Random BSODs while gaming

[Pinossaur]

Hello, this is going to be a long one, so brace yourselves
So it all started a couple weeks ago, when my pc randomly BSOD in the middle of my gaming session, specifically it was a whea_uncorrectable_error BSOD, which proceeded to happen a couple more times during the day. The next day, I decided to try and reseat both the CPU, GPU and RAM, which also did not help, tried to clear all overclocks, in case one of them was unstable, also did not work, and tried to underclock and undervolt the CPU, because I thought it could be that my CPU had degraded a lot somehow, but again, it didn't work, also, running the BIOS at stock settings just made the PC not even being able to boot to windows, always crashing at the welcome screen, which didn't happen before. Also alslo, reinstalling windows also didn't help.

For context, my CPU is a ryzen 5 3600x, which I daily drove for about 8 months, at 4.5GHz, with 1.3V, keeping it at 60ºC while stressing it, my GPU is a ASUS TUF GTX 1660, with 180MHz Clock and 800MHz Memory OC, and my RAM is a random no-brand stick of 16GB, rated at 2666MHz CL 18, which ran at 2800MHz CL15.

After all of the steps done previously, the problem persisted, until it stopped like 2 days later, allowing me to re-enable all my overclocks, without any instability problems, but it all started happening again today, and I wanted to know if any of you have gone through this, or have any tips that might help.

I have also tried stress testing the GPU and CPU separately, the GPU was always fine, and the CPU crashed almost always, but it didn't BSOD windows, or even stopped, AIDA64 just threw an error saying it was not stable. This happened at both stock speeds, underclocked and undervolted speeds, and with my overclock.

Again, thanks for any tips that might come through, and sorry if i'm doing something wrong or posting on the wrong place, as this is my first time posting on this forum. 

[LateLesley]

Random BSODs can nearly always be tied down to either unstable overclock, faulty memory, faulty cpu, or one part that most folk miss, faulty power. 

 

So with that in mind, which PSU are you using?

 

Set everything stock, get a copy of memtest86 (https://www.memtest86.com/) and test the memory through 3 passes at least. It'll take some time. 

 

Thats the first test I would do, if it fails, try it with just one stick at a time. If it still fails, i'd try different memory, and if I got it stable, i'd try clearing and updating the BIOS. I'd probably try a different PSU if it was still unstable. It's not easy if you don't have parts to hand to test with though. 

[Bombastinator]

Most common hardware causes of BSOD after boot are memory and PSU issues.  That is not to say this is automatically hardware though the AIDA thing implies it’s not something high level.  Memtests take a long time but at least they don’t cost money.

[Pinossaur]

4 minutes ago, LateLesley said:

Random BSODs can nearly always be tied down to either unstable overclock, faulty memory, faulty cpu, or one part that most folk miss, faulty power. 

 

So with that in mind, which PSU are you using?

 

Set everything stock, get a copy of memtest86 (https://www.memtest86.com/) and test the memory through 3 passes at least. It'll take some time. 

 

Thats the first test I would do, if it fails, try it with just one stick at a time. If it still fails, i'd try different memory, and if I got it stable, i'd try clearing and updating the BIOS. I'd probably try a different PSU if it was still unstable. It's not easy if you don't have parts to hand to test with though. 

Right now, if I set everything to stock, I can't even get to windows, the RAM I know it works fine, because my 16GB stick works properly on other computers, and I have 2 4GB dimms, which i'm not currently using, because I don't want to have mismatched RAM on my system, but it happens with any RAM, the PSU is a seasonic 500W, which might be the cumplit, but I don't think it would cause problems that only happens sometimes, if the PSU is not working properly, it should not work properly after X wattage, or it should not be working for long, another idea I had would be the SSD, but as it is like 2 months old, I didn't think it was the problem, also it is a kioxia exceria plus 500GB, but something that I did that is currently helping is dropping the GPU, Chipset and M.2 PCI speeds to gen2, which made so the pc wouldn't crash right away, but i'm still testing it, and might not be the problem, it would also not make that much sense, but who knows.

I'll update this thread after a few days (or whenever the problem comes back) to tell if dropping PCIe speed was indeed the fix, so that anyone else having this problem can also try it.  

[LateLesley]

A Seasonic PSU should be decent enough, as long as it's not too old. 

 

The problems with PSU isn't always just an overload of wattage, over time the capacitors in the PSU dry out, they tend to be rated for around 2000 hours or so, maybe 5000 or 10,000 hours if they are really good caps. Anyway, as they wear, they loose capacitance, and that can cause increased ripple on the supply lines, and dips as the CPU/GPU/Motherboard change from low to high load. IF you had an oscilloscope, even one of those very cheap ebay kit oscilloscopes, it would be enough to see if there is ripple and dips on the power lines. It's not always the problem, and Seasonic make pretty good PSUs, so i'm tempted to suspect it less, but it is an issue that can happen, especially as PSUs age 3-7 years old. 

 

You could probably eliminate the SSD as being the problem by running a live linux, or a diagnostic disc/USB like Hiren's Boot CD. (https://www.hirensbootcd.org) That would at least take that out of the equation. 

[Pinossaur]

8 minutes ago, LateLesley said:

A Seasonic PSU should be decent enough, as long as it's not too old. 

 

The problems with PSU isn't always just an overload of wattage, over time the capacitors in the PSU dry out, they tend to be rated for around 2000 hours or so, maybe 5000 or 10,000 hours if they are really good caps. Anyway, as they wear, they loose capacitance, and that can cause increased ripple on the supply lines, and dips as the CPU/GPU/Motherboard change from low to high load. IF you had an oscilloscope, even one of those very cheap ebay kit oscilloscopes, it would be enough to see if there is ripple and dips on the power lines. It's not always the problem, and Seasonic make pretty good PSUs, so i'm tempted to suspect it less, but it is an issue that can happen, especially as PSUs age 3-7 years old. 

 

You could probably eliminate the SSD as being the problem by running a live linux, or a diagnostic disc/USB like Hiren's Boot CD. (https://www.hirensbootcd.org) That would at least take that out of the equation. 

Well, this PSU is about 6 years old, so yeah, it could happen, i've dissassembled it once, to clean up, so it's not clogged up with dust, but yeah, it might be, i'll test the ssd later, but for now, it seems dropping pcie speed worked

[Bombastinator]

1 hour ago, Pinossaur said:

Right now, if I set everything to stock, I can't even get to windows, the RAM I know it works fine, because my 16GB stick works properly on other computers, and I have 2 4GB dimms, which i'm not currently using, because I don't want to have mismatched RAM on my system, but it happens with any RAM, the PSU is a seasonic 500W, which might be the cumplit, but I don't think it would cause problems that only happens sometimes, if the PSU is not working properly, it should not work properly after X wattage, or it should not be working for long, another idea I had would be the SSD, but as it is like 2 months old, I didn't think it was the problem, also it is a kioxia exceria plus 500GB, but something that I did that is currently helping is dropping the GPU, Chipset and M.2 PCI speeds to gen2, which made so the pc wouldn't crash right away, but i'm still testing it, and might not be the problem, it would also not make that much sense, but who knows.

I'll update this thread after a few days (or whenever the problem comes back) to tell if dropping PCIe speed was indeed the fix, so that anyone else having this problem can also try it.  

If you’ve swapped out the ram it’s not a ram problem.  3600+1660 shouldn’t stress a 500w either.  So it seems to be something low level which may be but probably isn’t hardware.  This won’t boot at default at all but will boot with problems with an overclock thing is not normal.  I’ve never heard of anything like that.

[Pinossaur]

Yep, I just tested everything I could, memtest runs perfectly, on any settings, and linux crashed as windows did, but only when a load is applied, no matter what load, either on the cpu, or the gpu, but whenever I drop the PCIe speed of the PCIE16x, the M.2 and the Chipset to gen 2, the crashing stops, I don't have enough time today to test what change is the one that is causing it, but the best bet is my ssd trying to run at gen 4, while only being able to run at gen 3, and somehow doing this, which again, makes absolutely no sense, given the symptoms, like, how could dropping pcie speed of any of those devices help when either the cpu or the gpu are under load. If it was only the cpu, maybe running at higher speed made the cpu unstable somehow, if it was only the gpu, maybe dropping pcie16x speed could help, but whatever, for now, the problem is fixed, and for anyone else that ever has this problem, try and turn down pcie gen

[Pinossaur]

Okay, a day later, without any changes, and the BSOD is happening again, but a bit diferently this time, the crash happened twice, both while playing F1 2018, at random times, where the game would freeze, my mouse would be okay, until like 3 seconds later, where my mouse would lock up, until eventually BSOD, followed by a system reboot, even before it ends collecting information, which makes impossible to get any relevant data from Event Viewer, once again, if anyone has any other ideas, they would be very welcome.

[Pinossaur]

5 hours ago, Pinossaur said:

Okay, a day later, without any changes, and the BSOD is happening again, but a bit diferently this time, the crash happened twice, both while playing F1 2018, at random times, where the game would freeze, my mouse would be okay, until like 3 seconds later, where my mouse would lock up, until eventually BSOD, followed by a system reboot, even before it ends collecting information, which makes impossible to get any relevant data from Event Viewer, once again, if anyone has any other ideas, they would be very welcome.

Yeah, at this point, it almost definetely has to be the PSU, after all things tested, with another disk, on linux and windows, with both CPU and GPU safely undervolted and underclocked, and with all devices at pcie gen 2, the crashes still occur.

[Bombastinator]

2 hours ago, Pinossaur said:

Yeah, at this point, it almost definetely has to be the PSU, after all things tested, with another disk, on linux and windows, with both CPU and GPU safely undervolted and underclocked, and with all devices at pcie gen 2, the crashes still occur.

There at his ultra weird “won’t work stick but will work ONLY withOC thing going on though.  I can see the PSU being a possible issue because without being able to turn the OC off the electrical draw basically cannot be effectively calculated.  The only way to even look is to measure total draw at the wall with a kill-a-watt or something. A measurement rather than a calculation.  At this point I’m not sure that there is a single problem. 

[Pinossaur]

12 hours ago, Bombastinator said:

There at his ultra weird “won’t work stick but will work ONLY withOC thing going on though.  I can see the PSU being a possible issue because without being able to turn the OC off the electrical draw basically cannot be effectively calculated.  The only way to even look is to measure total draw at the wall with a kill-a-watt or something. A measurement rather than a calculation.  At this point I’m not sure that there is a single problem. 

Well, gpuz tells me the gpu is drawing about 80W and the cpu about 60, with my undervolted settings, so the power supply not being enough is out of the question, the thing is if it being a 5 year old PSU going to make any difference on it's reliability, because the gpu is only a year old, and works fine on another pc, the cpu is 8 months old, but can't test it outside the motherboard, because other systems I have are intel, same with the motherboard, which is also 8 months old, the SSD is like 2 months old, but it also works fine on another system, and the RAM also knows works fine, so yeah, it's either something with the motherboard/cpu, or the power supply.