MPI并行计算与矩阵1

来源：互联网发布：人工智能需要哪些技术编辑：程序博客网时间：2024/05/18 05:31

#include<stdio.h>#include<mpi.h>#include<math.h>long n,     i;double sum,pi,mypi,x,h;int group_size,my_rank;int main(argc,argv)int argc;char *argv[];{int group_size,my_rank;MPI_Status status;MPI_Init(&argc,&argv);MPI_Comm_rank(MPI_COMM_WORLD,&my_rank);MPI_Comm_size(MPI_COMM_WORLD,&group_size);n=20;MPI_Bcast(&n,1,MPI_LONG,0,MPI_COMM_WORLD);h=1.0/(double)n;sum=0.0;for(i=my_rank+1;i<=n;i+=group_size){printf("i=%ld,processor=%d\n",i,my_rank);x=h*(i-0.5);sum=sum+4.0/(1.0+x*x);}mypi=h*sum;printf("the rank %ld,calculate mypi=%lf\n",my_rank,mypi);MPI_Reduce(&mypi,&pi,1,MPI_DOUBLE,MPI_SUM,0,MPI_COMM_WORLD);if(0==my_rank) printf("pi is :%lf\n",pi);fflush(stdout);MPI_Finalize();return 0;}

In order calculate pi

1)Every processor calculate part of pi

Then use MPI_Reduce() to sum all of each part

see the code above

2) If we remove the clause:if(0==my_rank), what will happen?

We see that each clause was run 4 times, and only when my_rank equal to 0, pi=3.14 otherwise it equals to 0.00!

any question?

0 0