Windows 10 Frequent, Semi-random BSODs (Thread stuck in device driver)

Hello Circinae,

thank you for that updated information. I also think it is annoying that we have to fight a regression introduced by AMD update.
For me with the newest driver and that Registry tweak i had only one crash yet.
I didn't check without the registry hack. It could also be the new driver or the crashes are still to come.

BR
 
Well, I can safely say that the newest Crimson driver update does not fix the BSODs at all. :p I only had Steam open, alt tabbed to open chrome (habit) and got a BSOD. I'll try that registry hack now.
 
Hey, sorry for the really late reply. I've been a bit busy with irl stuff, and I probably will be for another week.
For about 6-7 days I hadn't gotten a single BSOD, and then all of a sudden I got a couple the other day, one yesterday, and one this morning. I don't remember if I'd already said this, and I'm too lazy to check, but at least 90% of the time I'm alt tabbing when I get the BSOD. This morning's BSOD happened while I was just watching a video on youtube (in fullscreen), and I alt tabbed to check the time.

There may have been one or two BSODs where I wasn't alt tabbing, but the vast majority of them happened while alt tabbing (whether from Chrome to another application, or to/from a game).

Figured I'd upload the minidumps incase any of them have new information, but I'm not crossing any fingers. I had planned to upgrade my card to either an r9 390 or a 1080 in the next year or two (or possibly whatever the newer cards are), but I may have to do it sooner. Even BSODing while alt tabbing is causing me to lose some of my work. :(
I've never had BSOD issues before (had the card for over 2 years now), up until a couple months ago, which pushes me to agree that it's either my GPU dying (unlikely?), or an AMD driver software issue.
All dump files were the same Bugcheck EA.

As already mentioned it's either the card or the driver itself. I tried to see if others had experienced anything like this by doing a search and I did find one case where it was eventually discovered that the fix lay within the driver. The users issue eventually stopped via a driver update.

Have you followed AMD's instructions on how to install the driver to the letter?

Installing the AMD Radeon™ Software

  1. Ensure that any existing versions of AMD Radeon™ Software on the system have been uninstalled before proceeding with installing a new driver. For instructions on how to uninstall AMD Radeon Software please refer to KB article: How-To Uninstall AMD Radeon™ Software from a Windows® Based System.
  2. Close all opened applications including anti-virus, firewall, remote-access, or webcam software before attempting the install process.
  3. Double-click the downloaded file and select Install to begin file extraction process.
    Note: It is recommended to use the default destination folder for installation.
How-To Install AMD Radeon Software on a Windows Based System

Other than that the only other thing I can suggest is pop over to Guru3D and try going through their drivers:
Videocards - ATI Catalyst Windows 7 | 8 | 10

In fact two drivers were released last week by AMD, have you tried both?
Graphic Drivers
 
Yes, I have followed the instructions to the letter. I've removed/uninstalled all AMD drivers, ran DDU, restarted my computer, had no applications running, and installed the new drivers. (A total of 3 times now)
I have updated to to 16.11.2 this morning, and about 30 minutes later I had a BSOD, so the driver definitely has not fixed it.
I'm currently trying out Sven's idea using the RadeonMod program by Guru3D for the registry tweak, to disable Crimson Power Saving mode.
I'll report if I get any more BSODs.

Did you check out that thread I linked? The majority of the users posting there have almost the exact same issue that Sven, Circinae, and I are having, with our R9 290s. It seems that no one in that thread has found a solution either, though.


Edit: Going back to that thread I linked, the "more like this" sidebar has several more threads with the same problem. It seems like no one has found a solution, other than rolling back to older drivers, or.. buying an NVIDIA GPU. Rolling back to previous drivers is about as bad as having these BSODs, as most of the games I play are ones released recently, that don't have the optimization of AMD's drivers from 6+ months ago.

Edit2: Threads that have similar issues: Blue Screen Of Death THREAD_STUCK_IN_DEVICE_DRIVER | Community
THREAD_STUCK_IN_DEVICE_DRIVER | Community
thread_stuck_in_device_driver | Community
THREAD_STUCK_IN_DEVICE_DRIVER THREAD_STUCK_IN_D... | Community
AMD Radeon R9 200 Driver Crashes in Windows 10 | Community

I'm sure there's even more than that, but this issue is very clearly an AMD driver issue, and not one that any of us can correct just by updating. It seems our only solutions, at the moment, are to roll back to 15.11 or 15.12, or just buy an NVIDIA.


That said, I'm still testing this registry tweak, so I'll still report back if there's any more BSODs.
 
Last edited:
Hello SpaceKitkat,

i contacted AMD support regarding this issue. I will point them to our thread and the thread of the others to proof that it is not a fault of our card but a driver regression problem.
My computer after the registry fix is pretty reliable. I only had one crash after i wanted to leave a game after some hours of play.

BR

Sven
 
Last edited:
You could if you want the optimisations install a set that works and then update to the latest but only install the application profiles.
 
I can say with confidence that using Clock Blocker and changing the "Default" setting to "Block", then having it enabled stops the crashes, regardless of the driver version (I'm running the latest). Make sure the program is set to start when Windows starts and as I mentioned before, it will increase your gfx card's idle temperatures (from 46C to 51C for me) and obviously the energy consumption. You are able to give it custom setting for particular applications, so with more tweaking you can probably find a better set up.

EDIT: This was mainly aimed at SpaceKitKat as this is a third, albeit a less than perfect, alternative.
 
Hello Circinae,

this is a very valuable information as it proves the point that the current DVFS (dynamic voltage and frequency scaling) software implementation is broken and the root cause of the mess.
The old driver not crashing proves that it can work with our cards and the cards are not broken (AMD / XFX argumentation was: your card is broken RMA it, if out of warranty, which I am, you are on your own).
The summary is that it is a fault introduced by driver changes (driver regression) and has nothing to do with our cards.

AMD should analyze their changes and fix it.

It is interesting that your card goes to 51°C only. My card went to 90°.
Can you describe which settings you used with the C[l]ock blocker ? ;)
 
If you send me private message with your email i can cc you onto the conversation with AMD / XFX support.
BR
Sven
 
I'm still testing out the tweak in RadeonMod, but so far I haven't had any BSODs, and I've been stressing alt-tabbing a lot.
That said, it's only been one day. lol
 
Some nice posts here guy's and it looks like your on the right track.
 
..
I also tried your Radeon Mod registry change and I believe it would be effective if it actually worked. For me, it failed to actually deactivate the Crimson Power Saving and thus, the clock speeds were still all over the place, crashing my PC after about two minutes.
..
Hello Circinae,
i don't understand how it can not work. Didn't you run the Radeon Mod with Administrator rights ?
In my understanding it is just modifying the registry. After that you have to reboot of course.
BR

Sven
 
The RadeonMod Registry hack is not working for me.
I already had 3 BSOD since then.

Circinae please can you post your ClockBlocker settings. For me the heat is getting very large.
 
61 ° C with the fan running at 42% (which is already a lot).
That's not too bad actually. If it starts to get to temps like 90 - 100c then you may need to check the heatsink. Try removing the side panel and run the test again.
If the temps are a lot lower then you know it's possibly an airflow problem.
 
Last edited:
Back
Top