open TELEMAC-MASCARET The mathematically superior suite of solvers

Hey,

I would like to use weirs in parallel mode with the V6P2 version of Télémac.
In first, I tried to use it in parallel mode with the validation case but it doesn't working in parallel mode just in scalar mode
The message error is :

WITH PARALLELISM,
NO DIRECT SYSTEM SOLVER

PLANTE: PROGRAM STOPPED AFTER AN ERROR
EXITING PVM: CALLING P_EXIT

EXITING MPI

_____________
runcode::main:
D:\telemac\weirs_p:
|runCAS: fail to run
| C:\MPICH2\bin\mpiexec.exe -wdir D:\telemac\weirs_p\t2d_weirs.cas_2013-
04-24-13h43min15s -n 6 D:\telemac\weirs_p\t2d_weirs.cas_2013-04-24-13h43min15s\out_telemac2dv6p2.exe

I don't really understand this error. I need to used a specified solver for weirs?

When I tried on my own case, it's working too in scalar mode but in parallel mode the computation stops but doesn't crashed.

If somebody have a solution.

Thank for your help

Regards

Marine

Hello,

Just look in the steering file where you ask for a direct system solver. It must be :

SOLVER : 8

or another SOLVER with option 8 asked, try 1 instead of 8 and it should work.

Regards,

JMH

Thank you for your reply

The solver 1 seems to be better in my case. But with the validation case neither solver working in parallel mode, it is the same error message.
Do you have some advice about the solver option?
Do you have already test weirs in parallel mode? It is working?
I join my case with the output listing maybe I missed something?
In my case, something concerns me. When it read my liquid and sources files the values are write twice and it is not exactly the same, like this

ITERATION 1 TEMPS : 1.0000 S
SOURCES : Q(1) = 0.000000000000000E+000
SOURCES : Q(1) = 0.000000000000000E+000
SOURCES : Q(2) = 0.000000000000000E+000
SOURCES : Q(2) = 0.000000000000000E+000
LE FICHIER DES FRONTIERES LIQUIDES CONTIENT
576 LIGNES AVEC :
Q(2) Q(3) Q(4) Q(5) Q(6) Q(7) Q(8)
FRONTIERE LIQUIDE : Q(2) = 42.8700000000000
FRONTIERE LIQUIDE : Q(2) = 42.8707128514056
FRONTIERE LIQUIDE : Q(3) = 0.000000000000000E+000
FRONTIERE LIQUIDE : Q(3) = 0.000000000000000E+000
FRONTIERE LIQUIDE : Q(4) = 53.9600000000000
FRONTIERE LIQUIDE : Q(4) = 53.9613219544846
FRONTIERE LIQUIDE : Q(5) = 0.000000000000000E+000
FRONTIERE LIQUIDE : Q(5) = 0.000000000000000E+000
FRONTIERE LIQUIDE : Q(6) = 0.000000000000000E+000
FRONTIERE LIQUIDE : Q(6) = 0.000000000000000E+000
FRONTIERE LIQUIDE : Q(7) = 0.000000000000000E+000
FRONTIERE LIQUIDE : Q(7) = 0.000000000000000E+000
FRONTIERE LIQUIDE : Q(8) = 1.000000000000000E-002
FRONTIERE LIQUIDE : Q(8) = 1.000334672021419E-002

Thank for your help

Regards

Marine

Hello,

On the 037_weirs case in parallel: I see an error when I use a compiler that checks bounds, and I am currently correcting this (but you should not have the same error message when going from 8 to 1...).

Now your problem below. It is normal, if you look at function Q (library telemac2d) the subroutine read_fic_frliq is called successively for time AT-DT and time AT and the average is taken. The message is written in read_fic_frliq. This certainly could be clearer, it is the first time someone points out this.

With best regards,

Jean-Michel Hervouet