Thermal Paste Comparison - Eleven Compounds and One Liquid Metal
Вставка
- Опубліковано 28 тра 2024
- This thermal paste comparison is based on AMD Ryzen 9 7900X3D since in majority of my benchmarks I am using this CPU. Regarding that, through AIDA64 and Cinebench R23 we can notice some differences in terms of clock speeds, points scored and other segments which will give us the end results of eleven thermal compounds and one liquid metal.
Where to buy:
►Noctua NT-H2: amzn.to/3PnEMLQ
►Thermal Grizzly Kryonaut Extreme: geni.us/gYH5eA
►Arctic MX-6: geni.us/Nsg82S
@LevelUPGamingTech Thermal Paste War video on Intel CPU: • Thermal Paste Wars 🤘- ...
SUGGESTED PRODUCTS:
►CableMod Cables: geni.us/TAYTe
►GT Omega Chairs: bit.ly/GTOmegaChairs
►UPerfect Monitors: bit.ly/uperfectmonitors
►MOZA Racing Gear: bit.ly/MOZARacingGear
►Subscribe here: bit.ly/3erJB33
►Join my Membership to get access to perks:
/ pccrazydario
Chapters:
00:00 - 1:57 Introduction
1:58 - 3:00 - Benchtable and components
3:01 - 4:29 - Thermal Pastes Used
4:30 - 9:43 - Specifications of Compounds
9:44 - 14:01 - AIDA64 Comparison
14:02 - 18:32 - Cinebench R23 Comparison
18:33 - 19:42 - Compounds in Segments
19:43 - 21:15 - Winners
21:16 - 23:12 - Final Thoughts
My recording gear:
►Camera (video): Blackmagic PCC 6K Pro: geni.us/yZlAn
►Lenses: Samyang CineLens VDSLR MK2 kit: geni.us/nUoaK
►External monitor: ATOMOS SHINOBI: geni.us/IFGQ25k
►Storage: AngelBird Match Pack Blackmagic PCC 6K PRO: geni.us/0bQA
►Lighting System: NANLITE FS-150: geni.us/09WC3A
►Softboxes: NANLITE SB-PR-90-Q: geni.us/UT0RA
►RGB Tubes: NANLITE PavoTube II 6C: geni.us/HHd76
Music by Epidemic Sounds:
►Lofive - Yhello
NOTICE: Amazon links provided are linked to my Amazon Affiliate account and support the channel at no extra cost to you.
#noctua #thermalgrizzly #arctic - Наука та технологія
Great to see this video and round-up of current products. Thanks for your time and effort.
Thank you very much for the comment! Appreciate it! 🙌
Really helpful video. Thanks for taking the time to do it.
Glad it helped and thank you for the comment. :)
Only issue with this type of testing is you really need to run the system for around a week before you get a true result on the pastes. Not faulting the tester, it's just some don't "settle in" until they've been run for a while.
Imagine doing the tests for 11/12 weeks+ when you have only one set of components... Unfortunately, I don't have extra components to leave this set to lay around for these benchmarks. Bench table is already taken apart as I need the parts for other stuff sadly. But for some, I do agree even though in most scenarios, thermals would vary for 1-2 degrees in a week, while in 6 months, 3-5 degrees... Or even in this scenario, clock speed and scores.
@@PCCrazyDario Don't be hard on yourself, it's already an amazing video! however he has a point some paste degrade but we can't really bench as it would be expensive (time/money etc..)
@@alsammanlepro96 Much appreciated, and yes I do agree on the part they degrade over time, each model differently. Just saying why it can't be done :D At least not in my studio...
Plus they all on same testing conditions. Even if in a month they get cooler or hotter
This isn't accurate. Only a paste like Artic Silver 5 requires a curing time. The vast majority of TIM on the market has ZERO curing time, especially all these mainstream ones shown here.
I love seeing how thermal pastes compare, but I usually still use Noctua NT-H1 for ease of use and tearing things back apart. Thumbs up!
Actually, everytime I was building my custom liquid cooled builds, I either used Noctua NT-H1 or Thermal Grizzly that came with the blocks. My last workstation is basically still running with the supplied Thermal Grizzly one. Past ones were on Noctua, no complaints. And thanks for the comment. :)
@@PCCrazyDario which on is better, NT-H1 or NT-H2?
cheers for posting results, i went down the thermal paste rabbit hole a few years ago with a gtx1080 that would pump out after 6 months. tried noctua and a couple other premium pastes all pumped out within 3-6 months, artic silver 5 was the only tim that didn't and held up without any performance degradation over time (1year+)
i've heard the same about TG and noctua paste having pump-out issue on GPUs, currently i am using noctua coz that's all i could find a the time (either that or TG stuff, everything else was low on stock locally = .=)... apart from the artic silver 5, any other suggestion for gpu to prevent/slow down the pump-out issue?
Thermalright tfx, no pump out after a few months for me @@askaliu2943
All the comments are just people asking for other pastes haha. Thank you for testing these! I’m trying to figure out what I want to use for my new build. :)
Haha well it is good for me because next time I don't have to research. I just write all these down, buy them and do the benchmarks. Thank you for the comment. :)
Noctua NT-H2 is well priced and has excellent longevity. I used to Cryptomine and used NT-H2 on all my 3090s. They ran for several years non-stop with no change in temperature. Two of them are still with me in gaming PCs and temps, still as good as ever. I was a bit surprised when I first started using it as I thought it might be too thin for direct die application, but it has been fantastic. With CPUs, it's a great TIM, and thanks to this review, everyone can see its performance.
I would recommend what I use, HY-510. The stuff works great and is so cheap that it's laughable. I've even used it on the exposed die of a Radeon RX 580 2048SP GPU (those make A LOT more heat than any CPU) and it never missed a beat. Even my old A8-3500M Llano APU (with a Radeon HD 6620G IGP), which also is an exposed die part, works perfectly to this day (and I use it every day).
Super video. Would be nice to check out the Prolimatech PK-3 thermal compound as well.
One off Best termal paste i try. Glad u Test it. Thermal Grizzly Kryonaut Extreme
Thanks for the comment. :) Glad you like it.
Interesting results. I've not been able to recreate a scenario were kryonaut extreme was beaten by anyone so far.
They were so close to each other that I would even take "the beating" seriously. If I could split the 1st place, Kryonaut Extreme and NT-H2.
MX4 has always been my go to.
Still does perform quite well. ;)
Really really nice video! Obrigado por compartilhar esse teste!🇧🇷🤝🏼
Thank you very much for the comment! Appreciate it! 😉🙌
great job 👍🏻
Thank you! Appreciate it. :)
You deserve a sub for this❤
Too bad you were not able to test KPX. Hands down the best thermal paste I have ever used in my 35 years of building PC's. I have been running KPX on both my PC, my wife and kids PCs without any drying out or temps getting warmer over time. Temps have stayed the same as they were on day one. Great video and information on some of the other great thermal pastes out there!
If I manage to get a hold of some, I will write it down for next time. Thanks for the feedback! And glad you liked the video. Much appreciated.
Jayz2cents says the same thing about kpx but I've recently seen him praising thermal grizzly since der8auer sent him some.
@@cactuarking4138that lasted maybe two videos and he went back to only using KPX. Its hands down the best thermal paste. Please check his videos to see him using KPX
@@cactuarking4138 This is where I lose respect for Jayz2cents... Thermal Grizzly isn't the best.. but it is very expensive. Seems like he sold out.
Kryosheet is nice because you never have to remount it, and it's consistent. Great for cooler testing, but, proper paste will always perform better.
NT-H2 is the king though, but man, it's pricy. Nice to see that MX-6 is actually an upgrade. MX-4 to MX-5 was pretty marginal.
Also, you can see some worthwhile gains by lapping your CPU/Heatsink up to mirror polish. It allows you to use less paste, which in turn lowers the thermal resistance of the whole system.
Man, I totally agree on the Kryosheet. Such clean way to apply it on the CPU. No excessive cleaning afterwards.
MX-4 was pretty good back in the day, MX-5 had some issues, but they fixed all of it with MX-6 without a doubt. NT-H2 is in pair and in some instances better than MX-6 as well as Thermal Grizzly Kryonaut Extreme.
Yes, I do agree on that, but it has to be done perfectly if I am not mistaking. And I didn't do that in the video because not every one has steady hands if you know where I am getting at. :)
@@PCCrazyDario definitely. Lapping is not for the faint of heart, and you need to adhere to the correct process or you'll make it worse, not better. Not to mention, it takes alot of time and this video already took a ton of time.
Either way, very good work all around man, I appreciate you.
I still have a large syringe of MX-2 that I got for pretty cheap. All of our pcs are running great with it.
Oh wow, I do remember it but only the tube not the actual thermals unfortunately and what CPU I used it on. :P Awesome!
I've used the MX-4 for years now as it's been the best bang got your buck thermal paste, but after seeing these results, I think i'll upgrade to MX-6 that seems to be the best bang for your buck thermal paste now.
Nice!! :)
I started using the square pad like this i don't have to clean and put new paste every year the pads with noctua NH-U12A push and pull, 3 pc 5600x, 2 5800x3d i have big cases and very good air flow for what i get it's only li 3c max over noctua paste and mx4. I'm going to try one on a asap rog strix 3070ti oc gamer gpu see if it works if is under 85c on the hot spot I'll start using the pads on my gpu to .
You can't overclock an X3D CPU anyway so stock was a good choice! 😁
These differences are all margin of error when it comes to clock speeds and scores. For 99% of people (especially gamers) just buy whatever's cheapest. Hell, I got some thermal compound off of eBay called HY-510 and I got (get ready for this...) THREE HUNDRED GRAMS (10 tubes of 30g each) for only $25CAD. To be fair, I think it was a bit of a fire sale because while HY-510 is still dirt-cheap, I've never seen it for a price as low as I paid before or since. The stuff works great. Here's what I've used it on:
CPU: R7-1700
CPU: R5-3600X
CPU: R9-3900X
CPU: R7-5700X
CPU: R7-5800X3D
GPU: RX 580 2048SP (exposed die)
APU: A8-3500M (exposed die)
Every single one of these items still works perfectly today after many years without changing the paste. Sure, I might not win any CineBench or overclocking competitions but for everyday use and high-end gaming, HY-510 is perfect on CPUs, GPUs and APUs.
There are some applications where longevity of a paste is preferred over a couple degrees in thermal performance. For server use or crypto mining a paste that holds up overtime is much more valuable.
Yes, indeed. That is correct. Server usage as well in general. But those people tend not to watch this type of videos.
Yeah most don’t care that much and buy what is cheapest, but a few of us find our way here :)
Kingpin KPx gives me an additional degree lower than H2 - would be nice to see that tested to see if my results stand up for others.
I totally understand, but not sure if it is available here to purchase. Maybe in some future, I will expand the benchmarks with others I haven't been able to cover. This already took too much time altogether. But thanks for the feedback.
Lol I still use Artic MX-4 been using it for quite a long time. Just been my go to paste.
It's 2024, Ptm 7950, Thermalright TFX?
They are in the store if I am not mistaking. Everyone has some thermal paste to add of course. If I had more time, there would be more compounds in this benchmark for sure
No Cooler Master CryoFuze? So far, this is the best paste I've used.
I literally have content to cover every day and this already took three days to build, test and record. Imagine if I added Gelid pastes, CryoFuze and others. 2 more days for sure. Of course, I would prefer having at least 20-30 pastes for benchmark to make it even more relevant, but time was of the essence. Hope you understand and that this is at least solid. :)
I’d be curious which paste lasts the longest on a cpu.
For me personally, that would be quite tricky to do. First of all, I have one or two sets of components for testing and builds in general and secondly, with those sets of parts it would take me personally minimum of 10-11 years to do that, taking into consideration that one goes minimum of 1 year. :)
I wonder how well the GC-Extreme would work under these conditions. That's what I use when I have to replace stock thermal paste.
Written down as well as others people suggested. Next time! ;)
Poorly, because a few months ago they changed its composition and its efficiency decreased. I would recommend trying the GC-4 from this company.
I am reminded of all the recommendations to replace the pre installed TIM on AIOs etc and it was actually Dow-Corning, a really good thermal paste 😂 So much snake oil in the PC community.
I think this is great info for most people. Unfortunately what I need to know is a bit different. My hobby is flying FPV drones. These are racing drones that we fly fast and crash. One of the components we use is called an ESC and on it are small chips called mosfets. These get hot and need cooling. An ESC I bought came with a heatsink which I needed to take off and put back and I don't know which thermal paste to use. The thermal paste they used was the only thing holding this heat sink on. No one seems to cover this topic. I have found some stuff called Heatsink Plaster on amazon along with a few other options but don't want to create a situation where I can remove it again if needed. I can't find any useful information about this topic so far. If anyone has any input or advice I would be very grateful.
I exactly know what you are talking about since my dad builds RC planes and has mosfets to cool down as well. So what he does, orders the passive heatsinks with the sticky thermal pads from AliExpress OR grabs a passive heatsink, applies thermal paste and creates some sort of a clips that hold the passive heatsink to the mosfet. I actually have him some old passive heatsinks I had, he cut it in the exact size he needed and it worked. Hope this gives you some ideas at least. Or there is a thermal sticky tape. Don't know how it is exactly called. Maybe something like this: www.3m.com/3M/en_US/p/c/electronics-components/electronics-films-tapes/thermally-conductive-tapes/
But be prepared when you do this, you have a slight chance when removing the passive heatink, you'll remove the mosfet as well because of the tape strength unless you heat it up a bit with a heat gun.
@@PCCrazyDario Thank you. I think that might work as a solution. I think if I cut the thermal tape to the right size it might work.
My main problem is you didn't manually spread the paste, pea size in the middle is inconsistent in coverage.
I guess I should have recorded the aftermath when removing the CPU block, because each time it was spread properly.
I've started switching over to the Thermal Grizzly Kryosheet so that I don't have paste dryout anymore.
From a thermal conductivity perspective, it's not the best, but again, the theory is once I have the sheet in and it is held in place under the retention force of the HSF mounting mechanism, I don't have to worry about it.
Next time, I will have to do a re-run of the Kryosheet, because I have a hunch that it might perform better plus yeah, totally agree. There is no mess, even if you apply proper amount of thermal paste, there is still cleaning that needs to be done regardless.
@@PCCrazyDario
Umm....it might -- it might not.
Unlike thermal paste, it won't really be able to fill in the differences in the micro surface finish gaps between the two surfaces (top of IHS and bottom of the HSF (where the two come into contact with each other).
Suppose that the average themral conductivity for thermal paste is ~ 8 W/m.K and let's assume that the thermal conductivity for the graphene Kyrosheet is 64 W/m.K.
With the paste, it should be able to better fill in that difference in the microsurface finish gaps, which means a larger area can be used for heat transfer.
Conversely, with the Kyrosheet, since it can't necessarily really fill in those gaps, so the area that is available for heat transfer may be smaller, but because of its higher thermal conductivity, it is able to transfer more thermal power from the top of the IHS into the contact patch/plate that's at the bottom of the HSF.
So in terms of temps, it might end up being the same, but you won't have the thermal performance degradation as a function of time like you would with thermal paste.
Bro Gigabyte B650E Aorus Elite X AX Ice or GIGABYTE X670 GAMING X AX and why pls help me chose
Honestly, not sure because I haven't worked on Gigabyte motherboards since Z390 I think. :/ Sorry.
Data is sampling is too short interval of a study. Please do month parallels to extend your research study to better accuracy as these pastes change overtime while in use.
If I had spare parts laying around and didn't need them for next builds, reviews and stuff, I would. This is just way too time consuming for one person with limited access to components. Most of the parts you see in this video came out of my own pocket.
Its a shame that none of that list contain some unknown china brand like coolmoon mt9 or LK-17 17w/m.k for example just to see how they actually live up to the claim.. Personally use the coolmoon as i bought the cpu fan and was performing well atm
You do know that thermal conductivity can’t be actually calculated and that it has been debunked. Unfortunately, I don’t know if I can actually buy Coolmoon thermal paste here where I live or even in near countries.
I got a 4-5c reduction to max temp and about 10c less in gaming going from mx-6 to liquid metal.
Maybe if I ran another benchmark, maybe it would, but it also depends on the hardware at some point. I also saw some had 1-2 degrees difference, some had 3-4 degrees difference. So I guess it varies.
@PCCrazyDario so many variables like cooler contact and ambient temp can alter results as well. Either way, I'm happy with my results, haha.
@@justin169169 Ah yes, definitely. Cooler and room temperature are also key factors. Totally agree. Glad you are! :)
And where is it thermalright?
Most likely in the store somewhere... As well as GELID and many others...
Nice that MX6 came with the Arctic Liquid Cooler III. No need for me to upgrade paste.
Without a doubt. I wouldn't definitely change that. ;)
Where did you go wrong - let me count the ways.
1) You are using a Ryzen CPU, this means that the two hotspots of the 3900X3D CPU - namely the chiplets - are not centred they are off towards the edge of the IHS in the direction of the GPU.
Thus, when you put a blob of thermal paste on the centre of the IHS you will not get proper coverage of those hotspots - or rather you cannot guarantee that you have coverage, and parts of the IHS under which the chiplets are located could quite well not have any coverage (the paste will spread in a circular pattern).
The edges of the IHS going away from the centre of the IHS will most certainly not have coverage, and heat that is conducted away from the chiplets towards the edge of the IHS will not be cooled.
So the testing could already be tainted not only by the methodology you employed, but also the amount that you deposited would dictate how far of a spread you would get with the pressure of your cooler.
The one and only method for applying TIM to a Ryzen CPU is to spread the paste over the entire IHS.
2) Unless you ran CineBench R23 at above normal priority for the ten-minute test, you can basically take your results and throw them into the garbage bin.
There are other Operating System components and also applications seeking CPU attention, and you want CineBench to have the highest priority when it comes to doing a benchmark run.
If you don't do this, then it is probable that the results will be skewed on whether you got a clean 10-min run with fewer interruptions or not.
Under no circumstances should monitoring software or vanity apps such as RGB software and the like be running during a benchmark run.
3) You are using a cooler that only has two points of contact for tightening it onto the IHS, this is not recommended for benchmark comparisons involving TIM because it is all too easy to not properly align the cooler head - again introducing uncertainty into the mix.
4) Last but by no means least, for the likes of Thermal Grizzly Kryonaut Extreme, I know for a fact that it does not reach its peak cooling performance until after about a week or two of actually using the PC. With the Arctic Liquid Freezer III 360 cooler I recently purchased (replacing the Arctic Liquid Freezer II 360 which I had been using), the difference between the temps on my CPU tested with CB R23 immediately after applying it and testing one and a half weeks later made a difference of six degrees Celsius.
Man your Kyrosheet benches could be off, the pad does tend to slide around upon mounting, use a tiny bit of paste to prevent it from shifting when mounting.
I did mention something in the video that it is my first time using Kryosheet and Condoctonaut, and that it might be my fault for the results.
Also, thanks for the advice. Good for future reference. :)
@@PCCrazyDario Yes I totally understand, I'll be on the lookout for any updates you might have.
@@sir1junior It won't be that soon, but there will be one for sure. Thank you!
This took a lot of work
Thanks for noticing that. Appreciate it and the comment 🤗
You show Conductonaut, have „one Liquid Metal“ in the title, and yet you forgot to use it and show data on it? Wtf?
Add pls next time thermalright paste, kingpin kpx and Alphacool apex
Thanks for the suggestions! I will definitely write them down for next time. 😉👌
You must have hear olso Prolimatech, PK25g, PK-2 Nano Aluminum,
No IC Diamond???
Will add it to the list since everyone has been suggesting other brands as well, so I will include it in the next one. 😉 Thanks for the recommendation!
Shame you did not test GD900 the best bang for bug Thermal paste on the planet. 30G for about 4 dollars on aliexpress, i am flipping pc's and use TP a lot so i gets very expensive with the other brands
Thanks for the suggestion! I'll keep it in mind for the next time. ;)
GD900-1 and GD-2 pastes are good, but for processors with TDP
I9 12900KS Arctic Freezer II 420 MM im pass to Arctic Mx6 to Thermal Grizzly Kryonaut Extreme 13 degrés gain cooler in cinebench R23 ❤
Pffff man, that is outstanding for sure. Thanks for the additional information!
@@PCCrazyDario thx im make this change at 3 days
I'll add my obligatory complaint...
You didn't test IC Diamond !
But seriously, that is my goto paste.
In my experience, it lasts 7+ years without dryout.
Will add it as well to the list. Damn, I have such a long list now. You guys think I’ll have the time 😅😅
@@PCCrazyDario500 tubes of paste later....
@@anonamouse5917 Hahaha something like that. :P
Get some cheap DG900 stuff is great value for performance
Adde on the list! :)
should have used intel cpu to test.. these ryzens have much lower power consumption and they run hot either way
I explain everything in the first few minutes…
the best results i get is with thermalright tf9
I chanched my 4year old paste 1-2°c diffrence
Some paste do that as well. Longetivity and almost the same thermals throughout their lifespan...
Doesn't really matter some just over price their paste
I would honestly say, it goes both ways. Some are really good and affordable, some are really bad and affordable and vice versa... Shame I forgot to add the price chart! Ugh... My bad... :/
MX-4 is my choice.
KingPin KPX
Written down for future comparison. Thanks!
MX 6 Tito, bar kod nas je 10 EUR 4 grama
Čovječe, €10... Kaos! Ako se ne varam, i Amazon de ima oko €4.5 cijenu.
try ptm7950 next time
Deal! Written down! Thanks...
There are other phase change materials such as Upsiren PCM-1, Gelid HeatPhase Ultra and Thermalright Heilos.
No kpx :(
I know, but this was already handful with my schedule since I do everything myself. Next time, I promise. :D
@PCCrazyDario lol I didn't expect you to it's just the one I was looking at buying I could only imagine how long it took u to do this I appreciate you taking the time to do what u did and responding to my silly comment
I still use good ol' Arctic Silver 5. Tried and true.
Oh man, nice. Totally forgot about that one.