[Paraview-developers] "Unable to build k-d tree structure" error with D3 filter and PV 4.4.0 on large datasets

Schlottke-Lakemper, Michael m.schlottke-lakemper at aia.rwth-aachen.de
Tue Mar 8 09:36:51 EST 2016


Dear all,

We have recently come across the following error when using the D3 filter on large and distributed vtkUnstructuredGrids. We have our own reader that locally creates a vtkUnstructuredGrid by reading the according subdomain from a file in netcdf format and I would like to use the D3 filter to have halo cells available on all domains (e.g. for gradient computations). For small scale meshes containing about 260 x 10^6 cells there is no problem. However, when I want to execute the filter on 1.8 x 10^9 cells I get the following error messages:

Warning: In /dev/shm/domini/ParaView/4.4.0/intel-para-2015.07-mpi-devel/ParaView-v4.4.0-source/VTK/Filters/Parallel/vtkPKdTree.cxx, line 207
vtkPKdTree (0x2943d0c0): (process 490) memory allocation on my node (BreadthFirstDivide)

ERROR: In /dev/shm/domini/ParaView/4.4.0/intel-para-2015.07-mpi-devel/ParaView-v4.4.0-source/VTK/Filters/ParallelMPI/vtkDistributedDataFilter.cxx, line 796
vtkDistributedDataFilter (0x29433cd0): Unable to build k-d tree structure


the same is happening with other remote nodes, i.e.,

Warning: In /dev/shm/domini/ParaView/4.4.0/intel-para-2015.07-mpi-devel/ParaView-v4.4.0-source/VTK/Filters/Parallel/vtkPKdTree.cxx, line 207
vtkPKdTree (0x2a9fdf20): (process 137) memory allocation on a remote node (BreadthFirstDivide)


I run a server on 576 cores using ParaView version 4.4.0. The program does not crash after this error, however, it is not creating any output. I have tracked down the error to the following (potential) location. I noticed that this error seems to be generated in the function BreadthFirstDivision, which somehow returns the error code 1, which is then in AllCheckForFailure summarized to a value > 0 on all MPI ranks leading to the error message. I also noticed that this cannot be a VTKERROR as it is not printed to the terminal. I believe that the error might reside in the check for midpt < -1, which would also return an error code of 1 in BreatdhFirstDivision.

I hope you can help me with this problem. 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<mailto:m.schlottke-lakemper at aia.rwth-aachen.de>
Web: http://www.aia.rwth-aachen.de

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/paraview-developers/attachments/20160308/ec04cb04/attachment.html>


More information about the Paraview-developers mailing list