[Paraview] [EXTERNAL] extract CTH parts (UNCLASSIFIED)

[Angelini, Richard C (Rick) CIV USARMY ARL (US)]

Classification: UNCLASSIFIED
Caveats: NONE

Alan - yeah, the dataset opens correctly with < 80 servers.   It doesn't
matter how I distribute those 80 servers - I can do 16 processes on 5 nodes,
8 processes on 10 nodes and it works.   If I play the same game with > 80
nodes, I start seeing the holes.   I *doubt* that it's a bad node, but I'll
do some deeper investigating just to verify.   

-----Original Message-----
From: Scott, W Alan [mailto:wascott at sandia.gov] 
Sent: Monday, August 26, 2013 3:43 PM
To: Angelini, Richard C (Rick) CIV USARMY ARL (US); paraview at paraview.org
Subject: RE: [EXTERNAL] [Paraview] extract CTH parts (UNCLASSIFIED)

Bad node?

Try running the problem on a few nodes.  Next, try running the "big node"
problem.  As you shift the nodes you are using, does the issue follow the
underlying iron, or the same bad blocks in the dataset?

Dataset opens correctly with smaller number of nodes?

Want me to try it here?  80 does seem really, really small/easy/trivial for
big data...

Alan

-----Original Message-----
From: paraview-bounces at paraview.org [mailto:paraview-bounces at paraview.org]
On Behalf Of Angelini, Richard C (Rick) CIV USARMY ARL (US)
Sent: Monday, August 26, 2013 1:38 PM
To: paraview at paraview.org
Subject: [EXTERNAL] [Paraview] extract CTH parts (UNCLASSIFIED)

Classification: UNCLASSIFIED
Caveats: NONE

I'm having an issue with the extract CTH parts filter in ParaView 4.0.1.
The dataset consists of 1024 spcth files, regular structured mesh, 14
blocks.  The issue seems to be that once I exceed X number of server
processes (it seems to be about 80 servers), I start seeing holes in the
resulting geometry.  The more I exceed the magic number of servers, the more
holes I see in the resulting part.    If I create a clip using the
volume_fraction scalar variable(s),  the geometry of the clip is correct -
that is, the underlying data seems to be there.   So, I'm thinking that the
data is being properly distributed and all of the data is accounted for, but
that there's an issue with the "extract CTH parts" filter?    Doesn't seem
likely, but stranger things have happened.

Any ideas?

Classification: UNCLASSIFIED
Caveats: NONE




Classification: UNCLASSIFIED
Caveats: NONE


-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/x-pkcs7-signature
Size: 5575 bytes
Desc: not available
URL: <http://www.paraview.org/pipermail/paraview/attachments/20130826/a9be3c75/attachment-0001.bin>