open TELEMAC-MASCARET The mathematically superior suite of solvers

Hi,

I have a problem when I tried to run the test case(v6p0) in parallel with v6p2 Telemac2d

it stopped and had errors like:

## Erreur : Fin anormale : cd /home/telemac2d/tel2d_v6p0/test.fr/delwaq/cas.txt11996_tmp; mpirun -machinefile /home/TELEMAC/v6p2/hostfile -np 2 ./out11996_intel_64_10.exe :512
## Error : System command failed for /home/telemac2d/tel2d_v6p0/test.fr/delwaq/cas.txt11996_tmp/telemac2d.bat :512

then I entered the tmp folder and tried to manually execute the outxxxx.exe, it failed and had error:

forrtl: No such file or directory
forrtl: severe (29): file not found, unit 12, file /home/telemac2d/tel2d_v6p0/test.fr/delwaq/cas.txt11996_tmp/T2DIMP


I don't understand why it failed to create T2DIMP.
is this the problem of the version of the test case? because it's v6p0 but the Telemac is v6p2?

Regards,

Beck

Hi Beck,

Before investigating further, can you say if this problem also occurs with a scalar computation (PARALLEL PROCESSORS = 0) ?
Do you have the same problem with the test case 029_pildepon ?

Why do not you use the test cases in version 6.2 available via SVN in the Download section, then Sources: SVN menu on the right and read the section "The details of the TELEMAC SVN repository"? As the code has changed since version 6.0, if you use a 6.2 version, you will not be able to run every test case correctly for sure!

Anyway, after these first questions, if your problem has not been solved, please send the complete output listing, not only from
## Erreur : Fin anormale ... (or read it carefully)
The error may be more described before...

With the error you described in the temporary directory, one first problem must come from the liquid boundaries file (mnemo T2DIMP, see the reference manual or the dictionnary telemac2dv6p2.dico in the lib directory). It says that this file is not found. Are you sure you have it when running this case?

Regards,

Chi-Tuan

Hi Chi-Tuan,

Thanks for your response.
I've already downloaded the v6p2 test cases.
The 031_delwaq still had the previous problem, but with your help the problem is solved now.

Thanks a lot!

Sincerely,
Becky

Hi Beck,

I tested this test case today and the problem comes from the choice of the solver with parallel. Indeed, in the error message before
## Erreur : Fin anormale ...

it is written:
The direct solver cannot be used in parallel. For a parallel computation, you have to use (and install...) MUMPS if you want a direct solver, or change the solver into an iterative one (e.g. conjugate gradient with wave equation or GMRES with coupled equations).
So, there was a mistake in the 6.0 test case version. Nevertheless, you are advised to use the latest test cases (as told in my previous post).

Regards,

Chi-Tuan

PS: in the temporary directory, you could have read the file PE00001-00001.LOG where the standard output is written for one subdomain when running a parallel computation. At the end, you have the same error message as the one in the launching directory.