[Paraview-developers] Using vtkPUnstructuredGridGhostCellsGenerator

[Joachim Pouderoux]

Hi Michael,

Sorry for the delay to answer your last questions.
Here are my answers:

*- The docs indicate that the filter might perform faster if global point
> ids are available. Does anyone have an estimate about how much faster we
> are talking here?*
>

I don't have any number to provide here but basically providing a global
point ids array avoid the algorithms to transfer point coordinates - it
transfers point ids between mpi ranks and thus reduce the number of values
by 3. Also, without global ids, the algorithm needs to check if a neighbor
point exists on the current rank - to do that we have to create a point
locator and search for points.
That said, making a benchmark would be quite simple as the existing VTK
test (TestPUnstructuredGridGhostCellsGenerator.cxx) exercises the algorithm
with a tetrahedralized wavelet (default size of 50 can be changed to scale
the test) with and without the use of the precomputer global point ids.

We would be interested to know the result of this benchmark if you had to
opportunity to run it on a big cluster.


> * - How can I make global point ids available that are suitable for this
> filter?*
>

D3 does that but it does lot more than that of course. So no, there is no
(other) filter to compute them.
In general this information is only available from the data source - see
the code of the test mentionned earlier: we redefine the wavelet source to
add this information as, the data being structured, it can be directly
computed from the i-j-k coordinates).


*- Is it possible/feasible/sensible to call this filter directly from
> within our custom reader plugin, such that users do not need to apply the
> filter manually each time?*
>

Yes sure. I would recommand to do that - call the filter at the end of your
reader and let your reader produce the output of the filter.


Best,
Joachim



*Joachim Pouderoux*

*PhD, Technical Expert*
*Kitware SAS <http://www.kitware.fr>*


2016-04-11 11:20 GMT+02:00 Schlottke-Lakemper, Michael <
m.schlottke-lakemper at aia.rwth-aachen.de>:

> Hi folks,
>
> I have a few questions about the new ghost cell generator filter
> (vtkPUnstructuredGridGhostCellsGenerator) that I already asked in a
> different context. Since I haven’t really resolved any of the issues yet, I
> took the liberty to post the questions here again:
>
> - The docs indicate that the filter might perform faster if global point
> ids are available. Does anyone have an estimate about how much faster we
> are talking here?
> - How can I make global point ids available that are suitable for this
> filter?
> - Is it possible/feasible/sensible to call this filter directly from
> within our custom reader plugin, such that users do not need to apply the
> filter manually each time?
>
> Performance is critical in our case. We use a custom reader plugin to
> build our unstructured VTK grids using voxels. However, when using ~600
> cores to visualize a 1 billion-cell grid, the “Generate Ghost Cells” plugin
> ran for >15 minutes without results, after which we aborted since it is
> just not feasible anymore. Therefore I am very much interested in the
> expected performance impact of using global point ids.
>
> Thanks a lot in advance
>
> Michael
>
>
> --
> Michael Schlottke-Lakemper
>
> Chair of Fluid Mechanics and Institute of Aerodynamics
> RWTH Aachen University
> Wüllnerstraße 5a
> 52062 Aachen
> Germany
>
> Phone: +49 (241) 80 95188
> Fax: +49 (241) 80 92257
> Mail: m.schlottke-lakemper at aia.rwth-aachen.de
> Web: http://www.aia.rwth-aachen.de
>
>
> _______________________________________________
> Powered by www.kitware.com
>
> Visit other Kitware open-source projects at
> http://www.kitware.com/opensource/opensource.html
>
> Search the list archives at:
> http://markmail.org/search/?q=Paraview-developers
>
> Follow this link to subscribe/unsubscribe:
> http://public.kitware.com/mailman/listinfo/paraview-developers
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/paraview-developers/attachments/20160411/5150e4e9/attachment.html>