Newsgroups: comp.parallel.mpi
From: fachat@physik.tu-chemnitz.de (Andre Fachat)
Subject: Re: MPI_finalize crashes program
Organization: TU Chemnitz
Date: 13 Dec 1996 13:57:53 GMT
Message-ID: <58rnd1$59n@narses.hrz.tu-chemnitz.de>

I had a quite similar problem with an older mpich version,
where MPI_Finalize just hang on our Sun (Solaris 2.3),
when using more than two processes.
I upgraded to a new version of mpich, and the problem
went away (and after that upgrade I have it running on
Linux, HP and Solaris, and all work together smoothly :-)

hope it helps
Andr

Jorge Paloschi (j.paloschi@ic.ac.uk) wrote:
: Has anyone experienced this problem?
: When calling MPI_finalize (from a fortran code), sometimes the program
: crashes. 
: The error is:p7_12605:  p4_error: interrupt SIGBUS: 10
: This happens using version 1.0.13 of mpich on a cluster of SUN 
: workstations. It happens both in SunOS and Solaris.
: On Solaris, I have traced the failure until it is attempted to do
: MPI_Comm_free(&MPI_COMM_SELF)
: within MPI_Finalize.
: Any help?
: I shoul add that the call for MPI_Finalize is issued as the last
: executed sentence of the program.
: Thanks,

: Jorge

: PS:Up to that point everything runs perfect.

--
Andr Fachat                       |"I do not feel obliged to believe that the
Institute of physics,              | same God who has endowed us with sense,
Technische Universitt Chemnitz    | reason, and intellect has intended us to
http://www.tu-chemnitz.de/~fachat  | forego their use" -- Galileo Galilei

