Thermal Compound Test

So I was testing thermal compounds and bought the most prominent ones and tested every single one of them on my P375SM-A. Here are the contenders, I've also added 2 cheap ones to the mix:


So here are some information about every single Thermal compound and how they were:
Coolermaster htk-002
Now this one was really really strange compared to the other ones. It was really liquiddy, almost like water. It also had a rather interesting color which was white. And I don't mean light grey, I mean white, as in paper. Overall if your heatsink fit is not at least perfect, this thermal compound won't work at all. Very easy to apply and clean off.

- Thermal Conductivity 0.8 watts/meter-C

Artic cooling MX-2
This one actually had the perfect liquidity it was very easy to use and it seemed perfect for most heatsinks, color was typical grey. Easy to apply and clean off.

Thermal Conductivity 5.6 W/(mK)


Noctua Nt-H1
The noctua is a little bit on the thicker side making compatibilty with most systems, it has a more dark tone to it and was rather easy to apply and clean off.

Uknown Thermal Conductivity

IC Diamond
This thermal compound is like noctua on the darker side, however the thickness of this paste is insane. It's perfect for terrible heatsink fits, however applying it is a nightmare, spreading the paste is really really difficult, cleaning is is harder than other pasts however not to bad.

Thermal Conductivity 8.5 W/(mK)

Gelid GC-Extreme:
Pretty much like MX2, easy to apply and to clean, it was slightly thicker but no issues at all. Pastecolor is the typical light grey.

Thermal Conductivity 8.5 W/(mK)


Thermal grizzly
Light in color, and similair to Artic MX-2, easy to apply, cleaning is also very easy and compatibility with almost all kind of heatsinks, be it tight or bad (not terrible).

Thermal Conductivity 12.5 W/(mK)


Coolaboratory Liquid Pro
The first liquid metal to be relased. The tube is annoyingly terribly designed and getting liquid metal out of the tube is a main in the bum. Cleaning it is also a nightmare and applying it is also rather hard. It takes quite a while to spread because the liquid metal doens't like to stick on the surface of the heatsink or CPU die.

Thermal Conductivity 32.6 W/mk


Thermal Grizzly conduconaut
Pretty much the same as Coolaboratory, however the tube is much much better so chances of spilling liquid metal isn't as high, applying it is the same hassle, however removing it was easier, still a pain.

Thermal Conductivity 73 W/mk


Results:
I only did stresstest with XTU for 10minutes and ran it idle for a little time to give an idea what we got. I did run it 2x for each tim to ensure a somewhat reliable result.

Specs:
P375SM-A, 4940MX, 4ghz, 1.050Vcore ~60W TDP

Idle results:
Coolermaster htk-002: min 39c, avg 43c, max 45c
Arctic MX-2: min 34, avg 35, max 38
Noctua NT-H1: min 34, avg 35, max 38
ICDiamond: min 33, avg 35, max 37
Gelid CG Extreme min 34, avg 35, max 38
Thermal Grizzly min 33, avg 35, max 37
Liquid Pro: min 33, avg 35, max 37
Conduconaut: min 33, avg35, max 38

Now take these results with a grain of salt. Except for the coolermaster every tim basicially performed about the same. I already expected this kind from coolermaster since it really is incredibly fluid, I don't think it makes contacts on all areas of the CPU.

Now lets get to the interesting part:

XTU 10min Stresstest:
Coolermaster htk-002: avg 96c, max 98c Thermal throttle to 3.2ghz
Arctic MX-2: avg 79, max 87
Noctua NT-H1: avg 79, max 86
ICDiamond: avg 76, max 81
Gelid CG Extreme avg 74, max 79
Thermal Grizzly avg 77, max 82
Liquid Pro: min avg 78, max 85
Conduconaut: avg 77, max 85

Now this is interesting,I don't know why and how but for some reason the Gelid CG Extreme did perform noticable better than all the other thermal compounds, followed by ICDiamond and then the Liquid metals. All in all all temps were really realy similair to each other. However Gelid was the winner, never breaching 80c on any core, which was suprising, since I was thinking that either ICD would win or the liquid metals. Never used this paste before, but I guess I'll use it in the future.

 

Great info, thanks for sharing! Did you by any chance take pics of each application before you installed the heatsink and after you pulled it off? I'd be curious to see the amounts you applied and the shapes of the resulting 'squish' after taking the heatsink off.

 

I didn't really do that, all looked rather similair tho, I took the paste and spread it over the CPU so there is a thin layer on the entire die. When pressing the heatsink on the die, then it pushes the thermal paste to the sides, only a very thin layer will be left. Which is between the CPU and heatsink.

I didn't deem it neccecarry to make a picture of each application since I do know what I'm doing. However I suppose I could make pictures about it, since people do some weird shapes such as a dot in the middle or an x, which is kinda bad when you do direcly on the die, as opposed to a heatspreader.

 

Ok, yeah that's kinda what I was wondering. It seems like everyone has a theory on how to best apply paste (thick/thin/x/dot, etc). I do the thin layer as well since, in my experience, the other methods usually end up squeezing too much out the sides and makes a mess.

Aside from all that, I'm surprised the metals didn't have a noticeable performance difference since they are hyped as the superior choice when choosing paste.

 

It does make sense tho. My fit is good, but it's not perfect, neither is the heatsink huge nor is the fan anything special. I'm pretty sure liquid metal would destroy on desktop PC's, while on notebooks it isn't really worth it.

 

question concerning the metal pastes: did you apply them on both the cpu and the corresponding contact area on the heatsink? cuz thats required for optimal fit, just putting a thin layer on one side (cpu) is not sufficient to make proper contact due to the high fluidity and thus very superthin layer of liquid metal tim when spread out.

ICD being a tad worse than GC Extreme doesnt surprise me, GCE is definitely the superior paste. on bad contact heatsinks, however, ICD is king due to its high viscosity.

what really surprised me, aside from the above discussed liquid metal pastes, is the comparably bad performance of kryonaut! it it actually kryonaut that u tested? or was it either hydronaut or aeronaut? cuz u only mention Thermal Grizzly (manufacturer), but not the name of the paste. im assuming it is kryonaut due to the thermal conductivity of 12.5 W/mK that you stated.

In any case, based on personal experience and reviews, Kryonaut performs a tad better than GCE but should definitely not be 5C worse something must be going on there, I either suspect a really bad heatsink fit or improper / varying applications between the pastes.

Ure running bare die, correct? in that case id recommend the line method in the center of the chip along its long side. applying a thin layer across a chip has the disadvantage of creating a lot of air bubbles in the paste, thus causing higher thermal resistance. for larger surface areas such as an IHS i can recommend the X / cross method for complete coverage. Dot in the middle might be good for saving amounts of thermal paste but in the end it doesnt properly cover the far edges of the chip...

 

Yes, if you don't apply LM on both heatsink and die, the temps are out of control. As you can see on the photo, it's kryonaut alright. The fit isn't bad, it's above average but not optimized with lapping. Also I did a little mistake, the max temp was wrong. Whoops.

I did test x, dot and thin layer. All had same performance, thin layer does the least amount of mess. When doing a dot or especially an x, you have so much wasted paste.

But same here, I was kinda suprised it did worse than gelid and ICD, I expected it not to be worse, but i tested it twice, and twice it was worse. So there is that :/

 

the picture in the OP just shows a black syringe with Thermal Grizzly lettering on it thats why i asked.

so in my experience so far:

- ICD provides on average around 2-3 C higher temps than GCE
- GCE provides on average around 0.5-1 C higher temps than Kryonaut
- Kryonaut again is about 3-10 C worse than CLU / Conductonaut / Liquid Metal TIM (depending on heatsink fit and stress test applied)

 

Well it could be a fit thing really. As stated it's more or less a stock heatsink, no lapping etc. If anything I guess we can take that kryonaut is better on notebooks with a really good fit. I'll order another minitube and do the test again. The tube is quite old and maybe something worsen it, idk. I'll update.

I've never seen a huge improvement from liquid metal over pastes such as kryonaut, not even on my MSI GT 72 which has a very very good fit. Kryonaut performed about the same as liquid metal on all notebooks i've tested it on.

At first I was mad because I thought I did something wrong, but after doing research it appears I wasn't the only one who didn't see any improvement compared to Kryonaut when using Liquid metal.

 

Nicely done. I think forum mods should consider making this post a sticky - http://forum.notebookreview.com/thr...e-recommendations-update-thermal-pads.796820/

and then add the findings here so everything's in one place.

 

That would explain my rather cool (hehe) temperatures, as I've used Gelid GC Extreme.

 

GGC-E has worked about as well as ICD, oftentimes slightly better. Between those and Kryonaut, as far as non-metal/conductive TIMs go, you really can't go wrong.

 

Depends, On my P375SM-A and P157SM when the GPU heatsinks were stock, ICD has performed by far the best due to a abysmally bad heatsink fit. If anything any high quality thermal compound will do absolutely fine given that the heatsink fit isn't abysmal.

 

That's more or less true. ICD's viscosity can be helpful when there's a variable like poor HS fit. I was wondering, though─

Did this do anything for you over any other method (pea/BB, or cross method)? I can't imagine it was worth the trouble.

 

Never tried to be honest, I just like to spread it since it makes the least amount of mess. But yeah for all people care you can just use pea method and be over with it. Performance is the same either way, you'll have to do way more cleaning when repasting tho.

 

This supports my theory that LM needs very close tolerances to be fully effective.
If your HS ain't installed 99% aligned it sucks. Almost need pressure paper and some play to get it to work.

For ease of use and price, Gelid FTW. Still my favorite TIM.

 

I'm curious how these temperatures would change after the thermal pastes "cure". As most usually increase efficiency after they set.

 

Good point. My Conduconaut finally cured after a few weeks and now runs 5c cooler at idle. (Max is still high 80's though, ugh)

 

thats weird, conductonaut doesnt have a curing time iirc. same goes for basically all current high end pastes, including CLU, kryonaut, GCE and ICD....

 

It could just be an odd result since the machine is new. Might be anything from the paint coating on the cooling fins curing to the fan bushings wearing in and gaining a few rpm. Either way, I'm happy they came down.

 

correct, I have also not experiences any improvements over time.

 

Why does no one test Arctic Cooling MX-4?

I love the stuff, as it's one of the few pastes I've tried that doesn't dry out after a year or three.


Wow my sig is so out of date... lol

 

It is a good compound. Low mounting pressure tests have shown it's not far off from ICD or GCE in performance.

 

"not far off" is relative though, especially since 1 C difference between pastes is already quite big. so with an overall range of like 10C between absolute worst paste and best liquid metal, being in the middle of the pack (mx2, mx4) would "only" result in like 5C difference or less, depending on mounting pressure.

Sent from my Xiaomi Mi Max 2 (Oxygen) using Tapatalk

 

It degrades within 1-3 months in high cooling systems. Temps spiked 15+C when I used it with oc 680m SLI.

Bought Gelid GC-Extreme and have't seen it degrade so far. Good lasting performance at least 6 months with those oc 680m sli and oc 980m sli. I think it is safe to say it will last at least 1 year.
Even liquid metal started to solidify due to low mounting pressure.

Sent from my SM-A300FU using Tapatalk

 

I've never heard of any legitimate TIM degrading that quickly, regardless of cooling system. Sounds strange...

 

Both my Kryonaut and my Phobya Nanogrease Extreme degraded rapidly in my GT73VR.
But only 2 cores. The cores with less pressure. Weak static pressure heatsinks+imbalanced pressure=rapid degradation on direct die (high thermal stresses).
Fixed this by sanding the heatsink, 0.5mm thermal pads then switching to Conductonaut (grin) later.

 

I probably used old MX-4, but it still performed good after a repaste. Temps just went through the roof after few months of heavy usage and oc made it occur even sooner.

 

Tim don't degrade that fast.

 

Well it did. Must have been an old batch.

Sent from my SM-A300FU using Tapatalk

 

Probably they do have dates when they expire.

 

Yeah, 3M and Dow pastes have use by dates. I'm a little surprised the consumer TIM doesn't.

Well for reference, I'm using a 3 year old tube of MX4 on some stuff right now. I'm a weirdo and I've always been a thin film of paste install guy.

That T61, is running Windows 10 Pro, 1709, I just did my taxes on it, and without the undervolting NVS140M BIOS on it, it was installing Windows Update, and I had the cooler base off, it was idling around 40'C and peaking around 55'C.

I have an E8200 running at 3.23Ghz full load, after 4 days of full load, the cores are reporting 50-51C, with the fans strapped to the TRUE-120 running at less than 800RPM. I scrapped off the rock hard stock paste on a Asus HD 6850, bumped the core to 950Mhz, thrashed it for 2 days, and I was getting weirded out by the fan not revving over 40% and holding the temp at under 60'C. I did such a dirty lapping on it's IHS, I used 180grit emery cloth... So it's flat but not super smooth, nothing to catch a finger nail on.

That's running right now. It's running my wallet and a Monero miner. The CPU has been mounted for almost a month and a half. I did the HSF on a RX 560, clocked it to 1300/1900 beating on it, full load 55'C. The R7 370 I replaced the fan on, clocked at 1050/1500, runs at 70'C on it's own curve and closer to 60'C with the fan set at 60%.

These are nasty environments, and the MX-4 is old. I will update if the temps spike, on anything. I used to review TIMs back in the day, and I always hated how even using the exact same hardware getting repeatable results in mountings, was such a *****.

It's still shocking to me how many people use Arctic Silver 5, it was the grease that was tested by the US gov't and found that it wasn't 9w/mk, but actually 0.89w/mk.

I'm actually ordering some Coolaboratory Ultra, to put under a water block an FX. Maybe I should test MX-4 first and then see how much better the semi-metal is. I've lapped the FX flat and to 2000grit, so it's flat and smooth.

 

Conduconaut is better than ultra. imo, easier to clean and seems slightly more effective.

 

Do you plan on testing MX4 @Danishblunt ? I have a new tube sitting here I've been meaning to use. I don't trust Dell's factory TIM to last very long really.

 

I didn't plan on doing further testing. What my testing concludes is, that it's not only the tim thats important, but also the heatsink fit. So your results will vary from mine, however, when using a high quality tim, you can basicially expect around the same performance. So go for the MX4 if thats what you can get easlily.

 

I have found the thermal grizzly metal to be more consistent in composition too. I had an entire tube of ultra almost as thin as water.

 

Someone down to test a watercooled heatsink ?
https://item.taobao.com/item.htm?sp...zSMYk&id=561089705878&ns=1&abbucket=13#detail

 

Looks like a clean mod. I like it!

 

Dad-gummit, that looks like about 500% worth of awesome right there. I wish that would fit my P870DM-G. Maybe @Ted@HIDevolution and @Donald@HIDevolution can look into having that as an add-on upgrade. That would be sweet.

 

What's the non water cooled performance like with only two heat pipes per GPU?

 

wow! that kit seems amazing

they make a kit also for the p650/p670
https://item.taobao.com/item.htm?sp...952ddqWTo7r&id=561276241511&scene=taobao_shop

 

P751tm/p751dm2
https://item.taobao.com/item.htm?sp...422&pvid=8aed78c0-9970-4162-961e-92b0cc3fa40e






P775TM
https://item.taobao.com/item.htm?sp...001&pvid=8aed78c0-9970-4162-961e-92b0cc3fa40e

 

Yep their kit look very nice! These watercoolings deserve to be an option for new systems. What i like is that with that system we can have choice with air cooling, watercooling or both.
i ll buy it if i had 1080sli. Or a fresh Ampere mod in 870KMG

 

http://forum.notebookreview.com/thr...ers-welcome-too.810490/page-215#post-10712157

 

whoa thats some crazy clean modding right there, amazing! id love to see detailed reviews on that including plugged vs unplugged and comparison with stock heatsink

so what would be outside of the machine? just the pump? and what kinda plugs are we talking about here? is it plug/unplug and go like the asus GX colostomy bag series?

Sent from my Xiaomi Mi Max 2 (Oxygen) using Tapatalk

 

I'd love to see detailed reviews too

 

I don’t like the idea that laptops being more like a watercooled Asusbook. Still headroom for making a better normal air cooling. I don’t want have to drag with me an extra bag with the water-cooling dock everywhere I go.

 

we have 3 fans in p870, micro pump allready exist so im sure it's possible to fit a reservor+micropump instead of one fan.
And using other both fans for cooling it

 

is this a p150/p170?
https://item.taobao.com/item.htm?sp...952ddjI3e6y&id=563953904536&scene=taobao_shop


edit
https://www.notebookcheck.net/Review-Schenker-XMG-P502-PRO-Clevo-P150EM-Notebook.72895.0.html


https://www.notebookcheck.net/Review-Clevo-P170EM-Notebook.73442.0.html

 

u gotta be kidding me! they even offer that solution for the P1 series?!?! no way!

@Papusan thats exactly why id be interested in the unplugged performance of this mod. if u get better temps both with plugged and unplugged state then this would be very nice indeed

Sent from my Xiaomi Mi Max 2 (Oxygen) using Tapatalk