I first noticed this trying to play Baldur’s Gate 3. The screen would go black, I would hear a small “zap” noise coming from the PC tower, and then the PC would reboot.
The card and machine are bought used off of eBay, and I certainly recognize they may be of questionable quality. However, I would like to know with as much certainty and precision what the issue is.
You see I’m on a little journey of exploration trying to learn more about my system and debugging ever deeper more complex issues.
I’ve managed to create a minimal reproduction by running the Unigine Superposition benchmark. (UNIGINE Benchmarks)
I’m going to rerun it to see if I can get the precise scene at which the PC crashes.
I’ll try some of the previous benchmarks to see if they also cause the PC to crash. I also have an Radeon card coming in a few days so I can see if that fixes the issue.
Looking forward to collaborating!
nvidia-bug-report-clean.log.gz (341.1 KB)
I’ll keep track of test result here:
- Run 1: crashed around scene 10-11
- Run 2: got all the way through.
- Medium 1080P Preset
- FPS: Min 83, Avg 132.26, Max 175.25
- GPU Temp C: min 50, max 83
- Max Utilization 99%
- Run 3: success
- High 1080P Preset
- FPS: min 76.10, avg 99.39, max 132.7
- GPU Temp C: min 54.0, max 83.0
- GPU Utilization: max 99%
- Run 1: crashed at scene 1, PC sang a weird beep song, didn’t restart itself, had to poweroff. Tried powering on, sang the weird beep song again, did not boot. Waited a few minutes, tried again, worked.
- Run 2: success
- Profile: Basic
- FPS: min 37, max 249.6
- Run 1: totally fine, crushes it. This is an old benchmark from 2009?
- It seems like maybe the GPU needs to be “warmed up”? After going through several benchmarks I can’t get it to fail.
- The tests seem to be running windowed now? They were failing when they were fullscreen I think? Not sure why they swapped to being windowed.
Spontaneous reboots/shutdowns are only initiated by the mainboard as a safety measure due to overheating or lack of power. Please monitor cpu temperatures, if those are normal, your psu is likely broken.
Interesting… good info.
I’ll have to check the CPU to see how its paste is. I’ve been using mangohud which hasn’t really reported temps that were too high.
But I’ll pay more attention to that. Thanks for your help!