[Paraview] Fwd: Stream Tracer Scaling

[Berk Geveci]

Actually, if the integration is fairly load balanced across partitions, the
streamline integration can scale OK. It is highly dependent on how the
seeds are distributed and if the integration ends up clustering on a few
processes. Amir:

- How big is your seed source? Does it span multiple partitions? How many
seed points?
- Do the streamlines stay together or do the move around the domain more
independently?

Are you running on a single machine by the way? I have been thinking about
parallelizing the streamline filter over threads, which should scale better
on a single machine.

Best,
-berk


On Fri, Apr 18, 2014 at 10:39 AM, Andy Bauer <andy.bauer at kitware.com> wrote:

> Hi Amir,
>
> Please keep the discussion on the mailing list so that everyone that wants
> to can participate.
>
> You could potentially do the smaller velocity data set in serial and the
> larger data set in parallel but would probably take a bit of extra work
> with Python. From the Python shell you have access to the AlltoN filter
> which can create a copy of the data set on a reduced amount of processes.
>
> Regards,
> Andy
>
> ---------- Forwarded message ----------
> From: Amir A <amiramir12345 at outlook.com>
> Date: Thu, Apr 17, 2014 at 2:52 PM
> Subject: RE: [Paraview] Stream Tracer Scaling
> To: Andy Bauer <andy.bauer at kitware.com>
>
>
> So maybe I need to clarify a little bit. I have two data sets M and V. M
> is a very large dataset for the mesh structure (~8GB) and I only need to
> perform simple filters to it which scales very well in parallel. V is the
> velocity dataset which is only in part of the domain and is relatively
> small. I want to do streamtrace filter on V.
> The problem as I said is that if I do parallel M's filters will scale, but
> the streamtracer will become many times slower. So is it possible to only
> do streamtrace on one processor serially and the rest in paralllel?
>
> ------------------------------
> From: amiramir12345 at outlook.com
> To: andy.bauer at kitware.com
> Subject: RE: [Paraview] Stream Tracer Scaling
> Date: Thu, 17 Apr 2014 12:26:49 -0500
>
>
> Hi Andy:
>
> Thanks for your reply. So is it possible to run the StreamTracer in serial
> in a parallel setting? It would be great if this is possible, because I can
> keep scaling of other filters and avoid doing everything in serial.
>
> Best,
> -Amir
>
> ------------------------------
> Date: Thu, 17 Apr 2014 13:19:52 -0400
> Subject: Re: [Paraview] Stream Tracer Scaling
> From: andy.bauer at kitware.com
> To: amiramir12345 at outlook.com
> CC: paraview at paraview.org
>
>
> The main issue is that stream tracer just doesn't scale well. It's in
> essence a serial computation which is very difficult to parallelize well.
>
>
>
> _______________________________________________
> 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
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.paraview.org/pipermail/paraview/attachments/20140421/cda07185/attachment-0001.html>