Windows 10 and random BSODs

elaola · 2 Nov 2020 at 15:08

This is my hardware:

CPU: i7-7700k
GPU: GTX 1080ti
RAM: Corsair Vengeance LPX Black DDR4 3200 mhz (32GB)
MOBO: MSI Z270a-PRO
System Disk: Samsung 970 EVO 1TB NTFS

The problems started when I got my Samsung 970 EVO 1TB.

The message in the BSOD said:

stop code: DPC_Watchdog_Violation.

So I tried with updating the BIOS, SSD and the latest NVIDIA Drivers. After that it did not happened as frequently, but it did happen about 3-4 times every week. Then I got fed up with the BSOD and deleted the system partition and reinstalled Windows 10 completely. The BSOD disappeared and I really thought it was gone for good. But unfortunately as soon I touched the overclocking in the bios the BSOD came back. First I just tried with X.M.P. profile on which had worked before and that gave me the BSOD maybe once every 2-3 days.

The message in the BSOD said:

stop code: WHEA_UNCORRECTABLE_ERROR.

So I read on forums and got the advice to set down the RAM frequency. I tried very low, from default (2133) to (2400). It worked for another 2 days and when I was sorting pictures today (on another HDD, not systemdisk) the PC crashed (BSOD) with the same message as previous.

I made a memory test with TechPowerUp Memtest64 and no errors were found. I thought it also could be heating problems on my m.2 SSD, but turns out it has pretty stable temps around 40-44c degrees.

Now I have completely disabled the overclocking in BIOS and I really hope there is no more BSOD, but it doesn't feel good to run RAM that is supposed to run @ 3200, but is running @2133. Especially now when I've just bought my NOCTUA NH-D15 cooler. If no more BSOD occurs now, does this mean that I can never overclock? Is my CPU faulty?

This is what I have tried to do:

• Downloaded the latest Samsung Magician software.
• Ran SFC/Scannow and the DISM command check/restore/repair.
• Upgraded the BIOS.
• Updated the SSD and Graphic drivers
• Ran a memtest
• Checked temperatures

pp111 · 2 Nov 2020 at 15:20

Thing is with RAM that is is guaranteed to run at XMP but your motherboard is not. So selecting the XMP profile should, and usually does work, but it is not guaranteed. I am afraid it just looks like yours will not overclock at all.

elaola · 2 Nov 2020 at 16:56

pp111 said:
Thing is with RAM that is is guaranteed to run at XMP but your motherboard is not. So selecting the XMP profile should, and usually does work, but it is not guaranteed. I am afraid it just looks like yours will not overclock at all.

I hear you. So is there anything I should try out before buying a different motherboard? Can I be sure it is the motherboards fault?

SimonC · 3 Nov 2020 at 11:17

If I were you I would try and recreate the bsod in a controlled way, doing a stress test etc so its repeatable.

As the issues started when you fitted the new ssd logically its best to start from there, get it to blue screen, remove the sdd and try and blue screen it again.

You can run a diagnostics scan on the drive using magician see what that reports, maybe the new drive is faulty.

pc-guy · 3 Nov 2020 at 13:34

memtest64 is not a reliable test as it is run under windows environment. the trial version only able to allocate 2GB ram where the paid version can do the full lot i believe. bsically any memory windows & programmes have allocated, these cannot be test.

you are better off download memtest86 https://www.memtest86.com/download.htm. and run it off USB stick as bootable drive.

probably with RAM is that they are guaranteed to work with JEDEC and XMP profile anything in between is a bit of gamble.

https://answers.microsoft.com/en-us...iolation/100cb707-0716-40d4-88a8-7bf9dd8f836d
https://www.channelpro.co.uk/advice/9911/how-to-fix-wheauncorrectableerror-in-windows-10

take it you have been to the above links for some resolutions?

elaola · 3 Nov 2020 at 15:23

SimonC said:
If I were you I would try and recreate the bsod in a controlled way, doing a stress test etc so its repeatable.

As the issues started when you fitted the new ssd logically its best to start from there, get it to blue screen, remove the sdd and try and blue screen it again.

You can run a diagnostics scan on the drive using magician see what that reports, maybe the new drive is faulty.

I will try to do that. Thanks! Magician found no errors on the SSD or any other harddrive I have.

pc-guy said:
memtest64 is not a reliable test as it is run under windows environment. the trial version only able to allocate 2GB ram where the paid version can do the full lot i believe. bsically any memory windows & programmes have allocated, these cannot be test.

you are better off download memtest86 https://www.memtest86.com/download.htm. and run it off USB stick as bootable drive.

probably with RAM is that they are guaranteed to work with JEDEC and XMP profile anything in between is a bit of gamble.

https://answers.microsoft.com/en-us...iolation/100cb707-0716-40d4-88a8-7bf9dd8f836d
https://www.channelpro.co.uk/advice/9911/how-to-fix-wheauncorrectableerror-in-windows-10

take it you have been to the above links for some resolutions?

Alright. Thank you very much for clarifying that! I will download memtest86 and run it today. I heard that this can take a while so I am better off doing this on the night?

I have actually visited those links you wrote down and many others concerning this problem. But I did not get any clue to why I cannot get it to overclock, but I will do this memtest86 and see if there is a problem with my ram.

One thing worth mentioning, I have had no bluescreens since I turned off the X.M.P. profile. I am running at defaults and I have not turned off my PC either just to see how long it can stay so without any BSOD.

pc-guy · 3 Nov 2020 at 18:40

Memtest86 can take a long time. Usually default is 4 passes which is a good set cos sometimes a single pass may not pick up the error.

it does sound like a ram issue with respect to XMP setting. I suspect your board is capable of running 3200MhZ. Once you go pass the default and doing manual OC you need to make sure your Vram is set at 1.35V.

elaola · 5 Nov 2020 at 13:40

pc-guy said:
Memtest86 can take a long time. Usually default is 4 passes which is a good set cos sometimes a single pass may not pick up the error.

it does sound like a ram issue with respect to XMP setting. I suspect your board is capable of running 3200MhZ. Once you go pass the default and doing manual OC you need to make sure your Vram is set at 1.35V.

Ok! I understand. I have atm tested 3 of the 4 ram sticks and there is 0 errors after 4 passes. Gonna test the last ram stick tonight. Maybe it can also be a bad slot on the motherboard?

pc-guy · 5 Nov 2020 at 14:13

ya once you got each stick passed test out each dimm slots. if those are still ok then maybe your board cant handle it.

what PSU u using just so that you dont end up with some weird power behaviour due to dirty or unstable power

also in bios, try disable all the c-state and sprectrum spread and overvolt protection etc. after you tested out the dimms and see if you still end up wiht problems.

also checked SMART for the samsung 970? if you got a spare SSD to transfer your system and data across then do that and boot off that SSD and do a zero fill format on 970 and then install W10 on that 970 and copy over your data from the mirrored SSD.

you could have installed W10 on a potentially unstable system therefore there will be errors in windows files which can be the cause of issues.

liltimmy · 5 Nov 2020 at 21:50

Was your X.M.P profile running fine before you installed the SSD? Do you have any other disk drives?

PiKe · 5 Nov 2020 at 22:00

Try running the ram at be 3000mhz and see if the errors go away.

I've had issues with XMP which aren't revealed by memtest86

elaola · 6 Nov 2020 at 11:23

pc-guy said:
ya once you got each stick passed test out each dimm slots. if those are still ok then maybe your board cant handle it.

what PSU u using just so that you dont end up with some weird power behaviour due to dirty or unstable power

also in bios, try disable all the c-state and sprectrum spread and overvolt protection etc. after you tested out the dimms and see if you still end up wiht problems.

also checked SMART for the samsung 970? if you got a spare SSD to transfer your system and data across then do that and boot off that SSD and do a zero fill format on 970 and then install W10 on that 970 and copy over your data from the mirrored SSD.

you could have installed W10 on a potentially unstable system therefore there will be errors in windows files which can be the cause of issues.

All 4 ram sticks are now tested and fine. I have tested 3 out of 4 dimm slots on the motherboard and they are also fine. Going to test the last one today.

My PSU is an EVGA Supernova 850w G2 80 plus gold certified.

I have got a spare SSD and I could transfer my system over to that one. What do you mean by zero fill format? Is that a regular format or something else? Before I installed W10 I did a regular format.

liltimmy said:
Was your X.M.P profile running fine before you installed the SSD? Do you have any other disk drives?

To be honest. I am not sure if I did run the X.M.P profile before the M.2 SSD, but I did get many bluescreens. I have 2 other SSDs in my system, 3 others are regular 3,5 drives.

PiKe said:
Try running the ram at be 3000mhz and see if the errors go away.

I've had issues with XMP which aren't revealed by memtest86

OK, I will try that when I have checked the last dimm slot on the mobo.

liltimmy · 6 Nov 2020 at 16:09

elaola said:
To be honest. I am not sure if I did run the X.M.P profile before the M.2 SSD, but I did get many bluescreens. I have 2 other SSDs in my system, 3 others are regular 3,5 drives.

I feel like if you didnt get bluescreens before installing the SSD, and you now do then you should test with the SSD uninstalled to see if the BSODs persist. If so, try to return as it may be faulty.

elaola · 7 Nov 2020 at 12:55

liltimmy said:
I feel like if you didnt get bluescreens before installing the SSD, and you now do then you should test with the SSD uninstalled to see if the BSODs persist. If so, try to return as it may be faulty.

I will do that later.

All dimm slots have been tested with 0 errors and I also tried running test with one stick @ 3200 mhz with 0 errors. My i7 7700k will become very hot though now when playing games (90 degrees).

I have not installed my new cooler yet (NH-D15) cause I did not have any iso propyl alcohol to remove the old thermal paste with. There is no shops here that sell it physical so I have ordered it on the net.

Meanwhile I am going to try running now at 3200 mhz to see if I get any bluescreens anymore. If I do I will switch out my system disk for another SSD to install W10 on. It would be weird if it just runs now without errors. I have not done anything other than testing memorys with memtest86.

Edit: But I have also never tried to run games with only one stick as I do now. I will proceed with 2 sticks if it is stable. Thing is that the bluescreen could come after 2 days in use. Maybe I should stress-test it with Prime95? Just afraid that it will go over 100C with those tests.

elaola · 9 Nov 2020 at 11:19

No bluescreens at all since I turned on X.M.P with the ram running at their full potential. I do only use 2 sticks atm. Only problem is that the CPU gets a bit hot, but I know if I lower the values on the ram I will definitely get bluescreens. So I guess it is either 3200mhz or default 2133 mhz that works.

liltimmy · 10 Nov 2020 at 13:44

elaola said:
No bluescreens at all since I turned on X.M.P with the ram running at their full potential. I do only use 2 sticks atm. Only problem is that the CPU gets a bit hot, but I know if I lower the values on the ram I will definitely get bluescreens. So I guess it is either 3200mhz or default 2133 mhz that works.

that is strange that turnng on XMP would fix the bluescreens... How hot is your cpu running? hopefully that noctua cooler will resolve the temps when you're able to install it safely! (remember to run the machine for a bit to warm up the thermal paste before trying to remove the cpu + existing cooler)

pc-guy · 10 Nov 2020 at 23:22

Were you running 4 sticks before? Many motherboard don’t like 4 sticks.

to remove the thermal paste you need some makeup removal or nail varnish remover it is the same stuff. I use my wife’s all the time.

Cyber-Mav · 11 Nov 2020 at 06:59

Ssd firmware updated?

Also you should really be first running memtest for a few hours to see if thats the problem

elaola · 12 Nov 2020 at 12:36

pc-guy said:
Were you running 4 sticks before? Many motherboard don’t like 4 sticks.

to remove the thermal paste you need some makeup removal or nail varnish remover it is the same stuff. I use my wife’s all the time.

I removed the thermal paste with isoproyl-alcohol. Then I installed my new Noctua NH-D15 (used a drop of noctua thermal paste) and booted up with X.M.P. enabled and with 32GB installed ram. At first glance at the temps in idle mode, I can swear it has a higher temp. 35C compared to 32c. And the same spikes up to 70c, but this is probably cause this fans may run slower in idle than my previous ones. Anyway, I launched up Borderlands 3 and went nuts on all weapons to see if it will go up to 90C, but it never went higher than 73C in load! This Noctua cpu-cooler is a serious badass. 17C degrees cooler is crazy. And I haven't had a bluescreen for days. I think I can say that this is solved for now. Can update in a few weeks to say if everything is still going smooth!

Cyber-Mav said:
Ssd firmware updated?

Also you should really be first running memtest for a few hours to see if thats the problem

Already done!

read my previous answers.
__________________________

Edit: With several players and a bit longer play time, the cpu hit a bit higher temps than 73C, but it never went beyond 77C degrees and that is my highest compared to 91C, before with the old cooler.

Any feedback? Is this ok temps for idle, 34-37c? ^^

elaola · 7 Dec 2020 at 02:27

So for almost a month and no crashes. But tonight it crashed with the message: uncorrectable error. Was not playing or anything. Just sitting in discord and chatting when I got that bluescreen. And as always I have to turn the PC completely off. Then start and go into bios and choose the M.2 ssd 970 evo again as boot drive. So, what do you guys think this can be?

Edit: My temps are now 70C in idle and around 80C in load. What the hell is happening? The game is getting lag, like for 4 seconds and then goes to normal. Then the lag comes again.