[flotus1]

While trying to load a rather large Ensight case file in Paraview, it crashes writing the following message to console:

Qt has caught an exception thrown from an event handler. Throwing
exceptions from an event handler is not supported in Qt. You must
reimplement QApplication::notify() and catch all exceptions there.

terminate called after throwing an instance of 'std::bad_alloc'
what(): std::bad_alloc
Aborted

The case has a fortran binary format with around 350million hexa8 elements with 2 scalar results. Duplicate vertices were avoided to reduce file size and enhance rendering speed in ParaView.
The geometry file uses around 14.4 Gb of disk space, each of the scalar files has around 1.3 Gb size on disk.
Before clicking "apply" I removed the tick box for one of the two scalars to further reduce the memory requirement.

Smaller case files can be loaded without problems.
The workstation has 256 Gb of memory available. Watching the output of top I found out that paraview allocates about 3% of the memory after clicking apply and then crashes. ParaView version is 4.3.1 64 Bit.

Is there anything I can do about this issue?

[flotus1]

Seems like we are running out of ParaView experts.

[wyldckat]

Greetings Alex,

At least for me, the problem is not having enough free time to do it all

My guess is that 350 million hexahedral cells will result in roughly 6 times the number of points, which is 2100 million points. This is certainly overkill, because this would assume that each cell has its own 6 points, which should not happen as you've indicated.
If I'm not reading the number incorrectly, 2^31 = 2147483648 which is very near to the maximum limit of 2100 million points.

Edit: Forgot to mention that some 64-bit code sometimes doesn't fully support 64-bit arrays

Beyond this, I would need a test case and would have to investigate the source code... for which I don't have enough free time in the next few months Therefore, I suggest you ask about this on the ParaView user mailing list, where some of the ParaView developers usually answer to a few questions: http://www.paraview.org/mailing-lists/

Nonetheless, if you would like for anyone to look into this, you might want to provide a test case example for both here and on the mailing list.

Best regards,
Bruno

[flotus1]

I really appreciate that you took the time. I know we all have other things to do.

Your suggestion about the number of points led me towards the right direction.
The number of "potential" points (I prefer calling them vertices) seems to be the problem.
A geometry file with 266'607'582 hexa8 cells is no problem for ParaView (8*266'607'582 < 2^31).
A geometry file with 268'522'448 hexa8 cells causes ParaView to crash (8*268'522'448 > 2^31).
This is independent of the actual number of vertices in the geometry file which was only 272'819'068 in the second case.

[wyldckat]

Hi Alex,

I can't remember exactly right now, but I believe that the Ensight file formats can handle multi-block meshes. In other words, instead of having a single mesh, have one mesh per block.
This to say that it would then be possible to at least divide the mesh in half and have each half for a block.

After a quick search:

• A similar problem you've stumbled upon is also registered here: http://www.paraview.org/pipermail/pa...ry/006670.html - but an example case doesn't seem to have been provided for testing and debugging, hence it's unlikely the bug has been fixed.
• Near the end of the ParaView FAQ: http://www.paraview.org/Wiki/ParaVie...iew_support.3F - look for "EnSight Master Server files", which were designed for parallel processing.

Best regards,
Bruno

[joncotgreave]

I have found this the case running a smaller file size on a low spec machine.
In my case it was due to running out of memory.
Have you turned on the "Memory Inspector" to allow visulisation of you resource usage.
How much RAM does you computer have?