[Paraview] Any performance hints or tips for .pvtr reader?

[Moreland, Kenneth]

The parallel VTK XML file formats are honestly not that optimized for use on parallel file systems.  Inefficiencies in reading can come from many places and I don't think you have provided enough information to be totally sure on the problem (even if I was an expert, which I am not), but horrendously bad performance like that you are seeing usually happens when a bunch of processes make lots of small reads.  The file system gets inundated with tons of file requests.  In response, the read head on each cylinder starts moving back in forth trying to satisfy all of these requests, and your performance is now dead.

Efficient parallel reading has nice, organized requests that have large, contiguous reads that can be streamed off of the disk cylinders very efficiently.  The easiest way to do that is to simply match files to processes.  That is, organize the data so that each process only has to do one read from a single file.  The easiest way to do that is to read in your data into ParaView or VTK on 64 processes as you are doing now (pay the time price) and then write it back out in a new .pvtr file.  The new .pvtr file should contain 64 .vtr files.  Loading that back onto 64 processes should be quick.

-Ken


On 6/26/09 9:42 AM, "Jonathan Dursi" <ljdursi at scinet.utoronto.ca> wrote:

Hi:

Are there any performance gotchas I need to know about when dealing with
parallel xml vtk file formats?  Right now, running pvserver on 64 cores
(2GB/core, 1 pipe to disk for every 8 cores) it's taking literally hours
to read 16GB of data as a .pvtr file containing a list of 192 .vtr files.

        - Jonathan

--
Jonathan Dursi     <ljdursi at scinet.utoronto.ca>
_______________________________________________
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




   ****      Kenneth Moreland
    ***      Sandia National Laboratories
***********
*** *** ***  email: kmorel at sandia.gov
**  ***  **  phone: (505) 844-8919
    ***      web:   http://www.cs.unm.edu/~kmorel

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.paraview.org/pipermail/paraview/attachments/20090629/60942ec0/attachment-0001.htm>