open TELEMAC-MASCARET The mathematically superior suite of solvers

Hi,

I am writing a new subroutine to compute runoff from a defined rainfall.
This subroutine is called from prosou.f where the rain/evaporation implementation is.
The subroutine works in scalar. In parallel the subroutine is being compiled without error but the run is not starting (hanging after the ...\out_fortran.exe command).

My code is simple, with no argument and only

USE DECLARATIONS_TELEMAC2D, ONLY: PLUIE,DT,NPOIN,LT,AT,HN

plus some local variables. I even made a test with an empty subroutine called in parallel, and I have the same problem.

Is there something special to be done when creating subroutines for having them working in parallel?

I am on Win7 with gfortran and MPI. Using version 7.1.0.

Any help is greatly appreciated!

Best regards
PL

Hi,

Problem fixed. I mean, there was no problem, I had some trouble connecting to my company's network yesterday and I noticed that MPI can be affected in such a case.

So, after a good night sleep everything is working fine

Best regards
PL

Hello,

No, there is nothing special to do if the subroutine is in the FORTRAN FILE. The only possible reason would be that a parallel communication in this subroutine (like calling function P_DSUM or subroutine PARCOM) is not done by all processors, which would cause some processors to indefinitely wait for the others. For the same reason it would be necessary that the subroutine is called by all processors. These are the only reasons I can figure out.

With best regards,

Jean-Michel Hervouet