• Search Research Projects
  • Search Researchers
  • How to Use
  1. Back to previous page

FPGA implementation of low energy asynchronous convolutional neural network circuits

Research Project

Project/Area Number 18K11221
Research Category

Grant-in-Aid for Scientific Research (C)

Allocation TypeMulti-year Fund
Section一般
Review Section Basic Section 60040:Computer system-related
Research InstitutionThe University of Aizu

Principal Investigator

Saito Hiroshi  会津大学, コンピュータ理工学部, 上級准教授 (50361671)

Co-Investigator(Kenkyū-buntansha) 富岡 洋一  会津大学, コンピュータ理工学部, 上級准教授 (10574072)
Project Period (FY) 2018-04-01 – 2021-03-31
Project Status Completed (Fiscal Year 2020)
Budget Amount *help
¥3,510,000 (Direct Cost: ¥2,700,000、Indirect Cost: ¥810,000)
Fiscal Year 2020: ¥780,000 (Direct Cost: ¥600,000、Indirect Cost: ¥180,000)
Fiscal Year 2019: ¥780,000 (Direct Cost: ¥600,000、Indirect Cost: ¥180,000)
Fiscal Year 2018: ¥1,950,000 (Direct Cost: ¥1,500,000、Indirect Cost: ¥450,000)
Keywords非同期式回路 / FPGA / 畳み込みニューラルネットワーク
Outline of Final Research Achievements

In this research, to clarify the usefulness of asynchronous circuits for deep machine learning, we designed low energy convolutional neural network circuits for image classification on Field Programmable Gate Arrays (FPGAs) as asynchronous circuits. Also, we designed quantized binarized neural network circuits as asynchronous circuits. Compared to synchronous counterparts, the energy consumption of the binarized neural network circuits was reduced to half in the best case. Furthermore, to improve performance, we proposed a design method to implement asynchronous circuits on FPGAs using placement constraints.

Academic Significance and Societal Importance of the Research Achievements

深層学習を実現するにあたり、GPUを用いた場合、電力消費やコストが大きくなる。一方、CPUを用いた場合、性能が問題となる。こうした問題に対し、深層学習を行う回路をFPGAに実現する手法が提案されている。しかし、こうした回路は、クロック信号を基に回路全体を制御する同期式回路として実現されているため、クロック周りの消費電力が大きい。本研究で実現した非同期式回路によって、消費エネルギーを削減することで、電力要求が厳しいアプリケーションでの使用が期待できる。

Report

(4 results)
  • 2020 Annual Research Report   Final Research Report ( PDF )
  • 2019 Research-status Report
  • 2018 Research-status Report
  • Research Products

    (9 results)

All 2021 2020 2019 2018

All Journal Article (3 results) (of which Int'l Joint Research: 1 results,  Peer Reviewed: 3 results) Presentation (6 results) (of which Int'l Joint Research: 6 results)

  • [Journal Article] A Design Method for Designing Asynchronous Circuits on Commercial FPGAs Using Placement Constraints2020

    • Author(s)
      OTAKE Tatsuki、SAITO Hiroshi
    • Journal Title

      IEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences

      Volume: E103.A Issue: 12 Pages: 1427-1436

    • DOI

      10.1587/transfun.2020VLP0006

    • NAID

      130007948286

    • ISSN
      0916-8508, 1745-1337
    • Year and Date
      2020-12-01
    • Related Report
      2020 Annual Research Report
    • Peer Reviewed
  • [Journal Article] Optimization Methods during RTL Conversion from Synchronous RTL Models to Asynchronous RTL Models2020

    • Author(s)
      SEMBA Shogo、SAITO Hiroshi、TATSUOKA Masato、FUJIMURA Katsuya
    • Journal Title

      IEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences

      Volume: E103.A Issue: 12 Pages: 1417-1426

    • DOI

      10.1587/transfun.2020VLP0004

    • NAID

      130007948308

    • ISSN
      0916-8508, 1745-1337
    • Year and Date
      2020-12-01
    • Related Report
      2020 Annual Research Report
    • Peer Reviewed / Int'l Joint Research
  • [Journal Article] Conversion from Synchronous RTL Models to Asynchronous RTL Models2019

    • Author(s)
      S. Semba and H. Saito
    • Journal Title

      IEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences

      Volume: E102.A Issue: 7 Pages: 904-913

    • DOI

      10.1587/transfun.E102.A.904

    • NAID

      130007670864

    • ISSN
      0916-8508, 1745-1337
    • Year and Date
      2019-07-01
    • Related Report
      2019 Research-status Report
    • Peer Reviewed
  • [Presentation] Study on an RTL Conversion Method from Pipelined Synchronous RTL Models into Asynchronous RTL Models2021

    • Author(s)
      SEMBA Shogo、SAITO Hiroshi
    • Organizer
      The 23rd Workshop on Synthesis And System Integration of Mixed Information technologies
    • Related Report
      2020 Annual Research Report
    • Int'l Joint Research
  • [Presentation] Comparison of RTL Conversion and GL Conversion from Synchronous Circuits to Asynchronous Circuits2019

    • Author(s)
      S. Semba and H. Saito
    • Organizer
      IEEE International Symposium on Circuits and Systems
    • Related Report
      2019 Research-status Report
    • Int'l Joint Research
  • [Presentation] A Study on the Optimization of Asynchronous Circuits During RTL Conversion from Synchronous Circuits2019

    • Author(s)
      S. Semba and H. Saito
    • Organizer
      Workshop on Synthesis And System Integration of Mixed Information technologies
    • Related Report
      2019 Research-status Report
    • Int'l Joint Research
  • [Presentation] Design of Asynchronous Circuits on Commercial FPGAs Using Placement Constraints2019

    • Author(s)
      T. Otake and H. Saito
    • Organizer
      Workshop on Synthesis And System Integration of Mixed Information technologies
    • Related Report
      2019 Research-status Report
    • Int'l Joint Research
  • [Presentation] Design of Asynchronous CNN Circuits on Commercial FPGA from Synchronous CNN Circuits2019

    • Author(s)
      H. Kato and H. Saito
    • Organizer
      IEEE International Symposium on Embedded Multicore/Many-core Systems-on-Chip
    • Related Report
      2019 Research-status Report
    • Int'l Joint Research
  • [Presentation] Comparison of Pipelined Asynchronous Circuits Designed for FPGA2018

    • Author(s)
      Takuya Kudo and Hiroshi Saito
    • Organizer
      3rd International Conference on Applications in Information Technology
    • Related Report
      2018 Research-status Report
    • Int'l Joint Research

URL: 

Published: 2018-04-23   Modified: 2022-01-27  

Information User Guide FAQ News Terms of Use Attribution of KAKENHI

Powered by NII kakenhi