Random system crash - BSOD

[Nosfe]

Hello all! i've been trying to find the cause of thise problem and its driving me crazy, about 2 months ago my PC started to crash, at first it only happened once a week. About 1 week ago ive been having like 3-4 crashes per day, thise only happens when im playing games and GPU goes full load, the screen freezes, afters 5 seconds the mouse cursor freezes, the PC restart by it self and start up without image, and when i take a look to the GPU fans, they are not working, but the leds from the GPU does. I've been using HWInfo64 to monitoring the hardware but i really dont understand the log, so i was hoping that someone could help me.
I've tried update the GPU drivers to latest, and older ones, a clean windows installation (v1909 and v2004), undervolting the GPU, power limit from 10% to 50% and nothing worked.
My PC Hardware is (all relatively new, except the GPU):
-Ryzen 7 2700x
-Asus B450M-A Prime
-Crucial 16GB DDR4 3200Mhz Ballistix Sport LT
-Shappire RX480 Nitro+ 8GB (3 years olds+)
-Thermaltake 700w 80plus Smart

Attached the logs from when the crashes happens. I would really appreciate your help.

 

[Dalai]

Unfortunately such issues are often hard to diagnose, especially remotely. I see no obvious issue(s) from your logs. Temperatures look all fine, the fans are working, including the GPU fans (at least according to the logs) and voltages look OK as well. XMP was disabled but now it's enabled, and if the issue persists in both states, it apparently makes no difference.

Do the BSODs occur in all games or is it limited to specific ones? If it happens in any game, it's a more generic issue, maybe pointing towards the GPU. Maybe BlueScreenView can help you out what kind of driver was active at the time of the BSOD, but unfortunately this program's output is not always reliable.

Regards
Dalai

 

[Nosfe]

Hi, thanks for your help! the crash happens only when im playing Path of Exile and the GPU its on load, right now im testing Borderlands 3, and it hasn't crashed yet, then im gonna try and test with DOOM. I run BlueScreenView but it doesn't found any dump file.
Today i've tried changing the memory speed from 3200 from 2400 and the PSU start doing a wierd noice, so i put it back to 3200. Now it stoped, but that was the first time the PSU did that.

 

[Zach]

You can try to look on Windows “Event Viewer” what may cause the crashes. I’m not on my PC now but I think the path is “Windows Logs” —> “Application” or the other entries too. Windows is keeping track of events/errors for a lot days back.

 

[Nosfe]

So i've tried testing another games, and so far no crashes. I've played about 1 hour of Borderlands 3 and no crash, and about 5-6 hours of Doom and no crash, and no fps spikes. Borderlands was on medium graphics and Doom on ultra.
Event viewer from windows have a lot of error, i'll attached a screen of that, and the logs from HWinfo from when i was testing the those 2 games.
Tomorrow will continue testing Borderlands and Doom, and try another time with Path of Exile.

 

[Zach]

You can also check on Event Viewer, at Custom Views, Administrative Events, a summary of warnings and errors. Try to find those on a time of a crash.

 

[Nosfe]

Today i tried again Path of Exile, but thise time only standing in the same place using the same skill over and over again for about 1-2 min, then i stoped using skills and notice that the fans from the GPU didn't slow down untill the crash happens (about 20-30 seconds). I have the logs from HWinfo and the events from Windows, the crash is the one that happens at 3:28:55PM.
And then i tried Doom, played about 3 hours and didn't crash.

 

[Dalai]

Your most recent logs also don't show anything out of the ordinary. It's likely that HWiNFO doesn't catch the moment of the crash, so the logs don't contain information about it. Even if HWiNFO was able to read hardware information right before of the crash, it's very likely that the system doesn't write them to disk.

Both Borderlands and Doom use the Vulkan API, but Path of Exile does not (it's probably DirectX/Direct3D). Do you have any other game that doesn't use Vulkan? What do the SMART values of the drives look like? You can also use CrystalDiskInfo for that (I recommend the ZIP archive).

What I don't understand is that only the latest log contains the names and models of the drives. Why is this information missing from all other logs? You'd think that the available sensors would be consistent. Did you enable/disable monitoring/logging any of them?

Regards
Dalai

 

[Nosfe]

So i dont have any other game that render with DirectX/Direct3D, but im downloading Crysis Remastered to test right now, ill post what happen with the game later.

I have to say something more about PoE. It has Directx11 and Vulkan, i tried DirectX11 and the performance was terrible, as soon as the GPU starts to get load the FPS drops badly, so i have been running Vulkan and wasn't having the same problems with FPS, they drop a little but practicaly nothing like with DirectX11.

Tried CrystalDiskInfo and the disk are good, have the screens for that.
I have disable S.M.A.R.T because each time i power the PC an error about my disk "D" on POST didn't let me boot windows. But that was long ago, and wasn't having thise issue before.
Another thing i forgot to mention, rarely the screen flickers, the image goes black and 1 sec later comes again. Right now doesn't happens long ago.

 

[Nosfe]

So today i tested Crysis Remastered and it crashed (no windows event log). Tried DX11 and force DX9 and both crashed. After that tried Doom again with Vulkan and stay stable. So the problem might be, as you told me, something with DirectX.

Now the question is, there is something to try?

 

[Dalai]

No, I didn't say that it's an issue with DirectX/Direct3D. But apparently the crashes are far more likely to occur when using it instead of Vulkan. Since you already reinstalled Windows and tried different drivers, I don't think it's a driver issue. So, the only idea I have is to check for loose plugs on all connections and to verify that the GPU is fully inserted into the slot - maybe get the card out and reinsert it just to be sure. If everything is OK on that front, the only thing left then is to start swapping components one by one, PSU first, then the GPU. And test after swapping each component.

Sorry for not being more helpful, but as I said in the beginning such issues are (very) hard to diagnose.

Regards
Dalai

 

[Nosfe]

Ok so i'll check the hardware then, i can borrow some parts from my brothers PC.
Today i get another crash from Crysis, but thise time the PC didn't restart and event logs returned an error.

 

[Nosfe]

So i think the problem it's gone, after replacing GPU thermal paste (Arctic MX4) and VRM thermal pads (Arctic) the reboot stoped.
I wanted to thank you all for your patience and your help.
Now the temperatures are 30c min, 40-50c idle, and 60-69c on load, from 40-50c min and 70-80c on load and the FPS spikes are gone too.