Question GPU shows negative temps in Afterburner and Speedfan

May 13, 2022
5
1
15
Hi, I am absolutely clueless.

This has been happening for some time, but very randomly and it actually stopped for a while before coming back. Out of nowhere for no reason my GPU fan starts running at 5,7k RPM and temperatures are shown in Afterburner as -2147486948 °C (which is obviously ridicoulous. Once this "error" goes away, temperature is usually around normal +-40°C. This happens randomly, while under load or idle, doesnt influence it in any way. Not sure whether is Hardware based or software based. I have an older GPU I can switch my current for, but Im really busy at the moment since I have finals.

My specs:
Intel Core I7 2600
AMD RADEON HD 6970
RAM 8 GB
MoBo MSI ph61a-p35 (ms-7732)


Whole setup is about 10 years old, so a certain part might be failing. However Im absolutely blown away since Ive never encountered anything of this sort. I ve been having other issues as well, i might share them if they seem relevant to this in replies.I´d be glad if someone had a clue!
 
Last edited:
May 13, 2022
5
1
15
Sounds like failing temp sensor on the GPU. Not something that could be fixed I'm afraid (since the sensor is most likely on the die).
yeah, thats what Im thinking as well. But if it were failing wouldnt it just jump up and down? just showing random gibberish of a temperature? it always show this number when it starts going wild. Idk the GPU is a f** 6970 series (10 y.o. or so) so it would make sense... but this to me seems like a driver issue maybe? seeing this weird number and only this number? Dunno
 
seeing this weird number and only this number?
Probably the sensor when it fails it sends 'maximum' value which the software reads as negative most likely due to not expecting such high value - but that's only a suspicion. What's important here is when the reading goes wrong the fans go full speed, which means the reading is incorrect already on GPU level.
 
  • Like
Reactions: assassinmaster
May 13, 2022
5
1
15
Probably the sensor when it fails it sends 'maximum' value which the software reads as negative most likely due to not expecting such high value - but that's only a suspicion. What's important here is when the reading goes wrong the fans go full speed, which means the reading is incorrect already on GPU level.
yeah I suppose, I also had the idea of "maxing out" and software not knowing what to do with it. Seems like my suspicion has been confirmed... Dont know if other issues will persist or if i ve solved them as well, but they arent the topic of this tread. Thx mate
 
  • Like
Reactions: DRagor