I've been experiencing some stuttering in a very low-resource game, so I've decided to update my drivers from 19.x to the latest optional ones (20.5.1). Right after installing the drivers, I've updated my Windows 10 to the latest version (2004) - foolish of me to do both without a delay, I know.
Ever since I've done that, I've been experiencing issues with my PC:
-screen freezing and unfreezing a couple of seconds later while playing videos (in a video player, MPC-HC), also sometimes while scrolling in a browser (tested in Chrome)
-screen freezing permanently, mostly while playing videos on Youtube in Chrome
-while in-game, screen-freeze is followed by game crashing (tested on Witcher 3)
The temporary screen freeze is logged in the Event Viewer (the permanent screen freeze leaves no trace in Event Viewer) as Event 4101, with the description: Display driver amdkmdap stopped responding and has succesfully recovered.
Following the advice found on this topic, up to this point I have tried:
1. completely removing the drivers, using the Apps & Features, followed by running DDU in Safe-mode and using both AMD uninstall and DDU to prevent Windows from installing its own gpu updates, followed by installing the latest WHQL drivers (20.4.2) - same results
2. I've reinstalled Windows (clean install, with SSD formatting) - same results
3. using the same process as mentioned above (in step 1), I've tried installing the earliest available drivers for my GPU on AMD's site (19.10.2 Optional) - this actually made the problem worse, because screen-freezing now became permanent. At this point I've reverted to the latest WHQL drivers (20.4.2) yet again.
4. I changed TdrLevel in the registry to 0 - made the screen freeze permanent until I've deleted the value from the registry.
5. I ran memtest86 - reported 0 errors.
6. I ran a FurMark 4K stress test - while it doesn't necessarily cause a screen-freeze, the temps get pretty high (up to 92°C)
7. I tried several versions of the latest AMD drivers which do not mention any TDR-related issues for any gpu-s - same results.
8. Once I've gone through all of the prior steps, my PC decided not to output anything to the monitor. I've tried hooking it up to several monitors - no luck.
9. I've tried putting in an older GPU (of a different manufacturer) after scrubbing AMD drivers with DDU - the system became stable with no hiccups. I do not believe this to be a true test, since the other GPU is much older and noticably weaker than my AMD one.
10. Aforementioned high temps were bugging me, so I decided to take my GPU apart, clean up the old thermal paste and put on a bit of new one (if you are attempting to do this, take note that disassembling your GPU might void your warranty - please look into that first). After putting my GPU back together and hooking it up, the screen turned on and with the latest WHQL drivers installed (20.4.2), I had a pleasant, uninterrupted evening of Witcher 3. Sadly, on turning my PC off and back on the next day, the problems returned - first with the drivers (amdkmdap issue, temporary screen freeze) and then the more serious, permanent screen freeze issue. I've monitored the temps once again and there hasn't been a significant decrease in temperature.
11. On taking apart my PC once again, I've noticed that one of the pins of the 8-pin connector seems to have melted the PSU cable (just one of the cable connectors) at some point. I've looked this issue up and a common answer seems to be in faulty GPUs drawing too many amps for some reason. I'll look into this further and update if there's anything of note.
If anyone had similar issues and found a potential fix or has any sort of advice on how to proceed - it would be much appreciated.
GPU: R9 290 Tri-X OC
OS: Win 10 Pro v2004, OS build 19041.264