[vtkusers] Partitioning Data

[Terry Jordan]

No it creates 1 pvtu and 4 vtus but each vtu is approximately if not exactly the same size and contains the entire mesh.  To be clear I am not running this with MPI.  Do I have to?  And that is pretty much my entire pipeline.  How does the distributed data filter know how to break up the mesh?  Am I missing a call?

>>> "Berk Geveci" <berk.geveci at kitware.com> 12/29/2009 1:03 PM >>>
That should work (in theory). It is not?

On Tue, Dec 29, 2009 at 1:00 PM, Terry Jordan <Terry.Jordan at netl.doe.gov> wrote:
> I am trying this:
>
>    vtkDistributedDataFilter *dataFilter = vtkDistributedDataFilter::New();
>    dataFilter->SetInput(filteredGrid);
>    dataFilter->SetBoundaryMode( 0 );
>
>    vtkXMLPUnstructuredGridWriter *writer =
> vtkXMLPUnstructuredGridWriter::New();
>    writer->SetInput( dataFilter );
>       writer->SetFileName(vtkFilename);
>    writer->SetNumberOfPieces( 4 );
>    writer->SetStartPiece(0);
>    writer->SetEndPiece(3);
>       writer->Write();
>
> Do I have to use MPI?
>
>>>> "Berk Geveci" <berk.geveci at kitware.com> 12/29/2009 12:55 PM >>>
> Inserting a distributed data filter between the reader and the writer
> should theoretically work. Of course, you can easily run a ParaView
> (pvbatch) script with 4 processes to do the same thing.
>
> On Tue, Dec 29, 2009 at 9:07 AM, Terry Jordan <Terry.Jordan at netl.doe.gov>
> wrote:
>> Yes, I want to break the mesh down.  I want to partition the entire data
>> set
>> into 4 chunks.  Does anyone have a link to sample code?  I tried to use
>> the
>> distributed data filter, but was unsuccessful in its implementation.  I
>> assume the file format would still be a pvtu and 4 vtus.
>>
>>>>> "Berk Geveci" <berk.geveci at kitware.com> 12/28/2009 10:18 PM >>>
>> or the distributed data filter in VTK.
>>
>> On Mon, Dec 28, 2009 at 5:57 PM, Dominik Szczerba <dominik at itis.ethz.ch>
>> wrote:
>>> -----BEGIN PGP SIGNED MESSAGE-----
>>> Hash: SHA1
>>>
>>> I don't think this is intended to partition your mesh, or at least I
>>> would be highly (if positively) surprised.
>>>
>>> Do you not mean to partition a mesh for e.g. domain decomposition
>>> purposes or other parallel applications? Then metis or chaco is what you
>>> need here.
>>>
>>> Dominik
>>>
>>> Terry Jordan wrote:
>>>>
>>>> I am trying to partition a unstructured grid into 4 evenly sized
>>>> pieces.  Currently I am doing the following:
>>>>
>>>> vtkXMLPUnstructuredGridWriter *writer =
>>>> vtkXMLPUnstructuredGridWriter::New();
>>>> writer->SetInput( filteredGrid );
>>>> writer->SetFileName(vtkFilename);
>>>> writer->SetNumberOfPieces( 4 );
>>>> writer->SetStartPiece(0);
>>>> writer->SetEndPiece(3);
>>>> writer->Write();
>>>>
>>>> filteredGrid is a vtkUnstructuredGrid
>>>>
>>>> The result of this code is 4 copies of the data (1 pvtu and 4 vtu).
>>>> Each is identical to the original data.  What am I missing?  I am
>>>> running this code on a single cpu without the use of MPI.  Do I have to
>>>> use MPI?
>>>>
>>>> Thanks in advance.
>>>>
>>>>
>>>> ------------------------------------------------------------------------
>>>>
>>>> _______________________________________________
>>>> Powered by www.kitware.com 
>>>>
>>>> Visit other Kitware open-source projects at
>>>> http://www.kitware.com/opensource/opensource.html 
>>>>
>>>> Please keep messages on-topic and check the VTK FAQ at:
>>>> http://www.vtk.org/Wiki/VTK_FAQ 
>>>>
>>>> Follow this link to subscribe/unsubscribe:
>>>> http://www.vtk.org/mailman/listinfo/vtkusers 
>>>
>>> -----BEGIN PGP SIGNATURE-----
>>> Version: GnuPG v1.4.9 (GNU/Linux)
>>>
>>> iEYEARECAAYFAks5N84ACgkQ/EBMh9bUuzL60wCg45gdE2GhJbZT9zztDLMJMRiX
>>> RGoAoNjFFwEpY9IfusrMv2YWiOj/mrEg
>>> =Rjc+
>>> -----END PGP SIGNATURE-----
>>> _______________________________________________
>>> Powered by www.kitware.com 
>>>
>>> Visit other Kitware open-source projects at
>>> http://www.kitware.com/opensource/opensource.html 
>>>
>>> Please keep messages on-topic and check the VTK FAQ at:
>>> http://www.vtk.org/Wiki/VTK_FAQ 
>>>
>>> Follow this link to subscribe/unsubscribe:
>>> http://www.vtk.org/mailman/listinfo/vtkusers 
>>>
>>
>>
>
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.vtk.org/pipermail/vtkusers/attachments/20091229/218ef4d0/attachment.htm>