Question GPU temps is 90°C - Idle

[ThomasPP]

Hi everyone,

Something is going wrong with my graphic card : AMD Radeon R9 280

GPU Voltage is 0.950 V
Graphics clock is at 300 Hz
Memory is at 150 MHz
GPU utilization is 2% and Memory 2%.

BUT GPU temp is 92°C and fans are spinning at 100% (3,958 RPM according to HWinfo64)

Already checked :
-Fans are working and spinning (both)
-Pc is cleaned
-Thermal paste has been replaced
-Drivers and Bios are up to date

I am running out of idea.

Do you have any ?

Thanks a lot.

 

[TJS1986]

Hi everyone,

Something is going wrong with my graphic card : AMD Radeon R9 280

GPU Voltage is 0.950 V
Graphics clock is at 300 Hz
Memory is at 150 MHz
GPU utilization is 2% and Memory 2%.

BUT GPU temp is 92°C and fans are spinning at 100% (3,958 RPM according to HWinfo64)

Already checked :
-Fans are working and spinning (both)
-Pc is cleaned
-Thermal paste has been replaced
-Drivers and Bios are up to date

I am running out of idea.

Do you have any ?

Thanks a lot.

That Is a really Hot Temp and if it continues running that high your GPU could literally burn up(FIRE!!!)
I'm sure others will have more specific ideas but i would suggest Downloading MSI Afterburner and monitor your GPU through there and also set a Max Temp. I have my Max Temp set to 83 degrees Celsius and would be worried if it ever reached that. I'm confused as to why with fans running at 100% it's not cooling down? I would download that program i suggested and see if it's showing the same temp and also set your fan speeds and Max Temp, Clock Speeds, etc. through that.

 

[ThomasPP]

Thanks for your answer.

I did try MSI Afterburner.
This is what it looks like

Strange things :

1. afterburner can't read core Voltage
2. When I turn the fan speed to 50%, the temperature doesn't change.

Could the temp sensor be wrong or misread ?

 

[jay32267]

I find it hard to believe it is at 90C.
I'm leaning on a bad temp sensor.
Did it ever read correctly?
How warm does the card feel?
I'd shut down....let the system cool.....boot up.....go immediately to the temp readout software....and have a look and if it says 90C and the card isn't that warm...I think it's the sensor.

 

[ThomasPP]

Yes, I monitored the temp last summer when it was really hot outside (and inside) and nothing was wrong about it.

The fans are going crazy since boot, even in the bios, and the temp readout software says 90°C immediatly at start.

When the fans are spinning at 100% the card feels cooler than my hand.

When I turn the fans to 50%, I can feel the inductors from the power phase is (are? I can only reach one) hot. But this seems legit.


If it were really the sensor, is there anyway to bypass it? Actually the computer is turning off the graphic grad every time it reaches 100°C (seems obvious)

 

[jay32267]

Not sure about bypassing it.
Do you have another PC to try it in?
That would confirm it's the GPU or something else....but at this point I'm leaning on that it's the GPU itself.
I might consider clearing the BIOS (shutdown and remove the cmos battery for a few minutes)....perhaps something got corrupted. (just something to try)

 

[ThomasPP]

Hi,

So, I've cleared the BIOS, but it doesn't change anything.
I'm goiging to ask my friend who lives nearby if I can test the GPU with his PC.

By the way, a new error occured today. Here it is :


What could cause this ?

 

[ThomasPP]

Hi,

So I did try the failing graphics card inside my friend's PC.
Fans and the temp went crazy as well.

He lent me an old graphics card.
This old graphics card had a normal behavior in my PC.

I run a stress test to raise temp to 70°C before unplug it and i can confirm this old graphics card was way hotter than my failing one (which is supposed to be 90°C)

So I'm pretty sure now that something is failing with the temperature reading (may be the sensor).

Can I do anything to fix that ? Or at least bypass it ?

 

[TJS1986]

Hi,

So I did try the failing graphics card inside my friend's PC.
Fans and the temp went crazy as well.

He lent me an old graphics card.
This old graphics card had a normal behavior in my PC.

I run a stress test to raise temp to 70°C before unplug it and i can confirm this old graphics card was way hotter than my failing one (which is supposed to be 90°C)

So I'm pretty sure now that something is failing with the temperature reading (may be the sensor).

Can I do anything to fix that ? Or at least bypass it ?

I,m going to assume you have no active warranty on it? How old is the actual card? Did you buy it new?

 

[ThomasPP]

The actual graphics card is 6 years old, so yes no active warranty.

I guess it's time to buy a new one, but I'm frustrated that I must change it, just because of a failing sensor.
It feels so stupid.

 

[TJS1986]

Yeah, that sucks! I was wondering if you are going to stick with AMD or try a card from Nvidia? Personally I am a Nvidia guy myself, they are usually more expensive but the quality and longevity are unmatched. I just ordered an RTX 2060 KO 6GB to replace my GTX 1050TI 4GB. Nothing wrong with the 1050TI I just want something a bit more powerful as I am into Flight Simulation and the new Microsoft Flight Simulator 2020 is Releasing on August 18th and is very Graphic Intensive. I was very pleased with the 1050TI for the short time I had it though!