[DESKTOP] Need assistance with diagnosing intermittent suspected GPU Issue

[DrexLock]

I have a Kubuntu 20.04 system (originally 18.04) that overall am very happy with. However, I have a nagging intermittent issues that I am not sure how to properly troubleshoot and diagnose. Occasionally, when playing games (specifically CIV5 and City Skylines) I encounter a crash that I suspect is hardware related. The entire screen will freeze (except for the mouse) and I am unable to interact with anything. Audio continues to play normally for a while until eventually that begins to stammer and the GPU fans kick into high gear. The only way I have found to get back to a usable machine is to hard power off my tower by holding down the power button.

This crash can happen early on or after hours of play. Sometimes I can go for weeks without it happening, sometimes it happens multiple times in a day. It's been driving me nuts and I don't know whats the best way to diagnose the issue that I'm having and how to go about correcting it. This install has also had all the hardware except for the GPU changed out overtime, however this issue did occur before making a motherboard/cpu and ram change.

Any guidance on how to diagnose an issue like this would be appreciated.

[DrexLock]

Just rebooted from this same crash again. This time playing Minecraft. Any guidance on how to find the appropriate system logs to start troubleshooting would be appreciated.

[Snowhog]

System specs (your PC) would be helpful.

[claydoh]

Any guidance on how to find the appropriate system logs to start troubleshooting would be appreciated.

KSystemLog is a good tool for accessing different logs, and has search functions.
Also, running and checking a system monitor of some sort while gaming might also give clues -- ram, swap, cpu usage, GPU temps. There are desktop/panel widgets for these as well as using the standard KDE system monitor, as well as terminal ones such as top and htop
Memory tests, either from your bios, if it has such a tool, or the memtest option in your grub menu.

try ctrl-alt-F(2,3,4,, etc) to see if you can get to a TTY, at least to try rebooting, or running htop, and trying to 'kill' a rogue application or process.

And, yeah, some sort of system specs might be helpful

[DrexLock]

Sorry about that, my pc is an HP Envy 795, Core i7-9700, 32gb crucial ram, 1tb Samsung nvme ssd, Sapphire Flex HD7950. The original system this install was in was a HP Envy 1520T, i7-3770, 32 gb crucial ddr3, and 1tb Crucial sata ssd with the same Sapphire Flex HD 7950. I ran into this problem on both builds, the only hardware that carried over between the two is the GPU.

I tried Ksystemlog before but couldn't seem to find anything, I'll have to look again. I was using a mini DisplayPort to DVI cable to connect each monitor but swapped them to regular DVI for both last night. Not sure if a bad cable/port can create issues like this but figured its worth a shot. I remember having an issue with one of the mini DisplayPort ports getting finicky but it seemed to correct itself. Maybe this issue is how that problem is presenting itself now.

[Snowhog]

Did you have this issue when you were running 18.04?

[DrexLock]

Yes I believe so, it's been a while but I feel like it did. Part of me was hoping that once I can upgrade to 22.04 it'll correct the issue

[oshunluvr]

If it is indeed hardware, there will likely not be anything in any logs, at least not in my experience. It could be heat related. Since it's a desktop, you might try opening the case and putting an external fan pointed at it, then play your games and see if it still does it. You could also try the integrated graphics to verify it the video card and not something else.

[Jonas]

If nothing else, only to exclude the possibilities; you say that it only occurs when playing games? Civ and City skylines are native Linux compatible and "should" run well. Do you have any mods/addons when playing? I know these have caused me issues like freezing.
Does it happen when you run other GPU/CPU consuming tasks, videos or other 3D applications?

Possible to run steam (if that's your platform) through a terminal/konsole and get the readouts.

[DrexLock]

Thank you everyone for the guidance. The problem with this crash is its so hard to pin down, I've been unable to reliably recreate it. After some additional troubleshooting I believe that it is hardware related, specifically one of the mini DisplayPort ports on the GPU. I swapped back to DVI and had CIV up for a few hours today without issue. Weirdly, my tower regained the ability to sleep after so long of inactivity, something that randomly stopped working a while back. I'll continue to monitor but if it does occur again I'll update the thread with what I find.