[vtkusers] vtkMultiblockDataSet Ghost cells VTK paraview

[RIVERA, Orlando]

Hello,

I am working in a reader  base on  vtkMultiBlockDataSet  working in parallel Paraview , and I have a question about the vtk internals.

I have a 100s of  blocks, and  each block has X.Y.Z cells in  structured grids with data being cell centered.  Each block is a unit , all cells within this block have the same properties and follow the same physics , but cells in different blocks can have different properties  (important for the solver not for visualization)

I read and put each block as an element of  vtkMultiblockDataSet.  Blocks are distributed amongs processor (MPI-Ranks)

In Parallel I  have been using two approach to distribute the blocks among  processors


1)     Each Processor takes blocks at random , important  is that each processor has the same number of cells

Example:  block 1 has 100cells, block2- 500, block 3- 300 ,  block 4-100, block 5 -400



Then processor 0  has   blocks  1, 2 and 4  = (100+500+100)= 700 cells

Processor 1  has  blocks  3 and 5 = (300+400)=700 cells



2)      I use a Hilbert Curve (HSFC) to take in acount blocks’ position

i.e.,  all processor should have the same number of cells, but blocks which are close or next to each other  are also in the same processor.

then  processor 0  has   blocks  1, 2   = (100+500)= 600 cells

Processor 1  has  blocks  3,4 and 5 = (300+100+400)=800 cells , because  this blocks are connected together.

Since a have a  lot of blocks the imbalance is not such a problem. My problem here is first I have to find out block position (center). I.e., first read my geometric data, distribute blocks, each processor  rereads the data again  and corresponding fields.

With that information, can anyone tell me if paraview  generates gosht cells? and if yes for wich calculations ?  the gosht cells could be  be only located  at the boundaries of blocks (elements of vtkMultiBlockDataSet )

What happens when I do a cut, stream tracer, etc.  through several blocks,  that are not necessarely in the same processor ?
And what about parallel remote rendering ?  Data scatter like in 1)  would perform badly compared with  distribution 2)

Thanks in Advance

Orlando Rivera











--
MTU Aero Engines AG
Vorstand/Board of Management: Reiner Winkler, Vorsitzender/CEO; Dr. Rainer Martens, Michael Schreyögg, Dr. Stefan Weingartner
Vorsitzender des Aufsichtsrats/Chairman of the Supervisory Board: Klaus Eberhardt
Sitz der Gesellschaft/Registered Office: Muenchen
Handelsregister/Commercial Register: Muenchen HRB 157206

Diese E-Mail sowie ihre Anhaenge enthalten MTU-eigene vertrauliche oder rechtlich geschuetzte Informationen.
Wenn Sie nicht der beabsichtigte Empfaenger sind, informieren Sie bitte den Absender und loeschen Sie diese
E-Mail sowie die Anhaenge. Das unbefugte Speichern, Kopieren oder Weiterleiten ist nicht gestattet.

This e-mail and any attached documents are proprietary to MTU, confidential or protected by law.
If you are not the intended recipient, please advise the sender and delete this message and its attachments.
Any unauthorised storing, copying or distribution is prohibited.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.vtk.org/pipermail/vtkusers/attachments/20140127/47dd991c/attachment.html>