BSOD shortly after cold boot

GeekyDeaks · Feb 11, 2021

I have a peculiar issue. My machine will happily chug along all day flawlessly if I give it about 30 minutes to warm up before launching a sim. If I try to fire up something a little too quickly after power up, I can pretty much guarantee I will be greeted with a BSOD shortly after the VR headset kicks in.

Looking at the MEMORY.DMP, I see the following:

Code:

MEMORY_MANAGEMENT (1a)
    # Any other values for parameter 1 must be individually examined.
Arguments:
Arg1: 0000000000041792, A corrupt PTE has been detected. Parameter 2 contains the address of
    the PTE. Parameters 3/4 contain the low/high parts of the PTE.
Arg2: fffff5813332d010
Arg3: 0000000003000000
Arg4: 0000000000000000

I have taken the memory out and reseated them back in, but it's still the same. Also ran the Memory Diagnostic several times and it passed each one flawlessly.

I'm assuming this is thermal and there is probably a problem with my motherboard, but I used my old trick of blowing cold air on parts to try and identify the location and haven't been able to trigger it.

I would prefer not to randomly buy new parts in the hope it fixes it and actually identify the faulty component. Anyone have any bright ideas?

robt100 · Feb 11, 2021

Is it definitely system memory and not Vram?? Do you have a GPU Overclock? if si might be worth knocking 10% off the memory overclock to see if that helps. Overclocks that are at the bleeding edge of the limit can cause BSOD if tested from cold.

GeekyDeaks · Feb 11, 2021

Ah, unfortunately, I'm not overclocking anything. Good point on the GPU though, I didn't reseat that... I'll give it a go tonight.

blekenbleu · Feb 11, 2021

This site suggests that it may be a driver error and provides tests

GeekyDeaks · Feb 12, 2021

blekenbleu said:
This site suggests that it may be a driver error and provides tests

Activated 'Driver Verifier', shutdown the PC and let it cool for an hour. Fired it up and it reported a violation immediately! Perfect... I now have my evening entertainment sorted

GeekyDeaks · Feb 12, 2021

Ok, something is not right. I seem to be gradually told that every non-microsoft driver is faulty. I now have no network drivers, nor the logitech drivers so my wheel doesn't work fully. I'm now just removing the NVidia drivers...

I think I have a hardware fault... surely all those drivers cannot be faulty?

blekenbleu · Feb 12, 2021

GeekyDeaks said:
I think I have a hardware fault.

Probably time to move onto the second test

GeekyDeaks · Feb 12, 2021

unfortunately, I have already run that one several times. I'll try it again though

blekenbleu · Feb 12, 2021

GeekyDeaks said:
unfortunately, I have already run that one several times

Sadly, it does not abuse RAM in quite the same ways as does Windows.
Could you try running with less RAM?

Neilski · Feb 12, 2021

When you say "cold" boot, and also talk about letting it cool down for an hour, I wondering if you mean that the problem never occurs right after a "warm" boot.
To ask a clear question: let's say you get the BSOD and reboot instantly; can you then reliably trigger it again if you follow the same recipe?

GeekyDeaks · Feb 12, 2021

blekenbleu said:
Could you try running with less RAM?

Oh, do you mean take one stick out? I'll see if the board supports that

Neilski said:
When you say "cold" boot, and also talk about letting it cool down for an hour, I wondering if you mean that the problem never occurs right after a "warm" boot.

That seems to be the case. I cannot trigger it after the machine has been on for a few minutes. I have to really get in quick after the first power on.

Neilski said:
To ask a clear question: let's say you get the BSOD and reboot instantly; can you then reliably trigger it again if you follow the same recipe?

Not tried that. I'll give it a go tomorrow on first power up. Normally I spend 5-10mins running the memory dump through windbg, so I have not normally got it to trigger a second time after powering the machine on

blekenbleu · Feb 12, 2021

GeekyDeaks said:
Oh, do you mean take one stick out? I'll see if the board supports that

Most do, although running slower.
My thinking is that if one stick is flakey, then this could identify it.

GeekyDeaks · Feb 13, 2021

Fingers crossed, but it might be resolved.

Took your advice and pulled one memory stick out. Switched on and fired up ED, no BSOD. Let it cool for an hour and swapped the memory over. Switched on and fired up ED, no BSOD. Let it cool once more and put both memory back (but now in reversed slots to before), fired up ED and again, no BSOD. Left it to cool overnight and fired it up again this morning and still could not trigger the BSOD.

I don't remember when I reseated them before if I swapped them over, but I cannot be arsed to swap them over to check again

Thanks for the support!

blekenbleu · Feb 13, 2021

blekenbleu said:
Sadly, it does not abuse RAM in quite the same ways as does Windows.

Researching a bit more, the next time memory issues are suspected,
AIDA64 may be better at flushing them

RasmusP · Feb 15, 2021

GeekyDeaks said:
Fingers crossed, but it might be resolved.

Took your advice and pulled one memory stick out. Switched on and fired up ED, no BSOD. Let it cool for an hour and swapped the memory over. Switched on and fired up ED, no BSOD. Let it cool once more and put both memory back (but now in reversed slots to before), fired up ED and again, no BSOD. Left it to cool overnight and fired it up again this morning and still could not trigger the BSOD.

I don't remember when I reseated them before if I swapped them over, but I cannot be arsed to swap them over to check again

Thanks for the support!

Seems like it's sorted, nice!

My 2 cents after the initial post would be that your GPU might boost too high.
I had my GPUs causing driver crashes/crash to desktop when I put the heaters off over night.
At 20°c my 1070 likes to boost above 2100 MHz, which wasn't stable at all... When the room was at my standard 22°c and my PC would have run a bit in idle, the GPU would be around 40°c.
At 40° it would only boost to around 2060 MHz, still too high... During gaming it would actually run at 2000-2020 MHz, totally stable though! But at 70-80°c!

That's why my GPU OC curve looks like this now: