Newsgroups: comp.parallel.mpi
From: jcksndk1@micro.ma (Mr D K Jackson)
Subject: MPI_IO and MPI_HOST
Organization: IC
Date: 17 Mar 1998 19:03:59 +0000
Message-ID: <nqwsoohuokg.fsf@micro.ma>

Hi,

I seem to obtain strange numbers for MPI_IO and MPI_HOST.
Are these constants not generally implemented?

I'm fortunate enough to have access to several systems on which
I can use MPI, but I have the same problem on all of them - which
makes me think I'm probably doing something daft....

  fprintf(stdout," MPI_HOST  %d   MPI_PROC_NULL %d  MPI_ANY_SOURCE %d  MPI_IO %d\n",MPI_HOST,MPI_PROC_NULL,MPI_ANY_SOURCE,MPI_IO);

MPICH on an alpha cluster
This is node 3 of 5  on nano.mathphys, unix says nano
  MPI_HOST  24   MPI_PROC_NULL -1  MPI_ANY_SOURCE -2  MPI_IO 25

ANU on AP1000
This is node 5 of 6  on Host, 
  MPI_HOST  772392   MPI_PROC_NULL -2  MPI_ANY_SOURCE -1  MPI_IO 772424

Fujitsu implementation on AP3000
This is node 3 of 4  on cnet0012, unix says cnet0012
  MPI_HOST  308280   MPI_PROC_NULL -1  MPI_ANY_SOURCE -2  MPI_IO 308320

Any ideas?

cheers,
david