给出了微机网络并行计算环境和拥有共享内存的并行计算环境下大规模矩阵相乘的几种

并行算法,分析了这些算法的计算和通讯时间复杂性,在多台微机上安装并行消息传递机制MPI

(message passing interface) ,构成微机网络并行计算环境,在该环境下对所给算法作了数值试验,

结果表明所给出的算法非常有效,同时在程序没有改动的条件下,将微机网络并行环境下的程序移

植到具有共享内存的并行计算机IBMP650 上,结果表明,所给算法具有更高的加速比和并行效率。-Gives the PC network parallel computing environment and have a shared memory parallel computing environment, several large-scale parallel matrix multiplication algorithms, analyzed their computational complexity and communication time, in more than one computer to install the parallel message passing MPI (message passing interface), constitute a computer network parallel computing environment, in the context of the given algorithm for the numerical test results show that the algorithm is very effective given the same time, no changes in the program under the conditions of the PC network parallel environment ported to a shared memory parallel computer IBMP650, the results show that the given algorithm has higher speedup and parallel efficiency.