摘要
乐观时间同步机制能够显著提高并行仿真的性能,但是在乐观时间推进过程中会出现因果错误,需要根据保存的状态对事件进行回退,因此事件状态保存机制是影响乐观时间推进效率的重要因素。本文首先简要介绍了并行仿真中逻辑进程的执行过程,讨论了乐观时间推进的状态保存与回退机制;然后对全状态保存技术进行了建模和理论分析,并通过实验测试了全状态保存算法的性能,测试结果验证了理论分析的正确性。
The optimistic time synchronization mechanism can dramatically improve the parallel simulation performance, but the optimistic time advancement will lead to causality error, which needsthe event rollback according to the saved state,therefore the state saving mechanism is an important factor impactingthe optimistic time advancement performance. In this paper,we first introduce the logic process execution procedure of the parallel simulation and discuss the state saving and event rollback mechanism for the optimistic time advancement; then we model and theoretically analyze the copy state saving technique,and test its performance through experiment tovalidatethe correctness of our theoretical analysis.
出处
《计算机工程与科学》
CSCD
北大核心
2012年第9期47-50,共4页
Computer Engineering & Science
基金
国家自然科学基金资助项目(60803100
61103190)
关键词
并行仿真
回退机制
全状态保存
性能分析
parallel simulation
rollback mechanism
copy state saving
performance analysis