Re-Applying Thermal Paste Works Wonders

@anticupidon
As @Mufflore said, not, the thickness doesn't matter.
It will squish out the excess as soon as you put pressure on it.
If anything, more is better because you want COMPLETE coverage on the die. CPU or GPU.

@Mufflore

You should have used Liquid Metal for the 6700k, as I did. I have it since a year and a half by now after a second application (first one got permeated into the IHS.)

 

I know this gives a better result but I'm not happy using conductive liquid.
It permeating the lid is another issue, I pass on my kit to family and dont want them to have future problems they cant deal with.

 

Usually the oem paste is very thick almost dry so the manufacturer's must know what they are doing ( obviously as they spend big money on research and testing ) , wonder which paste out there is the oem stuff .

Also I always believed that lapping the cpu and the cooler , using only small amount of TIM gives the best results as you have metal on metal direct contact , that's what I would expect, the physics should back this theory but the only way to be sure is side by side testing .
There are some videos about this but I have never seen long term test with clear conclusion so who knows .

 

Liquid metal sounds great in theory , even with the risk if using conductive materials if done right , problem is that it seeps in to the metals and "Dryes" out eventualy , i dont like that and i dont want to change the TIM every year .

 

There is only 1 point where could be a problem on the 6700k and you can just either use nail vanish or liquid tape to cover it.
Other than that, there is no real risk.

The other points around can be left alone as long as you relid it with glue (I have used UHU for that. EXTREMELY Thin layer)

As I said, I have put the second coating of LM and this time it won't "dry" for a year and a half (It oxides, not dry) and the temps are still the same.

 

Thank you.
Now I remember, Kryonaut is very good for cold temps and is very easy to apply. It is a better application for IHS then direct die do to it's viscosity.
You would need something that is bit thicker but is just as thermal conductive as Kryonaut if you want to apply it to GPU "Die" to avoid thermal pumpout.

I nearly forgot about that. And we have a review here that tells you all you need to know.
https://www.guru3d.com/articles_pages/guru3d_thermal_paste_roundup_2019,1.html

If you change the order of paste based on thermal conductivity you have
Kryonaunt
Prolimatech Pk-3 Nano Aluminum
Prolimatech Pk-2 Nano Aluminum
Cooler Master Mastergel Maker

Afterwards just look at which are least easy to use. IE: medium, IMO.
As long as you are around 1 degree of the top performer you should be good.

 

I do wonder if Kryonaunt was the wrong thermal compound to be using as this is something XOC users have been using for chillers and Nitro Pots. It even reads that on the package.

 

TG now has a sub-zero compound, the Kryonaut extreme.
And Der8auer actually suggest not to use the normal Kryonaut for sub-zero applications.

So, yes, he did a bad call.

 

Shin Etsu you had to heat up the tube with hot water. Otherwise it was like softish putty. I loved Shin Etsu.

 

Yeah, same method I use to spread GC-Extreme and Hydronaut.
Otherwise, since they lack silicone, it's a pain to spread.

 

I take it we shouldn't be using Kryonaunt at all then. Not enough for sub zero and too watery to prevent pumpout for direct die use.

Kryo is the greek word for cold after all.

 

Yeah, the one you want from Thermal Grizzly for die applications is Hydronaut.

 

Ah I see.

 

Here is 1 week of using TG Kryonaunt.
I won't be using this on the die after today.

 

Yes, pumped out very badly.
Same happened when I used it for my 6700K delid (I didn't know crap about anything.) All the cores shoot up to 100c, therefore, thermally throttling.
When before when it was recently applied never went up to 60 or so.

 

That must be an AMD CPU, I read they are convex.
Intel are concave.
Perhaps this is why?

Although it didnt help my 6700K, after pumpout 1/2 the cores increased in temp by over 10C.
I think the lid underside isnt flat.

 

Thermalright TFX is the new shin etsu.
It is EXTREMELY thick, and will won't suffer pump out at all on laptops and GPU's, and is also rock solid on CPU's.
What's interesting is the consistency turns into a putty type material after it cures for awhile. And when that happens, you can rest assured that paste isn't going anywhere.
It's an extremely good paste but VERY expensive. It also should NOT be spread. The X long double diagonal method or the straight line across laptop CPU die method is best. Thermalright themselves recommends the 5 dot method like this:


But you can also do the long diagonal X method (Just don't apply too much or too little). The four very tiny dots are just to 100% ensure there's coverage at those cardinal end direction points.

 

Damn that's just crazy.. I've never seen anything like that before.

I've always used Arctic Silver 5, then used a little piece of plastic to spread in on thin, it has never ever spread outside of the contact space like what you are showing before, looks like that kryonaunt stuff is quite liquidy.

 

5700xt Nitro Die
Junction temps were peaking after a few days. If you look at the center there is thermal compound "missing" not sure how that happened. Funny thing with the 5700xt you lose FPS way before you see abnormal temps. Now with PK3 both edge temps and junction temps are the same at desktop. That's a 1st. However, in game it's about the same with edge temps but junction temps are in the low 70"s now. The fps is about 10 fps higher in mw.

When the paste pumps out edge temps can read mid 50C but junction temps are near or past 80C with 2100Mhz OC. I would suggest to anyone who is seeing 30C+ gap between edge and junction temps should consider re-applying thermal compound that is not marketed for chillers/pots or subzero temps.

That GPU is very, and I do mean, very sensitive to temps. I have, on 2 occasions, seen FPS dip south well before the junction temps reached 85C or so. It appears that 80C the gpu throttles performance even though I'm not seeing much diff. in clock rate.

 

Artic SIlver 5 is less bound to pump out, but, it will still end up pumping out.


Edit:

It was Artic Ceramique, not Silver. My bad.