可重构多核处理器设计方法及其关键技术研究

项目名称： 可重构多核处理器设计方法及其关键技术研究

项目编号： No.61274133

项目类型： 面上项目

立项/批准年度： 2013

项目学科： 无线电电子学、电信技术

项目作者： 郭东辉

作者单位： 厦门大学

项目金额： 90万元

中文摘要： 片上多核处理器(CMP)之间通讯要求其互联结构具备高的吞吐率和低的延迟特性，同时鉴于CMP芯片将来在嵌入式系统及移动计算中的实际应用还要求其低功耗设计，因此还得考虑CMP片上互联具备可重构特性，以支持片上网络及其计算资源具备动态开启功能。就设计可重构CMP芯片所涉及到的技术基础问题，本项目主要研究：1）建立支持网络拓扑自动构造的数学模型，设计满足CMP片上网络拓扑可重构要求的路由交换结构和电路实现方案；2）设计适用于可重构互联的 CMP多级缓存动态分配方案及其共享内存的一致性管理机制，提高片上内存使用率；3）分析不同拓扑配置下CMP各处理器间通讯时延与电路时序延迟的统计特性，设计基于竞态条件分析方法解决时序冲突与系统容错的流水线结构电路，提高CMP应用系统的可靠性。最后，利用FPGA搭建一款集成64个Leon3处理器的可重构CMP应用验证系统，为实际设计可重构CMP芯片奠定技术研究基础。

中文关键词： 片上网络多核系统；可重构片上网络模型；内存一致及系统可靠性；云存储共享安全；基因芯片数据分析

英文摘要： With the increasing complexity and capacity of computing in embedded systems, Chip Multi-processor(CMP) system is gradually replacing the single processor-based SoC system as the next-generation development mainstream. Meanwhile, the interconnects among the on-chip processor cores of CMP are becoming more and more important, and Network-on-Chip (NoC) is a promising solution for the interconnection of CMP because of its excellent parallelism,scalability,and high communication efficiency. However, NoC will consume extra energy which can be a serious concern especially when more cores are connected in a CMP along with the technology scaling. Adopting CMP in embedded systems, especially for mobile computing applications, requires low-power design. One promising technique is to design the CMP with reconfigurable interconnects in such a way that its on-chip computing resources can be turned on and off dynamically to save energy. In this proposal, we will focus on the key technologies and their basic theories of the reconfigurable CMP design, including reconfigurable routing and switching circuit design,shared memory allocation and consistency management,and fault-tolerance of race conditions. Gaussian Integer, Markov Chain, and Probability Distribution will be introduced to analyze the critical problems on reconfigura

英文关键词： NoC Multi-Core System；Reconfigurable Model of NoC；Memory Coherence and System Relability；Secure Sharing for Cloud Storage；Data Analysis for Gene Micro-array