open TELEMAC-MASCARET The mathematically superior suite of solvers

Good afternoon,

I allow myself continuing this post to submit my problem.
I am trying to compile (and then to run) telemac in parallel mode
(I think) I succeded in install Metis (~/metis-5.1.0).
However I am really unable to compile telemac. Although I read several post, the contruction of the systel*.cfg (for a parralel config) stays confuse (I think the error came from there).
Hence, i really would be grateful if someone could help me in improving the *cfg file

Configuration a HPC cluster (www.lncc.br/altix-xe/)
OS: GNU/Linux (Linux altix-xe.hpc.lncc.br 2.6.18-194.el5 #1 SMP Fri Apr 2 14:58:14 EDT 2010 x86_64 x86_64 x86_64 GNU/Linux)
Python 2.7.13
GNU Fortran (GCC) 4.1.2 20080704
gcc (GCC) 4.9.3
OpenMpi 1.10.7

Hi,

Can you try this file.

And run compileTELEMAC.py --clean

That should do the trick.

Thanks for the Help Yugi.

But it did not work
I noticed that some bin path were false (so I adapted). Anyway, it did not work.

I also noticed that
1. my configuration is
GNU Fortran (GCC) 4.1.2 20080704 (with gcc (GCC) 4.9.3 ??) whereas telemac tutorial specifies that Gfortran 4.6.3 is the minimum (but i don't know how to install it with restricted rights)
2. libs_all: maybe should be added /hpc/openmpi-1.10.7-gcc4.9/lib/libmpi.so

What do you think ?
(here the updated file and the error)

You posted the systel twice and not the error message.

As for the 2. you need it as mpif90 contains the link.

Indeed with the log error file this is better.
It seems that the cpp option gives some pb. but even removing this option (same pb of allocation in utils_med.F)

I think the Gfortran version should be more recent...

The issue is with the option -cpp which does not seem to exist with your version of gfortran.
You need to find to option that activated the preprocessor.

That will solve your problem.

Ok

I searched for an equivalent option of -cpp...i did not found. Anyway i load another environnement (GNU Fortran (GCC) 4.7.2).
The compilation succeeded !

Now I am (first) trying the code serial version, and i get this error (see the attached file).

I checked in the file:
[sebpinel@altix-xe ~]$ strings /lib64/libc.so.6 | grep ^GLIBC GLIBC_2.2.5
GLIBC_2.2.6
GLIBC_2.3
GLIBC_2.3.2
GLIBC_2.3.3
GLIBC_2.3.4
GLIBC_2.4
GLIBC_2.5
GLIBC_PRIVATE

Effectively. It doesnot exist...

The issue seems to be with your installation if numpy.
It seems to have been compiled with an older version of GLIBC.

The only solution i can see is to reinstall your python.
I would suggest using anaconda it is a python sitribution that contains a package handler.
Making it easy to install new python module if you have an internet access.
www.anaconda.com/download/#linux

Thanks Yugi !

Indeed there was a pb with numpu (unexpected weird path to numpy)

I succeeded in running the serial version.
Now, the parallel version is giving some trouble (I was expecting this). Normally, the geo file are good because i can run on my PC with the parrallel version.
Thus, the pb are with the systel / cluster configuration.

I already enable the debug option...but it changes nothing.
Program received signal 11 (SIGSEGV): Segmentation fault.

I am reading the forum about similar post...I accept any suggestion.


Config on a HPC cluster (www.lncc.br/altix-xe/)
OS: GNU/Linux (Linux altix-xe.hpc.lncc.br 2.6.18-194.el5 #1 SMP Fri Apr 2 14:58:14 EDT 2010 x86_64 x86_64 x86_64 GNU/Linux)
GNU Fortran (GCC) 4.7.2
Python 2.7.13
mpirun (Open MPI) 1.8.5

Just to tell that finally, i got the solution.
I re-built metis (with the 32 bits option), and now it can run.

Thanks for the advises