open TELEMAC-MASCARET The mathematically superior suite of solvers

I am modeling the case the study of an estuary by TELEMAC-2D, I have used the parallel mode in case of hydrodynamics computations and it was ok, but when I tried to use it with the transport (salinity transport) it is usually give me errors. Can you tell me what is the reason and is the case of transport need special arrangements in the case of parallel mode.

Hello,

No, parallelism also works with transport of a tracer. Check first that you have no specific Fortran that breaks parallelism, like assigning values with the node numbers (which are changed in parallelism), and if not send your parameter file.

With best regards,

Jean-Michel Hervouet

I am not using any fortran files with the computations. So, I attached the steering file , and the configuration file for the parallel computations, I haope that this what you mean

Hello,

I see no obvious mistake. Does it work in scalar mode with salinity ? (so that we can be sure that it is a problem in paralellism). Can you tell us also what is the kind of error you get ?

JMH

without parallel it works at least for two hours and then stops because of some problems with the boundary conditions, but with parallel it doesn't work at all. the error message is as follows:

forrtl: severe (157): Program Exception - access violation
Image PC Routine Line Source
out3728_win.exe 004C8960 Unknown Unknown Unknown
out3728_win.exe 0045738B Unknown Unknown Unknown
out3728_win.exe 00401714 Unknown Unknown Unknown
out3728_win.exe 006B0DBC Unknown Unknown Unknown
out3728_win.exe 006943B9 Unknown Unknown Unknown
kernel32.dll 7C817077 Unknown Unknown Unknown

job aborted:
rank: node: exit code[: error message]
0: localhost: 157: process 0 exited without calling finalize
1: localhost: 123
Duration of job : 2 seconds ( 0:0:2 ) (system=0.016 sec)

Hello,

Let us try a last thing before looking in deep at your case.

You have no Fortran file, so you use the default executable. If you have not built it it will make a problem (you can build it by running the command "maktel parallel install" in the sources of telemac-2d).

So try to put a Fortran file (just put a subroutine inside, like homere-telemac2d.f for example). It will bypass the default executable and we'll see.

With best regards,

Jean-Michel Hervouet

I found out the problem, it is the use of the following line in the steering file:

OPTION FOR LIQUID BOUNDARIES : 1;2

when I remove this the parallel computation works very well in all the case and when I add it I get errors. It is very strange for me, do you have explanation for that.

Hello,

This is the Thompson boundary conditions. It works in parallel only from version 6.1 on. As I have corrected a bug recently, make sure that you have the latest version through svn, subroutine thomps.f

With best regards,

Jean-Michel Hervouet