[Paraview] benchmarking through pvbatch in 5.0

[Utkarsh Ayachit]

Carson,

Can you share the data/script? I can dig in to see what could be
happening. BTW, Render() will never use LOD irrespective of
LODThreshold since Render() always triggers full-res render aka
StillRender.

Another thing to try is to run with "-dr" in both cases: pvpython and
ParaVIew GUI + Python shell. Do you still see different results
between the two?

Utkarsh

On Wed, Feb 24, 2016 at 1:00 PM, Carson Brownlee <carson at tacc.utexas.edu> wrote:
> Hello,
> we are using the same benchmark scripts through pvbatch/pvpython as we were in 4.x, however with larger models we are getting odd results.  When running a 336 million triangle isosurface of the LLNL Richmtmeyer meshkov dataset in a for loop that consists of:
> for i in range(0,numRuns):
>         Render()
>
> We get reported render times of 30 seconds through pvbatch/pvpython.  Using outputs in polydatamapper, I can see it’s triggering the polydatamapper on each call to Render and the timing is consistent with what you might see if LOD is changing for each of those calls.  We set the LODThreshold to be large enough to never be triggered but something appears to be ejecting the polydata representation and regenerating on each call to Render().  If we run this script through the GUI using RunScript in the python shell, we do not see this issue, it only shows up in batch.
>
> Any leads on what might have changed since 4.x that would trigger this?
> Carson Brownlee
> _______________________________________________
> 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
>
> Search the list archives at: http://markmail.org/search/?q=ParaView
>
> Follow this link to subscribe/unsubscribe:
> http://public.kitware.com/mailman/listinfo/paraview