open TELEMAC-MASCARET The mathematically superior suite of solvers

Hi,
I am running telemac2d and sisyphe simulations on two different systems and get different results. Both with version v6p1.

First one is Windows XP with g95, sources have been downloaded on august 1st then compiled.
Second one is Linux debian with gfortran, last compilation is June 28th.

For validation case 120_digue (telemac2d), when comparing to reference file, I get for instance max differences in velocity of 0.37 m/s for Windows and 1.47 m/s for Linux system.
For validation case 106_loire-ingrande (sisyphe and telemac2d), I get max differences in bed evolution of 6.4 m for Windows and 2 m for Linux.

For me, these differences seem large enough, but perhaps, such values are common when comparing to reference files using different systems/compilers?
Has anyone obtained the same sort of values in validation tests?

If not, could the difference come (1) from the systems/compilers (2) from the sources (3) Other ideas?

If anyone has some input about that, your help will much appreciated!
Thank you very much,

Caroline

Hello,

Something that we do behind the scene is to validate the TELEMAC system on multiple plateforms and compilers. By default, this done evey night for the version in development. We have not observed what you descibed.

However, even though we created a tag v6p1 and associated zip files (which you have installed) we have had to correct a few bug on v6p1 not long after first release. The SVN was updated but the zip file were only updated later. It is therefore possible that you have two different version.

We are sorry for this. We will try not to do this again with the coming v6p2. In any case we recommend that link your version(s) with our SVN repository so the update becomes super-easy.

So please make sure you have the latest / same version on both computers.

Hope this helps.

Sébastien.