Overheating problem strix 5700XT

[SavageNeo]

Specs:

Asus Strix B550-A

Asus Strix 5700XT

Phanteks P500A Digital (360mm aio top exhaust, 140mm fans back exhaust, 3x140mm fans front exhaust.)

SSDs, HDDs etc.

 

Hey! So suddenly my gpu started overheating. First when i mounted it to my new Case (P500A) my max gpu temp was 68C. now it goes to 99C

 

I replaced thermalpaste and updated gpus bios, did not improve anything. Also this is the "fixed" version also.

 

Now I think that the temp sensor broke. Since gpu temp hits like 94C but the gpu is fairly cold. Like it is easy to touch and can hold the hand over the PCB for minutes! Mem temps adn VRM temps are in their 70s and 60s.

 

Also quick look at the gpus PCB there seems to be some dried up liquid behind the PCB. Any idea? It cannot be from my AIO since there is only dried liquid in the gpu.

 

Also Hwinfo64 and Wattman shows the same values.

 

So your thoughts? Sensor broke or something else.

During Superposition

 

 

Ignore the dust: 

 

 

Edit: Limiting power to -30% on msi afterburner improves temps bu about 10-12C. limiting power by 50% improves around 15C. Still over 75C on any intensive game

Edit2: Default fan curve does not seem to recognise the high tempereture, becuase it still runs fans at 35% even if the temp is 95C.

Edit3: One day it was fine the second it was little under 30C higher than usual? I hit 73C at WWZ with 50% fan profile. Next day even at 100% fan profile it was over 85C. So temporary fix was to leave the fans at 60% and -30% power limit. I can still play at 144fps 1080p at max setting but my clock speed is now around 1.75ghz since the power limit. Time to get another gpu?

[AndreiArgeanu]

Your best choice for now is to set a flat fan curve until you find out what's the issue.

Is the card still under warranty?

[SavageNeo]

1 minute ago, AndreiArgeanu said:

Your best choice for now is to set a flat fan curve until you find out what's the issue.

Is the card still under warranty?

I have scratched it so not under warranty. Also at idle it behaves like normal. Hits 60C every 15mins or so, so the fans ramp up until the pgu is 55C, same again in 15mins. It was like that before and it is like that now. Seems very odd to me.

[AndreiArgeanu]

Just now, SavageNeo said:

I have scratched it so not under warranty. Also at idle it behaves like normal. Hits 60C every 15mins or so, so the fans ramp up until the pgu is 55C, same again in 15mins. It was like that before and it is like that now. Seems very odd to me.

Well if there's dried up liquid I suggest taking the card apart and giving the whole board a thorough cleaning with some isopropyl alcohol. Chances are something could have shorted or been damaged. While cleaning perhaps try to look for things that don't look like they should. 

[SavageNeo]

3 minutes ago, AndreiArgeanu said:

Your best choice for now is to set a flat fan curve

I Cannot do that. My fans Do not obey me. Like if I put the fans at 50% nothing happens when idle. But when I hit the gpu with a Load of even a little the fans ramp up to the 50%. Also odd. Cannot use even custom fan curve since my screen flashes. It has been always liek that since i Bios updted this GPU a year ago. ALso flashing the biso gain doesn ot help with the fans

[SavageNeo]

1 minute ago, AndreiArgeanu said:

Well if there's dried up liquid I suggest taking the card apart and giving the whole board a thorough cleaning with some isopropyl alcohol. Chances are something could have shorted or been damaged. While cleaning perhaps try to look for things that don't look like they should. 

Well I did apply new thermalpaste. Did not see anything out of the ordinary. I am not sure if it is liquid, but there was less dust so i assumed that there were dried up liquid.

[SavageNeo]

Updtae: If set the case fans to 0%. no fan moves expect the AIO fans. The results are the same as the fans would be at 100%. Like what? 15min stress test gpu hits 98C with no fans moving. Same with fans at 100%.

 

Seems like broken sensor. Also I did not see any thermalthrottling sicne the clock speed stayed at the normal 1.98-2.02Ghz like it always has been.

[SavageNeo]

Since it clims up to 93C is matter of secnods I have come to conclusion that it has to be broken sensor. GPU is not hot to touch, it is fairly cold. Not stuttering at games, Default fan curve acting liek the gpu is at 65-75C range.

[SavageNeo]

Update: Did DDU, that did not help. Used out of the box settings and gpu hit 100C and hotspot 105C While mem was around 70c(games), 80c (furmark), VRMS were in the low 60s.

The card did not thermathrottle. It ran at 2ghz with fans running at 70% this time. still 100c.

 

Did run Unigine heaven this time and got 2722 score and 110fps which is slightly over average. Sensor goen bad or something else?

 

Should i try to run the fans at 25% (lowest i can get them)  and try if the system crashes of gpu overheating to determine bad sensor?

[SavageNeo]

Update again: Tried with the 25% fan curve. Still did not go over 100C, hotspot still 105C. Odd, clock was still around the same: 2ghz!. WTF is going on? If i run the fans at 100% the gpu is 100C after a while. Same with 25% curve, Same clocks same scores. And  100% is very loud!

[VEXICUS]

Follow these steps ....

1. Clean the entire card with IPA, and let it dry.
2. Change the thermal paste and pads that are preapplied.
3. Flash a new bios for your card. You can watch a few videos regarding how to use nvflash. You can find bios for your card from techpowerup website.

If none of thiß fixes your card, then underclocking and undervolting will be ur best bet.