GPU Very hot after changing thermal paste

Trevor_4

Reputable
Sep 18, 2015
10
0
4,520
I decided to change the thermal paste on my Powercolor r9 390 after temps were hitting the 90s. The temps were still hitting the 90s and very quickly (I had to decrease the voltage by 100 millivolts). I've replaced it several more times and the problem is still occurring. What can I do?
 
Solution
I figured it out, I wasn't putting enough thermal paste on. I put on a lot this time, way more than should be necessary, and I'm reaching max temps of around 73-74 degrees. This is with the clock back to stock and the undervolt at 20 millivolts. Thanks for everyone's help!

Trevor_4

Reputable
Sep 18, 2015
10
0
4,520
I know that all of the fans are running, and they are running at about 95%. Even with the under-volt and a 30 mhz underclock, the GPU is still hitting 83 degrees.
 

atomicWAR

Glorious
Ambassador
sounds like a lemon...whether your fan is going bad or something else entirely in the VRM/Caps/etc. One thing i might try is running DDU in safe mode and then reinstalling the driver to rule out a software issue (like a bad fan profile for example). Be sure to back everything up just in case. If DDU doesn't clear things up you'll likely need to replace your fan/cooling solution or get a new GPU altogether.

http://www.guru3d.com/files-details/display-driver-uninstaller-download.html
 
Are you running with a closed or open case?

Ambient temps?

Are you running a stress test on the graphics card, or just seeing these temps while gaming?

Have you double checked that you reassembled the graphics card cooler correctly?

Did you use too much thermal paste; enough?

Were there thermal pads that were bridging gaps, did you remove them?

How well did you clean the cooler?

What were temps like under the same conditions when the card was new? Is this a new occurrence, or do you not know?
 

Trevor_4

Reputable
Sep 18, 2015
10
0
4,520
The conditions are the same, open case, using fire mark. All the pads have been put back where they were when I took it apart. Before I was having temperature problems, temps would be in the low 80s with clock at 1010mhz and stock voltage. I blew out the cooler fairly well with compressed air.
 

Trevor_4

Reputable
Sep 18, 2015
10
0
4,520
The temps are the same in the room as far as I can tell. I'm in the process of updating the drivers right now using DDU. I used afterburner to set the fan to 100%, but that doesn't make much of a difference.
 
Well, it sounds like there isn't much left here. Once the fans are at 100%, that's the best you can ask of your cooler in keeping the temperature of your graphics card down. One thing I would certainly do differently is, use the AMD driver software to set and verify fan speed, not Afterburner.

If I were faced with this situation personally, I would start trying to figure out if the cooler is properly getting the heat from the graphics chip away from the card. In my own situation, I have a heat pipe that gives me a great indication of whether heat is being pulled away from the chip itself. I can both touch the heat pipe with a finger (no, I'm not recommending you stick your fingers in your cooler,) and use an infrared thermometer to check temps on the heat sink. If the heat sink isn't hot, despite your GPU's reported high temperatures, your interface between card and cooler is likely the culprit.

The reason I asked about drivers is, sometimes driver updates cause temperature profiles for graphics cards to be adjusted. If you haven't updated recently, that obviously wouldn't be the case. It wasn't meant to mean that you necessarily need a driver update, although running the most recent driver is usually recommended.
 
Which paste are you using? Also, is there anything that might be stopping the cooler from seating fully onto the GPU?

Edit: When you've taken the cooler off to re-apply the paste, has the old paste from the previous time spread across the cooler and GPU in a nice even thin layer?
 

Trevor_4

Reputable
Sep 18, 2015
10
0
4,520
I figured it out, I wasn't putting enough thermal paste on. I put on a lot this time, way more than should be necessary, and I'm reaching max temps of around 73-74 degrees. This is with the clock back to stock and the undervolt at 20 millivolts. Thanks for everyone's help!
 
Solution
Ok, good to hear you found the problem. Whenever I hear of someone not adding enough paste, it always reminds me of what "JayzTwoCents" proved when he used a whole tube of paste. https://www.youtube.com/watch?v=rAid5G30-WM

 

ShadyHamster

Distinguished


I've always wondered why and where people got the idea of using to much paste was a bad thing, if you use to much once you put the cooler on any excess will be squished out with the mounting pressure, then once things heat up and the paste gets runny it will get squished out even more.
Luke from LTT did a video about this too, this time with CPUs, and the results showed little to no difference using to much paste.
I can understand with using electrically conductive pastes, but with normal pastes the only negative to using to much paste is you're just wasting it.
 

Karadjgne

Titan
Ambassador
On the other hand, using too much paste, especially the thicker viscosity ones can have a disastrous affect on temps. The thinner almost liquid pastes tend to run down around the sides, which isn't so much an issue with Intel lga, but can be a huge issue with connectivity with Amd pins.