网络并行计算是并行计算与分布式计算技术非常重要的发展方向之一,结合具体的数值试验,探讨了Windows操作系统下基于PVM的网络并行数值计算中影响PVM并行程序性能的几个重要因素,包括负载平衡、通信开销、网络性能、任务粒度、处理机个数、精度要求及处理机内存容量问题等,并提出了提高PVM并行程序性能的相应策略,以高效快速地实现问题的求解。
Network parallel computing is one of the important branches of parallel and distributed computing technology. With numerical experiments, several important factors affecting the performance of PVM parallel programs in PVM network parallel numerical computing under Windows operation systems are discussed, which relate to load-balancing, communication costs, network performance, task granularity, the number of processors, precision requirement of the problem to be solved and the computer's memory capacity, At the same time, the corresponding strategies for improving the performance of PVM parallel programs are also explored, so as to solve the problem efficiently and quickly.