13900k Best Thermal Paste Method

[Arthaei]

Hi chaps I’ve got some Artic MX-6 paste I intend to use on a new build I’m doing shortly, just wondering with the more rectangular shape of the Intel 13th gen what paste method works best.

I’m thinking of doing a pea or possible a smaller X in the centre. But a bit baffled some are pasting it over the whole thing or actually rubbing it all over with a finger.

A few YouTube vids are saying they’re all within 0.8c of each other, but wondering what you guys would recommend? Using it in conjunction with the Arctic Liquid Freezer II 360 cooler.

Cheers

 

[jigger]

I use as little as possible spread out as evenly as possible over the CPU. Something like a bank card is ideal. Would avoid using fingers.

 

[Armageus]

A few YouTube vids are saying they’re all within 0.8c of each other

Exactly this - regardless of what you do, it won't make much of a difference.

Personally I've always use the spread method as above - shiny cardboard folded in half is ideal as a spreader

Another couple of ideas:
- Noctua recommend a big dot in the middle, and four little dots at the corners

https://noctua.at/pub/media/blfa_files/manual/noctua_nh_l9i-17xx_chromax.black_manual_en_web.pdf

or

- Thin line lengthways along the middle of the die.

 

[Arthaei]

Ooh the Noctua one sounds like it would work well! I might give that a go or spread as mentioned. I know from other applications that Arctic themselves simply recommend pea method so it’s interesting. X also seems popular.

 

[Dave2150]

Make sure you use plenty, as the IHS is really huge. I manually spread MX6 as thin as I can over the IHS. It's excellent paste and easy to spread, so should be fine.

 

[MartinPrince]

Ooh the Noctua one sounds like it would work well! I might give that a go or spread as mentioned. I know from other applications that Arctic themselves simply recommend pea method so it’s interesting. X also seems popular.

If you see any old videos of thermal paste spread patterns after the pressure of a heatsink is applied you will see that the pea in the centre leaves the corners without paste, which is why for years I've used the Noctua method or variation of (large dot in the middle and 4 dots in the corner) for years, even before Noctua started recommending it. Spreading it with another utensil always seems like a bit of a waste to me as you always end up with wasted paste on the utensil itself.

 

[SkeeterUK]

Cling film around ur finger can work too when spreading i have heard.

But spreading is not the method im gona use as the best i think has always been pea sized and let the hsf spread it out. When i get around to finishing my 7600x setup.

About the corners, well heat isnt generated from there so not exactly needed, u just want the paste to fill the micro holes on the surface to help the contact, and the heat comes from the chips under the plate which is the centre and underneath for chiplet ones. So long as them bits are covered it should be ok (i think).

 

[Arthaei]

Thanks all. I think I’m going to go for the 5 dot method. My wife is a scientist and agreed it would work well as she works in manufacturing. She said the buttered toast/spreading method sounds likely to create air bubbles which obviously raises temps.

Cheers!

 

[Rroff]

With all due or undue respect ignore any advice which involves manually spreading it (most modern pastes are not designed to be manually spread although the last few years they've made them more tolerant of it as people keep persisting with it, while much older pastes used to be designed to be manually spread).

The optimal method depends on several factors and in many cases the difference is small 1-2C at most but some methods can result in bigger inefficiencies or the paste lasting a shorter length of time before needing to be reapplied. Also many people doing tests for this don't take into account that some pastes need multiple thermal cycles to reach best results while others are at their best instantly.

A small blob and letting the mounting pressure spread it you can rarely go wrong - it might not always produce the best most possible optimal result but it will almost if not always produce a good result.

After that read the application guide for the relevant paste, in this case https://www.youtube.com/watch?v=m7x2sUt0mqo though I'd say they put slightly too much on there.

The best application with the latest Intel CPUs is a short line lengthwise along the core using a little less paste but about the same in total as the video (Google 13900K delid to see the layout), for Ryzens you want to use the 5 dot method as their packages can be spread out.

The ARCTIC MX-6 is a carbon filler-based thermal paste and reaches its maximum performance without burn-in, directly after application and distribution. It achieves all of this without any reputed ingredients, such as expensive diamond dust or various precious metal particles. The silicone gel used as a carrier enables optimal distribution, which is achieved through even contact pressure, for example when mounting the heat sink.

EDIT: Additional note - there are some specific applications where hand spreading is advised such as certain server sockets and some GPUs.

 

[Arthaei]

Thank you, I think I might still try the 5 dot but make the middle blob a bit more of a line perhaps. Cheers!

 

[cjgardens]

Just squeeze the full tube on it then tighten the screws as much as possible.

I use a the rice method. This is the way!

I've got a 12700k and same cooler. I used the rice for that like I always do and when I took the cooler off to go full custom water ithe CPU was fully covered

 

[MartinPrince]

Thank you, I think I might still try the 5 dot but make the middle blob a bit more of a line perhaps. Cheers!

Yes, this is exactly what I do on Raptor Lake CPUs.

 

[uscool]

always spread keeps my OCD in check knowing its fully covered also stops me wasting too much

 

[Arthaei]

It’s funny seeing no-one can agree! I think as @Rroff says a lot of YouTubers don’t test for how it works evenly after multiple cycles which it’s designed to do as it spreads itself out with the heat generated.

So all the tests with a clear bit of plastic pressed down aren’t showing what actually happens over time and use.

And manually spreading it creates bubbles so it sounds as though it would appear covered however it’s not going to give best performance.

I’m going to stick with a decent line for Intel socket by the look of the die and a couple of tiny dots in the corners I think.

Thanks all

 

[Dave2150]

On LGA 1700 it's very unlikely you'll get as good results from a line, or 5 dot method, verses manually spreading the paste. The IHS is just too big. Experts/champion overclockers such as der8auer use the manual spreading method, regardless of the paste used.

Video is old, though shows the method >

If you look through his 12th, 13th gen or Ryzen videos, you'll always see the paste has been manually spread over the IHS. He's using the thermal grizzly paste, but same is applicable for arctic silver, MX6, whichever paste.

You'll get armchair "experts" here saying otherwise, though be wary as they don't know as much as der8auer

 

[Dave2150]

It’s funny seeing no-one can agree! I think as @Rroff says a lot of YouTubers don’t test for how it works evenly after multiple cycles which it’s designed to do as it spreads itself out with the heat generated.

So all the tests with a clear bit of plastic pressed down aren’t showing what actually happens over time and use.

And manually spreading it creates bubbles so it sounds as though it would appear covered however it’s not going to give best performance.

I’m going to stick with a decent line for Intel socket by the look of the die and a couple of tiny dots in the corners I think.

Thanks all

Each to his own I guess, mean we are talking only a few C difference - though with 13th gen and Ryzen 7000 series, even 1C difference means the difference between thermal throttling or not, so there's real performance left on the table by using the lazy dot/line method

 

[SkeeterUK]

End of the day, do what you want, you can always repaste a diff way if you fancy. Yes it means redoing the cooler etc but you do have the option to try.

 

[jigger]

Each to his own I guess, mean we are talking only a few C difference - though with 13th gen and Ryzen 7000 series, even 1C difference means the difference between thermal throttling or not, so there's real performance left on the table by using the lazy dot/line method

Probably not even a few degrees with a decent paste. Case selection is and system config is going to make the biggest difference.

The only mildly significant differences I’ve ever seen is between applying too much paste or just enough. Liquid metal is maybe the one single exception, but it’s a PIA.

As a principled man, I like to smear. pfft, dot and dab, draw lines. Men smear!

 

[Dave2150]

End of the day, do what you want, you can always repaste a diff way if you fancy. Yes it means redoing the cooler etc but you do have the option to try.

Yeap, removing the cooler is a good shout and will show how good/bad your application method is.

 

[Arthaei]

You guys are hilarious I have no idea what method to do now lol. I think @Rroff has made a very valid point which is that NONE of these videos show what ACTUALLY happens after the pc has been on and off a number of times actually running and heating. They just show how it spreads under glass/plastic COLD. So even though the 'pea' method and others initially don't appear 'the best' it doesn't actually show what happens with them over a number of cycles. Surely that is more important as the heat spreads it out further...?

Also...
Straight from Intel themselves:

"It might be tempting to spread the thermal paste onto the CPU yourself. We recommend letting the pressure from the base-plate or waterblock being installed do it for you. Incorrect manual application can cause air bubbles to form in the paste, which can negatively impact the thermal conductivity."

Is that just old hat now relevant to stock cooler?