PC crashing but stays on - Dying GPU or something else?

[TGO]

A few weeks ago, I was watching a video and suddenly my screen froze for a split second before just going dark with "No display detected", but my keyboard and PC were still on. Nothing happened for a short while even after hitting the power button, so I was forced to do a hard shutdown. I shrugged it off as a grid spike after an hour of investigating and as my brother had the exact same thing happen to his PC in the same area. But 2 days ago, I had 5-6 crashes in one day.

 

It started off when I was away on an errand and I locked my PC. Came back to the monitor on and the keyboard off, which I thought was odd, but as soon as I logged in I noticed that it had rebooted with a "unsafe shutdown" entry in event viewer.. And then a few hours later when I locked it, I once again came back to it restarted with only that error present. It didn't break lose until right after a gaming session I had with a friend, it kept crashing over and over just after ~2 minutes in Windows doing nothing! Before I decided to just kill it completely for the night, I checked event viewer and the reliability page which were littered with new errors. Notable ones:

• LiveKernelEvent 141
• nvlddmkm (Event 14) - I assume this is the Nvidia driver?
• "Desktop Window Manager stopped working"
• And then of course a dozen errors regarding "unsafe shutdown"

All this has me suspecting that the GPU (GTX 1080Ti) is on its way out, but is it possible that another component here is failing here? PSU, motherboard etc. 

The reason I'm wondering is because why else would the entire system be (seemingly) rendered frozen, not responding to the safe shut down and rebooting from the GPU failing? A good few of the crashes logged literally nothing but "Unsafe shut down" in Windows with nothing graphics related around it.

 

Relevant details to complement the rig on my profile:

First crash was 2 days after installing Nvidia's 531.68 driver, but today I did a clean install of 531.79

Latest version of Windows 10 22H2

Latest BIOS for my motherboard

[CommanderAlex]

Have you checked the temperatures on the graphics card? Now how old is your hardware just so we can better help you and specifications of the system?

[Agall]

39 minutes ago, TGO said:

A few weeks ago, I was watching a video and suddenly my screen froze for a split second before just going dark with "No display detected", but my keyboard and PC were still on. Nothing happened for a short while even after hitting the power button, so I was forced to do a hard shutdown. I shrugged it off as a grid spike after an hour of investigating and as my brother had the exact same thing happen to his PC in the same area. But 2 days ago, I had 5-6 crashes in one day.

 

It started off when I was away on an errand and I locked my PC. Came back to the monitor on and the keyboard off, which I thought was odd, but as soon as I logged in I noticed that it had rebooted with a "unsafe shutdown" entry in event viewer.. And then a few hours later when I locked it, I once again came back to it restarted with only that error present. It didn't break lose until right after a gaming session I had with a friend, it kept crashing over and over just after ~2 minutes in Windows doing nothing! Before I decided to just kill it completely for the night, I checked event viewer and the reliability page which were littered with new errors. Notable ones:

• LiveKernelEvent 141
• nvlddmkm (Event 14) - I assume this is the Nvidia driver?
• "Desktop Window Manager stopped working"
• And then of course a dozen errors regarding "unsafe shutdown"

All this has me suspecting that the GPU (GTX 1080Ti) is on its way out, but is it possible that another component here is failing here? PSU, motherboard etc. 

The reason I'm wondering is because why else would the entire system be (seemingly) rendered frozen, not responding to the safe shut down and rebooting from the GPU failing? A good few of the crashes logged literally nothing but "Unsafe shut down" in Windows with nothing graphics related around it.

 

Relevant details to complement the rig on my profile:

First crash was 2 days after installing Nvidia's 531.68 driver, but today I did a clean install of 531.79

Latest version of Windows 10 22H2

Latest BIOS for my motherboard

Try using the shortcut win+ctrl+shift+b which'll restart the video driver next time, see if that resolves it. Most likely a driver issue if so.

[TGO]

12 hours ago, CommanderAlex said:

Have you checked the temperatures on the graphics card? Now how old is your hardware just so we can better help you and specifications of the system?

It has been running quite hot for a few months now. Spiked at 92c before I realized how bad it was and adjusted the fan curve, can still spike at 90 though but with better clocks. But none of the crashes have occurred during intense gaming, I played Hitman 3 a short while before the first crash, and the one I played before it went really bad wasn’t demanding and only sent it to 60-65c. I bought and built it in June 2017.

 

12 hours ago, Agall said:

Try using the shortcut win+ctrl+shift+b which'll restart the video driver next time, see if that resolves it. Most likely a driver issue if so.

I forgot to add that I did try this during one of the crashes where the video I was watching kept on playing as the screen went out, but nothing happened and the audio cut out after a few seconds…

[CommanderAlex]

18 minutes ago, TGO said:

It has been running quite hot for a few months now. Spiked at 92c before I realized how bad it was and adjusted the fan curve, can still spike at 90 though but with better clocks. But none of the crashes have occurred during intense gaming, I played Hitman 3 a short while before the first crash, and the one I played before it went really bad wasn’t demanding and only sent it to 60-65c. I bought and built it in June 2017.

It's definitely time to repaste the graphics card as GPU manufacturers generally don't use the highest quality of thermal paste and only last a couple years. Something like MX-4 or Thermal Grizzly Kryonaut would work best. 

[TGO]

6 hours ago, CommanderAlex said:

It's definitely time to repaste the graphics card as GPU manufacturers generally don't use the highest quality of thermal paste and only last a couple years. Something like MX-4 or Thermal Grizzly Kryonaut would work best. 

Definitely, but I’ve only got ~2 weeks until my new rig arrives. But from the looks of things it seems like the damage is already done… It lived for 6 years, just had 2-3 more weeks left of service and then this happens.

[CommanderAlex]

2 hours ago, TGO said:

Definitely, but I’ve only got ~2 weeks until my new rig arrives. But from the looks of things it seems like the damage is already done… It lived for 6 years, just had 2-3 more weeks left of service and then this happens.

Oh that sucks. Typically though the thermal paste doesn't really last long enough to warrant 6 years of use. 

[TGO]

On 5/4/2023 at 10:45 PM, CommanderAlex said:

Oh that sucks. Typically though the thermal paste doesn't really last long enough to warrant 6 years of use. 

Even in such a high end card for its time? I guess every card is different (or Asus just takes shortcuts), but a friend of mine’s GTX 1060 is sitting at <40c idle temps despite being used even more than mine..

 

Back on the topic, do you think it’s only the GPU that’s causing the crashing? I enabled the iGPU and using that instead going forward. Still got the 1080Ti plugged in though, not sure if that’s a bad idea? 

[CommanderAlex]

10 minutes ago, TGO said:

Even in such a high end card for its time? I guess every card is different (or Asus just takes shortcuts), but a friend of mine’s GTX 1060 is sitting at <40c idle temps despite being used even more than mine..

 

Back on the topic, do you think it’s only the GPU that’s causing the crashing? I enabled the iGPU and using that instead going forward. Still got the 1080Ti plugged in though, not sure if that’s a bad idea? 

Yeah manufacturers will use the cheapest quality thermal paste just to last as long as the warranty. A GTX 1060 does produce as much heat as a 1080 Ti will since there are less transistors packed in and runs cooler as a result. 

 

With the GPU temperatures the way they are, I still do suspect the graphics card to be the problem. If you can try it out in another system and see if the problem persists still,then you'll know that that is the problem. 

[emothxughts]

On 5/4/2023 at 5:16 AM, TGO said:

A few weeks ago, I was watching a video and suddenly my screen froze for a split second before just going dark with "No display detected", but my keyboard and PC were still on. Nothing happened for a short while even after hitting the power button, so I was forced to do a hard shutdown. I shrugged it off as a grid spike after an hour of investigating and as my brother had the exact same thing happen to his PC in the same area. But 2 days ago, I had 5-6 crashes in one day.

I had a similar black screen situation with my late RX580 card. Like you, the only error I see in the Event Viewer afterward was Kernel Event 41, which simply means the PC wasn't shut down properly (that's what happens when you force restart the PC). I know mine is AMD not NVIDIA, but I knew for a fact that the problem is with the card and not the driver, as my emergency RX550 worked more or less perfectly in my PC, and the RX580 also crashed like that in a computer shop's test PC. Long story short, I deemed the card dead and sold it off as faulty for parts.

 

BTW, GPU-Z reports that my card ran at x8 PCIe lanes instead of the x16 that it's supposed to be. Not to be confused with the PCIe gen, the thing that goes from 1.1 to 3.0 or 4.0 when you run the PCIe render test. And yes, the RX580 card blackscreened when I start the test. Something thus must have been really borked in the card. Try running GPU-Z, see how many lanes does it report for your 1080ti, see if it also crashes on the render test.

[TGO]

2 hours ago, CommanderAlex said:

Yeah manufacturers will use the cheapest quality thermal paste just to last as long as the warranty. A GTX 1060 does produce as much heat as a 1080 Ti will since there are less transistors packed in and runs cooler as a result. 

 

With the GPU temperatures the way they are, I still do suspect the graphics card to be the problem. If you can try it out in another system and see if the problem persists still,then you'll know that that is the problem. 

In a way I understand why, but on the other it's such a small difference in margins on the higher end cards.

 

Either way it seems like something in the GPU gave up due to heat or age, and probably can't be saved considering that it crashed while just sitting in the desktop with "normal" temps. It's a total bummer since it was probably the most valuable component in my system if I'd sell it.. But right now I'm more concerned if it'll survive through transferring data to my new rig and then wiping it.

 

2 hours ago, emothxughts said:

I had a similar black screen situation with my late RX580 card. Like you, the only error I see in the Event Viewer afterward was Kernel Event 41, which simply means the PC wasn't shut down properly (that's what happens when you force restart the PC). I know mine is AMD not NVIDIA, but I knew for a fact that the problem is with the card and not the driver, as my emergency RX550 worked more or less perfectly in my PC, and the RX580 also crashed like that in a computer shop's test PC. Long story short, I deemed the card dead and sold it off as faulty for parts.

 

BTW, GPU-Z reports that my card ran at x8 PCIe lanes instead of the x16 that it's supposed to be. Not to be confused with the PCIe gen, the thing that goes from 1.1 to 3.0 or 4.0 when you run the PCIe render test. And yes, the RX580 card blackscreened when I start the test. Something thus must have been really borked in the card. Try running GPU-Z, see how many lanes does it report for your 1080ti, see if it also crashes on the render test.

I didn't see anything abnormal in there last time I checked, but I'll see what it says when I get back to it! If it was like that in your case, I suppose I should drop all my hope that it's a driver issue.. I just hope my new GPU will fare better in the future, especially with these prices.