open TELEMAC-MASCARET The mathematically superior suite of solvers

Hi

No problems on Windows XP for the partitioning step. I join the partel.log I obtained

The execution crash because I only had 2 processors ...

Maybe you could try to execute partel manually to see where is the problem...

Regards

Thanks for your answer...

We tried to execute partel manually and it crashed at the same point
(the last line that is written is : "The input file assumed to be
2D Selaphin").

Does the problem come from our installation ? The configuration is this one :

- Intel Core i7 CPU This email address is being protected from spambots. You need JavaScript enabled to view it. Ghz RAM 6Go (2.99 utilisable)
- Windows 7 32 Bits
- Visual Studio 2005 SP1
- Intel Fortran V11
- Telemac V6P0
- Mpich2 1.3.1
- Stawberry Perl 5.12.0


Moreover, TELEMAC is working correctly with a smaller model (1600 nodes) in parrallel mode !!!

Regards

Hi
Telemac is not concerned by your problem. This is only a PARTEL problem.
But if it works on other mesh, it seems to be a random error.
In my opinion, the only other things you could try is to compile yourself partel (which is in the sources directory or parallel) maybe with some print information to understand where is the crash ...

Regards

Hi,

+ I also encountered no problem when partitioning your geometry file (without sections).
=> I'm using gfortran on a Vista and tests have been made from 8 to 64 subdomains.

+ If you are interested, I can send you the partitioned files => You then will be able to run your case by executing the mpirun command in the temporary directory.

+ Did you compile Metis from sources?
=> I used their latest stable (4.0.1) windows binaries.

Best regards,

Fabien Decung

Hi

Thanks for your answer

We use the "standard" executable version of TELEMAC V6P0, so we don't have compile Metis from sources.

We have to :
- check Metis version
- compile partel to locate the error
- change our fortran compiler (do you know if there is an incompatibilty between Telemac V6P0 and Intel Fortran V11 ?)

Can you tell us your complete configuration ?

Regards

Hi

After compiling of Partel.exe, it works !

Thank you for your help

Hi,

I get an error when I do parallelism:
Error : NBMAXHALO too small

What does it mean?

It's weird because I already run this model with a .geo a bit different and parallelism worked ...

Please find below the partel.log

Thanks for your help

JA

+

---

+
PARTEL: Telemac Selafin Metisologic Partitioner

Rebekka Kopmann & Jacek A. Jankowski (BAW)
Jean-Michel Hervouet (LNHE)
Christophe Denis (SINETICS)
PARTEL (C) Copyright 2000-2002
Bundesanstalt fuer Wasserbau, Karlsruhe

METIS 4.0.1 (C) Copyright 1998
Regents of the University of Minnesota

BIEF 5.9 (C) Copyright 2008 EDF
+

---

+

=> This is a preliminary development version
Dated: Tue Jan 27 11:11:20 CET 2009

Maximum number of partitions: 100000

+

---

+


Selafin input name <input_name>: input:
T2DGEO




Boundary conditions file name : input:
T2DCLI




Number of partitions <nparts> [2 -100000]: input: 2

Partitioning options:

Partitioning method <pmethod> [1 or 2]: input: 1
variables are:

one-level mesh.
ndp nodes per element: 3
npoin number of mesh nodes: 163879
nelem number of mesh elements: 326162

The input file assumed to be 2D Selafin
timestep: 0.0000000E+00 s = 0.0000000E+00 h

there is 3 liquid boundaries:

boundary 1 :
begins at boundary point: 720 , with global number: 2217
and coordinates: 10096.00 11896.50
ends at boundary point: 727 , with global number: 841
and coordinates: 10149.50 11942.00

boundary 2 :
begins at boundary point: 1049 , with global number: 2216
and coordinates: 3540.500 15785.90
ends at boundary point: 1065 , with global number: 1933
and coordinates: 3414.700 15679.70

boundary 3 :
begins at boundary point: 1553 , with global number: 1009
and coordinates: 7032.200 1778.900
ends at boundary point: 1571 , with global number: 556
and coordinates: 7150.700 1642.000

there is 3 solid boundaries:

boundary 1 :
begins at boundary point: 1571 , with global number: 556
and coordinates: 7150.700 1642.000
ends at boundary point: 720 , with global number: 2217
and coordinates: 10096.00 11896.50

boundary 2 :
begins at boundary point: 727 , with global number: 841
and coordinates: 10149.50 11942.00
ends at boundary point: 1049 , with global number: 2216
and coordinates: 3540.500 15785.90

boundary 3 :
begins at boundary point: 1065 , with global number: 1933
and coordinates: 3414.700 15679.70
ends at boundary point: 1553 , with global number: 1009
and coordinates: 7032.200 1778.900
Using only METIS_PartMeshDual subroutine
The mesh partitioning step METIS starts
The mesh partitioning step has finished
Runtime of METIS 0.4380000 seconds
Error : NBMAXHALO too small

Hello,

Hum, our partel specialist is on holidays...

There are 3 possibilities :

1) a problem in your mesh
2) NBMAXHALO is really too small
3) a bug in partel

What you could test is splitting into 4 instead of 2. If you can do it change NBMAXHALO from 100000 to 200000 in partel and recompile the parallel library.
If the problem remains send you geometry file and we'll test it here.

With best regards,

Jean-Michel Hervouet

Thank you for your reply Jean-Michel.

My computer was only a 2 cores, so impossible to split in 4.

What means this NBMAXHALO ?

I finally solved my problem: I generated a new .geo with Matisse from the same .mat file... for an unknown reason the .geo size was different and I can now run my model with 2 processors parallels.
So you were right, it was a bug in my mesh, thanks.

Regards,

Julien

Hi Julien

Even your computer had only 2 cores, you could try to run telemac with 4 processors. The model will be splitted in 4 and Each processor will be charged by 2 telemac process.
This is not necessary really efficient but it works!

Regards
christophe