Hi Berk, I thought D3 wouldn't repartition the data but it was my mistake. I did some tests here and D3 worked properly, however, I was expecting that it would be possible to remove the internal surfaces, produced by D3, with "clean to grid" filter. I also tried to play with the "Boundary Mode" options with no success. The internal faces are still there.
Some further comments: I used a Xdmf reader in a temporal collection (12 time steps) of spatial collections (4 partitions) in a small unstructured mesh. The reader loads everything as an unique multiblock. It'll surely blows out the memory for a larger number of time steps and/or spatial partitions (I'm copying this message to xdmf list). ParaView 3.10.0-RC1 does not seem robust (yet) :-( it crashes all of a sudden and stop responding in random actions (I'll try to list some scenarios and let my dataset available to reproduce the errors). The client machine is a Windows 7 x64 connected to an offscreen MPI session with a Altix-ICE server (Linux x86_64). []'s Renato. On Wed, Mar 2, 2011 at 12:58 PM, Berk Geveci <[email protected]>wrote: > I am not sure that I follow. The output of D3 should be a dataset > re-partitioned to be load balanced whether the input is distributed or > not. Are you saying that the output of D3 is different based on > whether the input is distributed? > > As expected, D3 will probably produce a different partitioning than > the input. But that shouldn't be a problem, right? > > On Wed, Mar 2, 2011 at 7:54 AM, Renato Elias <[email protected]> wrote: > > Hi Berk, > > I already did such test. It really works but the dataset must be serial > and > > loaded in a parallel session. In this case, D3 will take care of the data > > distribution, load balance and ghost information. However, if the dataset > > read is already partitioned, D3 only creates a new partition and let the > > original distribution (in my case, performed by Metis) untouched. D3 is > not > > able to (re)partition unstructured grids (at least in my tests... or I'm > > doing something wrong). > > []'s > > Renato. > > > > On Tue, Mar 1, 2011 at 1:09 PM, Berk Geveci <[email protected]> > wrote: > >> > >> Great. Ping me in a 2-3 months - we should have started making changes > >> to the ghost level stuff by then. Until then, you should be able to > >> use D3 to redistribute data and generate (cell) ghost levels as > >> needed. So the following should work > >> > >> reader -> D3 -> extract surface > >> > >> -berk > >> > >> On Tue, Mar 1, 2011 at 10:57 AM, Renato Elias <[email protected]> > wrote: > >> >> Berk: Do you have the ability to mark a node as "owned" by one > >> >> partition > >> >> and as "ghost" on > >> >> other partitions? > >> > Yes! We classify processes as master and slaves according to their > >> > ranking > >> > numbers. After this we can assign the shared node to a master (which > >> > will > >> > take care of shared computations) and tell the slave process that this > >> > node > >> > is being shared with a master (and the slave process will consider it > as > >> > a > >> > "ghost" for computations). > >> > The ideas we used in our parallel solver were taken from the following > >> > article: > >> > Karanam, A. K., Jansen, K. E. and Whitinig, C. H., "Geometry Based > >> > Pre-processor for Parallel Fluid Dynamic Simulations Using a > >> > Hierarchical > >> > Basis", Engineering with Computers (24):17-26, 2008. > >> > This article was made available from the author > >> > in http://www.scorec.rpi.edu/REPORTS/2007-3.pdf (Figures 7 to 9 > explain > >> > the > >> > communication method between processes), > >> > Regards > >> > Renato. > >> > > >> > On Tue, Mar 1, 2011 at 12:24 PM, Berk Geveci <[email protected] > > > >> > wrote: > >> >> > >> >> Hi Renato, > >> >> > >> >> > I think I'm missing something.... you said cells only?! > >> >> > If I understood this subject correctly, a cell should be considered > >> >> > ghost if > >> >> > it's held by more than one partition/process, isn't it?! In this > >> >> > case, > >> >> > there'll be an overlapped layer of elements. The problem is that my > >> >> > MPI > >> >> > solver does not make use of this overlapped layer of > cells/elements. > >> >> > >> >> Yep. You understood correctly. Ghost cells are very common for finite > >> >> difference calculations but not as common for finite elements. > >> >> > >> >> > It only > >> >> > has nodes/points that are shared by processes. This explains why I > >> >> > asked > >> >> > by > >> >> > a ghost node (shared node would be more appropriated to define such > >> >> > kind > >> >> > of > >> >> > node). > >> >> > Can I consider a cell as ghost if it touches the parallel interface > >> >> > (without > >> >> > overlapping)? Would it work? > >> >> > >> >> Nope. Then you'd start seeing gaps. The right thing to do is for > >> >> ParaView to support ghost points (nodes) better. However, this is > >> >> non-trivial in some cases. For removing internal interfaces, it is > >> >> sufficient to mark points as ghosts. However, for accurately > >> >> performing statistics, you need to make sure that you count all > points > >> >> only once, which requires assigning ghost nodes to processes. So a > >> >> replicated node would be marked as ghost (a better word is shared) > and > >> >> also owned by a particular process. We are going to improve VTK's > >> >> ghost level support. This is something we'll support. However, it > will > >> >> be up to the simulation to produce the write output. Do you have the > >> >> ability to mark a node as "owned" by one partition and as "ghost" on > >> >> other partitions? > >> >> > >> >> Best, > >> >> -berk > >> > > >> > > >> > > >> > -- > >> > Renato N. Elias > >> > ============================================= > >> > Professor at Technology and Multilanguage Department (DTL) > >> > Federal Rural University of Rio de Janeiro (UFRRJ) > >> > Nova Iguaçu, RJ - Brazil > >> > ============================================= > >> > Researcher at High Performance Computing Center (NACAD) > >> > Federal University of Rio de Janeiro (UFRJ) > >> > Rio de Janeiro, Brazil > >> > > >> > > > > > > > > > -- > > Renato N. Elias > > ============================================= > > Professor at Technology and Multilanguage Department (DTL) > > Federal Rural University of Rio de Janeiro (UFRRJ) > > Nova Iguaçu, RJ - Brazil > > ============================================= > > Researcher at High Performance Computing Center (NACAD) > > Federal University of Rio de Janeiro (UFRJ) > > Rio de Janeiro, Brazil > > > > > -- Renato N. Elias ============================================= Professor at Technology and Multilanguage Department (DTL) Federal Rural University of Rio de Janeiro (UFRRJ) Nova Iguaçu, RJ - Brazil ============================================= Researcher at High Performance Computing Center (NACAD) Federal University of Rio de Janeiro (UFRJ) Rio de Janeiro, Brazil
_______________________________________________ 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 ParaView Wiki at: http://paraview.org/Wiki/ParaView Follow this link to subscribe/unsubscribe: http://www.paraview.org/mailman/listinfo/paraview
