摘要： 使用基于MPI并行编程方法对IFortran77串行计算程序进行并行处理,主要工作在循环级的可并行性研究和并行实现。文中给出并行代码在国家高性能计算中心(合肥)的曙光-1000机上运行的并行加速比以及相应的系统并行效率。

关键词: 并行计算, 消息传递, MPI编程, 加速比

Abstract: In this article, we make parallelizing processing to a Fortran77 serial routine by MPI-based parallel-programming, and put emphasis upon parallelability research and implement of cycling-layer in source programme.The parallelized code was debugged and executed on Dawn-1000 computer in National High Performance Computing Center at Hefei. The actual paralleling speed ratio and the relative paralleling efficiency are given. ;;;

中图分类号: 

• TP311.52