[prev in list] [next in list] [prev in thread] [next in thread]
List: paraview
Subject: Re: [Paraview] [BUG] segfault playing a simulation
From: Sergi Mateo Bellido <sergi.mateo.bellido () gmail ! com>
Date: 2015-08-31 7:27:23
Message-ID: 55E401DB.5030007 () gmail ! com
[Download RAW message or body]
[Attachment #2 (multipart/alternative)]
Hi Cory,
I'm stuck with this issue :(
1. Changing just the pvti file is not enough: I got a segfault using the
raw binary files.
2. I tried also adding a dummy slice at the end of all pieces to fix the
problem that I reported in one of my previous emails. So, instead of
changing the PVTI files I tried changing the VTI files.
2.1. First of all, I had to increase the number of tuples of the
VtkArray because I was adding the dummy slice. Since I didn't change
the whole extent and the piece extent of each VTI file, I got this
warning message:
Warning: In /opt/VTK-6.2.0/Common/DataModel/vtkDataSet.cxx, line 421
vtkImageData (0x3450390): Point array Var with 1 components, has
4080501 tuples but there are only 4040100 points
This didn't work either.
2.2. Afterwards, I tried modifying also the whole extent and the
piece extent. This seemed to work :)
Despite 2.2 seemed to work, I'm not sure if this is a valid solution or
just a workaround: If I open a VTI file directly (because I want to see
the data produced by a MPI process) the dummy slice is visualized.
From my point of view, I think that we're trying to address an issue of
the PVTI format modifying something that works fine (VTI files).
Best regards,
Sergi
On 08/12/2015 02:46 PM, Sergi Mateo Bellido wrote:
> Cory,
>
> I did not change the VTI files in the first experiment but it crashed
> anyway.
>
> Thanks!
> Sergi
>
> On 08/12/2015 01:55 PM, Cory Quammen wrote:
>> Sergi,
>>
>> Note that I did *not* change the whole extents of the VTI files after
>> I changed the listed extents in the PVTI files.
>>
>> Cory
>>
>> On Wed, Aug 12, 2015 at 1:47 AM, Sergi Mateo Bellido
>> <sergi.mateo.bellido@gmail.com
>> <mailto:sergi.mateo.bellido@gmail.com>> wrote:
>>
>> Cory,
>>
>> The ascii version is the more stable version and it usually
>> doesn't crash (even the original version, I reported that in one
>> of my first emails). I tried to different things without
>> regenerating the *.vti *.pvti (using the zipped raw version):
>>
>> 1. Changing the piece extent in all the *.pvti files to what you
>> proposed in your last email -> segfault
>> 2. Changes describe in 1. + changing the whole extension of the
>> VTI files that contain the second piece -> Paraview warns me
>> because the size of the second piece doesn't correspond with the
>> data.
>>
>> vtkXMLImageDataReader (0x419ef70): Cannot read point data array
>> "var" from PointData in piece 0. The data array in the element
>> may be too short.
>>
>> Following your advice, I will try to generate an extra dummy
>> slice in the first piece. I will let you know the results :)
>>
>> Best,
>> Sergi
>>
>>
>>
>> On 08/11/2015 11:57 PM, Cory Quammen wrote:
>>> Actually, you could try one other thing first:
>>>
>>> Change the extents of the pieces in the .pvti files to the
>>> following:
>>>
>>> <Piece Extent="0 200 0 200 0 100" Source="ascii_2_0.vti"/>
>>> <Piece Extent="0 200 0 200 100 200" Source="ascii_2_1.vti"/>
>>>
>>> I'm able to load the files and move between time steps fine
>>> doing this.
>>>
>>> Cory
>>>
>>> On Tue, Aug 11, 2015 at 5:10 PM, Cory Quammen
>>> <cory.quammen@kitware.com <mailto:cory.quammen@kitware.com>> wrote:
>>>
>>> Ideally both pieces would contain a copy of the slice, but I
>>> think for the first piece you could just write some dummy
>>> values into this extra slice. When you load it, the valid
>>> values from the first piece will overwrite the dummy values.
>>>
>>> I'm frankly surprised this extra slice would be needed
>>> because the ghost level is set to 0. But I'll admit I don't
>>> fully understand this format yet.
>>>
>>> Cory
>>>
>>> On Tue, Aug 11, 2015 at 3:37 PM, Sergi Mateo Bellido
>>> <sergi.mateo.bellido@gmail.com
>>> <mailto:sergi.mateo.bellido@gmail.com>> wrote:
>>>
>>> Hi Cory,
>>>
>>> This extra slice is the first slice of the second piece.
>>> Should both pieces contain this slice?
>>>
>>> In my case, each MPI process computes something and
>>> writes its portion of the mesh. All these portions are
>>> completely disjoint among them.
>>>
>>> Thanks!
>>> Sergi
>>>
>>>
>>>
>>>
>>> --
>>> Cory Quammen
>>> R&D Engineer
>>> Kitware, Inc.
>>>
>>>
>>>
>>>
>>> --
>>> Cory Quammen
>>> R&D Engineer
>>> Kitware, Inc.
>>
>>
>>
>>
>> --
>> Cory Quammen
>> R&D Engineer
>> Kitware, Inc.
>
[Attachment #5 (text/html)]
<html>
<head>
<meta content="text/html; charset=utf-8" http-equiv="Content-Type">
</head>
<body text="#000000" bgcolor="#FFFFFF">
<div class="moz-cite-prefix">Hi Cory,<br>
<br>
I'm stuck with this issue :(<br>
<br>
1. Changing just the pvti file is not enough: I got a segfault
using the raw binary files.<br>
2. I tried also adding a dummy slice at the end of all pieces to
fix the problem that I reported in one of my previous emails. So,
instead of changing the PVTI files I tried changing the VTI files.<br>
2.1. First of all, I had to increase the number of tuples of
the VtkArray because I was adding the dummy slice. Since I didn't
change the whole extent and the piece extent of each VTI file, I
got this warning message:<font face="Courier New, Courier,
monospace"><br>
Warning: In /opt/VTK-6.2.0/Common/DataModel/vtkDataSet.cxx,
line 421<br>
vtkImageData (0x3450390): Point array Var with 1 components,
has 4080501 tuples but there are only 4040100 points</font><br>
This didn't work either.<br>
2.2. Afterwards, I tried modifying also the whole extent and
the piece extent. This seemed to work :)<br>
<br>
Despite 2.2 seemed to work, I'm not sure if this is a valid
solution or just a workaround: If I open a VTI file directly
(because I want to see the data produced by a MPI process) the
dummy slice is visualized.<br>
<br>
From my point of view, I think that we're trying to address an
issue of the PVTI format modifying something that works fine (VTI
files).<br>
<br>
Best regards,<br>
<br>
Sergi<br>
<br>
On 08/12/2015 02:46 PM, Sergi Mateo Bellido wrote:<br>
</div>
<blockquote cite="mid:55CB402B.10300@gmail.com" type="cite">
<meta http-equiv="Context-Type" content="text/html; charset=utf-8">
<div class="moz-cite-prefix">Cory,<br>
<br>
I did not change the VTI files in the first experiment but it
crashed anyway.<br>
<br>
Thanks!<br>
Sergi<br>
<br>
On 08/12/2015 01:55 PM, Cory Quammen wrote:<br>
</div>
<blockquote
cite="mid:CAB5Fpx4Z-ZJAigMWtUcE_QnD5AXsO93K07KbFOeAPUk7Hysv5g@mail.gmail.com"
type="cite">
<div dir="ltr">Sergi,
<div><br>
</div>
<div>Note that I did *not* change the whole extents of the VTI
files after I changed the listed extents in the PVTI files.</div>
<div><br>
</div>
<div>Cory</div>
</div>
<div class="gmail_extra"><br>
<div class="gmail_quote">On Wed, Aug 12, 2015 at 1:47 AM,
Sergi Mateo Bellido <span dir="ltr"><<a
moz-do-not-send="true"
href="mailto:sergi.mateo.bellido@gmail.com"
target="_blank">sergi.mateo.bellido@gmail.com</a>></span>
wrote:<br>
<blockquote class="gmail_quote">
<div>
<div>Cory,<br>
<br>
The ascii version is the more stable version and it
usually doesn't crash (even the original version, I
reported that in one of my first emails). I tried to
different things without regenerating the *.vti *.pvti
(using the zipped raw version):<br>
<br>
1. Changing the piece extent in all the *.pvti files
to what you proposed in your last email -> segfault<br>
2. Changes describe in 1. + changing the whole
extension of the VTI files that contain the second
piece -> Paraview warns me because the size of the
second piece doesn't correspond with the data.<br>
<br>
vtkXMLImageDataReader (0x419ef70): Cannot read point
data array "var" from PointData in piece 0. The data
array in the element may be too short.<br>
<br>
Following your advice, I will try to generate an extra
dummy slice in the first piece. I will let you know
the results :)<br>
<br>
Best,<br>
Sergi
<div>
<div class="h5"><br>
<p><br>
</p>
<br>
On 08/11/2015 11:57 PM, Cory Quammen wrote:<br>
</div>
</div>
</div>
<div>
<div class="h5">
<blockquote type="cite">
<div dir="ltr">Actually, you could try one other
thing first:
<div><br>
</div>
<div>Change the extents of the pieces in the
.pvti files to the following:</div>
<div><br>
</div>
<div>
<div> <Piece Extent="0 200 0 200 0 100"
Source="ascii_2_0.vti"/></div>
<div> <Piece Extent="0 200 0 200 100
200" Source="ascii_2_1.vti"/></div>
</div>
<div><br>
</div>
<div>I'm able to load the files and move between
time steps fine doing this.</div>
<div><br>
</div>
<div>Cory</div>
</div>
<div class="gmail_extra"><br>
<div class="gmail_quote">On Tue, Aug 11, 2015 at
5:10 PM, Cory Quammen <span dir="ltr"><<a
moz-do-not-send="true"
href="mailto:cory.quammen@kitware.com"
target="_blank">cory.quammen@kitware.com</a>></span>
wrote:<br>
<blockquote class="gmail_quote">
<div dir="ltr">Ideally both pieces would
contain a copy of the slice, but I think
for the first piece you could just write
some dummy values into this extra slice.
When you load it, the valid values from
the first piece will overwrite the dummy
values.
<div><br>
</div>
<div>I'm frankly surprised this extra
slice would be needed because the ghost
level is set to 0. But I'll admit I
don't fully understand this format yet.</div>
<span>
<div><br>
</div>
<div>Cory</div>
</span></div>
<div class="gmail_extra">
<div>
<div><br>
<div class="gmail_quote">On Tue, Aug
11, 2015 at 3:37 PM, Sergi Mateo
Bellido <span dir="ltr"><<a
moz-do-not-send="true"
href="mailto:sergi.mateo.bellido@gmail.com"
target="_blank">sergi.mateo.bellido@gmail.com</a>></span>
wrote:<br>
<blockquote class="gmail_quote">
<p dir="ltr">Hi Cory,</p>
<p dir="ltr">This extra slice is
the first slice of the second
piece. Should both pieces
contain this slice?</p>
<p dir="ltr">In my case, each MPI
process computes something and
writes its portion of the mesh.
All these portions are
completely disjoint among them.</p>
<p dir="ltr">Thanks!<span><br>
Sergi</span></p>
</blockquote>
</div>
<br>
<br>
<div><br>
</div>
</div>
</div>
<span>-- <br>
<div>Cory Quammen<br>
R&D Engineer<br>
Kitware, Inc.</div>
</span></div>
</blockquote>
</div>
<br>
<br>
<div><br>
</div>
-- <br>
<div>Cory Quammen<br>
R&D Engineer<br>
Kitware, Inc.</div>
</div>
</blockquote>
<br>
</div>
</div>
</div>
</blockquote>
</div>
<br>
<br>
<div><br>
</div>
-- <br>
<div class="gmail_signature">Cory Quammen<br>
R&D Engineer<br>
Kitware, Inc.</div>
</div>
</blockquote>
<br>
</blockquote>
<br>
</body>
</html>
["tests.tar.gz" (application/gzip)]
_______________________________________________
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
[prev in list] [next in list] [prev in thread] [next in thread]
Configure |
About |
News |
Add a list |
Sponsored by KoreLogic