Arch Linux

adamnvrtil

Member

Registered: 2022-02-05

Posts: 5

GPU sensor is showing wrong temperatures.

Hello, I have an issue, I couldn't find any solution to yet.

Whenever I run any program, that uses atleast a bit of GPU power (using qutebrowser also) my system will kernel panic (I believe it's kernel panic, because my keyboard stops responding too), and my screen will go black. The system won't come back by itself, I always have to turn off my computer by holding down the power button for 10 seconds, and then turn it back again. The system boots just fine after this. I also experience black horizontal lines a lot.

We're talking about:

$ inxi -G
Graphics:
  Device-1: AMD Oland [Radeon HD 8570 / R5 430 OEM / R7 240/340 / Radeon 520
  OEM]
    driver: radeon v: kernel
  Display: server: X.Org 1.21.1.3 driver: loaded: ati,radeon
    unloaded: modesetting resolution: 1280x720~50Hz
  OpenGL: renderer: AMD OLAND (DRM 2.50.0 5.16.5-zen1-1-zen LLVM 13.0.0)
    v: 4.5 Mesa 21.3.5

$ uname -r
5.16.5-zen1-1-zen

I went ahead and tried to troubleshoot this problem, I first checked any missing drivers and those look pretty much fine to me.

$ pacman -Q | grep xf86-video
xf86-video-amdgpu 21.0.0-2
xf86-video-ati 1:19.1.0.r9.g5eba006e-2

As you can see, this isn't exactly the issue.

I thought, that it could be because of my gpu might be overheating, so I tried looking on it using `sensors`, and the temperatures looked okay to me (+55.0°C on average), so I tried benchmarking it using some utilities, and the temperatures were going crazy (+20°C every second), so I thought It's thermal paste and fan issue, so I went ahead, and replaced GPU fan and thermal paste, then I tried it again, and I had the same issue. So something came in my mind because of this, GPU temperature sensors are lying. Every time computer goes off (it goes off aprox. at 110°C), but when I turn my computer back on (that takes about 1-2 minutes), and I check the temperature using `sensors`, it's back on ~55°C.

There comes my question: What's wrong, is there's any solution to it? Could I force my system to not go off, when GPU gets to that GPU temperature? I know, it's risky, but I don't game or do anything intensive on this machine.

Also sorry for not putting any journal messages and etc., but I couldn't manage to get them. Also, I would like to put there video showing this, but I think, that administrators wouldn't like any embeded links there.

Thank you for your time.