open TELEMAC-MASCARET The mathematically superior suite of solvers

Hi,

I have been trying to run a simulation in parallel for several days, but I always obtain a segmentation fault and the error "LIT : ABNORMAL END OF FILE". This simulation (launched in v6p1) was however successful in Telemac2D and 3D with the scalar mode. I tried running this simulation on three different systems, all in Ubuntu 11.04 with gfortran v4.6.1. These systems are a Xeon 6x2.8Ghz (Win7/64-bit), a i7 4x2.2Ghz (Win7/64-bit) and a 2x1.6Ghz (Vista/32-bit). Since I have an problem with parallel computing, I later compiled mich2 then openmpi on the i7 machine, hoping that it may resolve the problem(without success).

Due to previous posts on the topic, I thought that my geometry file may not be correct. When I follow the steps mentioned in the discussion on "creating SELAFIN files and mapping bathymetry to a mesh in Blue Kenue", my IPOBO object does not follow the boundary of my mesh. In the discussion "Problems using TeleMac2D", Chris suggests (in point no.4) that the IPOBO points should only be on the boundary. However, when I drag a mesh to a Selafin object, the IPOBO points will simply follow the order of points ids of the mesh. I opened a few .slf files provided in the validation package, and the IPOBO objects do not follow the boundary. I also tried the 32-bit version of BlueKenue and obtained the same result. Initially, I built a .2dm file with SMS, imported it in BlueKenue, dragged the mesh object into a new Selafin object and renamed it as BOTTOM (or FOND). I also tried creating a .t3 mesh with triangulation, but the result was the same. I created a conlim file with Fudaa-Prepro and modified certain nodes to define an inlet and an outlet (I'm modelling river hydraulics and sediment transport). Are there some other (free) tools that can be used to build a .slf geometry file?

I installed the new version of partel.f, deleted all compiled folders (ubugfopenmpi) and rebuilt the project. In systel.cis-ubuntu.cfg, I ensured that -fconvert=big-endian and -lpthread appear after cmd_obj and cmd_exe. The .slf file that I generated with BlueKenue is of big-endian type.

I'm not sure what I could try at this point. Any ideas? I must have skipped an important step. Any help will be greatly appreciated.

Yannick

Hello,

For me and according to the listing it is an error during the partitioning, thus in partel.f, could you look what happens in partel.log ? Otherwise I could try the case here in version 6.1 or 6.2, to see if the file is correctly partitioned.

With best regards,

Jean-Michel Hervouet

Hi,
Thank you for your fast reply. The file partel_T3DGEO.log (in the temporary folder that was created during the simulation) is empty and the file partel_T3DGEO.par (in the same folder) contains:
T3DGEO
T3DCLI
2
1
0
I will send you the case at edf in a few minutes.
Yannick