[mpich-discuss] error in MPI_INIT on 64 bit machine

[Jayesh Krishna]

Hi,
 Did you try running the non-matlab program that spawns child processes without mpiexec (Launch your code by typing the executable name at the command prompt) ?
 Meanwhile, the version of MPICH2 that you use is old. I would recommend upgrading to the latest stable version, 1.3.1, of MPICH2 (available at http://www.mcs.anl.gov/research/projects/mpich2/downloads/index.php?s=downloads). Please uninstall the existing version of MPICH2 before installing the new version.

Regards,
Jayesh

----- Original Message -----
From: "Cugnon" <frederic.cugnon at samtech.com>
To: "Jayesh Krishna" <jayesh at mcs.anl.gov>
Sent: Tuesday, January 25, 2011 10:45:13 AM
Subject: RE: [mpich-discuss]  error in MPI_INIT on 64 bit machine

Thanks for this quick answer.

In fact, I can spawn my program from another master program (not matlab), I
can also spawn another slave program from matlab, only matlab spawning my
Samcef program doesn't work. So I'm not able to provide a standalone test
case. However, I've joined my test master (fred1.f) and slave (fred.f)
programs if it could help to find any reason for this error. MPI calls of
fred1.f are identical to MPI calls used in the Matlab S-function, while
fred.f makes same MPI calls than my Samcef program.

For your information I'm using mpich2-1.0.8-win-x86-64

Thanks for your help

-----Original Message-----
From: Jayesh Krishna [mailto:jayesh at mcs.anl.gov] 
Sent: Tuesday, January 25, 2011 5:27 PM
To: mpich-discuss at mcs.anl.gov
Cc: frederic cugnon
Subject: Re: [mpich-discuss] error in MPI_INIT on 64 bit machine

Hi,
 From the error messages it looks like the spawned process is not able to
connect back to the parent process (the parent process TCP/IP port is
blocked) OR the parent process crashed after spawning the child process. 
 Can you spawn your program from another MPI program (non-matlab/simulink
MPI program launched from the command line)?
 Try turning off any firewalls running on your system. I don't have Matlab
here - do you have a test case that reproduces the problem (preferably
without using Matlab)?

Regards,
Jayesh

----- Original Message -----
From: "Cugnon" <frederic.cugnon at samtech.com>
To: mpich-discuss at mcs.anl.gov
Sent: Tuesday, January 25, 2011 9:10:58 AM
Subject: [mpich-discuss]  error in MPI_INIT on 64 bit machine





Hi 



I have been using MPICH2 to implement co-simulation between Matlab Simulink
and my code. I have implemented a master-slave procedure, launching my code
from a Matlab S-function using MPI_Comm_spawn. Things are working perfectly
on an old Windows XP computer with 32 bit version of both softwares. When I
move to 64 bit versions on Windows Seven system, I have errors when the
slave process calls the MPI_Init function. Depending on the communication
channel that I use (sock or nemesis), I get slightly different error
messages when MPI_Init is called (see attached log files). 



Could anybody help me with this problem? 



Thanks 



Frédéric CUGNON 
Samtech SA 
Rue des Chasseurs-Ardennais 8 
4031 Angleur 
tel int-32-4-361.69.69 
fax int-32-4-361.69.80 
frederic.cugnon at samtech.com 


-- 
This message has been scanned for viruses and 
dangerous content by MailScanner , and is 
believed to be clean. 
_______________________________________________
mpich-discuss mailing list
mpich-discuss at mcs.anl.gov
https://lists.mcs.anl.gov/mailman/listinfo/mpich-discuss

-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.

-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.