error when using Mpi_reduce

22 Januari 2012 7:45

0
Hello,

When the program runs more than three processes, it gives an error:
- Assertion failed in file helper_fns.c at line 337: 0 memcpy argument memory ranges overlap, dst_=0x8accf0 src_=0x8ac848 len_=1200 internal ABORT - process 1 rank 1 in job 46 n1.blades.cluster caused collective abort of all ranks exit status of rank 1: killed by signal 9
I found the problem with using MPI_Reduce, without MPI_Reduce program works on any number of processes without any problems, then I have only used code from MPI_Reduce:
```
#include <iostream>
#include <mpi.h>
#include <stdlib.h>
using namespace std;
 
 
int Size = 100;
 
 
 
int main (int argc, char *argv[])
{
  
  int rank, size;
 
 
  MPI_Init (&argc, &argv);      /* starts MPI */
  MPI_Comm_rank (MPI_COMM_WORLD, &rank);        /* get current process id */
  MPI_Comm_size (MPI_COMM_WORLD, &size);        /* get number of processes */
    
    double *ms_total = new double[Size];
    double *ms_new = new double[Size];
 
 
  MPI_Bcast (&Size, 1, MPI_INT, 0, MPI_COMM_WORLD);
    srand(1);
  
  for(int i=0;i<Size;i++)
  ms_new[i]=rand()%100;
  
MPI_Barrier(MPI_COMM_WORLD);
  for( int i =0; i < Size; i++)
     {
      MPI_Reduce(&ms_new[i], &ms_total[i], Size, MPI_DOUBLE, MPI_SUM, 0, MPI_COMM_WORLD);
      }
      
      if(rank==0)
      {
for(int i=0; i< Size; i++)
cout << rank <<"  "<<ms_total[i]<<endl; 
     }
 
 
 
  std::cout <<  "Hello world from process" << rank << "\n";
  MPI_Finalize();
  return 0;
}
 
```
Now gives the following error:
- Assertion failed in file helper_fns.c at line 337: 0 memcpy argument memory ranges overlap, dst_=0x8ce010 src_=0x8cdcf8 len_=800 internal ABORT - process 1 Fatal error in PMPI_Reduce: Other MPI error, error stack: PMPI_Reduce(1198).................: MPI_Reduce(sbuf=MPI_IN_PLACE, rbuf=0xfbc9c8, count=100, MPI_DOUBLE, MPI_SUM, root=0, MPI_COMM_WORLD) failed MPIR_Reduce(764)..................: MPIR_Reduce_binomial(172).........: MPIC_Recv(83).....................: MPIC_Wait(513)....................: MPIDI_CH3I_Progress(150)..........: MPID_nem_mpich2_blocking_recv(948): MPID_nem_tcp_connpoll(1720).......: state_commrdy_handler(1556).......: MPID_nem_tcp_recv_handler(1446)...: socket closed rank 1 in job 13 n1.blades.cluster caused collective abort of all ranks exit status of rank 1: return code 1
What is this file helper_fns.c? and where I made a mistake in the program??

thanks in advance
- Diedit oleh CaptainV 22 Januari 2012 7:51
- Dipindahkan oleh DanielMothMicrosoft Employee 06 April 2012 7:41 (From:Parallel Computing in C++ and Native Code)
- Balas
- Kutip

Semua Balasan

22 Januari 2012 8:48

I found a bug in the program

#include <iostream>
#include <mpi.h>
#include <stdlib.h>
using namespace std;
 
 
int Size = 100;
 
 
 
int main (int argc, char *argv[])
{
  
  int rank, size;
 
 
  MPI_Init (&argc, &argv);      /* starts MPI */
  MPI_Comm_rank (MPI_COMM_WORLD, &rank);        /* get current process id */
  MPI_Comm_size (MPI_COMM_WORLD, &size);        /* get number of processes */
    
    double *ms_total = new double[Size];
    double *ms_new = new double[Size];
 
 
  MPI_Bcast (&Size, 1, MPI_INT, 0, MPI_COMM_WORLD);
    srand(1);
  
  for(int i=0;i<Size;i++)
  ms_new[i]=rand()%100;
  
MPI_Barrier(MPI_COMM_WORLD);
 
      MPI_Reduce(ms_new, ms_total, Size, MPI_DOUBLE, MPI_SUM, 0, MPI_COMM_WORLD);
      
      
      if(rank==0)
      {
for(int i=0; i< Size; i++)
cout << rank <<"  "<<ms_total[i]<<endl; 
     }
 
 
 
  std::cout <<  "Hello world from process" << rank << "\n";
  MPI_Finalize();
  return 0;
}

22 Januari 2012 22:44

0

For questions related to MPI (Message Passing Interface), the "Windows HPC Server Message Passing Interface (MPI)" forum at http://forums.community.microsoft.com/en-US/windowshpcmpi/threads/ is a good place to ask them.

Cheers

Daniel
http://www.danielmoth.com/Blog/
- Balas
- Kutip