Warning MumpsParallelSolver

This forum is for issues related to parallel processing
and OpenSees using the new interpreters OpenSeesSP and OpenSeesMP

Moderator: selimgunay

Post Reply
AHARSHIAN
Posts: 87
Joined: Tue Apr 03, 2012 3:34 am
Location: Germany, Weimar
Contact:

Warning MumpsParallelSolver

Post by AHARSHIAN » Fri Oct 24, 2014 6:22 am

Hi,

What is the following error, and how can I fix it?

Warning Mumps Parallel Solver :: Solve(void) -Error 13 returned in substitution dmumps

and after a couple of warnings, job aborts:

Thanks for your help
Best Regards,
A.H.Arshian

fmk
Site Admin
Posts: 5883
Joined: Fri Jun 11, 2004 2:33 pm
Location: UC Berkeley
Contact:

Re: Warning MumpsParallelSolver

Post by fmk » Mon Oct 27, 2014 8:00 am

looks like a memory issue (http://mumps.enseeiht.fr/doc/userguide_4.10.0.pdf) .. how big is the model?

AHARSHIAN
Posts: 87
Joined: Tue Apr 03, 2012 3:34 am
Location: Germany, Weimar
Contact:

Re: Warning MumpsParallelSolver

Post by AHARSHIAN » Mon Oct 27, 2014 11:46 am

Hi Frank,

All in all, I have 553 nonlinear elements with 2352 cross-sections (integration points), and 100 fibers at each cross-section plus 1728 ShellMITC4.

This is only a 3-story RC building, and I am going to work on larger models!

Thanks for your help,
Best Regards,
A.H.Arshian

fmk
Site Admin
Posts: 5883
Joined: Fri Jun 11, 2004 2:33 pm
Location: UC Berkeley
Contact:

Re: Warning MumpsParallelSolver

Post by fmk » Tue Oct 28, 2014 8:33 am

it's not a large model .. have you tried a different solver in sequential mode to see if it solves the problem (it could be your model is bad)?

AHARSHIAN
Posts: 87
Joined: Tue Apr 03, 2012 3:34 am
Location: Germany, Weimar
Contact:

Re: Warning MumpsParallelSolver

Post by AHARSHIAN » Tue Oct 28, 2014 9:20 am

Hi Frank,

Yes! I have tried on sequential OpenSees using UmfPack. It works fine, and no problem occurs even if I run it in a loop for 1000 of times. However, sequential approach takes a lot of time. Could it be a systematic problem of our machine?! Others are working on linux, and I am working on Windows. The machine i am running my simulations on has both linux and Windows. However, Windows is not the main OS.

Then again, I would like to emphasize that this problem happens randomly in the loop of analysis. What I mean is that it suns the model for a couple of times, and then such a problem happens. The only different between these analyses are input variables! It is a Sensitivity Analysis.
Best Regards,
A.H.Arshian

fmk
Site Admin
Posts: 5883
Joined: Fri Jun 11, 2004 2:33 pm
Location: UC Berkeley
Contact:

Re: Warning MumpsParallelSolver

Post by fmk » Thu Oct 30, 2014 8:15 am

i think there is an internal problem with Mumps somewhere, i have examples that work and ones that don't .. it only pops up when the Mumps solver is called more than once in a script (say for gravity and then transient) .. if i delete the lines that cause the problems i get a large memory leak and the next error that pops up when we want to use Mumps is that we run out of memory.

Post Reply