Question Help diagnosing a GPU that is working way too hard

Jun 3, 2020
22
4
15
Hi! I'm hoping to get some ideas about why a graphics card would be running way too fast and hot.

I have a brand new build and I have two of the exact same graphics cards, ASUS Rog Strict GTX 980. (Only using one card in the system, I just happen to have two on hand)

When I run one if them it works as it should. Cool and fairly quiet even under medium to heavy load. The other one, however, seems to be working really hard under the same conditions. As soon as I start a game the fans speed up to 100% and the temperature increases pretty rapidly. I only let it get up to 80 degrees c before I shut it down but my bet is that it would have kept climbing.

I can pretty much rule out any sort of driver or software issue as well as any issues with other hardware since the other identical card works fine.

I opened up the problem card to see if there was anything obvious. It was very clean on the inside. The only thing that looks a little off is that there is some excess thermal paste around the outside of the chip. Seems unlikely that that would cause the problem I am having, but could it?

Any ideas on what could be the cause and how might fix it?
Thanks!!!!!!
 
Jun 3, 2020
22
4
15
That's normal. My 1080Ti was like that when I took it apart:
View: https://imgur.com/3okdykr


The GTX 980 is getting up there in age... my guess would be the thermal paste has dried up and needs to be replaced.
Yep, that's pretty much how mine looked. Considering the age of the card, it definitely wasn't the original thermal paste. It still had a good bit of moisture to it. The card was very clean on the inside, so my guess is the person I bought it from cleaned it and re-applied the paste recently.

Only thing I could see was that the paste was caked up around the outside of the die and in the same spot on the heatsink, so maybe it was so thick that it was causing a bit of a gap between the two.

I cleaned it off and reapplied a proper amount and it made a pretty significant difference. It's still a little louder than my "good" card but in no longer sounds like a jet engine.

I ran Furmark for about 30 minutes on both for comparison. Card A (the "good" card) maxed out at 71° C with the fan speed at about 36%. Card B (the "not as good" card) got up to 76° C and the fans were a good bit faster at around 55%. I didn't run Furmark on the bad card before I cleaned it, so I don't know exactly how much of a difference it made, but it definitely helped a lot.

I would love to get card B running like card A. The extra noise of the fans is still noticeable. But, I don't know any other easy fixes aside from what I have already done (cleaning it and putting on new thermal paste). Any suggestions? Or do I accept that this as good as it's going to get and move on?
 

Phaaze88

Titan
Ambassador
You may need to replace the thermal pads. It's similar to thermal paste when taking the cooler apart.
Those imprints in the pads are going to leave air pockets unless you're a master at putting gpus back exactly like they were.
The paste certainly helped the gpu core though.

The temperatures reported are an average of 3: the gpu core, VRM, and memory. I do not know why Nvidia has it set that way.
 
  • Like
Reactions: nathandiffee