[MPICH] Errors in MPICH2
Rajeev Thakur
thakur at mcs.anl.gov
Sun Jan 28 11:39:39 CST 2007
Yes, that is a known bug (that MPI_Type_create_resized is not recognized by
ROMIO). You can do the same thing by using MPI_Type_struct with an explicit
LB and UB. We will fix it in a future release.
Rajeev
_____
From: owner-mpich-discuss at mcs.anl.gov
[mailto:owner-mpich-discuss at mcs.anl.gov] On Behalf Of Luiz Mendes
Sent: Sunday, January 28, 2007 5:49 AM
To: mpich-discuss
Subject: [MPICH] Errors in MPICH2
I was trying to run this example
#include "mpi.h"
#include <stdio.h>
#define BUFSIZE 100000
int main(int argc, char *argv[])
{
int i, myrank, numprocs;
MPI_Status status;
MPI_Init(&argc, &argv);
MPI_Comm_rank(MPI_COMM_WORLD, &myrank);
MPI_Comm_size(MPI_COMM_WORLD, &numprocs);
double start, finish, tempo;
MPI_Aint lb, extent;
MPI_Datatype etype, filetype, contig;
MPI_Offset disp;
MPI_File fh;
int buf[1000];
MPI_File_open(MPI_COMM_WORLD, "data.dat",MPI_MODE_CREATE |
MPI_MODE_RDWR, MPI_INFO_NULL, &fh);
MPI_Type_contiguous(2, MPI_INT, &contig);
lb = 0;
extent = 6 * sizeof(int);
MPI_Type_create_resized(contig, lb, extent, &filetype);
MPI_Type_commit(&filetype);
disp = 5 * sizeof(int); /* assume displacement in this file view
is of size equal to 5 integers */
etype = MPI_INT;
MPI_File_set_view(fh, disp, etype, filetype, "native",MPI_INFO_NULL);
start=MPI_Wtime();
MPI_File_write(fh, buf, 1000, MPI_INT, MPI_STATUS_IGNORE);
finish=MPI_Wtime();
printf("\nResult : %f\n",finish-start);
MPI_File_close(&fh);
MPI_Finalize();
return 1;
}
and it fails with this message below:
Error: Unsupported datatype passed to ADIOI_Count_contiguous_blocks
[cli_0]: aborting job:
application called MPI_Abort(MPI_COMM_WORLD, 1) - process 0
rank 0 in job 1 luizmendes_34376 caused collective abort of all ranks
exit status of rank 0: return code 1
what it could be?
thanks
luiz mendes
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/mpich-discuss/attachments/20070128/0bb285f4/attachment.htm>
More information about the mpich-discuss
mailing list