[SOLVED] Graphical issues and "device removed" crashes

rq2

Aug 17, 2020
3
0
10
Specs:
ryzen 5 2600
msi b450 tomahawk max motherboard
sapphire nitro+ rx580
16gb 3000mhz corsair vengeance lpx ram (kit of 2x8)
corsair tx750m psu (750w)
500gb m.2 form factor sata ssd with windows 10 and fedora 32, and 1tb mechanical drive with an ext4 partition and a shared btrfs game library

I've been having issues for a while with games displaying artifacts and occasionally crashing.
Some games work fine, but those most affected are Control and Remnant from the ashes, both running in windows.
Here are some screenshots of the issues, some are permanent, but some come randomly and go with a restart of the game: [url=https://postimg.cc/gallery/MKTyVMx]https://postimg.cc/gallery/MKTyVMx [/URL]
I have also had problems with games crashing:
When Control crashes, the message is "<something> due to device removed", and Remnant from the ashes displays the unreal engine crash dialog, with "Unreal Engine is exiting due to D3D device being lost. (Error: 0x887A0006 - 'HUNG')".
The crashes always start with the game freezing but in Control the sound effects playing at the time continue in a loop.
Sometimes my monitor will also briefly switch off due to lack of input, and in even rarer cases the whole windows desktop environment has crashed when it switches back on.

I would put this down to something software related on Windows, but I am getting colour banding in Fedora (similar to one of the screenshots I linked) though in various applications other than games, for example when watching youtube videos in firefox, which appear fine with no banding on other devices. I also seem to remember seeing the same dotted pattern as in some of the screenshots appear in the sky when playing minecraft with shaders, which was running natively on fedora, but it could be down to a bug in the third party shaders and hasn't been an issue for a while.
I have also had l4d2 crash once on Fedora, causing a black screen and a ticking sound from (I think) the GPU fans.

What I have tried:
  • disabling XMP
  • removing the ram one stick at a time
  • moving my GPU to a different slot
  • swapping my GPU for an old nvidia gt630 I had
  • removing any unused peripherals
  • connecting my PC to a TV via hdmi and checking for the banding and artifacts there
  • reinstalling and reconfiguring the radeon drivers and software under windows (before I thought I may be dealing with a hardware issue)
  • checking GPU and CPU temperatures; GPU stays at around 60C rarely surpassing 70, and CPU stays at around 50-60C

I still don't really know if I am dealing with a software or hardware problem, and I don't have replacements for any other components to test with
Any help is greatly appreciated
 
Solution
It can be the GPU, or the Mboard. But considering swap the GPU and reinstall driver not fixed the issue, the Mboard may be the culprit. The only way to make sure it try change the Mboard, or test the GPU on another system. If the GPU is working fine on another system, so the Mboard is toast.

rq2

Aug 17, 2020
3
0
10
As an update;
I have tried a fesh installation of only windows and control on another drive which still had the same problems,
I have also discovered that disabling antialiasing in Control changes the glittery pattern (seen in the last screenshot) to a very dark but perfectly even grid of dots, barely visible.

I had a rather extreme crash in Remnant also, it started with some flickering in the top left of the screen, and then the entire monitor flickered before switching off, and the voice call I was in at the time also went dead, leaving me to restart my PC.

It is likely I was wrong about the colour banding being in Fedora too, I made a mistake regarding the video playback I mentioned.

GTAV is added to my list of games with issues, the dotted pattern can be seen on textures in the distance as they load in, maybe its related to some sort of view distance fog rendering in that game?

Given that at least part of this problem is completely predictable, it would surely be some sort of memory problem? Yet, I ran the windows memory test which yielded nothing, and I don't think it would be gpu memory if switching out the gpu didn't change the problem in the slightest.
 
It can be the GPU, or the Mboard. But considering swap the GPU and reinstall driver not fixed the issue, the Mboard may be the culprit. The only way to make sure it try change the Mboard, or test the GPU on another system. If the GPU is working fine on another system, so the Mboard is toast.
 
Solution

rq2

Aug 17, 2020
3
0
10
It can be the GPU, or the Mboard. But considering swap the GPU and reinstall driver not fixed the issue, the Mboard may be the culprit. The only way to make sure it try change the Mboard, or test the GPU on another system. If the GPU is working fine on another system, so the Mboard is toast.
Thanks for the reply!
I have been gathering together some old parts to test the GPU in another system, but unfortunately I can't do a proper test as the power supply would have to come too (its the only one powerful enough and with the correct connectors that I have).
I too suspect the motherboard, but do you think there is any way these issues could be caused by a component other than the GPU or motherboard?