Skip to main navigation Skip to search Skip to main content

Work-in-Progress: Toward Energy-efficient Near STT-MRAM Processing Architecture for Neural Networks

  • Yueting Li
  • , Bingluo Zhao
  • , Xinyi Xu
  • , Yundong Zhang
  • , Jun Wang*
  • , Weisheng Zhao*
  • *Corresponding author for this work
  • Beihang University
  • Vimicro Corporation

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

The size of parameters in artificial neural network (NN) applications grows quickly from a handful to the GB-level. The data transmission poses a key challenge for NN, and either neuron is removed or data compression reduces pressure on memory access but cannot successfully decrease data traffic. Therefore, we propose the near spin-transfer-torque magnetic random processing architecture for developing energy-efficient NNs. Our approach provides system architects with a preliminary scheme to obtain real-time transmission that near memory controller directly compresses non-zero elements, and encodes the corresponding index depending on the kernel size. Furthermore, it adjusts the number of multiplication accumulators and avoids unnecessary hardware overheads during computation. The preliminary experimental results demonstrated this design verified with weights that currently achieve up to 3.05x speedup and 29.6% power compared with the unoptimized one.

Original languageEnglish
Title of host publicationProceedings - 2022 International Conference on Hardware/Software Codesign and System Synthesis, CODES+ISSS 2022
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages13-14
Number of pages2
ISBN (Electronic)9781665472944
DOIs
StatePublished - 2022
Event2022 International Conference on Hardware/Software Codesign and System Synthesis, CODES+ISSS 2022 - Shanghai, China
Duration: 7 Oct 202214 Oct 2022

Publication series

NameProceedings - 2022 International Conference on Hardware/Software Codesign and System Synthesis, CODES+ISSS 2022

Conference

Conference2022 International Conference on Hardware/Software Codesign and System Synthesis, CODES+ISSS 2022
Country/TerritoryChina
CityShanghai
Period7/10/2214/10/22

Keywords

  • Energy-efficient
  • Near-memory Processing
  • Neural Network
  • STT-MRAM

Fingerprint

Dive into the research topics of 'Work-in-Progress: Toward Energy-efficient Near STT-MRAM Processing Architecture for Neural Networks'. Together they form a unique fingerprint.

Cite this