Gpu overheating

[Srdjanx]

Hi, i need help. When i turn on my laptop, gpu is imediately at 120c i cleaned all the dust and i changed thermal paste but its still same. Fan speed is 25% and gpu usage is less than 10%. I updated drivers downloaded older driver but nothing, is my gpu dead or is there a Way to fix it?

I have to add that laptop turns off after five minutes so i uninstall driver and then it wont turn off but gets really hot in gpu area

 

[P4-630]

i cleaned all the dust and i changed thermal paste but its still same.

Are you sure you tighten the screws good enough?
Seems to me the heatsink doesn't make good contact to the die.

102C? Maybe it died already...

 

[Srdjanx]

Are you sure you tighten the screws good enough?
Seems to me the heatsink doesn't make good contact to the die.

102C? Maybe it died already...

Yes i tightened it. In hw monitor it says 119c and it goes up to 130c. It is integrated gpu i dont know but if it is totaly dead it would not show anything on screen

 

[P4-630]

Yes i tightened it. In hw monitor it says 119c and it goes up to 130c. It is integrated gpu i dont know but if it is totaly dead it would not show anything on screen

What are the specs?
The laptop has probably a iGPU in the CPU.
130C? no dedicated GPU? Maybe faulty sensor or hw monitor is just wrong, if it really was that hot I think it would be fried already.

What does GPU-Z say, also post some screenshots.

 

[Srdjanx]

I have intel core 2 duo t9400 and amd ati mobility Radeon hd3430 it is quite old laptop but i need it. Cpu and gpu are separated on motherboard. I cant use gpu-Z because i cant use original gpu drivers, laptop shuts off when i enter program and without drivers gpu-Z doesn't recognize any gpu

 

[droid-I]

I think you should try re-do the heatsink+fan mounting. Does the GPU and CPU share the same heatsink+fan?
As a temporary workaround could try if there is anything in advanced power setting to reduce the load/increase cooling effect.
Anything blocking cooler air intake?
Since you did clean the cooling, might have expected at least a few degrees lower temp, but seems it did not change?

 

[Srdjanx]

I was regularely using laptop and one day i turned it on and it just powers off.
Nothing is blocking air intake and yes gpu and cpu share one heat sink. I tried low power mode but nothing. I reseated heat sink several times but nothing. Before two monts i upgraded cpu, can that be an issue? I mean i used it many times after. I tried with old cpu but its same

 

[Toothless]

Time to move on with something younger and prettier.

 

[Srdjanx]

Time to move on with something younger and prettier.
I am planning to buy a new one but right now im broke so i just wanna know is there a chance to fix it or is it dead.
Btw i am usint that laptop for over 9 years.

 

[Jose Jeswin]

As a final resort, you can try adding copper shims to the CPU and GPU chips...i have successfully used the shims to save a 12 year old dell studio 15 laptop a year ago...copper shims can compensate for any gaps between the chip and the heatsink better than thermal paste..i applied thermal paste to both surfaces of the shims and mounted them as usual..

Here are two pics from gamers nexus showing shims mounted to a GPU

its worth a try...if you have any doubts, there are youtube videos showing how to mount copper shims..

 

[droid-I]

Does this not have switchable graphics? (Hybrid GPU)

https://www.amd.com/en/support/kb/faq/dh-017

If the fan is at only 25% when it shuts down does not seem correct. A utility to the control fan perhaps might help. Other than that, since shutdowns began all of a sudden, could well be some component has failed.

There is also this, but not sure if that is any help since problem seem be with GPU :

The ThrottleStop Guide (2024): Lower Temperatures, Increase Performance and Battery Life

Author's Note: I wrote the first edition of this ThrottleStop guide for UltrabookReview as part of a short undervolting/tweaking guide several years ago. I

www.ultrabookreview.com

Come to think of it might also try underclocking the GPU :

Why and How to Underclock a GPU - Everything You Need to Know - The Tech Lounge

While overclocking is the more popular option, there are some cases where you might want to underclock your GPU. So if you want to learn how to underclock your GPU, or whether you might even need to do so, then this is the right article for you.

www.thetechlounge.com

Since it is a laptop, chances are limited if any of the suggested programs in the above article may work, worth a try though.
The other thing is, need the correct GPU driver for utilities to work.

 

[Srdjanx]

Hi again, i managged to enter gpu-Z and here are the photos

 

[droid-I]

Ok, so the driver seem to work and GPU-Z loads. But this not much help. Not aware what utility you use to read the temps, but you can try HWinfo.
https://www.hwinfo.com/download/
Skip the lansweeper, click the green download box instead.
Check the "sensors" section within hwinfo.

 

[Srdjanx]

Here is hwinfo

 

[droid-I]

Load is 0%, GPU fan @ 100%, temp 126'C , does not look good. Fan actually spinning at full rpm? Other than that, something is broken.
What is the make and modelno of the laptop?

 

[Srdjanx]

cpu and gpu share one fan and it is spining like at 25% of speed

 

[Zach_01]

The fan is driven by CPU temp usually. Is CPU temp ok?

Do you have the latest HWiNFO version? You can always ask at HWiNFO forums the software developers if your reading is true or false. They’re going to ask you for a debug file.

 

[Srdjanx]

I tried several softvares for temperature and it is always the same

 

[toilet pepper]

Maybe this will help the others... does it actually get hot? Do you feel warm air from the vents? Does the heatsink get hot?

 

[Srdjanx]

It gets hot in the gpu area but not hot like 130c.

 

[Valantar]

It gets hot in the gpu area but not hot like 130c.

If you touch the GPU cold plate or heatpipes connected to this, does it burn your fingers? If not, there are two possibilities:
- There is insufficient (given the temperatures reported: essentially zero) contact between the cold plate and the GPU, or
- The thermal readings are wrong, meaning that there is something broken in the GPU itself.

My money is on the latter, as >120C should cause an immediate system shutdown/crash. Even if the GPU is old enough to not have built-in thermal protection, that's hot enough to cause it to just crash or even break outright.