Question PC Locked up, could move cursor but nothing else

[iWillz]

So, I built a new pc last week, I have a 6900XT and a 5800x and I put an undervolt from 1175mV to 1110mV on the gpu 3 days ago, it was running perfectly fine under heavy gaming loads until today where I was just watching youtube and suddenly the pc just locks up, everything goes to 1fps and wouldn't let me press shut down so I had to do it directly from the power button on the pc. Do you guys think this is probably an issue with the gpu undervolt I had done or is there other places I should be looking at this problem? After I restarted the pc it does seem the gpu settings manually made by me were reverted to stock settings, cpu didn't reset so just want to know your guys thoughts on this.

 

[Aeacus]

I put an undervolt from 1175mV to 1110mV on the gpu

Why?

If you don't need all the power 6900XT offers, then why not buy weaker GPU, e.g 6800XT. You could've saved some money.

Do you guys think this is probably an issue with the gpu undervolt I had done or is there other places I should be looking at this problem? After I restarted the pc it does seem the gpu settings manually made by me were reverted to stock settings

When you starve the GPU, and then put a heavy load on it, which you did, then you can consider yourself lucky that GPU didn't bite the dust on you and instead reverted the undervolt.

 

[iWillz]

Why?

If you don't need all the power 6900XT offers, then why not buy weaker GPU, e.g 6800XT. You could've saved some money.

I did it in an attempt to reduce thermals on the gpu which it did, from what I've read the 6900XT gets fed too much voltage for what it outputs which is why they run really hot (Not my expert opinion, just what I've read on a few forums) Also in my post I said that it ran fine under max load with the undervolt for a few days, thermals and everything were completely fine it just happened randomly under light load which seemed really odd.

When you starve the GPU, and then put a heavy load on it, which you did, then you can consider yourself lucky that GPU didn't bite the dust on you and instead reverted the undervolt.

I've actually read quite the opposite for undervolting a gpu, everything that i've seen about the topic is that it is completely safe to undervolt them, I read that you should change it in increments of 10 and if you still notice artifacting or crashes then you change it back 10 until it stabilises. Am I wrong in all of this and the internet are all dumb?

 

[Aeacus]

GPU undervolt, as such, is only meant to be used, when stock clocks, mostly due to the age of GPU, aren't stable anymore. So, to keep using GPU, undervolt can remedy the situation. But it is a short term fix, only used until dying GPU can be replaced.

But to undervolt it, just to reduce thermals, is stupid.

Better increase the case fan RPMs if you need better cooling for your GPU. Or get better/more case fans. Also, depending on which 6900XT you got, defines how good of a cooler it has. For example, GIGABYTE AORUS Radeon RX 6900 XT Master 16G has the best air cooler on it, out of all 6900XT GPUs out there,
source: https://tech4gamers.com/best-rx-6900-xt/

Or like i said above, rather than getting this power hungry GPU, with can have high-ish thermals (especially if the cooling in your PC case is inadequate), better option would've been going with less power hungry and and better thermals GPU.

Am I wrong in all of this and the internet are all dumb?

I'll say this: when you undervolt your GPU, and your GPU happens to die due to this, then go after all those who said that undervolt was safe and ask the money for new GPU from them. (Since warranty will not cover GPU dying due to your undervolt.) Also, any technical support that you need regarding issues from GPU undervolt, you can ask from those who say it is safe to do so. Deal?

 

[iWillz]

GPU undervolt, as such, is only meant to be used, when stock clocks, mostly due to the age of GPU, aren't stable anymore. So, to keep using GPU, undervolt can remedy the situation. But it is a short term fix, only used until dying GPU can be replaced.

But to undervolt it, just to reduce thermals, is stupid.

Better increase the case fan RPMs if you need better cooling for your GPU. Or get better/more case fans. Also, depending on which 6900XT you got, defines how good of a cooler it has. For example, GIGABYTE AORUS Radeon RX 6900 XT Master 16G has the best air cooler on it, out of all 6900XT GPUs out there,
source: https://tech4gamers.com/best-rx-6900-xt/

Or like i said above, rather than getting this power hungry GPU, with can have high-ish thermals (especially if the cooling in your PC case is inadequate), better option would've been going with less power hungry and and better thermals GPU.



I'll say this: when you undervolt your GPU, and your GPU happens to die due to this, then go after all those who said that undervolt was safe and ask the money for new GPU from them. (Since warranty will not cover GPU dying due to your undervolt.) Also, any technical support that you need regarding issues from GPU undervolt, you can ask from those who say it is safe to do so. Deal?

Thanks a lot for the replies, it is appreciated and my comments weren't meant to discredit your knowledge on undervolting, it was just things I had researched prior to actually doing it, it wasn't as if the thermals were bad enough that I couldn't run it I just thought that 90-100c was a little high, taking into account what you've suggested I won't try to mess with it anymore I'll just run it at stock from now on however besides from that do you think that the issue that my pc practically felt like it was running at 1 fps was due to the undervolt I had on it or do you think it would have stemmed from elsewhere because I was under the impression that the pc doesn't really utilize much gpu resources unless you're gaming, keep in mind the issue happened when I had only 2 chrome tabs open with a youtube video playing and that when I pressed the power off button on windows it did nothing so I had to press the physical one on the case.

 

[Aeacus]

do you think that the issue that my pc practically felt like it was running at 1 fps was due to the undervolt I had on it or do you think it would have stemmed from elsewhere

Hard to say.

Undervolting can make GPU unreliable and that unreliable can kick out at any point. Also, the fact that GPU automatically reverted your undervolt back to stock, has indication towards GPU freezing up and in turn, freezing your whole system as well, whereby only manual restart helped.

I just thought that 90-100c was a little high

This points towards inadequate cooling within your PC, rather than GPU issue in itself. Since when you don't have proper, if any, airflow in your PC case, GPU will heat up the air inside the PC and without hot air having a way out, with cold air having a way in, it will lead to such high temps and GPU thermal throttle as well.

Reference 6900XT (AMD Radeon), in a properly cooled PC case, hovers around 75C at full load,
review: https://www.tomshardware.com/reviews/amd-radeon-rx-6900-xt-review/4

While aftermarket GPUs have better power delivery and cooling than the reference GPU. So, when you have aftermarket 6900XT and you're seeing such high temps, then issue is with inadequate cooling within your PC.

 

[iWillz]

While aftermarket GPUs have better power delivery and cooling than the reference GPU. So, when you have aftermarket 6900XT and you're seeing such high temps, then issue is with inadequate cooling within your PC.

That would make sense however it confuses me because I built in a corsair 5000d airflow with 6 intake be quiet silent wings 3 fans, 3 exhaust mounted on the 360mm rad at the top and 1 more silent wings 3 exhaust on the rear all running at 1000rpm and ramps up depending on the system temperature sensor. I don't think my system temps have ever exceeded 45 degrees, it seems to be only the GPU temps that are high, for reference I played red dead redemption 2 at max settings 1440p today for 3 hours and the gpu maxed out at 87 degrees celcius after making a custom fan curve for it so it does seem a little bit better. My 5800x also maxes out at around 70 degrees, that is fitted with a corsair h150i elite capellix.

This is what my HWMonitor looks like

 

[Rogue Leader]

Its important to note the two temperatures you have there "Global" and "Hotspot".

Global is your real temperature, and keeping that in the 70-80c range tops is to your benefit.

Hotspot is the reading of the hottest temp sensor on your GPU, that you want to keep under about 95c.

Your temps are excellent based on what you have posted here. I wouldn't worry about them. If you do want to experiment with undervolting, I suggest getting the paid version of 3DMark and stress testing your system to be sure its safe. Adrenalin also has its own built in "undervolt" setting. TBH this automatic mode does it as well as you can and will likely work fine. Instead of you sitting there and bumping voltage down 10mV and testing over and over wasting a day. Honestly with those temps you could probably slow your fan curve a bit if you want to keep the system a little quieter, obviously test this to see how it performs.

I disagree with what Aeacus said, undervolting is a legitimate technique to get sometimes lower temps and power consumption at the sacrifice of little to no performance, but it does require that your GPU itself casn handle it. Some can and some can't, not every chip is the same. Also undervolting is not going to break anything. Thats not to say that its good to keep crashing your system over and over again, or run the GPU with artifacting and other issues long term, but unless you do something really dumb (drop voltage very low, and clocks very high and somehow get it very hot, and even then) you're not going to break anything.

 

[Aeacus]

I disagree with what Aeacus said, undervolting is a legitimate technique to get sometimes lower temps and power consumption at the sacrifice of little to no performance, but it does require that your GPU itself casn handle it. Some can and some can't, not every chip is the same. Also undervolting is not going to break anything. Thats not to say that its good to keep crashing your system over and over again, or run the GPU with artifacting and other issues long term, but unless you do something really dumb (drop voltage very low, and clocks very high and somehow get it very hot, and even then) you're not going to break anything.

While this can be true, GPU undervolting is more in the ballpark of power user, rather than standard user. There are few specific reasons why to undervolt GPU, but most reasons why people undervolt GPU, are either to do with lowering thermals or power consumption. Both can be fixed with far better options. E.g thermal issue is usually due to inadequate cooling within PC. And power consumption can be remedied either by not buying such power hungry GPU, or buying more efficient PSU.

So, unless knowing exactly what you're doing and why, GPU undervolt, CPU downclock, RAM timings loosening and the like, are better to be reserved as one of the last options of what to do, given that one has tried to remedy the issue by other means beforehand.

 

[Rogue Leader]

While this can be true, GPU undervolting is more in the ballpark of power user, rather than standard user. There are few specific reasons why to undervolt GPU, but most reasons why people undervolt GPU, are either to do with lowering thermals or power consumption. Both can be fixed with far better options. E.g thermal issue is usually due to inadequate cooling within PC. And power consumption can be remedied either by not buying such power hungry GPU, or buying more efficient PSU.

So, unless knowing exactly what you're doing and why, GPU undervolt, CPU downclock, RAM timings loosening and the like, are better to be reserved as one of the last options of what to do, given that one has tried to remedy the issue by other means beforehand.

Any undervolting, overclocking, etc is a power user move yes, one that should be approached methodically with testing. It is likely that he didn't have a problem to begin with so he didn't need to undervolt. His crash I still believe is unrelated to the undervolting. It was at the desktop watching a video, not under gaming. Your solution of not buying such a hungry GPU or just replacing the PSU are throwing money at a problem (if it even is) when you don't need to in all cases as there are tools out there to remedy, if you're willing to take the time to learn and do it right. And again AMD literally bakes in automatic undervolting, overclocking, etc into the drivers that has been thoroughly tested and found to work.

The issue here is he undervolted and assumed it caused his problem. He probably didn't need to try that to begin with, and his problem is probably something else entirely.

@iWillz can you look through event viewer for Critical issues and Errors that happened around the time of your crash? From there we may be able to determine a culprit. You will see Kernel Power 43 in there, which is the act of you cutting power to the system, so you can ignore that one.

 

[iWillz]

@iWillz can you look through event viewer for Critical issues and Errors that happened around the time of your crash? From there we may be able to determine a culprit. You will see Kernel Power 43 in there, which is the act of you cutting power to the system, so you can ignore that one.

This was the only critial error at the time it stopped responding, basically the pc was still active, I could still go into file explorer and start menu however it wouldn't do anything when I pressed shut down from the start menu, was only able to shut it down from the pc tower. Also around the time of pc restart it is logged in the application logs things to do with my avast antivirus which is the 2nd image.

 

[Rogue Leader]

Ok now I am 100% sure your undervolting had nothing to do with it. That said at this point its not undervolted so thats no longer a concern.

I am thinking there may have been something going on with Avast Antivirus, but it is hard to tell.

What I would like you to do is go to Windows / Run / Command prompt and right click and run as Administrator. At the command prompt run this:

DISM.exe /Online /Cleanup-image /Restorehealth

It may take like 10 min or so to complete. Once that is complete run this:

sfc /scannow

After that just use the PC normally and see if it happens again. If it does, try uninstalling Avast and using the built in Windows AV and see if it happens again. This is one of those things that may take some time to diagnose. Report back when you've done this.

 

[iWillz]

Ok now I am 100% sure your undervolting had nothing to do with it. That said at this point its not undervolted so thats no longer a concern.

I am thinking there may have been something going on with Avast Antivirus, but it is hard to tell.

What I would like you to do is go to Windows / Run / Command prompt and right click and run as Administrator. At the command prompt run this:

DISM.exe /Online /Cleanup-image /Restorehealth

It may take like 10 min or so to complete. Once that is complete run this:

sfc /scannow

After that just use the PC normally and see if it happens again. If it does, try uninstalling Avast and using the built in Windows AV and see if it happens again. This is one of those things that may take some time to diagnose. Report back when you've done this.

Thanks a lot for the responses, I've ran the command, didn't take long but I'll leave you with the other 2 avast errors that occured at the same time, will report back if the issue happens again.