HNPU: An Adaptive DNN Training Processor Utilizing Stochastic Dynamic Fixed-Point and Active Bit-Precision Searching

Author:

Han DonghyeonORCID,Im DongseokORCID,Park GwangtaeORCID,Kim YoungwooORCID,Song SeokchanORCID,Lee JuhyoungORCID,Yoo Hoi-JunORCID

Funder

Institute for Information and Communications Technology Promotion (IITP) Grant

Korea Government (MSIP)

Publisher

Institute of Electrical and Electronics Engineers (IEEE)

Subject

Electrical and Electronic Engineering

Cited by 40 articles. 订阅此论文施引文献 订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献

1. Energy and Bandwidth Efficient Sparse Programmable Dataflow Accelerator;IEEE Transactions on Circuits and Systems I: Regular Papers;2024-09

2. A Dynamic Execution Neural Network Processor for Fine-Grained Mixed-Precision Model Training Based on Online Quantization Sensitivity Analysis;IEEE Journal of Solid-State Circuits;2024-09

3. DQ-STP: An Efficient Sparse On-Device Training Processor Based on Low-Rank Decomposition and Quantization for DNN;IEEE Transactions on Circuits and Systems I: Regular Papers;2024-04

4. CAMEL: Co-Designing AI Models and eDRAMs for Efficient On-Device Learning;2024 IEEE International Symposium on High-Performance Computer Architecture (HPCA);2024-03-02

5. LUTein: Dense-Sparse Bit-Slice Architecture With Radix-4 LUT-Based Slice-Tensor Processing Units;2024 IEEE International Symposium on High-Performance Computer Architecture (HPCA);2024-03-02

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