摘要
面向物联网与人工智能的深度结合,考虑到终端在功耗、成本、生命周期等维度的特性需求,文章提出了一种可重构架构流处理器,该处理器基于动态重构模型,适合于承载深度学习算法,从而为智能物联网提供低成本、低功耗、高性能、高资源效费比、长生命周期等,既利于智能物联网产业的应用,也有利于促成我国在半导体领域上的产业突破。
Internet of things(IoT)and Artificial Intelligence(AI)are merging into Artificial Internet of Things(AIoT).A reconfigurable stream processor is proposed to fulfill the special requirements of AIoT in the aspects of power,cost,life cycle and etc.The processor is based on the dynamic edge computing model,and it is suitable for the calculation of deep learning.It can provide low cost,low power,high efficiency,high resource efficiency ratio,long life cycle,and it is thus conducive to IoT application,and essential for China to be the leader of intelligent IoT semiconductors.
作者
钱雨
肖立波
文钧雷
焦捷
徐昕
张涛
Qian Yu;Xiao Libo;Wen Junlei;Jiao Jie;Xu Xin;Zhang Tao(China Industrial Development Research Center,School of Economic Management,Qinghua University,Beijing 100084,China;UVMIC Technology(Zhejiang)Co.,Ltd.,Jiaxing 314100,China)
出处
《信息通信技术》
2019年第4期63-69,共7页
Information and communications Technologies
关键词
智能物联网
可重构架构流处理器
数据驱动
Artificial Internet of Things(AIoT)
Reconfigurable Stream Processor
Data-driven