An Emerging NVM CIM Accelerator With Shared-Path Transpose Read and Bit-Interleaving Weight Storage for Efficient On-Chip Training in Edge Devices

Author:

Guo Zhiwang1ORCID,Chen Deyang1,Zhao Chenyang1ORCID,Fang Jinbei1,Jiang Jingwen1,Liu Yixuan1,Tian Haidong2,Xiong Xiankui2,Zhou Keji3ORCID,Xue Xiaoyong1ORCID,Liu Qi3ORCID,Zeng Xiaoyang1

Affiliation:

1. State Key Laboratory of ASIC and System, School of Microelectronics, Fudan University, Shanghai, China

2. State Key Laboratory of Mobile Network and Mobile Communication Multimedia Technology, ZTE, Shenzhen, China

3. Frontier Institute of Chip and System, Fudan University, Shanghai, China

Funder

National Natural Science Foundation of China

Science and Technology Commission of Shanghai Municipality

Opening Project of Zhejiang Laboratory

Publisher

Institute of Electrical and Electronics Engineers (IEEE)

Subject

Electrical and Electronic Engineering

Reference18 articles.

1. MINT: Mixed-Precision RRAM-Based IN-Memory Training Architecture

2. Benchmark Non-volatile and Volatile Memory Based Hybrid Precision Synapses for In-situ Deep Neural Network Training

3. A Ferroelectric-based Volatile/Non-volatile Dual-mode Buffer Memory for Deep Neural Network Accelerators

4. CIMAT: A Compute-In-Memory Architecture for On-chip Training Based on Transpose SRAM Arrays

5. A 14nm 100Kb 2T1R Transpose RRAM with150X resistance ratio enhancement and 27.% reduction on energy-latency product using low-power near threshold read operation and fast data-line current stabling scheme;wang;Proc Symp VLSI Technol,2021

同舟云学术

1.学者识别学者识别

2.学术分析学术分析

3.人才评估人才评估

"同舟云学术"是以全球学者为主线,采集、加工和组织学术论文而形成的新型学术文献查询和分析系统,可以对全球学者进行文献检索和人才价值评估。用户可以通过关注某些学科领域的顶尖人物而持续追踪该领域的学科进展和研究前沿。经过近期的数据扩容,当前同舟云学术共收录了国内外主流学术期刊6万余种,收集的期刊论文及会议论文总量共计约1.5亿篇,并以每天添加12000余篇中外论文的速度递增。我们也可以为用户提供个性化、定制化的学者数据。欢迎来电咨询!咨询电话:010-8811{复制后删除}0370

www.globalauthorid.com

TOP

Copyright © 2019-2024 北京同舟云网络信息技术有限公司
京公网安备11010802033243号  京ICP备18003416号-3