open TELEMAC-MASCARET The mathematically superior suite of solvers

Hello,

I just installed Telemac2D v6p1 (scalar), Strawberry Perl v.5.12.1 and gnu fortran 4.9.2 on Windows 7, 64-bit. I used the full installer "Setup_TELEMAC_v6p1_windows_intel" and modified the systel.ini and martel.bat files to use with gfortran instead of Intel.

When I run a simulation, it ends normally, but there is a warning:
" Note: The following floating-point exceptions are signalling:
IEEE_Underflow_Flag
IEEE_Denormal "

This warning is also present even if I do not use my own Fortran code so I do not think it's due to uninitialized variables on my part. The results from the simulation showing these floating-point exceptions present oscillations in the water levels in the lower portion of the model domain.

What could be causing these exceptions? Could it be that I have not installed Telemac properly? Any help on this would be greatly appreciated!

As a side note, I will be using Telemac2D to continue simulations run on another computer with an older version (v6p0 or prior) and Intel Fortran. That is why I chose not to install the latest version of Telemac.

Thanks!

Hello,

Underflow exceptions are not a real problem. Suppose that you square 1.D-200, it will trigger an exception and the result will not be 1.D-400 but probably the smallest number that can be represented.

I do not know what is IEEE_Denormal but probably something related to underflows also.

With best regards,

Jean-Michel Hervouet

Hello,

I also used to get the IEEE_Underflow_Flag errors (and some others...) when I switched from gfortran 4.8.3 to 4.9.2 keeping the same Telemac installation. At the beginning I was worried and switched back to 4.8.3 but eventually I started using 4.9.2 and just ignored them. Computations run properly in any case.

Regards,
Costas

Hi,

The question is old but an explanation of this note at the end of the program can be found here: stackoverflow.com/questions/44308577/iee...normal-in-fortran-77

Best regards. JBF

I ran into the same problem while running computations on a double precision 'SELAFIND' mesh. I set the keywords GEOMETRY FILE FORMAT and RESULTS FILE FORMAT to 'SELAFIND' and the IEEE exceptions went away. Also the nodes in my results mesh were no longer skewed by a couple of cm which was the case when I didn't define the file format.