Computer GPU temps goes from 30 to 90 degrees in 2 seconds after i open any game

[solo_mid]

Hello guys, im facing a big problem and im trying to understand whats going on.
 

This month i had too many computer force-restarts in PUBG specifically & APEX and sometimes in DOTA 2 and Black Desert Online. Most of the times happened in PUBG the exact moment while i was clicking menu categories like customizing, carrer etc. or in-gameplay my computer crashed and then force-restart with a lot of BSOD. Also sometimes when i put psu switch on ( 0 ) and after in ( I ), causing power outage in my room but not everytime.

 

After that i decided to take some actions. I did MemTest64 full passes no problem found. disk analyze not problem found. Next i removed the amd drivers, roll back (nothing), i let windows decide (nothing), cmd: sfc /SCANNOW (nothing), updated all drivers (nothing), Bios latest update (nothing), i did 2 times windows 10 clean install (nothing), changed room to try my pc in another plug (nothing). CPU is fine, temps fine. Then opened gpu changed thermal pads and thermal paste, i saw some progress with temps in idle 27 - 30 celsius (80 - 86 farenheit) but when i start a game my temps goes directly from 30 to 89 celcius (192 farenheit) instantly in 2-3 seconds and then computer is force-restarting with a small delay (on boot i see black screen for 10 seconds more than normal boot). I have 2 .txt files from GPU-Z. One from the moment i open DOTA 2 then crash, and the other im doing userbenchmark test. Both crash. When i do userbenchmark test, the moment it shows the 2 fluffy-hairy circles with the black background my computer restarts and stucks there with fans 100% and black screen and nothing happens, so i need to turn off and turn on again.

 

Maybe my PSU is dying i have it 4+ years and i dont believe that is GPU. The problem is that i can't test anything right now, for example to ask someone for another psu just for a test, or give someone my gpu to test it in their pc. Not possible right now. So i need your help if you ever faced something like this. THANKS!

 

PC Specs:

Windows 10 Pro 21H1
CPU : Ryzen 5 3500X
GPU : RX 580 Gigabyte Aorus 4GB ( Andrenaline Driver: 21.6.1 (17-Jun-21))
MOBO : MSI B450 TOMAHAWK MAX ( Latest BIOS version )
RAM : Kingston HyperX DDR4 3200 C16 2x8GB
PSU : Thermaltake SE 730W (4 years+)
 

dota test.txt userbenchmark test.txt

[emosun]

6 minutes ago, solo_mid said:

when i start a game my temps goes directly from 30 to 89 celcius (192 farenheit) instantly in 2-3 seconds

your gpu isn't making significant contact with the heat sink. Meaning you didnt reassemble the gpu correctly when you took it apart. it has enough contact to stay alive when idle but not enough for any load.

the thermal mass of the heatsink should prevent instant jumps like that but if its not making any contact then yea it'll jump as if no mass is present.

[solo_mid]

4 minutes ago, emosun said:

your gpu isn't making significant contact with the heat sink. Meaning you didnt reassemble the gpu correctly when you took it apart. it has enough contact to stay alive when idle but not enough for any load.

So i need to tighten the screws more?

[emosun]

2 minutes ago, solo_mid said:

So i need to tighten the screws?

You need to reassemble the gpu correctly so the heatsink is making contact with the gpu , whatever that entails in your situation.  

9 minutes ago, solo_mid said:

changed thermal pads

You could have easily used the wrong size thermal pads and now the gpu isn;t making contact anymore.

your gpu wasn't previously overheating so disassembling it was a mistake. It will be hard to diagnose what your original issue was as now youve added another issue on top of it

[solo_mid]

13 minutes ago, emosun said:

You need to reassemble the gpu correctly so the heatsink is making contact with the gpu , whatever that entails in your situation.  

You could have easily used the wrong size thermal pads and now the gpu isn;t making contact anymore.

your gpu wasn't previously overheating so disassembling it was a mistake. It will be hard to diagnose what your original issue was as now youve added another issue on top of it

 

[emosun]

what the heck is this

theres too much thermal paste on that gpu ..... and it's not even pushed to the sides meaning the gpu wasn't making any contact with the heatsink at all....

Thermal paste isn;t bathroom caulk..... it's not there to make up the gap. You use a small amount of it to seal the micro imperfections in the metal or silicon. No wonder it's overheating like that.....

[solo_mid]

21 minutes ago, emosun said:

what the heck is this

theres too much thermal paste on that gpu ..... and it's not even pushed to the sides meaning the gpu wasn't making any contact with the heatsink at all....

Thermal paste isn;t bathroom caulk..... it's not there to make up the gap. You use a small amount of it to seal the micro imperfections in the metal or silicon. No wonder it's overheating like that.....

I removed the most of it but again, in games goes 90 degrees and idle 30

[Hairless Monkey Boy]

1 minute ago, solo_mid said:

I removed the most of it but again, in games goes 90 degrees and idle 30

I agree with @emosun, the GPU is not making good contact with the cooler. Disassemble and reassemble, this time correctly.

[emosun]

3 minutes ago, solo_mid said:

I removed the most of it but again, in games goes 90 degrees and idle 30

then it's not making any contact. thermal mass smooths out those temperature changes. If the gpu is only contacting paste and not the thermal mass of the cooler then thats what happens. The temp of the gpu shoots up because it has nowhere to go.

Whatever you have to do to get that heatsink to contact the gpu do it.

[solo_mid]

Just now, HairlessMonkeyBoy said:

I agree with @emosun, the GPU is not making good contact with the cooler. Disassemble and reassemble, this time correctly.

I did, 4 mins ago sir

[solo_mid]

Maybe 1mm thickness is too much?

[emosun]

Just now, solo_mid said:

Maybe 1mm thickness is too much?

 

47 minutes ago, emosun said:

You could have easily used the wrong size thermal pads and now the gpu isn;t making contact anymore.

 

[Mark Kaine]

1 hour ago, solo_mid said:

Maybe 1mm thickness is too much?

thats the question… or its the.1.5mm one?

no idea how to figure that out, but i guess info is out there somewhere.

 

Also the thermal paste was ok, for gpus its better  too much than too little… problem is it really doesnt look like its making contact.

 

i mean i know this isnt much of help, im just confirming that you really need to find out the correct  thermal pad size for this gpu. 

 

 

[solo_mid]

I removed all pads to test it and temps doesnt spike like before. Temps are very good and stable. I ordered 0.5mm pads. Propably thermal pads was too thick so the chip didnt make contact with heatsink.

[solo_mid]

Thank you but that was the half of the problem. I must test now if it crashes again in games.