THREAD_STUCK_IN_DEVICE_DRIVER atikmdag.sys randomly on moderate load

Hi there,

I have been getting this BSOD for my newly built PC that I built last month.

Things that I have tried:

1. Fully updated Windows
2. Tried both AMD Adrenalin 20.4.2 and 20.2.2, with DDU in safe mode between each try
3. Run MemTest86, all 4 passes passed with no error
4. Enabled/disabled FreeSync both in monitor and in Adrenalin settings
5. Enabled/disabled XMP profile
6. Got advices to update my motherboard BIOS, but it is already P3.70 out of the box and AFAIK it is the latest version
7. Installed AMD Chipset driver for B450 motherboard

First time I got the BSOD is when I'm playing a rather heavy games, but recently it's been happening even during moderately heavy load, like 10+ Chrome tabs, 2D drafting in AutoCAD, or even just by opening Steam, not even starting any game, but it never happened during light load situation like working with Office documents and file explorer only.

At very rare chance I could play the heavy games for 8+ hours without encountering any BSOD or even any sluggishness (which usually happen 5-6 seconds before BSOD). But most of the time, if I encountered the BSOD and rebooted using the reset button, it is guaranteed that I will either get the BSOD again if I open any game, even light ones like Dota 2, or the PC simply freezes and monitor displays 'no signal'. The subsequent restart would randomly puts me in one of these 2 situations, BSOD or freezes. At this point I usually just get frustated and not using my PC again until the next day (this PC is mostly for games, I have a laptop for work).

During games I usually run MSI afterburner to display GPU and CPU temperature, and none exceed 65C even after 8+ hours of heavy gaming. Most of times, the BSOD or freeze still do happens just after I launched the game, while the temperature are still below 55C.

I already filled system specs in profile and attached V2 Log collector result. Any help would be appreciated, thanks.

Windows 10 Version Number is 18363.836

As you've stated, your BIOS/UEFI version 3.70 is the latest available for your motherboard.

Please ensure that you have checked Windows Update manually and ensure that you have ALL available updates.

WER section of the MSINFO32 report shows that most of the BSOD's are STOP 0xEA (9) and they are due to AMD video driver
There are also other video related errors in that section.
The 5 uploaded memory dumps are all 0xEA's and all blame AMD video.

As you've already tried replacing the video drivers (and used DDU to do it), we can presume it's not a software problem.
So, the next step is to try hardware diagnostics. I suggest you start with these free one's:

- OCCT (all tests): OCBASE / OCCT Run until the temps level out.
Each test is different, so you'll have to keep an eye on the temps while running each test.

- Furmark (run until temp's level out or there are problems - whichever comes first): FurMark > Home

- Prime95 Stress Testing (run for at least an hour - or until the temp's level out. If you get errors you can stop the testing): GIMPS - Free Prime95 software downloads - PrimeNet

Beyond that, I'm not very familiar with the Vega graphics. I did find this link while searching the web: (Solved) - Crashing video driver Ryzen 5 2500U Vega 8 HP x360 laptop

See rule #4:

https://www.tenforums.com/misc.php?do=showrules

No piracy or discussion of piracy allowed at all. Such as software, music, videos and other intellectual property violations (e.g. downloading youtube videos locally etc) - it is forum policy that no help shall be given to people who knowingly steal software or services.


Please uninstall any pirated / cracked software and any tools / methods used to bypass the activation of software.

jdc1 said:

WER section of the MSINFO32 report shows that most of the BSOD's are STOP 0xEA (9) and they are due to AMD video driver
There are also other video related errors in that section.
The 5 uploaded memory dumps are all 0xEA's and all blame AMD video.

I see.. I suspected as much it's about AMD graphic driver.

jdc1 said:

Please ensure that you have checked Windows Update manually and ensure that you have ALL available updates.

Windows Update says it is fully updated, just checked.

jdc1 said:

As you've already tried replacing the video drivers (and used DDU to do it), we can presume it's not a software problem.
So, the next step is to try hardware diagnostics. I suggest you start with these free one's:

- OCCT (all tests): OCBASE / OCCT Run until the temps level out.
Each test is different, so you'll have to keep an eye on the temps while running each test.

Temperatures leveled out after a few minutes, but I ran each test 15 minutes anyway, one after another immediately, without restart. Here are the results:
1. CPU OCCT
Attachment 279601
2. CPU LINPACK
Attachment 279602
3. GPU 3D
I couldn't do this test at all, tried it four times. 3 seconds into it then my PC automatically restarted. It is different than when in game though, in game the PC either freezes then no display, or BSOD, but this test put the PC to restart.
4. GPU Memtest
Attachment 279603
5. Power Supply
Attachment 279604

jdc1 said:

Beyond that, I'm not very familiar with the Vega graphics. I did find this link while searching the web: (Solved) - Crashing video driver Ryzen 5 2500U Vega 8 HP x360 laptop

I don't think this is related, that post is for Ryzen mobile processor and is a different model and generation than mine.

jdc1 said:

- Furmark (run until temp's level out or there are problems - whichever comes first): FurMark > Home

- Prime95 Stress Testing (run for at least an hour - or until the temp's level out. If you get errors you can stop the testing): GIMPS - Free Prime95 software downloads - PrimeNet

Will do them tomorrow, gotta leave for work now.

- - - Updated - - -

zbook said:

See rule #4:

https://www.tenforums.com/misc.php?do=showrules

No piracy or discussion of piracy allowed at all. Such as software, music, videos and other intellectual property violations (e.g. downloading youtube videos locally etc) - it is forum policy that no help shall be given to people who knowingly steal software or services.


Please uninstall any pirated / cracked software and any tools / methods used to bypass the activation of software.

I understand. Will comply ASAP.

Thanks zbook! The system won't let me thank you again until I thank someone else!

jdc1 said:

- OCCT (all tests): OCBASE / OCCT Run until the temps level out.
Each test is different, so you'll have to keep an eye on the temps while running each test.

- Furmark (run until temp's level out or there are problems - whichever comes first): FurMark > Home

- Prime95 Stress Testing (run for at least an hour - or until the temp's level out. If you get errors you can stop the testing): GIMPS - Free Prime95 software downloads - PrimeNet

Alright, I decided to reinstall Windows again, but this time I installed nothing. Not ASRock APP, not AMD Chipset driver, not even AMD Adrenalin. Every driver is supplied by Windows Update, which I made sure already on latest update.

1. I tried to run OCCT again, and it's still the same. All test passed with no errors (OCCT, CPULINPACK, MEMTEST, POWER SUPPLY), ran each for 15 minutes. Except GPU test, it still restarted my PC after 3 seconds of running it.

2. Ran Furmark test for 15 minutes at default preset, stable temperature, no restart, no freezes, no BSOD:
Attachment 279738
Also tried at 1080p preset, score:
Attachment 279739

3. Ran Prime95 for 15 minutes, everything seems to be normal:
Attachment 279740

As I didn't installed Adrenalin, I don't how which AMD driver's version Windows Update gave me. I attached the latest V2 log, which include restart initiated by OCCT GPU test.

Will try games again, would take some days to re-download.

- - - Updated - - -

Tried to run a light game anyway, and it BSOD at the title screen (game is Warframe). New log attached.
Attachment 279766

If the GPU test restarted the system, then the GPU imay be damaged.
Combined with the previous test results, it's very likely that the GPU is damaged.
As it appears you're using the integrated Vega graphics (on the CPU), it may be that the CPU is defective/damaged

Unfortunately, the only sure way that I know of to test this is by trying another CPU.
Is your CPU still under warranty? If so, contact the reseller or AMD to arrange for a return (RMA).

Notice the number of times I said "may be" or "likely"
I'm not real comfortable with asking you to spend any money doing this.
If unable to get another CPU (for free) to test with, then I'd suggest further testing until you're certain that it's a hardware problem with the CPU.
FWIW - if you do purchase another CPU, see if you can purchase it from a shop that'll let you return it for your money back if it's not needed.


Good luck!

jdc1 said:

If the GPU test restarted the system, then the GPU imay be damaged.
Combined with the previous test results, it's very likely that the GPU is damaged.
As it appears you're using the integrated Vega graphics (on the CPU), it may be that the CPU is defective/damaged

Even though FurMark GPU test run without problem? Any other free GPU testing software that is worth trying, or are they all basically the same?

jdc1 said:

Unfortunately, the only sure way that I know of to test this is by trying another CPU.
Is your CPU still under warranty? If so, contact the reseller or AMD to arrange for a return (RMA).

Notice the number of times I said "may be" or "likely"
I'm not real comfortable with asking you to spend any money doing this.
If unable to get another CPU (for free) to test with, then I'd suggest further testing until you're certain that it's a hardware problem with the CPU.
FWIW - if you do purchase another CPU, see if you can purchase it from a shop that'll let you return it for your money back if it's not needed.


Good luck!

Aww just my luck.. on my first PC too, and on the most expensive part .

Getting a satisfactory RMA isn't the easiest thing to do where I live, might have to shell out over $80 just for a $200 CPU, and weeks of multiple trips to the city, without guaranteed result.
This is going to be an adventure :).

Is there any further testing that I can do? For example to test for motherboard or PSU failure? I am pretty confident my RAM sticks are fine since they passed MemTest 86 with flying colors (twice, recently I ran it again), but I'm not so sure about the motherboard and PSU.
If I can prove there are problems with them too, I might as well try to issue RMA for them when I visit the city.

The logs displayed: SL_GEN_STATE_INVALID_LICENSE

Please activate the Windows license then upload a new V2.

Use Activation Troubleshooter in Windows 10
https://support.microsoft.com/en-us/...troubleshooter
https://support.microsoft.com/en-us/...-phone-numbers
https://support.microsoft.com/en-us/...ws-10-activate
https://support.microsoft.com/en-in/...ws-10-activate

zbook said:

The logs displayed: SL_GEN_STATE_INVALID_LICENSE

Please activate the Windows license then upload a new V2.

Use Activation Troubleshooter in Windows 10
https://support.microsoft.com/en-us/...troubleshooter
https://support.microsoft.com/en-us/...-phone-numbers
https://support.microsoft.com/en-us/...ws-10-activate
https://support.microsoft.com/en-in/...ws-10-activate

Hi, there.
Thanks for the link. New V2 log attached.