It seems that you're writting parallel geometry in a serial way. Partitioned grid, as far as I know, must be stored as spatial collections in xdmf format. In other words, each processors stores its own grid portion (a pair xdmf/hdf5 file) and rank0, for instance, writes a xdmf file describing how to "glue" the grid pieces using a spatial collection clause.
[]'s Renato. 2009/9/2 weaponfire2005 <[email protected]> > Hi: > Well, I post my .xmf file(the attachment). In fact，the structure of > the xmf file is simple. It only contains some <Grid>. I use a "for()" > loop to write this file. The raw data is HDF5 data format. Is something > wrong? > Thanks for help. > > > > > 2009-09-01，"Berk Geveci" [email protected]> ： > >Can you post an example dataset? Just the xmf file would be sufficient. > > > >-berk > > > > <[email protected]%3e+%ef%bc%9a%3ecan+you+post+an+example+dataset?+just+the+xmf+file+would+be+sufficient.%3e%3e-berk%3e%3e> > > Re:Re: [Paraview] Paraview store all data in every server node? > >> Hi: > >> I used vtkXdmfReader to read my data(ParaView version is 3.4.0), and I > >> also found some information about my problem(XDMF bad memory use) from > >> the mail list. But I didn't find a final resolution for it. > >> My problem is, for example: > >> run: mpirun -np 4 ./pvserver > >> after reading data and rendering isosurface, each > >> node(process) consume 2GB memory > >> and run: mpirun -np 16 ./pvserver > >> after reading data and rendering isosurface, each > >> node(process) also consume 2GB memory > >> For this problem, I could't visualize large-scale dataset. Because if a > >> dataset could be visualized by one node, It also could be visualized by 16 > >> nodes, but if it could't be visualized by one node, it also could't by 16 > >> nodes(or more). > >> Maybe I should try ParaView 3.6.1. I will cry if the problem still > >> exist > >> in version 3.6.1...... > > >> > >> 2009-08-30，"Andy Bauer" <[email protected]> ： > >> > >> The problem may be the reader that you're using. Some of the readers in > >> VTK > >> are not parallel so even if you run multiple processes each process will > >> still try to load the entire dataset in that case. > >> > >> Andy > >> > >> 2009/8/30 weaponfire2005 [email protected]>: > >>> > >>> Hi all: > >>> First thanks for Dave.Demarle's help. > >>> I use "mpirun -np 4 ./pvserver" to launch a server with four > >>> nodes, visualizing a 800*800*300 grid. It seems that each process consume > >>> too much memory. So I extend the number of nodes from 4 to 8 using "mpirun > >>> -np 8 ./pvserver" , but each process on 8 nodes server consume the same > >>> memory as 4 nodes server. The situation of 16 processes(16 nodes) is same > >>> as > >>> above. > >>> I thought that with the number of processes' growth, each process on > >>> server would share less raw data, so the quantity of memory each process > >>> consuming would decline. It proved that I was wrong. Could someone tell me > >>> the reason? > >>> Thanks for your help:) > >>> Pan Wang > >>> > >>>>> _______________________________________________ > >>> 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 > >>> > >> > >> > >> <http://www.paraview.org/mailman/listinfo/paraview%3E%3E%3E%3E%3E%3E%3E%3E%3E>>> > >> 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 > >> > >> > > > > ------------------------------ > 没有广告的终身免费邮箱,www.yeah.net <http://www.yeah.net/?from=footer> > _______________________________________________ > 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 > > -- Renato N. Elias =================================== High Performance Computing Center (NACAD) Federal University of Rio de Janeiro (UFRJ) Rio de Janeiro, Brazil Sent from Rio De Janeiro, RJ, 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
