[SOLVED] (Gigabyte 3060ti gaming oc pro) GPU gets hotter after replacing thermal pads and paste.

Oct 30, 2022
3
0
10
Got my RTX 3060 ti gaming oc pro for almost 2 years now. I'm living in a tropical country so it gets very hot every once in a while.
Decided to replace the paste (Noctua nt-h2) and pads (Deepcool ek720) to somehow help the temps and prolong it's life.

After all that, the temps were significantly lower only on idle, from 45c now 35-37c.
However, when I play a game, even in the title menu the fans ramps up to 100% and the temps are way hotter from my old thermal compounds and paste of 76c-79c to now new 86c-92c.
I always quit right away when the fan speed and thermal shoots up to avoid damages.

Now, I'm thinking maybe it's the size or quality of the thermal pads (which I now newly ordered Gelid-ultimate) since I trust thermal paste I applied since I also use it for my CPU.
I also made sure that the GPU dye is covered and reach full contact on the heatsink. So my conclusion is I just ordered the wrong size/thickness of the pad and just found out that thermal conductivity of this pad is just 6.0 W/m·K.

So my question is if VRAMs and VRMs affect the reading of the GPU temp overall? since this card has no friggin sensors for VRAM temps.
Also, VRAMs of this card is significantly close to the GPU main chip.

Thank you so much in advance.
 
Last edited:
Solution
Pads should've been a last resort. You've now landed in thermal pad trial and error limbo until you find ones that work.
Thermal pads aren't just about their depth, but also the degree of hardness. Manufacturers don't have a universal standard on pad hardness between the different models, and that info isn't readily available.

A caliper can discern pad depth, not hardness, but the w/mK rating gives a hint: the higher it is, the harder they tend to be.

When the gpu core temperature gets worse like that, it means cooler mating with the PCB has been warped, which is from using pads that are too thick, too hard, or both, and the vram temperature is fine.

Though it can't be read with software, the VRM has a high temperature tolerance...
Pads should've been a last resort. You've now landed in thermal pad trial and error limbo until you find ones that work.
Thermal pads aren't just about their depth, but also the degree of hardness. Manufacturers don't have a universal standard on pad hardness between the different models, and that info isn't readily available.

A caliper can discern pad depth, not hardness, but the w/mK rating gives a hint: the higher it is, the harder they tend to be.

When the gpu core temperature gets worse like that, it means cooler mating with the PCB has been warped, which is from using pads that are too thick, too hard, or both, and the vram temperature is fine.

Though it can't be read with software, the VRM has a high temperature tolerance, like 120C.
Does hwinfo nor Gpu-Z(sensors tab) not work for Memory Junction?
 
  • Like
Reactions: JMNugget
Solution
Pads should've been a last resort. You've now landed in thermal pad trial and error limbo until you find ones that work.
Thermal pads aren't just about their depth, but also the degree of hardness. Manufacturers don't have a universal standard on pad hardness between the different models, and that info isn't readily available.

A caliper can discern pad depth, not hardness, but the w/mK rating gives a hint: the higher it is, the harder they tend to be.

When the gpu core temperature gets worse like that, it means cooler mating with the PCB has been warped, which is from using pads that are too thick, too hard, or both, and the vram temperature is fine.

Though it can't be read with software, the VRM has a high temperature tolerance, like 120C.
Does hwinfo nor Gpu-Z(sensors tab) not work for Memory Junction?

Appreciate the reply!

That might be the case right now, I should've ask more from here.

I actually installed those apps for monitoring because of this, still no sensor. I've done everything software-wise such as uninstalling using DDU, rollingback driver versions.

The old original pad has a gooey/sticky texture and what I bought and installed was more brittle and like clay.

Would you recommend some thermal pads that worked for you? or might have the same texture as my old ones?
 
I can't recommend any, as I've never had to deal with it. I've put a Kraken G12 on my 1080Ti and didn't bother with pads and sinks, as the research I'd gathered told me I could cool the memory and VRM on air alone.
What I've disclosed earlier is all from reading about it. The pad hardness enigma is often what sours pad swap experiences.
Unless someone can chime in with their successful swap on that card, you're stuck figuring out on your own.
 
I can't recommend any, as I've never had to deal with it. I've put a Kraken G12 on my 1080Ti and didn't bother with pads and sinks, as the research I'd gathered told me I could cool the memory and VRM on air alone.
What I've disclosed earlier is all from reading about it. The pad hardness enigma is often what sours pad swap experiences.
Unless someone can chime in with their successful swap on that card, you're stuck figuring out on your own.

Jeez, I'm so dumb. By your reply, it made so much sense. The pads are too thick and hard! that's the way my wife likes it, but not this card.
I used my extra .5 pads laying around. The answers were in front of me all along. Now The GPU is now colder than ever.

Thank you.

Gigabyte 3060ti gaming oc pro
VRAMS : 0.5mm
VRMS:1mm.

Just in case somebody needs the info.
 
Jeez, I'm so dumb. By your reply, it made so much sense. The pads are too thick and hard! that's the way my wife likes it, but not this card.
I used my extra .5 pads laying around. The answers were in front of me all along. Now The GPU is now colder than ever.

Thank you.

Gigabyte 3060ti gaming oc pro
VRAMS : 0.5mm
VRMS:1mm.

Just in case somebody needs the info.
Which tropical country do you live? Im in brasil and my gpu is getting some hot peaks. Can you recomend the pads you r using?