论文部分内容阅读
负载共享技术在并行处理中是至关重要的 .通过对大量负载共享文献的考察发现 ,以前的研究都是基于一定系统负载而进行平衡算法的设计 ,它们很少考虑到所选定的系统负载与程序运行时间之间的准确关系 .为了确定系统负载对并行程序运行的影响 ,首先确定了影响并行程序运行的两个重要系统因素 :CPU负载和网络负载 .为了不失一般性 ,也为了简化网络负载的测量 ,选用 NAS PVM并行 Benchmark作为实验测试对象 ;为了得到程序运行时 CPU负载信息 ,采用守护进程跟踪计算结点的当前 CPU负载 ;为了准确记录程序运行时间 ,设计了一个作业提交平台 .经过反复大量的实验和数据分析 ,分别得到了并行程序执行时间与 CPU负载之间、执行时间与网络负载之间关系的两个重要结论 .
Load sharing techniques are of paramount importance in parallel processing.Through a review of a large number of load sharing literature, it has been found in previous studies that the design of balancing algorithms based on a certain system load rarely takes into account the chosen system load And the program running time.In order to determine the impact of system load on the parallel program operation, we first determine the two important system factors that affect the parallel program running: CPU load and network load.In order to lose the generality, but also to simplify In order to obtain CPU load information during program running, the daemon is used to track the current CPU load of the node. In order to accurately record the running time of the program, a job submission platform is designed. After repeated large amounts of experiments and data analysis, two important conclusions were obtained respectively: the execution time of parallel programs and CPU load, execution time and network load.