[SOLVED] Windows 10 ntoskrnl.exe+3f72e0 BSOD

[OuToU]

Hello, around half a year ago I upgraded my CPU from a Ryzen 5 1600 to a Ryzen 5 3600X, changed RAM from a 2666MHz 4x4GB kit to a 3200MHz 2x8GB kit of G.Skill Ripjaws V and changed graphics card from a GTX 1070 to a RX 6700 XT, all within around a week of each other. All bought second hand. After getting the new parts I started getting random BSOD crashes, using BluescreenView to look at dumps they always pointed at ntoskrnl.exe. I didn't bother to start looking into it more until around a month or so ago.

I ran memtest86 on both RAM sticks seperately, and one of them gave an error (Test 9 [Modulo 20, ones & zeros]) USED PC WITH THE SINGLE "GOOD" STICK FOR ABOUT A WEEK WAITING FOR NEW KIT TO ARRIVE BUT STILL GOT BSODS which kind of leads me to believe that the problem isn't with RAM. Got the brand new identical kit, problem still persisted. Reinstalled Windows 10 21H2. No use. Bought a brand new motherboard (Gigabyte B450 Aorus Pro), problem still persists. After Windows reinstall I noticed the crash code is always the same (ntoskrnl.exe+3f72e0) where as before it was ntoskrnl.exe followed by different codes. Crashes happen regardless of what the PC is doing, gaming, idling, watching YouTube.

Tried to update BIOS to the newest version (F62d) which seems to have been pulled from the motherboads support page, it is now showing F61 as the newest.
Tried disabling fast boot from Windows power settings and BIOS.
Tried disabling XMP.
Tried installing chipset drivers via download from AMD website as opposed to motherboard manufacturers.
Graphics card drivers (version 21.10.2) were installed after clean Windows install with no internet connection.

Currently testing disabling Global C-state control via BIOS as I read that had helped someone.

PC Specs in their entirety:
GPU: AMD RX 6700 XT
CPU: AMD Ryzen 5 3600X
RAM: 2x8GB G.Skill Ripjaws V 3200MHz (F4-3200C16D-16GVKB) (On the supported memory list for the motherboard)
Motherboard: Gigabyte B450 Aorus Pro
PSU: Corsair TX550M
Storage: Windows installed on a 120GB Drevo X1 SSD
1x 1TB WD Blue Sata SSD
1x 500GB M.2 Crucial MX500
1x 1TB Seagate HDD

Connecting to the internet via USB tethering if that matters.

Attached is a picture from CrystalDiskInfo.
Uploaded dump files here: https://drive.google.com/file/d/1yjh88iybZC8curfwmJ4mY9bIabJFLkdk/view?usp=sharing

https://imgur.com/Cn9JJIN

 

[Colif]

since it survives new mb and new ram, it should be something else

have you tried installing windows on another ssd? 44% health would make me think about not using it. 120gb is getting a bit small for windows these days, we generally suggest 250 at very least. Just gives space to move.

Try running prime 95 - - https://www.guru3d.com/files-details/prime95-download.html
Prime 95 how to Guide: http://www.playtool.com/pages/prime95/prime95.html

Prime 95 checks CPU & Ram

 

[OuToU]

No I haven't tested another SSD, 120GB seems plenty for me, still got ~55GB free. One of Prime95 workers crashes after only a couple minutes.

[Thu Jan 27 19:53:39 2022]
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected running 960K FFT size, consult stress.txt file.
[Thu Jan 27 19:58:39 2022]
FATAL ERROR: Rounding was 0.4655136047, expected less than 0.4
Hardware failure detected running 960K FFT size, consult stress.txt file.
[Thu Jan 27 20:01:15 2022]
FATAL ERROR: Rounding was 2.836753831e+13, expected less than 0.4
Hardware failure detected running 960K FFT size, consult stress.txt file.

 

[Colif]

rounding errors can be fixed by giving more voltage to the ram in bios
they will cause BSOD so no point looking at dumps unless they continue after the rounding errors are fixed.

stress.txt is in the folder you installed prime 95 to, if you interested.

 

[OuToU]

Alright, how much should I start with? +0.025?

 

[Colif]

i would increase it as little as you can each time. I don't know the exact amounts as I suspect it changes per PC.

 

[OuToU]

Alright, here is what I did:
Started off from 1.2v>1.225>(noticed base voltage is 1.35)>1.375>1.4v>1.41v>1.425v>1.45v>1.475v>now at 1.5v, ran the torture test for 20 minutes with no workers crashing, every lower voltage crashed after max 5min. Going to keep using the system as normal and see if BSODs still occur. I'll update the post in the future!

Here are the results of the testing if anyone is interested:

[Thu Jan 27 20:15:03 2022]
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected running 960K FFT size, consult stress.txt file.
[Thu Jan 27 20:20:43 2022]
FATAL ERROR: Rounding was 8.598231346e+10, expected less than 0.4
Hardware failure detected running 960K FFT size, consult stress.txt file.
[Thu Jan 27 20:24:15 2022]
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected running 960K FFT size, consult stress.txt file.
[Thu Jan 27 20:26:44 2022]
FATAL ERROR: Rounding was 100.3535183, expected less than 0.4
Hardware failure detected running 960K FFT size, consult stress.txt file.
[Thu Jan 27 20:29:48 2022]
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected running 960K FFT size, consult stress.txt file.
[Thu Jan 27 20:30:51 2022]
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected running 960K FFT size, consult stress.txt file.
[Thu Jan 27 20:36:28 2022]
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected running 960K FFT size, consult stress.txt file.
[Thu Jan 27 20:40:22 2022]
FATAL ERROR: Rounding was 2.025540101e+18, expected less than 0.4
Hardware failure detected running 960K FFT size, consult stress.txt file.
[Thu Jan 27 20:48:51 2022]
Self-test 960K passed!
Self-test 960K passed!
Self-test 960K passed!
Self-test 960K passed!
Self-test 960K passed!
Self-test 960K passed!
[Thu Jan 27 20:54:44 2022]
Self-test 4K (thread 2 of 2) passed!
Self-test 4K (thread 1 of 2) passed!
Self-test 4K (thread 2 of 2) passed!
Self-test 4K (thread 2 of 2) passed!
Self-test 4K (thread 1 of 2) passed!
Self-test 4K (thread 1 of 2) passed!
Self-test 4K (thread 1 of 2) passed!
Self-test 4K (thread 2 of 2) passed!
Self-test 4K (thread 2 of 2) passed!
Self-test 4K (thread 1 of 2) passed!
Self-test 4K (thread 1 of 2) passed!
Self-test 4K (thread 2 of 2) passed!

 

[OuToU]

Update after 1 week, no bluescreens at all. I'll update again if the problems come back, but for now it seems like upping the RAM voltage did the trick.

 

[OuToU]

Another update, bluescreens started again. Happened three times thus far, first one was a new one: amdkmdag.sys+48b4, second one just happened, but no dump was created because "the system tried to repair itself but couldn't" and then bluescreened when I clicked boot to Windows, which is where the second dump with the same old ntoskrnl.exe+3f72e0 error code was created.
Also I've added another 2x8GB of the same Ripjaws V RAM just today, before the second BSOD but I don't see any why that would have an effect, they started happening before that as I said.

https://drive.google.com/file/d/13NMl6fcU2rqW8czka5HLa_fF-tCL95f_/view?usp=sharing
@Colif Would really appreciate if you could analyze the dumps, I have no experience doing that and no idea where to even start. Thanks in advance

 

[gardenman]

Hi, I ran the dump files through the debugger and got the following information: https://jsfiddle.net/uy2c758v/show This link is for anyone wanting to help. You do not have to view it. It is safe to "run the fiddle" as the page asks.

Spoiler: Dump File Results

File information:	020822-6078-01.dmp (Feb 8 2022 - 15:19:51)
Bugcheck:	CRITICAL_PROCESS_DIED (EF)
Probably caused by:	memory_corruption (Process running at time of crash: smss.exe)
Uptime:	0 Day(s), 0 Hour(s), 00 Min(s), and 02 Sec(s)

File information:	020622-7890-01.dmp (Feb 6 2022 - 15:01:53)
Bugcheck:	IRQL_NOT_LESS_OR_EQUAL (A)
Probably caused by:	memory_corruption (Process running at time of crash: System)
Uptime:	0 Day(s), 4 Hour(s), 49 Min(s), and 16 Sec(s)

File information:	012722-9015-01.dmp (Jan 26 2022 - 17:44:34)
Bugcheck:	IRQL_NOT_LESS_OR_EQUAL (A)
Probably caused by:	memory_corruption (Process running at time of crash: firefox.exe)
Uptime:	0 Day(s), 12 Hour(s), 23 Min(s), and 52 Sec(s)

File information:	012722-7156-01.dmp (Jan 27 2022 - 11:09:15)
Bugcheck:	KMODE_EXCEPTION_NOT_HANDLED (1E)
Probably caused by:	memory_corruption (Process running at time of crash: System)
Uptime:	0 Day(s), 5 Hour(s), 05 Min(s), and 47 Sec(s)

File information:	012222-7906-01.dmp (Jan 22 2022 - 14:46:36)
Bugcheck:	IRQL_NOT_LESS_OR_EQUAL (A)
Probably caused by:	memory_corruption (Process running at time of crash: System)
Uptime:	1 Day(s), 7 Hour(s), 12 Min(s), and 50 Sec(s)

Possible Motherboard page: https://www.gigabyte.com/Motherboard/B450-AORUS-PRO-rev-10
You recently updated to the newest BIOS available.

This information can be used by others to help you. Someone else will post with more information. Please wait for additional answers. Good luck.

 

[Colif]

Can you check bios and make sure ram voltages are still how they were? its possible adding new stick might have reset them?
wonder if it still passes Prime95

I wonder if cause is the CPU, its about only part you didn't replace. I would get PC looked at by a store, they can swap a known working CPU in and see if it has same problems. I feel that saves money. Hate to suggest it and find I was wrong.

amdkmdag.sys

is the AMD GPU drivers.

Oct 05 2021

amdkmdag.sys

AMD Graphics driver

can try running DDU in safe mode and reinstall AMD drivers - https://forums.tomshardware.com/faq...n-install-of-your-video-card-drivers.2402269/

 

[OuToU]

Yes, voltages are still at 1.5 volts, SOC voltage was around 1.1v. Just toggled off AMD Cool n' Quiet and AMD Core Performance Boost and changed power supply idle control from Auto to Typical Current Idle. I will run DDU and reinstall GPU drivers. I'll also try running Prime95 for a while and see if anything happens. I'll report back once again

 

[OuToU]

Ok, took the PC to a local repair shop and they replaced the motherboard. Originally they thought my SSD might be the issue so they tried to install Windows onto a new one and it kept crashing on them until they changed the motherboard. Will keep using the system as normal once again and report back in case anyone runs into similar problems.

 

[Colif]

well, mb is one thing we can't test for. So repair shop was only option really to find that, without blindly replacing board in hope its part to blame. I seen people replace almost entire PC and its the one part they haven't swapped that was cause all along

Hopefully it was the fix for you

 

[OuToU]

Motherboard was not the fix, although there was clearly something wrong with the old one since they couldn't even install Windows with it. Currently loaning a Ryzen 5600X from the shop for testing. I read online that early batch Ryzen 3000 series CPUs have caused BSODs, according to the sellers posting the timeline for that would match so hopefully that is it. Will keep updating.

 

[Colif]

i have what you call a late batch 3000 series, an R5 3600XT.

 

[OuToU]

Used the loaner CPU for roughly 72 hours straight with no problems at all, installed the 3600X back and within a couple hours got a bluescreen so I guess there's my problem. Time to order a new CPU. Thanks for your help @Colif and @gardenman