MPI-based Parallel Implementation Of Randomized Algorithms

With the development of computer technology,the demand of computing large set data becomes a serious problem.More and more domain of science,such as biology,mete-orology,medical,DNA sequencing,are requring using computers and numerical simulation method to resolve problems met in the science fileds and egineering.In general,the stong requriment has push an heavy pressureon the computers’ capability.These requirements and needs have large set data to be computed and use shorter time, the exchangment bewteen the data are also very large.While single computer can deal with daily use,sharing the good capa and reliablity,it doesn’t satisfiy the requirment of cloud computing and large statilite cloud picture.In order to meet the needs,proving the more powerful technolgy—parrallel computing,using many single computers to form the computing cluster,to provide com-puting capability to compute mass data and satisfy the high computing performance’s need.In this article,analysis the apply of MPI in the parallel computing technology,analysis and fullfill the implement of random algorithms using MPI.MPI is the standard of message passing programming,it takes the advantage of nice transplatation, high computing capabil-ity,flexible implement.MPI is independent of specific language,defining the library API and binding of C/C++program language.Now MPI is the most efficient and popular parallel computing platform.Giving the discussion of parallel computing,introduce the structure of parallel com-puting system,most used parallel computer and theory of parallel computing and parallel alogrithm.Then,bring the short introduction of MPI,build up the MPI’s running environ-ment,fulfill serveral common alogrithms. By implementing the parallel computing for these algorithms,running the MPI programming,count up and analysis the program run-ning time,calculating the MPI running efficiency,proving the high performance of parallel computing.