并行任务自动划分及调度算法SMPS  

An Algorithm for Automatic Parallel Allocation and Scheduling of A Simulation Task SMPS

作　　者：刘小峰[1] 李伯虎[1] 

机构地区：[1]北京计算机应用及仿真技术研究所

出　　处：《系统仿真学报》1996年第1期1-5,共5页Journal of System Simulation

摘　　要：随着ＶＬＳＩ技术的发展，多处理机并行处理技术有了新的进展。本文基于ＩＣＳＬⅡ连续系统仿真语言［１］，提出了一个高效的并行任务自动划分及调度算法（ＳＭＰＳ）。它首先采用启发式算法对任务进行预调度，得到一个次优结果。然后采用改进的分枝定界算法对该调度进行优化，从而得到最优调度。该算法既具有简单易行性的特点，又不失调度结果的最优性，它为并行仿真语言的实现提供了技术基础。With the derelopment of VLSI technology, parallel multiprocessor technique has got new progress. In this paper, a new algorithm for automatic parallel allocation and scheduling of a simulation task(SMPS) based on a continuous system simulation language ICSL Ⅱ [1]is presented. At first, a heuristic algorithm is used for prescheduling of a simulation task, and a suboptimal scheduling result is got. Then an improved branceh and bound algorithm is used to optimize the schedule above to implement an optimal schedule. This algorithm has advantages on simplicity and feasibility, in addition, its scheduling result is optimal. It is certainly a technical foundation for the implementation of parallel simulation language.

关 键 词：多处理机 任务划分 调度算法 SMPS 并行处理 

分 类 号：TP36[自动化与计算机技术—计算机系统结构]