嵌入式智能处理技术设计实现

Design and Implement of Embedded Intelligent Processing Technology

分析了嵌入式智能计算的相关设计技术,介绍了嵌入式智能计算3类算法的特征、需求和运行场景,包括知识驱动类、智能优化类和深度学习类。针对每种智能算法的处理要求,提出了智能处理部件的选型要求,分析了不同智能处理部件的性能及特点,提出了一种嵌入式智能处理单元架构,由模块支持单元(MSU)、处理单元(PU)、路由单元(RU)、网络接口单元(NIU)、电源支持元件(PSE)等5个部分组成,设计实现了嵌入式智能计算模块,具备高速的数据传输能力和智能计算能力,可以满足嵌入式智能计算的应用要求。

This paper analyzes the related design technology of embedded intelligent computing,and introduces the characteristics,requirements and operation scenarios of three kinds of embedded intelligent computing algorithms,including knowledge-driven class,intelligent optimization class and deep learning class.According to the processing requirements of each intelligent algorithm,the selection requirements of intelligent processing components are proposed,the performance and characteristics of different intelligent processing components are analyzed.The article introduces the selection and requirements of intelligent processors,and analyzes the performance and characteristics of difference types of intelligent processors.The paper proposes an embedded intelligent processing unit architecture,composed of module support unit(MSU),processing unit(PU),routing unit(RU),network interface unit(NIU),power supply support element(PSE).The embedded intelligent computing module is designed and implemented,which has high-speed data transmission capability and intelligent computing capability,and can meet the application requirements of embedded intelligent computing.