<html>
  <head>
    <meta content="text/html; charset=ISO-8859-1"
      http-equiv="Content-Type">
  </head>
  <body text="#000000" bgcolor="#FFFFFF">
    Hi Leo,<br>
    <br>
    Thanks, yes please send your fixes, or you could also push them to
    github. Which ever you prefer.<br>
    <br>
    Burlen<br>
    <br>
    On 06/08/2012 09:10 AM, Yuanxin Liu wrote:
    <blockquote
cite="mid:CAH+HHJ8swmVcSHa+q2zKEP0Ejz02jXkBQPAdiegNckatxxmhoQ@mail.gmail.com"
      type="cite">Hi, <br>
      &nbsp; I have recently gotten Burlen's code and updated it to work with
      the latest ParaView.&nbsp; Aside from vtkstd, there are also a few
      backward incompatible VTK changes ( see the VTK6.0 section on the
      VTK wiki).&nbsp;&nbsp; But it is not too much work. I will be happy send
      either of you my code changes if you need a reference.<br>
      <br>
      Leo<br>
      <br>
      <br>
      <div class="gmail_quote">On Fri, Jun 8, 2012 at 10:25 AM, Stephan
        Rogge <span dir="ltr">&lt;<a moz-do-not-send="true"
            href="mailto:Stephan.Rogge@tu-cottbus.de" target="_blank">Stephan.Rogge@tu-cottbus.de</a>&gt;</span>
        wrote:<br>
        <blockquote class="gmail_quote" style="margin:0 0 0
          .8ex;border-left:1px #ccc solid;padding-left:1ex">Someone told
          me that you have to clear your build directory completely and<br>
          start a fresh PV build.<br>
          <br>
          Stephan<br>
          <br>
          -----Urspr&uuml;ngliche Nachricht-----<br>
          Von: burlen [mailto:<a moz-do-not-send="true"
            href="mailto:burlen.loring@gmail.com" target="_blank">burlen.loring@gmail.com</a>]<br>
          Gesendet: Freitag, 8. Juni 2012 16:21<br>
          <div>
            <div>An: Stephan Rogge<br>
              Cc: 'Yuanxin Liu'; <a moz-do-not-send="true"
                href="mailto:paraview@paraview.org" target="_blank">paraview@paraview.org</a><br>
              Betreff: Re: [Paraview] Parallel Streamtracer<br>
              <br>
              Hi Stephan,<br>
              <br>
              Oh, thanks for the update, I wasn't aware of these
              changes. I have been<br>
              working with 3.14.1.<br>
              <br>
              Burlen<br>
              <br>
              On 06/08/2012 01:47 AM, Stephan Rogge wrote:<br>
              &gt; Hello Burlen,<br>
              &gt;<br>
              &gt; thank you very much for your post. I really would
              like to test your<br>
              &gt; plugin and so I've start to build it. Unfortunately
              I've got a lot of<br>
              &gt; compiler errors (e.g. vtkstd isn't used in PV master
              anymore). Which<br>
              &gt; PV version is the base for your plugin?<br>
              &gt;<br>
              &gt; Regards,<br>
              &gt; Stephan<br>
              &gt;<br>
              &gt; -----Urspr&uuml;ngliche Nachricht-----<br>
              &gt; Von: Burlen Loring [mailto:<a moz-do-not-send="true"
                href="mailto:bloring@lbl.gov" target="_blank">bloring@lbl.gov</a>]<br>
              &gt; Gesendet: Donnerstag, 7. Juni 2012 17:54<br>
              &gt; An: Stephan Rogge<br>
              &gt; Cc: 'Yuanxin Liu'; <a moz-do-not-send="true"
                href="mailto:paraview@paraview.org" target="_blank">paraview@paraview.org</a><br>
              &gt; Betreff: Re: [Paraview] Parallel Streamtracer<br>
              &gt;<br>
              &gt; Hi Stephan,<br>
              &gt;<br>
              &gt; I've experienced the scaling behavior that you report
              when I was<br>
              &gt; working on a project that required generating
              millions of streamlines<br>
              &gt; for a topological mapping algorithm interactively in
              ParaView. To get<br>
              &gt; the required scaling I wrote a stream tracer that
              uses a load on<br>
              &gt; demand approach with tunable block cache so that all
              ranks could<br>
              &gt; integrate any streamline and stay busy throughout the
              entire<br>
              &gt; computation. It was very effective on our data and
              I've used it to<br>
              &gt; integrate 30 Million streamlines in about 10min on
              256 cores. If you<br>
              &gt; really need better scalability than the distributed
              data tracing<br>
              &gt; approach implemented in PV, you might take a look at
              our work. The<br>
              &gt; down side of our approach is that in order to provide
              the demand<br>
              &gt; loading the reader has to implement a vtk object that
              provides an api<br>
              &gt; giving the integrator direct access to I/O
              functionality. In case you're<br>
              interested the stream tracer is class is vtkSQFieldTracer
              and our reader is<br>
              vtkSQBOVReader.<br>
              &gt; The latest release could be found here<br>
              &gt; <a moz-do-not-send="true"
href="https://github.com/burlen/SciberQuestToolKit/tarball/SQTK-20120531"
                target="_blank">https://github.com/burlen/SciberQuestToolKit/tarball/SQTK-20120531</a><br>
              &gt;<br>
              &gt; Burlen<br>
              &gt;<br>
              &gt; On 06/04/2012 02:21 AM, Stephan Rogge wrote:<br>
              &gt;&gt; Hello Leo,<br>
              &gt;&gt;<br>
              &gt;&gt; ok, I took the "disk_out_ref.ex2" example data
              set and did some time<br>
              &gt;&gt; measurements. Remember, my machine has 4 Cores +
              HyperThreading.<br>
              &gt;&gt;<br>
              &gt;&gt; My first observation is that PV seems to have a
              problem with<br>
              &gt;&gt; distributing the data when the Multi-Core option
              (GUI) is enabled.<br>
              &gt;&gt; When PV is started with builtin Multi-Core I was
              not able to apply a<br>
              &gt;&gt; stream tracer with more than 1000 seed points (PV
              is freezing and<br>
              &gt;&gt; never comes back). Otherwise, when pvserver
              processes has been<br>
              &gt;&gt; started manually I was able to set up to 100.000
              seed points. Is it a<br>
              bug?<br>
              &gt;&gt;<br>
              &gt;&gt; Now let's have a look on the scaling performance.
              As you suggested,<br>
              &gt;&gt; I've used the D3 filter for distributing the data
              along the processes.<br>
              &gt;&gt; The stream tracer execution time for 10.000 seed
              points:<br>
              &gt;&gt;<br>
              &gt;&gt; ## &nbsp; Bulitin: 10.063 seconds<br>
              &gt;&gt; ## &nbsp; 1 MPI-Process (no D3): 10.162 seconds<br>
              &gt;&gt; ## &nbsp; 4 MPI-Processes: 15.615 seconds<br>
              &gt;&gt; ## &nbsp; 8 MPI-Processes: 14.103 seconds<br>
              &gt;&gt;<br>
              &gt;&gt; and 100.000 seed points:<br>
              &gt;&gt;<br>
              &gt;&gt; ## &nbsp; Bulitin: 100.603 seconds<br>
              &gt;&gt; ## &nbsp; 1 MPI-Process (no D3): 100.967 seconds<br>
              &gt;&gt; ## &nbsp; 4 MPI-Processes: 168.1 seconds<br>
              &gt;&gt; ## &nbsp; 8 MPI-Processes: 171.325 seconds<br>
              &gt;&gt;<br>
              &gt;&gt; I cannot see any positive scaling behavior here.
              Maybe is this<br>
              &gt;&gt; example not appropriate for scaling measurements?<br>
              &gt;&gt;<br>
              &gt;&gt; One more thing: I've visualized the vtkProcessId
              and saw that the<br>
              &gt;&gt; whole vector field is partitioned. I thought,
              that each streamline is<br>
              &gt;&gt; integrated in its own process. But it seems that
              this is not the case.<br>
              &gt;&gt; This could explain my scaling issues: In cases of
              small vector fields<br>
              &gt;&gt; the overhead of synchronization becomes too large
              and decreases the<br>
              &gt; overall performance.<br>
              &gt;&gt; My suggestion is to have a parallel StreamTracer
              which is built for a<br>
              &gt;&gt; single machine with several threads. Could be
              worth to randomly<br>
              &gt;&gt; distribute the seeds over all available (local)
              processes? Of course,<br>
              &gt;&gt; each process have access on the whole vector
              field.<br>
              &gt;&gt;<br>
              &gt;&gt; Cheers,<br>
              &gt;&gt; Stephan<br>
              &gt;&gt;<br>
              &gt;&gt;<br>
              &gt;&gt;<br>
              &gt;&gt; Von: Yuanxin Liu [mailto:<a
                moz-do-not-send="true" href="mailto:leo.liu@kitware.com"
                target="_blank">leo.liu@kitware.com</a>]<br>
              &gt;&gt; Gesendet: Freitag, 1. Juni 2012 16:13<br>
              &gt;&gt; An: Stephan Rogge<br>
              &gt;&gt; Cc: Andy Bauer; <a moz-do-not-send="true"
                href="mailto:paraview@paraview.org" target="_blank">paraview@paraview.org</a><br>
              &gt;&gt; Betreff: Re: [Paraview] Parallel Streamtracer<br>
              &gt;&gt;<br>
              &gt;&gt; Hi, Stephan,<br>
              &gt;&gt; &nbsp; &nbsp; I did measure the performance at some point
              and was able to get<br>
              &gt;&gt; fairly decent speed up with more processors. So I
              am surprised you<br>
              &gt;&gt; are seeing huge latency.<br>
              &gt;&gt;<br>
              &gt;&gt; &nbsp; &nbsp; Of course, the performance is sensitive to
              the input. &nbsp;It is also<br>
              &gt;&gt; sensitive to how readers distribute data. So, one
              thing you might<br>
              &gt;&gt; want to try is to attach the "D3" filter to the
              reader.<br>
              &gt;&gt;<br>
              &gt;&gt; &nbsp; &nbsp; If that doesn't help, &nbsp;I will be happy to get
              your data and take<br>
              &gt;&gt; a<br>
              &gt; look.<br>
              &gt;&gt; Leo<br>
              &gt;&gt;<br>
              &gt;&gt; On Fri, Jun 1, 2012 at 1:54 AM, Stephan<br>
              &gt;&gt; Rogge&lt;<a moz-do-not-send="true"
                href="mailto:Stephan.Rogge@tu-cottbus.de"
                target="_blank">Stephan.Rogge@tu-cottbus.de</a>&gt;<br>
              &gt;&gt; wrote:<br>
              &gt;&gt; Leo,<br>
              &gt;&gt;<br>
              &gt;&gt; As I mentioned in my initial post of this thread:
              I used the<br>
              &gt;&gt; up-to-date master branch of ParaView. Which means
              I have already used<br>
              &gt;&gt; your implementation.<br>
              &gt;&gt;<br>
              &gt;&gt; I can imagine, to parallelize this algorithm can
              be very tough. And I<br>
              &gt;&gt; can see that distribute the calculation over 8
              processes does not<br>
              &gt;&gt; lead to a nice scaling.<br>
              &gt;&gt;<br>
              &gt;&gt; But I don't understand this huge amount of
              latency when using the<br>
              &gt;&gt; StreamTracer in a Cave-Mode with two view ports
              and two pvserver<br>
              &gt;&gt; processes on the same machine (extra machine for
              the client). I guess<br>
              &gt;&gt; the tracer filter is applied for each viewport
              separately? This would<br>
              &gt;&gt; be ok as long as both filter executions run
              parallel. And I doubt<br>
              &gt;&gt; that<br>
              &gt; this is the case.<br>
              &gt;&gt; Can you help to clarify my problem?<br>
              &gt;&gt;<br>
              &gt;&gt; Regards,<br>
              &gt;&gt; Stephan<br>
              &gt;&gt;<br>
              &gt;&gt;<br>
              &gt;&gt; Von: Yuanxin Liu [mailto:<a
                moz-do-not-send="true" href="mailto:leo.liu@kitware.com"
                target="_blank">leo.liu@kitware.com</a>]<br>
              &gt;&gt; Gesendet: Donnerstag, 31. Mai 2012 21:33<br>
              &gt;&gt; An: Stephan Rogge<br>
              &gt;&gt; Cc: Andy Bauer; <a moz-do-not-send="true"
                href="mailto:paraview@paraview.org" target="_blank">paraview@paraview.org</a><br>
              &gt;&gt; Betreff: Re: [Paraview] Parallel Streamtracer<br>
              &gt;&gt;<br>
              &gt;&gt; It is in the current VTK and ParaView master.
              &nbsp;The class is<br>
              &gt;&gt; vtkPStreamTracer.<br>
              &gt;&gt;<br>
              &gt;&gt; Leo<br>
              &gt;&gt; On Thu, May 31, 2012 at 3:31 PM, Stephan<br>
              &gt;&gt; Rogge&lt;<a moz-do-not-send="true"
                href="mailto:stephan.rogge@tu-cottbus.de"
                target="_blank">stephan.rogge@tu-cottbus.de</a>&gt;<br>
              &gt;&gt; wrote:<br>
              &gt;&gt; Hi, Andy and Leo,<br>
              &gt;&gt;<br>
              &gt;&gt; thanks for your replies.<br>
              &gt;&gt;<br>
              &gt;&gt; Is it possible to get this new implementation? I
              would to give it a try.<br>
              &gt;&gt;<br>
              &gt;&gt; Regards,<br>
              &gt;&gt; Stephan<br>
              &gt;&gt;<br>
              &gt;&gt; Am 31.05.2012 um 17:48 schrieb Yuanxin Liu&lt;<a
                moz-do-not-send="true" href="mailto:leo.liu@kitware.com"
                target="_blank">leo.liu@kitware.com</a>&gt;:<br>
              &gt;&gt; Hi, Stephan,<br>
              &gt;&gt; &nbsp; &nbsp; &nbsp;The previous implementation only has serial
              performance: &nbsp;It<br>
              &gt;&gt; traces the streamlines one at a time and never
              starts a new<br>
              &gt;&gt; streamline until the previous one finishes. &nbsp;With
              communication<br>
              &gt;&gt; overhead, it is not surprising it got slower.<br>
              &gt;&gt;<br>
              &gt;&gt; &nbsp; &nbsp; My new implementation is able to let the
              processes working on<br>
              &gt;&gt; different streamlines simultaneously and should
              scale much better.<br>
              &gt;&gt;<br>
              &gt;&gt; Leo<br>
              &gt;&gt;<br>
              &gt;&gt; On Thu, May 31, 2012 at 11:27 AM, Andy Bauer&lt;<a
                moz-do-not-send="true"
                href="mailto:andy.bauer@kitware.com" target="_blank">andy.bauer@kitware.com</a>&gt;<br>
              &gt; wrote:<br>
              &gt;&gt; Hi Stephan,<br>
              &gt;&gt;<br>
              &gt;&gt; The parallel stream tracer uses the partitioning
              of the grid to<br>
              &gt;&gt; determine which process does the integration.
              When the streamline<br>
              &gt;&gt; exits the subdomain of a process there is a
              search to see if it<br>
              &gt;&gt; enters a subdomain assigned to any other
              processes before figuring it<br>
              &gt;&gt; whether it has left the entire domain.<br>
              &gt;&gt;<br>
              &gt;&gt; Leo, copied here, has been improving the
              streamline implementation<br>
              &gt;&gt; inside of VTK so you may want to get his newer
              version. It is a<br>
              &gt;&gt; pretty tough algorithm to parallelize efficiently
              without making any<br>
              &gt;&gt; assumptions on the flow or partitioning.<br>
              &gt;&gt;<br>
              &gt;&gt; Andy<br>
              &gt;&gt;<br>
              &gt;&gt; On Thu, May 31, 2012 at 4:16 AM, Stephan<br>
              &gt;&gt; Rogge&lt;<a moz-do-not-send="true"
                href="mailto:Stephan.Rogge@tu-cottbus.de"
                target="_blank">Stephan.Rogge@tu-cottbus.de</a>&gt;<br>
              &gt;&gt; wrote:<br>
              &gt;&gt; Hello,<br>
              &gt;&gt;<br>
              &gt;&gt; I have a question related to the parallelism of
              the stream tracer: As<br>
              &gt;&gt; I understand the code right, each line
              integration (trace) is<br>
              &gt;&gt; processed in an own MPI process. Right?<br>
              &gt;&gt;<br>
              &gt;&gt; To test the scalability of the Stream tracer I've
              load a structured<br>
              &gt;&gt; (curvilinear) grid and applied the filter with a
              Seed resolution of<br>
              &gt;&gt; 1500 and check the timings in a single and
              multi-thread (Multi Core<br>
              &gt;&gt; enabled in PV<br>
              &gt;&gt; GUI) situation.<br>
              &gt;&gt;<br>
              &gt;&gt; I was really surprised that multi core slows done
              the execution time<br>
              &gt;&gt; to 4 seconds. The single core takes only 1.2
              seconds. Data migration<br>
              &gt;&gt; cannot be the explanation for that behavior (0.5
              seconds). What is<br>
              &gt;&gt; the<br>
              &gt; problem here?<br>
              &gt;&gt; Please see attached some statistics...<br>
              &gt;&gt;<br>
              &gt;&gt; Data:<br>
              &gt;&gt; * Structured (Curvilinear) Grid<br>
              &gt;&gt; * 244030 Cells<br>
              &gt;&gt; * 37 MB Memory<br>
              &gt;&gt;<br>
              &gt;&gt; System:<br>
              &gt;&gt; * Intel i7-2600K (4 Cores + HT = 8 Threads)<br>
              &gt;&gt; * 16 GB Ram<br>
              &gt;&gt; * Windows 7 64 Bit<br>
              &gt;&gt; * ParaView (master-branch, 64 bit compilation)<br>
              &gt;&gt;<br>
              &gt;&gt; #################################<br>
              &gt;&gt; Single Thread (Seed resolution 1500):<br>
              &gt;&gt; #################################<br>
              &gt;&gt;<br>
              &gt;&gt; Local Process<br>
              &gt;&gt; Still Render, &nbsp;0.014 seconds<br>
              &gt;&gt; RenderView::Update, &nbsp;1.222 seconds<br>
              &gt;&gt; &nbsp; &nbsp; &nbsp;vtkPVView::Update, &nbsp;1.222 seconds<br>
              &gt;&gt; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;Execute vtkStreamTracer id: 2184, &nbsp;1.214
              seconds Still<br>
              &gt;&gt; Render,<br>
              &gt;&gt; 0.015 seconds<br>
              &gt;&gt;<br>
              &gt;&gt; #################################<br>
              &gt;&gt; Eight Threads (Seed resolution 1500):<br>
              &gt;&gt; #################################<br>
              &gt;&gt;<br>
              &gt;&gt; Local Process<br>
              &gt;&gt; Still Render, &nbsp;0.029 seconds<br>
              &gt;&gt; RenderView::Update, &nbsp;4.134 seconds<br>
              &gt;&gt; vtkSMDataDeliveryManager: Deliver Geome, &nbsp;0.619
              seconds<br>
              &gt;&gt; &nbsp; &nbsp; &nbsp;FullRes Data Migration, &nbsp;0.619 seconds Still
              Render, &nbsp;0.042<br>
              &gt;&gt; seconds<br>
              &gt;&gt; &nbsp; &nbsp; &nbsp;OpenGL Dev Render, &nbsp;0.01 seconds<br>
              &gt;&gt;<br>
              &gt;&gt;<br>
              &gt;&gt; Render Server, Process 0<br>
              &gt;&gt; RenderView::Update, &nbsp;4.134 seconds<br>
              &gt;&gt; &nbsp; &nbsp; &nbsp;vtkPVView::Update, &nbsp;4.132 seconds<br>
              &gt;&gt; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;Execute vtkStreamTracer id: 2193, &nbsp;3.941
              seconds FullRes<br>
              &gt;&gt; Data Migration, &nbsp;0.567 seconds<br>
              &gt;&gt; &nbsp; &nbsp; &nbsp;Dataserver gathering to 0, &nbsp;0.318 seconds<br>
              &gt;&gt; &nbsp; &nbsp; &nbsp;Dataserver sending to client, &nbsp;0.243 seconds<br>
              &gt;&gt;<br>
              &gt;&gt; Render Server, Process 1<br>
              &gt;&gt; Execute vtkStreamTracer id: 2193, &nbsp;3.939 seconds<br>
              &gt;&gt;<br>
              &gt;&gt; Render Server, Process 2<br>
              &gt;&gt; Execute vtkStreamTracer id: 2193, &nbsp;3.938 seconds<br>
              &gt;&gt;<br>
              &gt;&gt; Render Server, Process 3<br>
              &gt;&gt; Execute vtkStreamTracer id: 2193, &nbsp;4.12 seconds<br>
              &gt;&gt;<br>
              &gt;&gt; Render Server, Process 4<br>
              &gt;&gt; Execute vtkStreamTracer id: 2193, &nbsp;3.938 seconds<br>
              &gt;&gt;<br>
              &gt;&gt; Render Server, Process 5<br>
              &gt;&gt; Execute vtkStreamTracer id: 2193, &nbsp;3.939 seconds<br>
              &gt;&gt;<br>
              &gt;&gt; Render Server, Process 6<br>
              &gt;&gt; Execute vtkStreamTracer id: 2193, &nbsp;3.938 seconds<br>
              &gt;&gt;<br>
              &gt;&gt; Render Server, Process 7<br>
              &gt;&gt; Execute vtkStreamTracer id: 2193, &nbsp;3.939 seconds<br>
              &gt;&gt;<br>
              &gt;&gt; Cheers,<br>
              &gt;&gt; Stephan<br>
              &gt;&gt;<br>
              &gt;&gt;<br>
              &gt;&gt; _______________________________________________<br>
              &gt;&gt; Powered by <a moz-do-not-send="true"
                href="http://www.kitware.com" target="_blank">www.kitware.com</a><br>
              &gt;&gt;<br>
              &gt;&gt; Visit other Kitware open-source projects at<br>
              &gt;&gt; <a moz-do-not-send="true"
                href="http://www.kitware.com/opensource/opensource.html"
                target="_blank">http://www.kitware.com/opensource/opensource.html</a><br>
              &gt;&gt;<br>
              &gt;&gt; Please keep messages on-topic and check the
              ParaView Wiki at:<br>
              &gt;&gt; <a moz-do-not-send="true"
                href="http://paraview.org/Wiki/ParaView" target="_blank">http://paraview.org/Wiki/ParaView</a><br>
              &gt;&gt;<br>
              &gt;&gt; Follow this link to subscribe/unsubscribe:<br>
              &gt;&gt; <a moz-do-not-send="true"
                href="http://www.paraview.org/mailman/listinfo/paraview"
                target="_blank">http://www.paraview.org/mailman/listinfo/paraview</a><br>
              &gt;&gt;<br>
              &gt;&gt;<br>
              &gt;&gt;<br>
              &gt;&gt;<br>
              &gt;&gt;<br>
              &gt;&gt; _______________________________________________<br>
              &gt;&gt; Powered by <a moz-do-not-send="true"
                href="http://www.kitware.com" target="_blank">www.kitware.com</a><br>
              &gt;&gt;<br>
              &gt;&gt; Visit other Kitware open-source projects at<br>
              &gt;&gt; <a moz-do-not-send="true"
                href="http://www.kitware.com/opensource/opensource.html"
                target="_blank">http://www.kitware.com/opensource/opensource.html</a><br>
              &gt;&gt;<br>
              &gt;&gt; Please keep messages on-topic and check the
              ParaView Wiki at:<br>
              &gt;&gt; <a moz-do-not-send="true"
                href="http://paraview.org/Wiki/ParaView" target="_blank">http://paraview.org/Wiki/ParaView</a><br>
              &gt;&gt;<br>
              &gt;&gt; Follow this link to subscribe/unsubscribe:<br>
              &gt;&gt; <a moz-do-not-send="true"
                href="http://www.paraview.org/mailman/listinfo/paraview"
                target="_blank">http://www.paraview.org/mailman/listinfo/paraview</a><br>
              &gt;<br>
              <br>
              <br>
            </div>
          </div>
        </blockquote>
      </div>
      <br>
      <br>
      <fieldset class="mimeAttachmentHeader"></fieldset>
      <br>
      <pre wrap="">_______________________________________________
Powered by <a class="moz-txt-link-abbreviated" href="http://www.kitware.com">www.kitware.com</a>

Visit other Kitware open-source projects at <a class="moz-txt-link-freetext" href="http://www.kitware.com/opensource/opensource.html">http://www.kitware.com/opensource/opensource.html</a>

Please keep messages on-topic and check the ParaView Wiki at: <a class="moz-txt-link-freetext" href="http://paraview.org/Wiki/ParaView">http://paraview.org/Wiki/ParaView</a>

Follow this link to subscribe/unsubscribe:
<a class="moz-txt-link-freetext" href="http://www.paraview.org/mailman/listinfo/paraview">http://www.paraview.org/mailman/listinfo/paraview</a>
</pre>
    </blockquote>
    <br>
  </body>
</html>