troubleshooting computer freezing with screen artifacts

[Fungusworld8000]

Hello hello,

 

I have a problem with my ~6 yr old gaming PC that I built. I switched to Debian Linux a few months ago and overall I've really enjoyed the operating system, but it has made my PC act up in ways that have me at my wit's end.

 

Roughly every half hour on average, my PC randomly freezes with either audio looping, very loud buzzing, or no sound at all (mostly depending on whether or not there was sound at the moment of the freeze) with the whole screen getting jumbled in these very rapidly flickering visual artifacts and the system becoming fully unresponsive (see picture)

 

Now this does seem like a hardware problem, not least because software-wise, I have essentially the exact same setup running on my laptop with no problems. My suspicion this whole time was that either my graphics card is busted, or that there is a driver problem, but through installing some other distros with preinstalled drivers (and experiencing the exact same crashes, sometimes during install) I can almost exclude the drivers being at fault, and through reinstalling windows to see if it would provoke the same crashes, it seems like the Hardware is stable as well (I played some games on windows and it didnt crash in the 2-3 hours I tested it). I also at some point unplugged my GPU and did a fresh install with only using CPU Graphics and the crashes looked about the same, although they were notably less frequent. This problem had me stumped, until I remembered reading that Linux tends to draw more power than Windows. I did some research and read on this forum, that my PSU should be avoided, even stating in the tier list that its use is "dangerous under several circumstances", although Im not sure I understand why that's the case.

This seems like the last convincing cause for this problem that I could look into fixing, especially since the crashes seem to correlate with tasks that you'd expect to be more power hungry, like launching steam, playing games, screen-sharing stuff on discord, etc... and it being mostly fine when doing more minor things like just watching something or listening to music.

Although I am a little scared of spending ~200-300$ on a good PSU only to see the problem unchanged.

 

Now I come here in humble search of an answer to three questions:

 

Is there any scenario in which I don't necessarily have to replace my current PSU asap?

 

What are the odds, that the PSU is directly or indirectly causing these crashes?

 

What else could be causing this?

 

OS: Debian GNU/Linux 12.5

Mainboard: ASUS Strix Z270F Gaming

CPU: i7 7700k

GPU: GTX 1060
RAM: Corsair Vengeance LPX, 16 GB, 2 sticks

PSU: Corsair VS650

Storage: 1x 256GB SSD, 2x HDD with 1 TB and 4 TB respectively

 

 

Thank you very much for your time. I would appreciate some help as I'm out of ideas, and do not want to go back to Windows.

[Levent]

This is the textbook definition of VRAM failure. Your PSU has nothing to do with it, its your graphics card.

[Fungusworld8000]

1 minute ago, Levent said:

This is the textbook definition of VRAM failure. Your PSU has nothing to do with it, its your graphics card.

 

But how is that possible if the exact same crashes occur when i fully remove the graphics card + fresh install? Is it an onboard graphics problem of the CPU?

[Levent]

2 minutes ago, Fungusworld8000 said:

 

But how is that possible if the exact same crashes occur when i fully remove the graphics card + fresh install? Is it an onboard graphics problem of the CPU?

At that point RAM would be my suspect. Honestly speaking I doubt there is just one thing wrong here.

[Fungusworld8000]

Just now, Levent said:

At that point RAM would be my suspect. Honestly speaking I doubt there is just one thing wrong here.

I ran every test I could using a memtest USB and it found 0 problems with CPU or RAM. Is it possible that it missed something? I do have some fresh RAM on hand that I could test when I get around to it. Currently my main suspects through excluding everything else are the motherboard or the PSU, although if it was a motherboard problem You'd expect the exact same instability when using windows, no?

[Fungusworld8000]

update: I ran an OCCT power supply stress test on my windows install. At around the 3 minute mark, it begins to detect hundreds of thousands of "errors" per second and keeps counting up, but the test keeps going and the PC doesn't freeze. I left it running for about 6 minutes and it ends up with around 50 million errors. I assume this means I should get a replacement anyway but I find it super odd that the freezing only happens on linux. Is it possible that Windows is just more optimized to handle voltage inconsistencies than linux? I don't think that makes a whole lot of sense and I'm even more confused now than before.

[Levent]

25 minutes ago, Fungusworld8000 said:

update: I ran an OCCT power supply stress test on my windows install. At around the 3 minute mark, it begins to detect hundreds of thousands of "errors" per second and keeps counting up, but the test keeps going and the PC doesn't freeze. I left it running for about 6 minutes and it ends up with around 50 million errors. I assume this means I should get a replacement anyway but I find it super odd that the freezing only happens on linux. Is it possible that Windows is just more optimized to handle voltage inconsistencies than linux? I don't think that makes a whole lot of sense and I'm even more confused now than before.

PSU instability will result in computer shutdowns not errors. Your have problems somewhere else, either combined heat from CPU and GPU caused your RAMs to be unstable or your VRMs are unstable.

[Fungusworld8000]

I've had the thought that it may be overheating related before, and decided to display cpu & gpu core temperatures at all times on my second screen, although this was a moment ago. If i remember correctly it didn't seem like there was any correlation between the temperatures and the freezing. The computer could at times get relatively hot without freezing and vice versa. Granted this was not a very precise way of measuring for this as I was essentially just waiting and didn't constantly check temps. I usually couldn't read them anymore as soon as the system froze, so maybe there were some spikes that i didn't pick up on early enough to notice a correlation.

I also ran furmark (on linux) several times to see if it would make my pc crash and it failed to do so even after having it running for around half an hour and if I remember correctly it did make my GPU get quite hot (around 80°).

I don't think any of this was very reliable testing though. How would I go about testing that hypothesis in a reliable way, let alone fixing the problem if that's the case?

Thank you for your help by the way!

[Fungusworld8000]

On 7/9/2024 at 11:52 AM, Levent said:

PSU instability will result in computer shutdowns not errors. Your have problems somewhere else, either combined heat from CPU and GPU caused your RAMs to be unstable or your VRMs are unstable.

 

I retested it by having a temp monitor of my cpu cores on my second screen while I was watching stuff on the primary monitor. When the crashes occured, from what I could see all cores were consistently around 40°-50° C, which i think is a very normal temperature. To me that suggests that there can hardly be such levels of heat in there as to cause instability. I think unstable GPU VRMs causing this problem is not very likely, seeing that the freezing looks the same after fully removing my GPU. That leaves CPU VRMs, and I happen to know next to nothing about VRMs. Is there any way I can check whether these are faulty other than straight up replacing the whole motherboard? I don't know whether this makes a difference but I've never really overclocked my PC. Since I suspected some power shenanigans to be going on I set my power consumption to the most low power one a while ago in the UEFI of my ASUS mainboard, to no avail, it didn't even change the frequency of the freezing.

[Mark Kaine]

just from error description this is typical ram... try a different set / only 1 stick etc. also check if there's bios/chipset driver updates. 

[Fungusworld8000]

I've actually figured out what the problem is and fixed it; or at least it seems like I did, I haven't had any crashes at all and the system has been stable and running smoothly since I did this.

The solution was surprisingly easy, I just hadn't thought of it. I found it in a comment on this reddit post. It turns out that old ASUS mainboards have a bug in the BIOS that conflicts with the Linux kernel when it tries to switch between power states and causes this kind of freezing.

 

I simply went into UEFI, selected "tools" and then did an update over the network.

Thanks everybody for your help!