I have been battling for weeks now with random black screens for a few seconds (haven't had this for a while though), wattman settings restored to default due to system failure, as well as black screens when gaming that I have to hard reset to get out of and this is often followed by Windows getting stuck in a BSOD repair loop - reliability reports say the amd drivers crash (thread stuck in device driver) - basically the same issue that many people are having.
I totally accept the possibility of a hardware failure as this is a second hand machine but, as there are so many people with literally exactly the same issues, I'd like to really be able to rule out software issues before I give up. When it does crash and is stuck in BSOD / repair loop and I just shut it down and leave it till the next day, Windows boots up normally and isn't stuck in the loop anymore, but it does complain about the driver recovery and losing wattman settings (although these never seem to stay anyway) so maybe it is an overheating / hardware problem.
I have tried just about every solution I have found online and in this forum, apart from changing PCIe settings in BIOS and undervolting because I don't really know what I'm doing, I guess I will try it if nothing simpler fixes it. Or maybe I'm just not using the right combination of solutions.
1.) What has worked for you if you've had these issues?
2.) What's up with the automatic settings in Adrenaline? The default fan curve seems super low and the card gets pretty hot pretty quickly (80 deg C in 3mins and rising testing with OCCT) with the fan not going much above 2200rpm. Surely this is wrong? If so then are the other auto-detected/default settings also wrong and is this why people seem to be fixing it by undervolting, etc?
3.) Many people seem to have solved this by not using the AMD Adrenaline Software at all and just using the drivers alone. I want to try this but I still need to adjust the fan curve - what's the best way to do this? > edit: Tried drivers only, no AMD software, still had exactly the same black screen crash (thread stuck in device driver) at GPU temp of 68 deg C 20mins into a game. Used ASUS GPU Tweak to adjust fan curve. So it's defs not the software for me or the GPU getting too hot. Nothing else with a sensor was getting hot either but can't rule out something else is.
Windows 10 Professional
Intel ® Core(TM) i5-7400 @3GHz
16G RAM (2 x 8G, unmatched but this shouldn't cause huge issues?)
AMD/ATI RX 570 4G
Hard drives: 2 x SSD:
SK hynix SC308 SATA 512G – where everything but 1 game is installed
TOSHIBA MK3265GSX 240G – just ARK Survival Evolved because it's big.
Motherboard: ASUS H110M-R
Power supply: 600W Vortex 80 plus Bronze https://www.raidmax.com/product_d.php?lang=en&tb=1&id=812
Monitor: 1 x Dell SE2416H connected with HDMI
Solved! Go to Solution.
So it's very, very hard to see but with some really close-up examination and some macro photos I found some pins on one of the chips on the card appear to be damaged. I am going to guess that this is probably the cause of my issues - see image.
I think this may have progressively become worse, particularly at high temperatures as I began experiencing 'glitchy' screens as time went on and this wasn't an issue at the start. I hope that driver/software didn't contribute to this.
I am giving up and will buy a new card when I have the money
I do still think there are driver / AMD software issues based on my issues as well the responses and solutions here and the weird fan curves and I really sympathise with the frustrations of anyone experiencing the black screens so if it is helpful to someone here is a summary of the solutions for black screen / BSOD that worked for some:
1.) Clean up drivers with AMD Cleaner and/or DDU, also remove windows drivers while in safe mode. Then install the AMD drivers and software (video here https://www.youtube.com/watch?v=Qh2xEUXWeqs ). - didn't really work for me but it's probably the best place to start
2.) Follow the AMD troubleshooting guide here Troubleshooting Common AMD Graphics Driver Installation Problems - didn't make any difference to me but it's good to check your file system etc to eliminate that
3.) Make sure anything that might conflict with the AMD software is uninstalled (eg. MSI Afterburner or ASUS GPU Tweak) - this is defs a well know issue so do this before you do more complicated things
4.) Use 2019 drivers instead of the 2020 ones
5.) Use a HDMI Cable or If You Still Wanna Use Displayport Disable FreeSync then Change Your Monitor Refresh Rate to 120hz From OSD Options.
6.) Every time you start the PC go into Adrenalin Settings / Display and change the Pixel Display from RGB Full to RGB 4:4:4 or any of the others and back to Full RGB
7.) Try this TDR fix Graphics driver stopped responding and has recovered....TDR fix
8.) If you're getting the Driver Verifier Detected Violation error try this FIX: Driver Verifier Detected Violation error in Windows 10
9.) Switch to a higher resolution monitor
10.) Undervolt the card
11.) Swap out the PSU - this seems to have worked for a lot of people and seems unrelated to the power of the PSU or any specific specs they have in common as far as I can tell so no idea why this would work, but it has helped some people
I have the same problem, the auto fan speed curves just broken. GPU is getting very hot 83c while gaming and fans just turning in 1500rpm, not enough to blow the hot air out. I have to set it manuallly thru adrenaline driver.
I think AMD must fix this issue to updated driver
Adrenaline Driver 20.4.2
Win 10 pro 2004
Asrock B550 PG velocita
XFX RX 480 8gb
Kingston Fury 16gb ram 3200
Seasonic G550 Gold psu
Please try again on version 20.9.2 if you have any problem
I have been using 20.9.2 and also tried 20.9.1 (as well as a whole bunch of older drivers), but I will try it again. I haven't tried 20.9.3 yet, I see it's just come out so I will try that too and report back
20.9.3 is out? I don't see a link. Also shouldn't it be 20.10.1? AMD names its drivers as YY.MM.revision
I 've been getting a black screen with a reboot seemingly at random since August 22.
Have you run sfc/scannow? It supposedly repaired some files when I ran it an hour ago.
Have you looked at event viewer to see what the crash consisted of? Mine was "Critical event 41 Kernel-power."
I also ran a program "Who Crashed" and it's pointing to AMD Drivers specifically atikmpag.sys. I also tried this work around.
So I'm waiting for another crash. I'll probably swap out this Evga power supply next that I bought at Newegg. I looked at the reviews on Newegg and Amazon for this particular PSU and users were reporting traits I've noticed with mine.
Sorry, you're totally right, I thought I saw something about it on another forum but it was a typo, lol.
Hi! Thanks for the TDR tip, I'm going to look into it
Yes I have run SFC (and every other step of this guide Troubleshooting Common AMD Graphics Driver Installation Problems ), tested RAM, stress tested CPU and GPU (could not replicate the issue but it happens at random times so maybe I just had to leave it running longer or it's a power issue and only running the GPU flat out isn't drawing enough to crash). In reliability monitor /event viewer I get hardware error code 141 "thread stuck in device driver" and when I take a look at the crash dump with Blue Screen Viewer the AMD drivers, amdkmdag.sys crash and then direct x crashes dxgkrnl.sys
Love to hear if you have any success with swapping out the PSU
Last night I cleaned everything with DDU and then tried to install 20.9.2 again - crashed during install (right at the end at 100%) - caused all sorts of chaos and BSOD. I eventually ended up doing a clean windows install and got as far as cleaning with AMD driver cleaner (just in case) and I will try and install amd drivers again after work
If you do the TDR work-around it might be in hexadecimal. No big deal. I just changed it from hexadecimal "1" to "8" hexadecimal.
Might try the PSU swap tonight. I haven't had a crash in 3 days.