Hi, debugging random crashes on an unknown remote system is hard, e.g. I had to debug a crash that was present only on a certain CPU model recently.
Typically when you submitted the issue, it gets in the bug database, it just has a "cannot reproduce" status. If you find a way how to reproduce the crash it might get solved faster.
It's usually best to send crash logs, houdini info.txt (from the menu), hip file, steps to reproduce (if you have an obvious one), any additional info that might be relevant.
E.g. figuring out the steps that result in a crash helps a lot even if they cannot reproduce it on their system. Record a video of these steps and attach it to the report.
A combination of your preferences might have an effect too (e.g. what desktop is on, UI scale, etc.).
Windows could log an event for the crash, check Event Log.
If you have PDB files (symbols) installed for that build (be sure you downloaded/installed pdb for matching build) your crash log might have additional information that is useful for devs. The easiest is to install pdb files from Houdini Launcher.
You can configure Windows to produce crash dump files (e.g.
https://www.meziantou.net/tip-automatically-create-a-crash-dump-file-on-error.htm [
www.meziantou.net] )
These have memory state/stack trace that might help to narrow the issue.
If the Houdini hangs on the crash, you can save the dump file from Task Manager (Right-click on the process -> Create Dump File).
If you have Visual Studio installed it will try to debug the crash (and you can save the dump file, but it might be too late)