code_saturne User's Forum

Hello, Yes, to overcome this, we often add: if (ifacel(1,iel).le.ncel) then ...add face contribution... endif in Fortran, or if (cs_glob_mesh->i_face_cells[face_id][0] < cs_glob_mesh->n_cells) { ... } in C. This works well when looping over all interior faces, but when looping on a subset of those f...

Hello, The MED SALOME module is a higher-level tool than the MED I/O library, which is in a separate path... ): So you should replace (in your setup file): med yes no /home/julien/salome/Salome-V7_6_0-x86_64/modules/MED_V7_6_0 with: med yes no /home/julien/salome/Salome-V7_6_0-x86_64/tools/Medfichie...

Hello, Yes, to have the CFDSTUDY module, you need to set the salome = line (and then run SALOME using "code_saturne salome", not directly run the main SALOME application). If you have an error when you do this, could you post it ? For Syrthes, you can use Scotch instead of Metis 5.1 if you...

Hello, This is normal (there was another post about this a few months ago). In 4.2, you have 2 modes: preprocessing and computation. You can switch from one to the other with the icons on the toolbar (a mesh or a colored cavity flow). You should be able to choose output formats as usual, on the &quo...

Hello, You have almost a factor 2 performance difference between 8 and 32 procs. It would be interesting to test at 16, as you probably have an improvement from 8 to 16 and almost no improvement from 16 to 32 (assuming the bandwith is saturated), but it can be interesting to check. I do not know whi...

Hello, This might also be a bug in our code... When porting from Qt4 to Qt5 (currently available only on trunk, so will be in 4.3 sometimes in July), we discovered some minor bugs in the Qt4 version, which lead to visible bugs on some architectures/Qt versions and not on some others). I'll need to c...

Hello,

It seems you simply forgot the "/" before "home" (should be "/home").

Best regards,

Yvan

Hello, You only need to reinstall the code, not its prerequisites, and use the same directory structure, but there is otherwise no "easy way". In may case, I usually untar the new version side by side with the previous version (assuming you want the bug-fix version 4.0.5, not the dev versi...

Hello, The mass flux sign is given relative to a face's normal. For interior faces, there is no guaranteed way of determining orientation relative to the geometry (it depends on the initial numbering). To determine the mass flow rate, you need to compare the face normal to your "reference"...

Hello, The speedup seems OK from 5 to 16 procs (a bit less than optimal, though), and rather small from 16 to 32 procs. This is not so surprising, as the code is mostly bandwidth limited, and on most Xeon cores, performance increase from half node to full node is much less than 2 (more like 20%) as ...