一种层次的、混合并行离散事件仿真算法

A HIERARCHICAL AND HYBRID ALGORITHM OF PARALLEL DISCRETE-EVENT SIMULATION

并行仿真算法是并行离散事件仿真中的核心问题 ,对于具体的应用系统 ,采用不同的并行仿真算法将导致其仿真性能大的差异 .提出了一种针对于分布环境中特定应用系统仿真的层次的、混合并行离散事件仿真算法 .测试和应用表明 :和通常的保守机制或者乐观机制相比 ,能够较大地提高仿真效率 ,并且具有良好的可扩展性 .首先给出了在通信开销不可忽略的环境下 ,保守机制和乐观机制的性能测试结果和两者适用情况的分析 ,然后根据测试结果和具体应用系统的特点 ,提出了层次的、混合并行离散事件仿真算法 ,给出了 L P级和组级算法等 。

Parallel simulation algorithm is the key issue in parallel discrete event simulation. For special applications, different parallel simulation algorithms would lead to obvious differences of performance. A hierarchical and hybrid parallel simulation algorithm for special applications under the distributed environment is proposed in this paper. Compared with the general conservative and optimistic mechanisms, the test and application of the new algorithm show that it can improve the efficiency and scalability of simulation obviously. Firstly, a test of conservative and optimistic mechanisms under the environment that the communication overhead cannot be ignored is presented. Then a hierarchical and hybrid algorithm, including the LP level algorithm and the group level algorithm, is proposed based on the test result and the characteristics of applications. Finally, the performance of the new algorithm is tested.