[SOLVED] GPU Thermal Throttles a few days after re-applying thermal paste

[Tioym]

Hello Readers,
Graphics Card: SAPPHIRE PULSE RX 6700 XT 12GB
So the problem is quite self-explanatory. I have re-pasted my Graphics card 3 times now. Everytime I repaste, my GPU temperatures plummet during load, around 80-85c (HOTSPOT). But everytime I do this, it does fine for a while but the temperatures slowly climb back up to 100-110c (HOTSPOT) after around 2 weeks of the repaste.

What I think is happening is that the thermal paste seems to seep out of the sides because whenever I re-open the card to re-paste, there is very little thermal paste in the middle and most of it is just squished out the sides. I may be applying too much thermal paste but I'm not sure if that would be the main issue.
It could also be that the mounting pressure is not enough to keep the paste trapped inside so it oozes out when its hot(under load).

At this point I have some more thermal paste but I cant open my card up every 2-3 weeks and would like a solution that has a high chance of working

 

[Phaaze88]

It's normal for some paste to get pushed off the edges from mounting pressure, but some pastes are not viscous enough for direct die, due to the smoother surface, and thus they kinda slide off. In comparison, you can use just about any paste on a cpu IHS.

What paste are you using? It might not be the one for the job, or it's possible you've not reassembled the card evenly.
One of those 2, I figure.

 

[Tioym]

It's normal for some paste to get pushed off the edges from mounting pressure, but some pastes are not viscous enough for direct die, due to the smoother surface, and thus they kinda slide off. In comparison, you can use just about any paste on a cpu IHS.

What paste are you using? It might not be the one for the job, or it's possible you've not reassembled the card evenly.
One of those 2, I figure.

I'm using corsair's XTM50 it's pretty cheap. I have some cryonaut lying around haven't used it because dont feel like wasting it. I'm guessing I should try the cryonaut?

 

[faalin]

Did you replace the thermal pads on the vram? if you did not replace them, where any of them torn? Wrong size pads or even trying to put two pieces of a torn pad back together could raise the cooler just enough to cause issues.

There is what is called pump out, where during the heating and cooling cycles the paste will warm up and be pushed out from between the dye and cooler. Thermal paste is only there to fill in the micro scratches on the surface of the dye and cooler, there is not suppose to be a thick layer between the two.

 

[Phaaze88]

I'm using corsair's XTM50 it's pretty cheap. I have some cryonaut lying around haven't used it because dont feel like wasting it. I'm guessing I should try the cryonaut?

How much it cost doesn't matter at all; cost doesn't determine viscosity.
The compounds that are fluid in texture can be used on cpu IHSs, but are not as useful on gpu dies. The opposite, which are towards the sticky end, can also be used on cpu IHSs - cpu IHS accepts just about anything - are also useful on gpu dies.

I'm not familiar with XTM50, so I don't know where it falls in terms of viscosity, but I'm sure it's fine for cpus.
You can try the Kryonaut - I've read of success with using it on gpus, but I don't recommend Kryonaut because of 2 reasons:
1)It's stated on TG's website, but it degrades/dries out faster at sustained temperatures around 80C = more frequent repastes.
2)The cost per gram is high, and gets even higher if sustained temperatures can't be kept low enough.

 

[Tioym]

How much it cost doesn't matter at all; cost doesn't determine viscosity.
The compounds that are fluid in texture can be used on cpu IHSs, but are not as useful on gpu dies. The opposite, which are towards the sticky end, can also be used on cpu IHSs - cpu IHS accepts just about anything - are also useful on gpu dies.

I'm not familiar with XTM50, so I don't know where it falls in terms of viscosity, but I'm sure it's fine for cpus.
You can try the Kryonaut - I've read of success with using it on gpus, but I don't recommend Kryonaut because of 2 reasons:
1)It's stated on TG's website, but it degrades/dries out faster at sustained temperatures around 80C = more frequent repastes.
2)The cost per gram is high, and gets even higher if sustained temperatures can't be kept low enough.

You are right the corsair one is a lot less viscous, its way easier to spread so that may be why it just flows out the sides. is there any other paste you recommend? For now I'll try the kryonaut and see what happens.

 

[Tioym]

Did you replace the thermal pads on the vram? if you did not replace them, where any of them torn? Wrong size pads or even trying to put two pieces of a torn pad back together could raise the cooler just enough to cause issues.

There is what is called pump out, where during the heating and cooling cycles the paste will warm up and be pushed out from between the dye and cooler. Thermal paste is only there to fill in the micro scratches on the surface of the dye and cooler, there is not suppose to be a thick layer between the two.

The pads are not torn, they looked fine to me.

 

[Phaaze88]

You are right the corsair one is a lot less viscous, its way easier to spread so that may be why it just flows out the sides. is there any other paste you recommend? For now I'll try the kryonaut and see what happens.

I've been using NT-H2 on my ancient 1080Ti for 3.5years, so I can personally vouch for that one.
Arctic MX-4 works, Gelid GC-Extreme and Prolimatech PK-3 as well.

 

[Eugene$$]

try phase change like ptm 7950 see if it helps

 

[Tioym]

How much it cost doesn't matter at all; cost doesn't determine viscosity.
The compounds that are fluid in texture can be used on cpu IHSs, but are not as useful on gpu dies. The opposite, which are towards the sticky end, can also be used on cpu IHSs - cpu IHS accepts just about anything - are also useful on gpu dies.

I'm not familiar with XTM50, so I don't know where it falls in terms of viscosity, but I'm sure it's fine for cpus.
You can try the Kryonaut - I've read of success with using it on gpus, but I don't recommend Kryonaut because of 2 reasons:
1)It's stated on TG's website, but it degrades/dries out faster at sustained temperatures around 80C = more frequent repastes.
2)The cost per gram is high, and gets even higher if sustained temperatures can't be kept low enough.

This was it. Have had kryonaut on there for a while and no signs of deteriorating temps. It was probably the viscosity, I dont think I made a mistake while closing the GPU previous times.