为实现局部性强弱分析的定量化,分析了进程工作集发生迁移时,现有工作集管理算法因未考虑局部性强弱因素而反复调整驻留集大小从而导致系统发生颠簸的现象,提出了采用采样周期内使用位求和结果和缺页次数作为指标的局部性强弱量化计算方法,并对该方法进行了算法设计和系统实施.结果表明,文中量化计算方法在不影响系统性能的情况下,减轻了系统进程切换时各进程工作集大小的波动,降低了系统资源需求,说明该算法是有效的.
As the existing working-set management algorithms ignore the locality strength in the procedure of wor-king set adjustment,the resident set size is frequently adjusted so that the thrashing problem is inevitable.In order to solve this problem and to quantitatively analyze the locality strength,a quantitative approach to locality strength is proposed,which uses the sum of use bits and the number of page faults in a sampling period as two calculation indexes.Then,the corresponding algorithm of this approach is designed and the approach is systematically implemented.Experimental results show that the proposed quantitative approach helps to alleviate the thrashing of wor-king set size in process switching and reduce the demands for system resources without degrading the performances of the whole system.