Drivers & Software

[Kurashinno]

Some time in the last month, games began crashing after some lengthy play sessions. Then it went down to 30 min, 20 min, and now it's less than 1 min. All stress tests, benchmarks, and stability tests pass perfectly with amazing temperatures on my custom loop rig. I'm running a 5800X3D with a RX6900XT Tuf. The symptoms are always the same: The game will run perfectly for a while, then video freezes. This is followed by a "disconnect" sound. (I think this is the driver shutting down and disabling the GPU). Then the sound in the game sputters and stops. The video never comes back at this point, but if I unplug and replug USB devices, I can still hear the "disconnect/connect" sound. Upon hard restarting the computer via the tower, going into the Device Manager, the GPU is disabled. Right click, enable. Then an error message pops up from AMD software saying it cannot load the drivers for this device. In Device Manager, it says enabled but not working properly. I cannot open AMD software at this point, saying that an incompatible driver disallows this, error PA300 or something. Back to Device Manager, update drivers, "you are running the most recent drivers." Now I go to my downloads where I keep a few of the most recent versions of the Adrenalin software ready to reinstall and repair. It does its thing, repairs (most of the time), now the GPU is enabled and functioning normally. I restart to clear everything. If I fire up any game, it simply happens all over again, and again, and again. The PC works fine when I'm not gaming and perfectly during benchmarks, stress tests, and stability tests using OCCT, Geek Bench, etc. 

I've tried using DDU to clear old drivers, rolling back, install a new version of windows 10, 11, installing the newest drivers, taking apart everything in the loop to make sure the thermals are ok and everything is installed ok, asking AI to help me troubleshoot, just about everything I can think of. 

Does any of this look familiar to anyone? Are there steps I haven't thought of? Is my GPU broken? Any thoughts, experiences, or insights would be super helpful. Thanks all!

-Kura

[johnnyenglish]

Does the card have dual BIOS? If it does, try the other one.

How about Heaven stress test? Does it crash?

What is your PSU?

[Kurashinno]

It does have dual bios. I'll flip the switch and see what happens. I ran the heaven test a few days ago under max settings and it was fine. It seems to have gotten worse since then. I should try running it again. I have the EVGA Supernova 850 GT PSU. Thanks for the reply btw

[Kurashinno]

Ok, I flipped the switch for the other bios. Then I reinstalled Heaven stress test and ran it with the same settings as before. It crashed before the loading screen finished. The time interval is getting smaller and smaller. Now things crash almost instantly. 

[infohills]

While your stress tests have passed, there could still be a hardware issue that only manifests under certain conditions. Try swapping out components if possible (e.g., RAM, GPU) to see if the issue persists with different hardware configurations. Even though you've mentioned good temperatures, it's worth monitoring them closely during gaming sessions to see if there are any spikes or unusual behavior that could indicate a cooling issue.

[Kurashinno · Accepted solution]

Alright, guys. Looks like it was at least one of the thermal pads on the GPU. The ones provided by EKWB were all 1mm. That was fine for the VRAM and the lower VRMs. There were two raised VRMs, bigger taller, with some 3 letter stylized logo on them, grey in color. Anyway, the raised ones needed a 1.5mm thermal pad to make proper contact on the heatsink. So I got some thermal grizzly minus 8 1.5mm pads, put them on and bam. We're back in business.