[MPICH] mpich2 with MatLab R2007b

Jayesh Krishna jayesh at mcs.anl.gov
Wed Dec 19 14:43:08 CST 2007


Hi,
 From the matlab docs it looks like the developers moved to the "pure
launcher" matlab.exe so that users can execute "post launch" commands (batch
file commands after the launch of matlab) in their batch scripts. Please
check with matlab support team on whether you can achieve your debugging
setup with the existing version of MATLAB.
 
regds,

jayesh


  _____  

From: owner-mpich-discuss at mcs.anl.gov
[mailto:owner-mpich-discuss at mcs.anl.gov] On Behalf Of peter_raeth at juno.com
Sent: Wednesday, December 19, 2007 11:03 AM
To: mpich-discuss at mcs.anl.gov
Subject: [MPICH] mpich2 with MatLab R2007b


On our Windows desktops we are running MatLab R2007b. With that version, we
seem to have lost a significant capability. Let me explain. 
 
Under R2007a it was possible to successfully run the following DOS command:

mpiexec -l -exitcodes -localroot -pwdfile MPI_Options.txt -n 2 -machinefile
MPI_MachineList.txt "C:\Program_Files\Matlab\bin\matlab.bat" -check_malloc
 

Since one machine was listed in the machinefile and told to run two
processes, two MatLab IDE instances appeared. Through manual activity, we
could step through the codes running as supervisor and worker processes. It
was a wonderful debugging and demonstration capability. mpiexec terminated
normally once the codes completed and the IDEs were ended.
 
Under R2007b, MathWorks eliminated the bat file in favor of an exe file.
Now, when the equivalent DOS command is run, mpiexec terminates immediately.
Yet, the two IDE instances come up. However, they are not running within the
cluster communications universe. This appears to be because matlab.exe is
just causing the real matlab executable to be run.
 
My question: Is there a way to retain the capability we enjoyed under
R2007a? Is there some flag we can set for R2007b?  Although a bat file
appears in the R2007b bin directory, the bat file is deprecated. Plus, it
does not enable the capability we enjoyed under R2007a.
 
Many thanks for whatever insights you can offer.
 
 
Best,
 
Peter.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/mpich-discuss/attachments/20071219/cd79694d/attachment.htm>


More information about the mpich-discuss mailing list