That is odd, how just that one game is triggering such a fault, due to such a BIOS bug. I had thought something in it was hitting a driver fault (and a failed gpu reset is always fatal, or soon to be… I’ve had cards/drivers that did that, even in Windows).
And you can reproduce it too. That’s fortunate. I’m glad you discovered it.
But weird things can happen. Years ago I bought an AMD R9 380 card. At the time I had to use the proprietary driver (“fglrx”) on Linux, but I did most of my gaming on Windows back then (and Wine was an abomination to me back then, it wasn’t until Proton that I started to change my mind). I had no end of trouble with that card in Windows, due to “clever” power management, dynamic clocking and stuff. I had a coincidental utility that just happened to completely override that and overclock to the specified profile. The “MSI Gaming App” of all things.
That settled down with catalyst drivers and amdgpu was viable for it on Linux, but a few years later I was having odd lockups (full, hard power off kind). It was happening in Linux, and Windows. One place it was never happening was while gaming. It was always doing desktop stuff, click a menu etc. It would happen a few times a day most days. Never reproducible (it wasn’t the things I was doing that were causing it).
Know what it was? Gkrellm polling sensors 10 times a second. I had everything on there, voltages, fans, temperatures (it87 chip), core temps, memory, disk i/o, and shit I don’t remember. I don’t know if it was the GPU sensor or what (the it87 motherboard driver in kernel was known to be starting to rot) or ultimately a BIOS bug. In Windows, I was running a free sensor app probably using similar code (lmsensors project etc.) and loadable driver and the same problem was happening there with similar circumstances (never while gaming etc.) The common denominator. It took me a long time to connect those dots.
I stopped using sensor displays, to this day. If I want to have a peek, I just type “sensors”.