It does not only increase the difficulty of parallel programming , but also add the burthen of the programmer , and therefore , limits the development of the parallel computing . based on the studying of the parallel computing environment on the heterogeneous cluster system , the paper proposed the key elements of the integrated development environment , which should be programmer friendly , convenient and practical 尤其在异构机群系统环境下，程序员开发与测试并行程序时，不得不面对机群的异构性，而必需掌握较多的系统操作及机群管理等专业知识，从而增加了并行编程的难度，阻碍了机群系统的应用，限制了基于mpi的并行计算的发展。
This paper will analyze and compare advantages and disadvantages of these two programming languages : java and fortran , clarify the practical significance of implementing supporting fortran parallel programming in the jme , and expatiates on several key problems and how to solve these problems : keeping working nodes isomorphic , parallel programming , distributing and executing task , collecting results , etc . this paper also discusses how to implement user management . at last this paper gives users some advice on programming in the jme on the basis of analyzing results of testing jme 本文分析比较了java和fortran两种程序设计语言的利弊，阐述了jme元计算环境支持fortran并行程序设计的意义，并针对实现这种支持的保证工作机同构、并行编程支持、程序的分布执行、结果回收等几个关键问题做了详细阐述；然后就用户管理功能的实现做了讨论；最后在分析jme系统测试结果的基础上，对用户编程给出了一些建议。
Also , the relactions between the best block size for matrix transpose and the size and associativity of the processor ' s cache is formulized . for parallel optimization , several programming models available on a numa system , such as lightweight processes ( sproc ) , posix threads , openmp and mpi , are compared , and their speedup and coding complexity are analyzed 对于sar成像处理的并行优化，本文对比了在numa架构上可用的几种并行编程模型：轻量级进程、 posix线程、 openmp和mpi ，针对numa架构和sar成像处理的特点从加速比、编程复杂度等多个方面进行了讨论。
Considering that wavelet - based image segmentation has very high computational complexity , especially for large images , in order to speed up the process of the large images segmentation , this paper put emphasis on exploring the computational parallelism in wavelet - based image segmentation . the parallelism of the mpi are in - depth analyzed , data parition of the large image also are designed for reducing the data transmission , and parallel wavelet transform algorithms are designed . this paper implement a parallel wavelet - based image segmentation using mpi on cluster machines with the better segmentation rate 为了解决分割的快速性问题，本文把另外一个重点放在应用mpi并行处理技术对基于小波的纹理图像分割进行并行化处理上，研究了mpi并行编程机制，分析了应用其进行并行图像分割的可行性。进而主要以减少并行处理里中的通信量为目的，分别对并行系统框架设计，图像数据划分，小波变换算法并行优化设计进行了研究。
It preestablish a parallel design pattern for each parallel project based on master / slave architecture , and the communication between computers is transparent to programmers . in another words , when programmers are developing parallel programs , they do n ' t need to learn any complicated primitives and write any codes related with communication between parallel tasks . just only set some parameters in the visual interfaces that provided by platform , and the codes will be generated by our system automatically and inserted to the fittest place in the files 系统为每一个并行计算工程设定了一个基于master slave架构的并行编程模型，并且使得各结点机之间的通信对于开发人员透明化，也就是说，开发人员在开发并行程序时，无需了解纷繁复杂的通信原语，不涉及任务之间通信的代码，只需要在并行程序开发平台提供的可视化界面上对数据的传送进行设定，通信代码将自动生成并插入到程序合适的地方。
