父节点在MPI_Barrier上失速

时间:2011-08-02 19:02:30

标签: c mpi

我是MPI的新手,我在运行雪豹的Mac OS X节点集群上实施mpirun时遇到了一些麻烦。我遇到的问题涉及MPI_Barrier()。我有一个简单的功能如下所示,工作正常。但是,当我在条件结束时添加MPI_Barrier()命令时,父节点会停止。其中一个处理器移过屏障命令,而另一个继续卡住。我希望有人可以给我一个快速的解决方法,或者可能会建议我的网络环境阻止命令正常运行。

int main (int argc, char **argv)
{
    int me, np, q, sendto;
    double t0, t1;
    MPI_Status status;
    MPI_Init(&argc, &argv);
    t0 = MPI_Wtime();
    MPI_Comm_size(MPI_COMM_WORLD,&np);
    MPI_Commm_rank(MPI_COMM_WORLD,&me);
    if (np%2 == 1) return 0;
    if (me%2 == 1) {sendto = me - 1;}
    else {sendto = me + 1;}
    if (me%2 == 0) {
        MPI_Send(&me, 1, MPI_INT, sendto, me, MPI_COMM_WORLD);
        MPIRecv(&q, 1, MPI_INT, sendto, sendto, MPI_COMM_WORLD, &status);
    } else {
        MPI_Recv(&q, 1, MPI_INT, sendto, sendto, MPI_COMM_WORLD,&status);
        MPI_Send(&me,1,MPI_INT,sendto,me,MPI_COMM_WORLD);
    }
    printf("Send %d to proc %d, received %d from proc %d\n",me,sendto,q,sendto);
    t1 = MPI_Wtime();
    printf("Timing is %g",t1-10);
    MPI_Finalize();
    return 0;
}

1 个答案:

答案 0 :(得分:0)

除了修复一些拼写错误之外,我让您的代码无故障地运行。我在IBM集群上使用openmpi和gnu c编译器。我不确定你把障碍放在哪里。我猜对了吗?

#include "mpi.h"
#include "stdio.h"

int main (int argc, char **argv)
{
int me, np, q, sendto;
double t0, t1;
MPI_Status status;
MPI_Init(&argc, &argv);
t0 = MPI_Wtime();
MPI_Comm_size(MPI_COMM_WORLD,&np);
MPI_Comm_rank(MPI_COMM_WORLD,&me);
if (np%2 == 1) {
  MPI_Finalize();
  return 0;
}
if (me%2 == 1) {sendto = me - 1;}
else {sendto = me + 1;}
if (me%2 == 0) {
  MPI_Send(&me, 1, MPI_INT, sendto, me, MPI_COMM_WORLD);
  MPI_Recv(&q, 1, MPI_INT, sendto, sendto, MPI_COMM_WORLD, &status);
} else {
  MPI_Recv(&q, 1, MPI_INT, sendto, sendto, MPI_COMM_WORLD,&status);
  MPI_Send(&me,1,MPI_INT,sendto,me,MPI_COMM_WORLD);
}
MPI_Barrier(MPI_COMM_WORLD);
printf("Send %d to proc %d, received %d from proc %d\n",
   me,sendto,q,sendto);
t1 = MPI_Wtime();
printf("Timing is %g\n",t1-t0);
MPI_Finalize();
return 0;
}