[Gmsh] Reclassify 2D does not work

Christophe Geuzaine cgeuzaine at ulg.ac.be
Thu Apr 11 12:43:18 CEST 2013


Hi Marin - Many thanks for your the detailed feedback! Let's keep in touch to see how we can address some of these issues.

Christophe


On 08 Apr 2013, at 14:07, Martin Vymazal <martin.vymazal at vki.ac.be> wrote:

> Dear prof. Geuzaine,
> 
> unfortunately I never tried to use fltk or vtk myself as a library and I 
> don't know their internals. I can only provide some observations as a user of 
> both.
> 
> I'm wondering what happens in gmsh for example when a user is trying to 
> rotate a large 3D mesh with both surface and volume elements. Since you don't 
> 'see' the underlying volume elements, you only need to redraw the surface. 
> When I switch off all volumes explicitly in visibility window, gmsh is 
> suddenly more responsive. I guess vtk can do this automatically, but I wonder 
> what's the level of sophistication of such algorithm which decides what (not) 
> to draw as the view angle changes.
> 
> Gmsh is also very slow when you decide to delete a few elements from the mesh 
> through the gui. You select an element, confirm your choice and have to wait 
> for some seconds before you actually see it disappear from the screen.
> 
> I wonder if some of these issues are directly related to fltk, because for 
> the Reclassify2D problem, just waiting for the reclassify dialog to pop up is 
> very long. When you want to move the dialog on the screen (without actually 
> manipulating the mesh), you again have to wait (easily 5-10 seconds depending 
> on how big is the mesh).
> 
> The gui of gmsh built from source usually behaves better than a pre-built 
> binary. Gmsh is more responsive in my linux laptop with nvidia card and closed 
> source drivers than in my office computer, which has cheap on-board intel GPU, 
> but faster cpu and twice as much ram (8Gb).
> 
> Please don't take this as criticism, I don't know myself what is the cause of 
> these problems or how to solve them. This is as much as I can say about the 
> speed issues paraview vs. gmsh from user point of view.
> 
> Best regards,
> 
>  Martin Vymazal
> 
> 
>>> 
>>> It's hard to process meshes with one opensource package, I usually use the
>>> combination Gmsh-Paraview-Meshlab. I like gmsh, but its gui becomes
>>> extremely slow with large meshes (say ~ 10^6 elements). ParaView is much
>>> more responsive.
>> Hi Martin - do you know how Paraview achieves this? Does it display the same
>> amount of information?
>>> Best regards,
> 
> 
> _______________________________________________
> gmsh mailing list
> gmsh at geuz.org
> http://www.geuz.org/mailman/listinfo/gmsh

-- 
Prof. Christophe Geuzaine
University of Liege, Electrical Engineering and Computer Science 
http://www.montefiore.ulg.ac.be/~geuzaine