open TELEMAC-MASCARET The mathematically superior suite of solvers

Hi I am unale to run tomawac applications in parallel, it runs in scalar but fails in paralle2, 3 7 etc processors), I have kept the application quite simple with no princi file, all depths are +ive etc and no spectrum outputs for point locations.

Please note The parallel set up works for Telemac2d applications

I attach the various files and error message

In the tmp file it produces the sub-domain geo files correctly

I would appreciate any assistance in this regard

Regards
Tony C.

Hello,

You are right, we have serious problems with parallelism in Tomawac version 6.2, especially in the option with currents. We decided to rewrite all the parallel part, which was too complicated. It is now done but it will be released with the next version.

With best regards,

Jean-Michel Hervouet

Hello again,

However your case fails at the early stage of opening the files, which may be another problem (our problem happens later in the method of characteristics).

JMH

Hello,

Do you have a default parallel executable of TOMAWAC? Have you compiled TOMAWAC in parallel (then you should see that a tomawacv6p2_MP.exe exits)?
If not, I think the problem is that you are trying to use a default executable in parallel that does not exist.

I can give you 2 solutions:
- compile TOMAWAC in parallel (you will then have a tomawacv6p2_MP.exe in your tomawac/toma_v6p2/$HOSTTYPE directory,
- write an empty FORTRAN file (e.g. empty princi.f) and activate FORTRAN FILE = princi.f then a compilation of an executable of TOMAWAC which will not be different from the default one if it exists will be done.

Hope this helps,

Chi-Tuan

Hi Chi-Tuam,
Thanks for your suggestions

I previously compiled a local princi file with this run and same error message occured

I performed the pyton install as parallel for all modules

when I search for the tomawacv6p2_MP.exe, I only have tomawacv6p2.exe in the tomawac/toma_v6p2/$HOSTTYPE directory
but that is also the case with telemac2d directory, there is only telemac2dv6p2.exe and this runs successfully in parallel.

Regards
Tony C.

I have tried tomawac in parallel on another Window7 machine with ifort compiler and same problem as previously posted aborting before commencing computation. I have tried the other validation cases and these run in scalar mode but similarly fail in parallel as previously posted.

I wonder has anybody got tomawac v6.2 working in parallel mode on a windows 7 machine?

tony c.

Hello,

as said by Jean-Michel before, we have serious problems with the parallel version of Tomawac 6.2. We are currently rewriting all the parallelism in TOMAWAC: this will be delivered in version 6.3.
At the moment, if you need to run Tomawac in parallel, you can try to run the version 6.1. If your mesh is not too big, it should work. If this is not the case, we have a patch that we can send you and that should help you.

Regards,

Elodie

Thanks Jean Michael and Elodie I misinterpreted the initial reply and wanted to confirm that Tomawac for Windows 7 was not currently capable of running in parallel rather than wasting time trying to get it to work. I Look forward to version 6.3 and should survive till then for parallel. Keep up the good work its an excellent resource.

Regards
Tony C.

Hi Tony,

Just for your information, your case starts correctly (starts running) on my Linux ifort desktop. Please see the attached file. In particular, the "Memory organisation" has been passed.

Just to be sure, can you try to run your computation in serial (PROCESSORS PARALLEL = 0) please? And see if it works or not.

The problem is that I cannot reproduce your problem. It may come from your installation.

Regards,

Chi-Tuan

Thanks Chi Tuaan,
I have no problem in Scalar mode its the windows parallel installation or application that the problem is as I have followed the recommended configurations etc and parallell works for the other modules without any problem.

the Linux system seems to be much more consistent and stable for the parallel applications

Regards

Tony C.