计算机与现代化 ›› 2011, Vol. 193 ›› Issue (9): 105-108.doi: 10.3969/j.issn.1006-2475.2011.09.028

• 计算机仿真 • 上一篇    下一篇

基于Simics的分布式一致性协议仿真

郑志硕1,郑存陆2,曹宏徙3   

  1. 1.广东第二师范学院物理系,广东 广州 510310; 2.华南师范大学计算机学院,广东 广州 510631;3.中南大学信息科学与工程学院,湖南 长沙 410083
  • 收稿日期:2011-03-21 修回日期:1900-01-01 出版日期:2011-09-22 发布日期:2011-09-22

A Simple Distributed Directorybased Protocol for Simics

ZHENG Zhi-shuo1, ZHENG Cun-lu2, CAO Hong-xi3   

  1. 1.Department of Physics, Guangdong University of Education, Guangzhou 510310, China;2.School of Computer Science, South China Normal University, Guangzhou 510631,China;3.Information Science and Engineering College, Central South University, Changsha 410083, China
  • Received:2011-03-21 Revised:1900-01-01 Online:2011-09-22 Published:2011-09-22

摘要: 用于多种计算机系统和指令系统仿真的Virtutech Simics只提供一个简单的顺序扁平侦听式高速缓存一致性(Snooping Cache Coherence Protocol)模型支持MESI协议, 从而制约了可仿真的并行处理器个数。以下将基于目录的分布式高速缓存一致性协议(Distributed Directorybased Cache Coherence Protocol)模型应用于Simics中并给出基于Simics的分布式一致性协议的仿真结果。这一结果证实分布式协议能降低事件总数,减少网络中的事件。本文提出一个简单的基于目录的分布式高速缓存一致性协议,从而解决制约Simics的可扩放性问题。

关键词: 计算机系统仿真, 全系统仿真平台, 片上众核, 高速缓存一致性协议, 基于目录的高速缓存一致性协议

Abstract: As a fullsystem simulation technology allowing one to completely simulate the functionality of computer systems with different memory hierarchy configurations and different number of processors, Vitutech Simics has an inorder ‘flat’ cache model which only provides with a simple snooping cache coherence protocol support for a simple MESI protocol between caches with a unified shared memory. This limits the number of processors that can be scalable to simulate for the logical bus contention. This paper presents a simple distributed directorybased cache coherence protocol that will solve scalability problem of Simics.

Key words: computer simulation, Simics, chip multiprocessor, cache coherence protocol, directorybased cache coherence protocol

中图分类号: