Paraview fails to open *vtu solution on Windows 10 laptop

Hello,

I am currently using Paraview 5.8.1 on a Windows 10 laptop. The computer is a Dell Precision 3551, which has the following specifications:

  • Windows 10 Pro 64 bits
  • 32Gb RAM (Intel Core i7, 2 x 16Gb RAM)
  • Nvidia Quadro P620 (4 Go GDDR5 SDRAM)

The solution I want to open is a *vtu file from CONVERGE software.
It has a size of approx 5Gb, and counts approx 5millions cells.
When I click File>Open, and tick the field I want to see to see, Paraview starts to load the data before crashing at approx 40% of progess bar (I just tick one field at a time).

I have no idea about the problem, please help!

Best regards

Maxime

Precision: I could perfectly open this solution from my former company supercomputing server, meaning that the solution is probably not bugged (no NaN or things like that…)
Unfortunately, I cannot access it anymore :frowning:

Can you monitor your memory while charging the dataset ?

I can see memory usage through Memory Inspector which tells me that Paraview uses 3% of total memory when I open my field.
How can I monitor the memory?

Maxime

Use the Windows Task manager accessible through Ctrl-Alt-Del

Sorry for the misunderstanding.

The Windows Taks Manager prompts that Paraview is using about 15% of processor memory until it crashes…

Maxime

You mention being able to open your dataset on another computer before, was it the same version of ParaView ?

I do not recall perfectly the exact version, but I’m pretty sure it was on a Paraview v5*.
Also, I saw on Paraview Download page that some MPI versions are available. My version (which causes crashes on laptop) is not an MPI one.
What is the difference between an MPI and a non MPI version?

Maxime

the MPI version support a parallel server, but afaik, you are not using the pvserver, so this is not the issue here.

I’d suggest that you try with a few other versions of ParaView to see if this issue is related to your version of ParaView or your setup.

I just tried 3 other versions (v5.8.0, v5.7.0, v5.4.0), but the problem remains…
My setup thus seems the cause.

Maxime

It will be hard to help you in this case. In your situation I’d build paraview in debug and run with a debugger to understand what is going on.

Re:
I launched Paraview from a Linux emulator command line (MobaXterm).
It allowed me to have a feedback from paraview when loading my data, which is the following: Segmentation fault

How to use debugging tools on Paraview?

Maxime

You need to build ParaView yourself and use a debugging tool. Not a trivial task.
https://gitlab.kitware.com/paraview/paraview/blob/master/Documentation/dev/build.md#windows

Seems a bit tricky and time consuming indeed…
I will try to find another alternative

Thank you

Maxime

Not sure if it is going to help, but also make sure ParaView is using your high performance graphics card (NVIDIA) rather than the on board graphics card.
https://www.top-password.com/blog/enable-high-performance-gpu-for-windows-10-apps-or-games/

1 Like

Hi Dan,
Unfortunately, it did not solve the problem.
After several tests, it apprears that I encounter this problem only with large files…
Smaller files are openable with no problem.
Is it possible to limit the number of digits Paraview reads from the solution, so the memory needed decreases?
Best regards
Maxime

If your data has several attribute arrays you could read fewer to save memory. That is the control there is to save memory when reading a data file. If you have a very large data reading it on a HPC machine is the solution (which results in data being chunked, and individual chunks being read on each node)