Hi, Pavan, Thanks a lot, that is most confusing part for me in MPI. I would like to clear this type of problems. So the blocking begins to send message to itself, but there is no receive posted. Thus, the MPI block it. Am I right here? Thus, I need a nonblocking receive to wait the message here. Can I use blocking receive? Hi Jeff, The weblink is https://computing.llnl.gov/tutorials/mpi/#Derived_Data_Types the code is second example. Thank you Thanks a lot! Sufeng On Tue, Jul 2, 2013 at 4:48 PM, Pavan Balaji <[email protected]> wrote:

This is an incorrect program. You need to post an Irecv before sending to yourself.

-- Pavan

On 07/02/2013 04:39 PM, Sufeng Niu wrote:

...

Hi,

I try to familiar with mpi self defined type api, and I directly run a very simple code from LLNL webpage:

//----------------------------**------------------------------** ------------------------------**-------

#include"mpi.h" #include <stdio.h> #define SIZE 4

int main(argc,argv) int argc; char *argv[]; { int numtasks, rank, source=0, dest, tag=1, i; float a[SIZE][SIZE] = {1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0, 12.0, 13.0, 14.0, 15.0, 16.0}; float b[SIZE];

MPI_Status stat; MPI_Datatype columntype;

MPI_Init(&argc,&argv); MPI_Comm_rank(MPI_COMM_WORLD, &rank); MPI_Comm_size(MPI_COMM_WORLD, &numtasks);

MPI_Type_vector(SIZE, 1, SIZE, MPI_FLOAT, &columntype); MPI_Type_commit(&columntype);

if (numtasks == SIZE) { if (rank == 0) { for (i=0; i<numtasks; i++) MPI_Send(&a[0][i], 1, columntype, i, tag, MPI_COMM_WORLD); }

MPI_Recv(b, SIZE, MPI_FLOAT, source, tag, MPI_COMM_WORLD, &stat); printf("rank= %d b= %3.1f %3.1f %3.1f %3.1f\n", rank,b[0],b[1],b[2],b[3]); } else printf("Must specify %d processors. Terminating.\n",SIZE);

MPI_Type_free(&columntype); MPI_Finalize(); }

//----------------------------**------------------------------** ------------------------------**-------

However, it doesn't work. the program just suspend there. I figure out the problem is at line MPI_Send, and MPI_Recv. the MPI_Send just send it to itself. but I don't know why it doesn't work. Is it caused by dead lock? why?

Thank you very much!

-- Best Regards, Sufeng Niu ECASP lab, ECE department, Illinois Institute of Technology Tel: 312-731-7219

______________________________**_________________ discuss mailing list [email protected] To manage subscription options or unsubscribe: https://lists.mpich.org/**mailman/listinfo/discuss<https://lists.mpich.org/mailman/listinfo/discuss>

-- Pavan Balaji http://www.mcs.anl.gov/~balaji

-- Best Regards, Sufeng Niu ECASP lab, ECE department, Illinois Institute of Technology Tel: 312-731-7219