[visit-developers] Ghost zones on rectilinear data?

Meredith, Jeremy S. jsmeredith at ornl.gov
Tue Oct 14 13:45:39 EDT 2008


For posterity's sake:

 

- Yes, avtRealDims is what I was looking for.  My confusion stemmed from
the fact that the data set I was looking at had ghost *nodes*, not ghost
*zones*, and so avtRealDims wouldn't show that.

 

- A followup problem I had was the vtkExtractRectilnearGrid doesn't
adjust avtRealDims, it passes it along blindly.  This is fixed inside
the avtGhostZoneFilter where it simply removes the array afterwards.  I
thought about modifying vtkExtractRectilinearGrid to adjust avtRealDims,
but since we know it should encompass only real data, having it
extracted is probably even better.

 

- The real reason I wanted this was that it turns out that the generic
execution path inside vtkDataSetRemoveGhostCells  (triggered in the
rgrid case when avtRealDims is missing) always converted the input
dataset to unstructured grid, even if it consisted of only real zones.
I'm adding code to short circuit in that case, should be a nice speedup
and memory savings in the situations I'm looking at.

 

--

Jeremy Meredith

Oak Ridge National Laboratory

 

From: visit-developers-bounces at email.ornl.gov
[mailto:visit-developers-bounces at email.ornl.gov] On Behalf Of Meredith,
Jeremy S.
Sent: Monday, October 13, 2008 4:51 PM
To: VisIt Developers
Subject: [visit-developers] Ghost zones on rectilinear data?

 

Question: I know for rectilinear data we still use the avtGhostZone
array to signify ghost zones.

And, I know that we have some things like "base index" and "group #"
encoded into field data.

 

But do we have anything that signifies the begin/end index of ghost
zones in a logical index fashion?  And if so, can I trust it later in
the pipeline?

 

(I see avtRealDims, but that doesn't appear to contain the ghost zone
indexes -- at least not correctly as far as I can tell -- bigsil.silo
has ghost zones, but its avtRealDims is [0 8 0 8 0 8] for all domains.
Either that, or I'm not understanding what avtRealDims is telling me.)

 

 

--

Jeremy Meredith

Oak Ridge National Laboratory

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://email.ornl.gov/pipermail/visit-developers/attachments/20081014/22f04c6d/attachment.html>


More information about the visit-developers mailing list