  givenname: Alban
  surname: Bourge
  fullname: Bourge, Alban
  email: Alban.Bourge@univ-grenoble-alpes.fr
  organization: Grenoble INP, Univ. Grenoble Alpes, Grenoble, France
– sequence: 3
  givenname: Frederic
  surname: Petrot
  fullname: Petrot, Frederic
  email: Fredaic.Petrot@univ-grenoble-alpes.fr
  organization: Grenoble INP, Univ. Grenoble Alpes, Grenoble, France
– sequence: 4
  givenname: Hande
  surname: Alemdar
  fullname: Alemdar, Hande
  email: Hande.Alemdar@univ-grenoble-alpes.fr
  organization: Grenoble INP, Univ. Grenoble Alpes, Grenoble, France
– sequence: 5
  givenname: Nicholas
  surname: Caldwell
  fullname: Caldwell, Nicholas
  email: Nicholas.Caldwell@univ-grenoble-alpes.fr
  organization: Grenoble INP, Univ. Grenoble Alpes, Grenoble, France
– sequence: 6
  givenname: Vincent
  surname: Leroy
  fullname: Leroy, Vincent
  email: Vincent.Leroy@univ-grenoble-alpes.fr
  organization: Grenoble INP, Univ. Grenoble Alpes, Grenoble, France
BookMark eNotkNtKw0AYhFdRsK19APFmXyB1z4fLUmwVAhZU8K78Tf610XS3bFKlb2_Azs0w38UwzJhcxRSRkDvOZkJ67h-W63ImGLczx7Rxml2QsWeeSaaEE5dkxL0yBVfO3ZBp132xQVpZp82IfLxW0MK2RbprPnfFAXNIeQ-xQgq52jU9Vv0xIx0orVL8Se2xb1KElvaYI-QTjXjMQ4zY_6b83dEU6XK9mt-S6wBth9OzT8j78vFt8VSUL6vnxbwsGqF4X3BrgjUSLXBvWQhGG811rQILTjIwQtdGWKmF9UpDzWFr6xqVAwf11jojJ-T-v7dBxM0hN_th1OZ8hPwDtXRUMg
ContentType Conference Proceeding
DBID 6IE
6IL
CBEJK
RIE
RIL
DOI 10.23919/FPL.2017.8056850
DatabaseName IEEE Electronic Library (IEL) Conference Proceedings
IEEE Xplore POP ALL
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 9090304282
9789090304281
EISSN 1946-1488
EndPage 7
ExternalDocumentID 8056850
Genre orig-research
GroupedDBID 6IE
6IF
6IL
6IN
AAWTH
ABLEC
ADZIZ
ALMA_UNASSIGNED_HOLDINGS
BEFXN
BFFAM
BGNUA
BKEBE
BPEOZ
CBEJK
CHZPO
IEGSK
OCL
RIE
RIL
ID FETCH-LOGICAL-i241t-176f763e7a1970ff656515d4f0f830a625d6273527945ad1ab7dde48a8adb7863
IEDL.DBID RIE
ISICitedReferencesCount 35
ISICitedReferencesURI http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=000426989400094&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 02:28:39 EDT 2025
IsPeerReviewed false
IsScholarly false
Language English
LinkModel DirectLink
MergedId FETCHMERGED-LOGICAL-i241t-176f763e7a1970ff656515d4f0f830a625d6273527945ad1ab7dde48a8adb7863
PageCount 7
ParticipantIDs ieee_primary_8056850
PublicationCentury 2000
PublicationDate 2017-09
PublicationDateYYYYMMDD 2017-09-01
PublicationDate_xml – month: 09
  year: 2017
  text: 2017-09
PublicationDecade 2010
PublicationTitle International Conference on Field-programmable Logic and Applications
PublicationTitleAbbrev FPL
PublicationYear 2017
Publisher Ghent University
Publisher_xml – name: Ghent University
SSID ssj0000547856
Score 1.9653509
Snippet Thanks to their excellent performances on typical artificial intelligence problems, deep neural networks have drawn a lot of interest lately. However, this...
SourceID ieee
SourceType Publisher
StartPage 1
SubjectTerms Field programmable gate arrays
Hardware
Neural networks
Neurons
Parallel processing
Random access memory
Throughput
Title Scalable high-performance architecture for convolutional ternary neural networks on FPGA
URI https://ieeexplore.ieee.org/document/8056850
WOSCitedRecordID wos000426989400094&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/eLvHCXMwlV1LS8NAEB5q8eBJpRXf7MGjaZNms4-jiNGDlIAKvZVJMgsFSUtf4L93NgmtghdvyUIITDYz38588w3AnUVLjIYoyGUsA6kUBiaX7AyVK7FgvBDHzbAJPR6bycRmHbjf9cIQUU0-o4G_rGv55bzY-FTZ0HC0Nv6AfqC1bnq1dvmU0AtTJaopXI5iG9lhmr167pYetM_9GqBSx4_0-H9vPoH-vhFPZLsQcwodqnoweWO7-o4n4bWGg8We-i9-lgUErwpPKm83F36KOvm3_BJexJJvq4YCvhLzSqTZ80MfPtKn98eXoB2QEMw48K69uKNj_0AaI6tD5xibMTwppQudiUPko02pGJ4kI_7pEiwjzDV7M2nQYJlro-Iz6Fbzis5ByKhgoBhGlKCWThYWyeRGEVpJkqy-gJ63ynTRaGBMW4Nc_r18BUfe8A0X6xq66-WGbuCw2K5nq-Vt_eG-ATY4muo
linkProvider IEEE
linkToHtml http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV1LS8NAEB5KFfSk0orW1x48mjZpNvs4ihgr1hKwQm9lk8xCQdLSh-C_dzYJrYIXb8lCCMwmM9_OfPMNwK02GgkNoZfykHtcCOOplJMzFDY3GeGFMKyGTcjRSE0mOmnA3bYXBhFL8hl23WVZy8_n2calynqKorVyB_S9iPN-UHVrbTMqvpOmikRVuuyHOtC9OBk69pbs1k_-GqFSRpD46H_vPob2rhWPJdsgcwINLFoweSPLup4n5tSGvcWO_M9-FgYYrTJHK68_L_PByvTf8os5GUu6LSoS-IrNCxYnT_dteI8fxw8Drx6R4M0o9K6dvKMlD4HSBFr61hI6I4CSc-tbFfqGDje5IIAS9em3i0wemFSSP-PKKJOnUonwFJrFvMAzYDzICCr6AUZGcsszbVClSqDRHDlqeQ4tZ5XpolLBmNYG6fy9fAMHg_HrcDp8Hr1cwKHbhIqZdQnN9XKDV7Cffa5nq-V1uYnfc3meMQ
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=proceeding&rft.title=International+Conference+on+Field-programmable+Logic+and+Applications&rft.atitle=Scalable+high-performance+architecture+for+convolutional+ternary+neural+networks+on+FPGA&rft.au=Prost-Boucle%2C+Adrien&rft.au=Bourge%2C+Alban&rft.au=Petrot%2C+Frederic&rft.au=Alemdar%2C+Hande&rft.date=2017-09-01&rft.pub=Ghent+University&rft.eissn=1946-1488&rft.spage=1&rft.epage=7&rft_id=info:doi/10.23919%2FFPL.2017.8056850&rft.externalDocID=8056850