Is it possible for you to post this data or a smaller version somewhere? I need an Ensight file similar to your to experiment. It doesn't have to be public, just accessible by Kitware.<br><br>-berk<br><br><br><div><span class="gmail_quote">
On 22 Jan 2008 14:04:17 -0000, <b class="gmail_sendername">Raashid Baig</b> <<a href="mailto:raashid.b@rediffmail.com">raashid.b@rediffmail.com</a>> wrote:</span><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<p>
I have a big problem while trying to visualize large multi-block<br>
Unstructured grid dataset which is in the form of ASCII Ensight files.<br>
(16 million cells 2.1GB disk usage).<br>
<br>
When I start ParaView(3.1.2, complied on a Linux cluster with OpenMPI)<br>
it starts without any problem, but ParaView fails while trying to load<br>
this file. pvserver crashes when the memory usage on the headnode of<br>
the cluster is about 20% to total available memory. Also on all other<br>
nodes, besides the head node ParaView does not seem to consume any<br>
memory. I have reported the problem on the ParaView mailing list on<br>
15-01-2007.<br>
<br>
However when I start the pvserver only on one node, the ParaView loads<br>
the Ensight file without crashing, since the single node does not<br>
sufficiently handle this large amount of data, I tried to save the<br>
loaded data in some other file format using ParaView. But for this<br>
data set only the option of VTK Multi-Block files is available, on<br>
saving that I get 1 .vtm file and 11 separate *.vtu files.<br>
<br>
Now when I restart the ParaView on 10 nodes and try loading this vtm<br>
file, it loads properly but all the load(of memory) falls only on the<br>
headnode of the cluster. If I perform any operation (Velocity glyphs<br>
for example) all the load again falls only on headnode of the<br>
cluster, this results in a very slow performance as well as risk of<br>
crashing the entire cluster for memory expensive operations (like<br>
isosurfaces).<br>
<br>
According the ParaView Wiki:<br>
<br>
"""Currently there are only a few readers that truly work in parallel:<br>
VTK files (not legacy), partitioned legacy VTK files, ParaView data<br>
files, HDF5 files, EnSight master server files, and raw (binary) files<br>
can be read in parallel."""<br>
<br>
Now what can I do with the current situation, there is no Ensight<br>
Master-Server/HDF5 writer in VTK. Both VTK-XML and VTK-legacy file<br>
formats don't support the multi-block data and the existing *.vtm file<br>
does not seem to work properly when ParaView runs in parallel mode.<br>
<br>
What can be the possible solution for the existing situations, can VTK<br>
be of some help ? Or I have to create a new Parallel Reader for<br>
ParaView.<br>
<br>
</p>
<br><br>
<table style="font-family: Verdana; font-size: 11px; line-height: 15px;" border="0" cellpadding="0" cellspacing="0" height="57" width="644"><tbody><tr><td><a href="http://adworks.rediff.com/cgi-bin/AdWorks/click.cgi/www.rediff.com/signature-home.htm/1050715198@Middle5/2027411_2020222/2027231/1?PARTNER=3&OAS_QUERY=null" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">
<img alt="Tata Aig" border="0"></a></td></tr></tbody></table><br>_______________________________________________<br>ParaView mailing list<br><a onclick="return top.js.OpenExtLink(window,event,this)" href="mailto:ParaView@paraview.org">
ParaView@paraview.org</a><br><a onclick="return top.js.OpenExtLink(window,event,this)" href="http://www.paraview.org/mailman/listinfo/paraview" target="_blank">http://www.paraview.org/mailman/listinfo/paraview</a><br><br>
</blockquote></div><br>