Which thermal paste has the absolute best performance?

[DeadlyPilot]

So far I'm thinking the MX4. Anything out there better than it cooling wise?

[Kobathor]

Lots of them are pretty similar. I personally use Noctua NT-H1.

[MoonlightSylv]

I don't know if it's "The best" but Cryorig CP7 (which came with my H7) gives me 28c~ idle CPU temps with an H7.

Edited June 23, 2016 by ShadowTechXTS

[LeapFrogMasterRace]

I always refer to this http://overclocking.guide/thermal-paste-roundup-2015-47-products-tested-with-air-cooling-and-liquid-nitrogen-ln2/6/

[SageOfSpice]

I'm using GC Extreme which is considered one of the best non-liquid metal pastes, and it's only reduced my CPU temp by about 3 degrees over AS5.

[byalexandr]

I like Arctic Silver 5.

[manikyath]

the thing about "absolute best" is that in 99% of cases it's not worth the money for the little benefit it gives over "lower end" pastes.

 

i just have a tube of MX-4 because its non-conductive, so i can use it on GPUs as well.

[ZestyJalapeno]

The liquid metal stuff is for nutters. Get a spill of that anywhere but the heatspreader (excess seeping into the pin array) and you're looking at money down the toilet .Get MX4. Cheap, effective, and foolproof.

[TheRandomness]

Toothpaste.

Mayonnaise.

Sem- I won't go there

 

Literally, there's like a 1 or 2 degree difference, don't spend so much.

[manikyath]

2 minutes ago, Dogeystyle said:

The liquid metal stuff is for nutters. Get a spill of that anywhere but the heatspreader (excess seeping into the pin array) and you're looking at money down the toilet .Get MX4. Cheap, effective, and foolproof.

non-conductive paste is a worth "loss in performance" for the security it brings.

[manikyath]

2 minutes ago, TheRandomness said:

Toothpaste.

Mayonnaise.

Sem- I won't go there

 

Literally, there's like a 1 or 2 degree difference, don't spend so much.

well, mayo..

[SageOfSpice]

3 minutes ago, manikyath said:

well, mayo..

Sorry, but LIQUID METAL coming below AS5?

 

That's the most rigged list I've ever seen.

[Delicieuxz]

I just ordered some IC Diamond thermal compound, after seeing it mentioned in a thread here, and then watching a LTT video where Linus says its his preferred compound (although the video was a few years old):

 

 

 

When I get a new GPU, I'm going to replace the stock paste with it, and hopefully shave a few degrees celsius off the operating temperature.

[ZestyJalapeno]

9 minutes ago, manikyath said:

 

i just have a tube of MX-4 because its non-conductive, so i can use it on GPUs as well.

AS5 and Noctua are also non-conductive. MX-4's trick card is that it's non-capacitative as well being non-conductive. You could apply it like a monkey shat it out on the die and still be good. Hell, I don't even replace the old paste and it's fine.

[aisle9]

The difference between best and worst amongst the higher-end stuff is 1-2C, maybe 3C. If you're using the envelope of Super Mega Thermal Paste that came with your $10 CPU cooler from China, well, that might be a different story. I just go with MX-4 because it's readily available, it performs admirably, it doesn't break down as readily as other pastes seem to, and it's not electronically capacitive at all.

 

Where thermal paste really makes a difference is delidding. Using liquid metal underneath a 212 EVO will do nothing but void your chip's warranty, buy you a degree or two and maybe frag your whole system if you're lucky. Delidding to add MX-4 or whatever Linus used will do nothing but void your warranty, but delidding to use a liquid metal paste can yield incredible results.

[manikyath]

4 minutes ago, SageOfSpice said:

Sorry, but LIQUID METAL coming below AS5?

 

That's the most rigged list I've ever seen.

it is not "rigged"

 

but i wont claim it was a propper test either. i mean SURE mayo is not as good as MX-4 but it is to prove a point: it matters so little that you may as well just dont care what you put in there.

 

ignoring no thermal compound and chocolate (which.. why did they even try...) the difference between the lowest crappiest paste and the best one out the bunch is absolutely minimal.

 

EDIT: actually, the only good quote ever from my local computer shop:

 

"the difference between the best good thermal paste, and the absolute worst good thermal paste is a few degrees. it is not worth paying 3x as much."

[aisle9]

4 minutes ago, SageOfSpice said:

Sorry, but LIQUID METAL coming below AS5?

 

That's the most rigged list I've ever seen.

Nope. If used on the CPU in contact with the heat spreader, liquid metal is marginally better, at best, than a high-quality standard paste. Delidding is where liquid metal shines.

4 minutes ago, Delicieuxz said:

When I get a new GPU, I'm going to replace the stock paste with it, and hopefully shave a few degrees celsius off the operating temperature.

Proceed with caution. IC Diamond scratched the crap out of my 4790K. I only use it if I know that I will never, ever expose the CPU the PC in question again.

[SageOfSpice]

2 minutes ago, aisle9 said:

Nope. If used on the CPU in contact with the heat spreader, liquid metal is marginally better, at best, than a high-quality standard paste. Delidding is where liquid metal shines.

Proceed with caution. IC Diamond scratched the crap out of my 4790K. I only use it if I know that I will never, ever expose the CPU the PC in question again.

How does that work? The thermal conductivity of liquid metal is INSANE. The flow characteristics are INSANE.

 

Yeah, I can see them being similar, but there should be absolutely NO way that AS5 beats liquid metal. No way.

[aisle9]

2 minutes ago, SageOfSpice said:

How does that work? The thermal conductivity of liquid metal is INSANE. The flow characteristics are INSANE.

 

Yeah, I can see them being similar, but there should be absolutely NO way that AS5 beats liquid metal. No way.

I don't fully understand it either, but that's not the first time I've seen that result. I'd speculate that it has something to do with how finicky the application of liquid metal pastes is, and how easy it is to screw up.

[Delicieuxz]

4 minutes ago, aisle9 said:

Proceed with caution. IC Diamond scratched the crap out of my 4790K. I only use it if I know that I will never, ever expose the CPU the PC in question again.

Thanks for the head up. I read that putting the tube of paste in a plastic bag, and submerging that bag in hot water for a bit before applying the paste will lower its viscosity, and allow it to spread more easily. I hope that doing this will also protect against scratches.

[manikyath]

4 minutes ago, SageOfSpice said:

How does that work? The thermal conductivity of liquid metal is INSANE. The flow characteristics are INSANE.

 

Yeah, I can see them being similar, but there should be absolutely NO way that AS5 beats liquid metal. No way.

realise that on the other side of that die there's regular normal thermal paste.

 

you're basicly using perfect surface silver audio cables, plugged into a coathanger.

[Vengeance_K1ng]

MX4 is what i use,  This is idle with Speed Step for the clocks.

 

[Zyndo]

MX-4 and AS5 are two reasonably cheap and reliable thermal pastes. I personally use AS5 atm, and overclocks on a 6700k are fine temps. Linus has mentioned in the past that his favorite is IC diamond, but like many others are saying in this thread, the difference is 1-3 degrees at most (unless you buy SUPER cheap paste).

 

One thing to note is I've heard that AS5 has a "cure" time. so if you do a lot of swapping, AS5 probably isn't the way to go (it can be 2 or so degrees higher until this cure time is over). but other than that, they're all pretty much the same.

[SageOfSpice]

5 minutes ago, manikyath said:

realise that on the other side of that die there's regular normal thermal paste.

 

you're basicly using perfect surface silver audio cables, plugged into a coathanger.

That's an irrelevant point though, they aren't swapping the paste under the lid for the AS5 test, right?

 

I actually answered my own question. I went and checked the composition of Liquid Ultra, and it's composed mostly of gallium. Gallium has a thermal conductivity that is 1/10th that of silver. So despite having the advantage in surface area and purity, it does actually lose in it's ability to transfer heat.

 

I feel stupid.

[manikyath]

1 minute ago, SageOfSpice said:

That's an irrelevant point though, they aren't swapping the paste under the lid for the AS5 test, right?

 

I actually answered my own question. I went and checked the composition of Liquid Ultra, and it's composed mostly of gallium. Gallium has a thermal conductivity that is 1/10th that of silver. So despite having the advantage in surface area and purity, it does actually lose in it's ability to transfer heat.

 

I feel stupid.

the thing i'm getting at here is, whatever you put on top of that heatspreader (especially towards the high end) is extremely limited by what's underneath the heatspreader. you cant get more perfect than what intel decided to stuff in there.

 

your amazing liquid metal paste is a 5 lane highway, the AS5 is a 4 lane highway, the paste under the lid is a city road.

 

for as much lanes as you add behind that city road, it'll eventually just reduce into margin of error, resutling into an imperfect result.