SpinLiM: Spin Orbit Torque Memory for Ternary Neural Networks Based on the Logic-in-Memory Architecture

Lichuan Luo; He Zhang; Jinyu Bai; Youguang Zhang; Wang Kang; Weisheng Zhao

doi:10.23919/DATE51398.2021.9474022

SpinLiM: Spin Orbit Torque Memory for Ternary Neural Networks Based on the Logic-in-Memory Architecture

Lichuan Luo
, He Zhang
, Jinyu Bai
, Youguang Zhang
, Wang Kang
, Weisheng Zhao

Beihang University

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

8 Scopus citations

Abstract

Logic-in-memory architecture based on spintronic memories shows fascinating prospects in neural networks (NNs) for its high energy efficiency and good endurance. In this work, we leveraged two magnetic tunnel junctions (MTJs), which are driven by the interplay of field-free spin orbit torque (SOT) and spin transfer torque (STT) effects, to achieve a novel statefullogic-in-memory paradigm for ternary multiplication operations. Based on this paradigm, we further proposed a highly parallel array structure to serve for ternary neural networks (TNNs). Our results demonstrate the advantage of our design in power consumption compared with CPU, GPU and other state-of-the-art works.

Original language	English
Title of host publication	Proceedings of the 2021 Design, Automation and Test in Europe, DATE 2021
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	1865-1870
Number of pages	6
ISBN (Electronic)	9783981926354
DOIs	https://doi.org/10.23919/DATE51398.2021.9474022
State	Published - 1 Feb 2021
Event	2021 Design, Automation and Test in Europe Conference and Exhibition, DATE 2021 - Virtual, Online Duration: 1 Feb 2021 → 5 Feb 2021

Publication series

Name	Proceedings -Design, Automation and Test in Europe, DATE
Volume	2021-February
ISSN (Print)	1530-1591

Conference

Conference	2021 Design, Automation and Test in Europe Conference and Exhibition, DATE 2021
City	Virtual, Online
Period	1/02/21 → 5/02/21

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

SDG 7 Affordable and Clean Energy

Keywords

Stateful logic-in-memory
magnetic tunnel junction
spin orbit torque memory
ternary neural network

Access to Document

10.23919/DATE51398.2021.9474022

Cite this

Luo, L., Zhang, H., Bai, J., Zhang, Y., Kang, W., & Zhao, W. (2021). SpinLiM: Spin Orbit Torque Memory for Ternary Neural Networks Based on the Logic-in-Memory Architecture. In Proceedings of the 2021 Design, Automation and Test in Europe, DATE 2021 (pp. 1865-1870). Article 9474022 (Proceedings -Design, Automation and Test in Europe, DATE; Vol. 2021-February). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.23919/DATE51398.2021.9474022

@inproceedings{4eb2118a285340f5b532e301bde889a5,

title = "SpinLiM: Spin Orbit Torque Memory for Ternary Neural Networks Based on the Logic-in-Memory Architecture",

abstract = "Logic-in-memory architecture based on spintronic memories shows fascinating prospects in neural networks (NNs) for its high energy efficiency and good endurance. In this work, we leveraged two magnetic tunnel junctions (MTJs), which are driven by the interplay of field-free spin orbit torque (SOT) and spin transfer torque (STT) effects, to achieve a novel statefullogic-in-memory paradigm for ternary multiplication operations. Based on this paradigm, we further proposed a highly parallel array structure to serve for ternary neural networks (TNNs). Our results demonstrate the advantage of our design in power consumption compared with CPU, GPU and other state-of-the-art works.",

keywords = "Stateful logic-in-memory, magnetic tunnel junction, spin orbit torque memory, ternary neural network",

author = "Lichuan Luo and He Zhang and Jinyu Bai and Youguang Zhang and Wang Kang and Weisheng Zhao",

note = "Publisher Copyright: {\textcopyright} 2021 EDAA.; 2021 Design, Automation and Test in Europe Conference and Exhibition, DATE 2021 ; Conference date: 01-02-2021 Through 05-02-2021",

year = "2021",

month = feb,

day = "1",

doi = "10.23919/DATE51398.2021.9474022",

language = "英语",

series = "Proceedings -Design, Automation and Test in Europe, DATE",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

pages = "1865--1870",

booktitle = "Proceedings of the 2021 Design, Automation and Test in Europe, DATE 2021",

address = "美国",

}

Luo, L, Zhang, H, Bai, J, Zhang, Y , Kang, W & Zhao, W 2021, SpinLiM: Spin Orbit Torque Memory for Ternary Neural Networks Based on the Logic-in-Memory Architecture. in Proceedings of the 2021 Design, Automation and Test in Europe, DATE 2021., 9474022, Proceedings -Design, Automation and Test in Europe, DATE, vol. 2021-February, Institute of Electrical and Electronics Engineers Inc., pp. 1865-1870, 2021 Design, Automation and Test in Europe Conference and Exhibition, DATE 2021, Virtual, Online, 1/02/21. https://doi.org/10.23919/DATE51398.2021.9474022

SpinLiM: Spin Orbit Torque Memory for Ternary Neural Networks Based on the Logic-in-Memory Architecture. / Luo, Lichuan; Zhang, He; Bai, Jinyu et al.
Proceedings of the 2021 Design, Automation and Test in Europe, DATE 2021. Institute of Electrical and Electronics Engineers Inc., 2021. p. 1865-1870 9474022 (Proceedings -Design, Automation and Test in Europe, DATE; Vol. 2021-February).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - SpinLiM

T2 - 2021 Design, Automation and Test in Europe Conference and Exhibition, DATE 2021

AU - Luo, Lichuan

AU - Zhang, He

AU - Bai, Jinyu

AU - Zhang, Youguang

AU - Kang, Wang

AU - Zhao, Weisheng

PY - 2021/2/1

Y1 - 2021/2/1

N2 - Logic-in-memory architecture based on spintronic memories shows fascinating prospects in neural networks (NNs) for its high energy efficiency and good endurance. In this work, we leveraged two magnetic tunnel junctions (MTJs), which are driven by the interplay of field-free spin orbit torque (SOT) and spin transfer torque (STT) effects, to achieve a novel statefullogic-in-memory paradigm for ternary multiplication operations. Based on this paradigm, we further proposed a highly parallel array structure to serve for ternary neural networks (TNNs). Our results demonstrate the advantage of our design in power consumption compared with CPU, GPU and other state-of-the-art works.

AB - Logic-in-memory architecture based on spintronic memories shows fascinating prospects in neural networks (NNs) for its high energy efficiency and good endurance. In this work, we leveraged two magnetic tunnel junctions (MTJs), which are driven by the interplay of field-free spin orbit torque (SOT) and spin transfer torque (STT) effects, to achieve a novel statefullogic-in-memory paradigm for ternary multiplication operations. Based on this paradigm, we further proposed a highly parallel array structure to serve for ternary neural networks (TNNs). Our results demonstrate the advantage of our design in power consumption compared with CPU, GPU and other state-of-the-art works.

KW - Stateful logic-in-memory

KW - magnetic tunnel junction

KW - spin orbit torque memory

KW - ternary neural network

UR - https://www.scopus.com/pages/publications/85111027410

U2 - 10.23919/DATE51398.2021.9474022

DO - 10.23919/DATE51398.2021.9474022

M3 - 会议稿件

AN - SCOPUS:85111027410

T3 - Proceedings -Design, Automation and Test in Europe, DATE

SP - 1865

EP - 1870

BT - Proceedings of the 2021 Design, Automation and Test in Europe, DATE 2021

PB - Institute of Electrical and Electronics Engineers Inc.

Y2 - 1 February 2021 through 5 February 2021

ER -

Luo L, Zhang H, Bai J, Zhang Y , Kang W , Zhao W. SpinLiM: Spin Orbit Torque Memory for Ternary Neural Networks Based on the Logic-in-Memory Architecture. In Proceedings of the 2021 Design, Automation and Test in Europe, DATE 2021. Institute of Electrical and Electronics Engineers Inc. 2021. p. 1865-1870. 9474022. (Proceedings -Design, Automation and Test in Europe, DATE). doi: 10.23919/DATE51398.2021.9474022

SpinLiM: Spin Orbit Torque Memory for Ternary Neural Networks Based on the Logic-in-Memory Architecture

Abstract

Publication series

Conference

UN SDGs

Keywords

Access to Document

Other files and links

Cite this