You asked for it....NO THERMAL PASTE TEST ua-cam.com/video/VAh_p90y_DU/v-deo.html Remember to LIKE and SUBSCRIBE! Thank you everyone for the continued support!
Even though the results are probably alright it doesn't seem like you've controlled for power consumption/boost which casts a huge doubt on these results. Plus we're talking about a pretty low power consuption, you'll have a bigger differential with overclock.
Well, all this may be true for the IHS chips. But in laptops/notebooks with direct die contact partial spread can cause more problems, like not cooling properly some of the cores, or maybe even chipping the silicone with no material in between chip and heatsink...
You didn't use enough paste. I can clearly still see motherboard. Cover the entire motherboard, including the RAM. It wouldn't be a bad idea to cover the USB ports as well.
As someone who has never done this before, your conclusion has definitely eased my mind. Put some thermal paste on it, make sure it's enough, and everything will be fine. Definitely a load off. Thank you for the video.
Aparently putting more than enough can be bad as well as putting not enough, so good luck getting a good balance xd today and yesterday I changed my consoles paste and used different methods, somehow I still feel like they’re boiling :c
@@meappeam6917 Actually, adding too much thermal paste is a lot better than little or none, as it covers the same area as acquitte thermal paste application, so it doesn't matter that much. The worst that could happen in most is that it goes off of the IHS, which is not bad. Most extreme, basically impossible, is that it goes off the IHS into the CPU socket contacting the pins, though I've found sources claiming that it is still possible to normally run a PC with non-conductive thermal paste even in the CPU socket (this does not mean to attempt to put thermal paste in your CPU socket). Though, putting too little can defiantly impact performance significantly (as per seen in the video) If you feel as if your consoles are still "boiling", make sure you have a good thermal solution (I would recommend Arctic MX-6 or Thermal Grizzly Kryonaught), or perhaps look to get better / higher quality fans that would move more CF/m (Cubic Ft / minute)
Despite this video being two years old, it’s still comforting. I built my very first pc last week and forgot about all application methods for thermal paste… but I still used thermal paste!! And not just a tiny dot either! My heart is at ease knowing I did it right! 😂
Well done bro, I just built my first rig and I didn’t install the cooler properly and now it overheats. I ordere a new cooler and hope I dont mess it again
@@drakata27yeah same, using stock wraith stealth with the ryzen 5 5600 non x.. 95*C on full use…. it‘s so loud. Ordered a pure rock 2 today. Hope it‘s better than the awful loud stock one..
Not being a tech person, I'm very happy to find your video, and want to say: Finally! A thermal paste demo that actually clarifies the most important considerations and helps me understand what I'm doing! Thank you! :)
Pfft! Everyone knows the best way to apply thermal paste is to swallow a dollop and then regurgitate it onto the CPU lid. I call it the "Feedin' my baby bird" method.
Gotta admire your commitment in testing those thermal paste combos. I was just curious on how to put a thermal paste properly. Now that I watched your vid, I learned a lot. You exceeded my expectations.
Subscribing just for the amount of effort it must've took to make this video. Mounting/ unmounting applying thermal paste/ removing it with alcohol. 20+ testing 4 to 5 times Sheesh!! 😬 you have my respect 💯
This guy saying "It'll be fine" in that calming voice is very nice to hear as someone who's almost constantly afraid to break the whole computer with my sheer presence
Well I just bent pins on my 2700x removing the stupid wraith prism that had molded to the cpu with its thermal paste preapplication like the Thing making a new creature
old school trick, you get a plastic baggy, use it as a glove, apply paste to baggy covered finger tip and dab the paste on the chip evenly with a very light coating. The goal is to fill in valleys and provide good contact and not cake it on to the point where it acts more like insulation. Great video by the way! I guess my method is more like the buttered toast method.
well he wants you to think that it's thermal paste and not ask any questions and considering what else it could be I am just going to assume that it is thermal paste cause the alternative is just gross
I'm someone else who has not done this and will need to on my new build very soon. Thank you for the experiment! This definitely puts my mind at ease. It seems like there is so much controversy about this topic that makes people nervous that have never done it. When in reality its a simple process and no need to worry so much, just do it!
Let's be clear. The paste is not the primary medium for heat transfer. Metal-to-metal is still far more efficient. The paste (dielectric grease with suspended silver shavings) is used because it is orders of magnitude more efficient than AIR. You want the most/closest metal-to-metal contact for proper heat transfer, and the paste is there to fill in all of the microscopic gaps for the rest. Too little or uneven paste causes the cooler to ride on a "cushion" of paste and leaves air gaps that insulate rather than transfer heat. Too much paste is going to make for less-efficient transfer of heat. You want to have the thinnest possible layer of paste with the most even distribution. So the "buttered toast" method is important, but so is the THICKNESS of paste. Only so much can be pressed out under the spring load of the cooler. This was demonstrated with the "small dot" method. Once the tension of the cooler mount matched the resistance of the paste to compression, it stopped flowing outward, but you could have spread an equal amount of paste across the entirety of the surface and achieved great results.
Depends on the paste thermal conductivity. Thickness of the paste that goes with the viscosity of the paste, what is it made of. Cpu cooler mounting if its screwed one, it will squeeze the paste to the limit, so it will make the thinnest paste spread possible. If its just press fitted one, like stock intel cooler, then hell yeah you are dealing with cooler just touching thermal paste, that is touching cpu. Thermal paste isnt just liquid with silver shavings in it. By the use of it it will become chemically something like the solid pads you can buy, but fused without air bubbles to the cooler and cpu face. So its more to it than you think.
you are thinking to hard on this. Curb the OCD, the pressure and weight of the heatsink/screws squeezes the paste out so its not riding on some crazy 1/4" of goo. Its quite thin its better to put on too much vs too thin. a tiny dot will not cover the square edges so a X or 5 dot is the best 2nd is the credit card spread or maybe tied.
@@Physics072 The truth is that nowadays all CPUs come from the factory with an integral heat spreader on them, so if you're thick and sloppy on the Wal-Mart rig you're fixing for aunt Betty, it will be perfectly fine. If you like to actually push rigs with high computation loads and run exotic watercooling setups like I have for 20+ years, then yes, doing it right pays dividends
I'm an electrical engineer with over 25 years of experience (with power semiconductors). Simply put, use what you can control. In the case of a CPU, you can control what Nick calls (or perhaps someone else) "buttered toast" - this is the only way you can be reasonably certain (Schrödinger's paste?) is the cover the IHS and be certain the when you tighten down the water block/air cooler with *EVEN* pressure. PERIOD!
I hate Electrical Engineer's ,just kidding I use to wire House's back in the day. so what's a Electrical Engineer doing playing around with Electronics and CPU paste And semi Conductor's , Maybe you like Tesla Coil's for Breakfast , I don t know , probably not . LOL
@@madmad8582 Power semiconductors like SCRs/IGBTs/GTOs/Diodes, etc. Mostly for robotics, AC & DC drives and high current applications like plating. Playing with computers is just a by-product of industrial automation which is what I've done most of my life.
I am a rocket scientist and before that I was a brain surgeon. I'd say 2 bottles of thermal paste will do, you want it to be as cool as possible, the motherboard wouldn't mind either.
The issue here being long term when the thermal starts to dry up (especially with cheap/crappy thermal pastes). Penta Dot/Butter Toast method means more contact points so would probably be more reliable in the long term.
@@KevinJDildonik only the 1% can upgrade their pcs all the time or at least less than twice a decade. Not everyone is an american video editor with disposable income.
@@KevinJDildonik I was gonna say it’s not an expensive or difficult process you probably still have 19/20ths of a tube from when you built the thing just redo it if your temps start creeping that’s always worked for me
@@StarmenRock if you built it, you definitely did not use an entire tube of thermal paste. Just unscrew the cooler clean everything up and reapply. If you have the tube from when you built the computer then it’s literally free to do this
it used to be much more important in the days of the original Athlon and Pentium 2/3 when you'd put the cooler directly in contact with the silicon - the risk of creating an air bubble and then have that area of the chip overheating were high. Ever since most chips started coming with a heat spreader, the risk has gone down.
Not to forget the risk of damaging the edges of the silicon during mounting of the cooler :) Btw, graphic cards are still bare silicon, but I never had issues with it. Repasted my Gpu yesterday.
I have always went with the peanut spread method earlier, but have decided to go with the penta dot on my Ryzen 5X 5600X CPU. AIO Thermal Take TH 360 ARGB Liquid cooling is going to be doing a better job than the prism AMD stock cooler. Thank you for this video!
I watched 5 videos about this topic in a row and your video was the only on convincing. A lot of people, a lot of opinions, only one guy approving his methods. Good job 🙂
True. took your advise and worked on my G750 cause its been 8 years past since I bought it. I chose Thermal Grizzly Kryonaut, did the spread on both CPU and GPU, cleaned out the fans and their heatsinks, and well running good as new. Temps dropped to 15C, and now I'm happy. Thank you again for a nice video as well with your advise.
As long as you use enough paste to cover the cpu without the paste running down onto the motherboard it really doesn't matter what method you use. What does matter is the more even the pressure applied as you secure the cooler to the motherboard, the more even the thermal paste spread will be....
Yeah, too much heatsink paste on the die could be worse than too little. I normally swivel the heatsink back and forth before clamping it down. I think this helps spread out the paste a little better.
They keep the grooves to get direct contact with the heatpipes, rather than adding more distance with an additional flat plate, as the pipes move more heat, faster, than solid metal. Did you consider filling (buttering/scraping) the bottom of the cooler plate just to fill in the gaps before applying, or using a cooler with a flat plate?
In my earlier years I used to be a machinist, and I kept some of my tools. All of these methods will work, but I prefer to spread a thin layer covering the entire processor, but...Most copper cold plates have lines with minute grooves left over from the machining process. I blue the cold plate of my AIO, then pull my machinist straitedge across it, you will usually find a few low spots. I then start with the medium grit lapping stone, and lap it until all bluing is gone (low spots). I blue it again, pull the straightedge across it, if it's flat and the pockets gone, I lap it with the fine side. I then lap it a couple more times, each time with a finer stone. I then finish with an Arkansas stone that's super smooth (looks like amber marble), this leaves the cold plate super flat and with a mirror finish. Now the contact area between the CPU and cold plate is almost a 100% with no paste used. I spread a very thin layer just in case of any flatness Inconsistencies the processor might have. At idle with a 280 AIO, all my systems have run 2 - 10 degrees (F) over the ambient temperature of the room. IE, 68 degree room temp, CPU at idle (70 - 78 degrees F) About (5) systems with both AMD and Intel processors.
you underestimate how much I min max my shit in every video game I play and I will absolutely be using the Penta Dot method and I will stab people without warning in its defense
My understanding is that the function of thermal compounds is to fill the tiny surface irregularities in the IHS, and NOT to create a thick thermally conductive layer between the cooler heatsink and the CPU. Arctic advises that a rice grain-sized dollop properly applied (to the smaller Intel die surface area) will spread out as CPU heating cycles cause it to migrate into an optimum thin layer. I'd like to see a test that puts the CPU through 20-30 heating/cooling cycles for each method. I bet the 'spread' would look quite different than after a single stress test.
Yup, the paste is less conductive than the cooler and IHS, so you don't need it to be bathed in it. If it's too thick it works more as insulator. Any excess gets pushed out by the mounting pressure anyway.
I did get some good results by applying paste, then applying the cooler and then removing the cooler and wiping the paste off the cooler and leave what was left on the CPU. This was a few years ago and I don't remember the stats but it was not bad at all. The thing is it's really got down to +-1 degrees either way, is it worth worrying about?
Just bought some thermal paste and was wondering about this. Very interesting to see that it makes almost zero difference how it's applied as long as you're using enough of it. Spreading in out across the entire CPU certainly seems like it would be the best way though as it's the most logical, although perhaps you'd want just a little bit more in the center for optimal effect if you're really obsessed with getting that extra less then one degree cooler, lol.
@@zxcv97 i've always used the buttered bread method, i'll add the pea in the middle. i had the pump die on my water cooled computer last week, due to the thin thermaltake cpu block being blocked. i have some proper coolant in the mail, and a newer cpu block for my old 1155 m/b.
Ok I applaud you for taking the time to do this test. It certainly is a help. Personally I have been building PC since 98'. Been using the Butter Spread since 2003. The only diff is that I put masking tape around the edge of the CPU approx 1/8" overlap, all around, onto the die surface. Then apply a bit of paste onto the die. Spread with a credit card or plastic always keeping the edges of the card on the tape. Simply remove the masking tape when finished. In this way you can have a very uniform result without too much overkill on the paste. I have always had great results. Again, thank you for your time.
Funny I had my first pc with 3 screws because one broke, so I still overclocked the cpu from 3.6 to 4.0 Ghz, it would stay 80 95 in games, and in the summer I had to down clock it so it doesnt thermal throttle. The funniest part is that when I'd hit with my leg the pc case, the fan would fall, because all the screws were loose so this would happen frequently and I never changed the paste. I used the pc like that for about 3 years and when I bought the new one, I gave the old one to my cousin, so he used it for 2 years more. That cpu is still working btw.
The only thing I find truly remarkable about this is the number of times you uninstall and reinstall your cooler without praying to god that the computer will turn back on
Love your tests. I remember when I was little I cleaned by PC and took the heatsink off. I replaced it like nothing happened and ran the PC like normal. Didn't even know i did anything wrong.
I’ve been in the electronics industry for 45 years, now retired. I would go with the “buttered toast” every time. I’ve repaired every household appliances inc audio amps to computers. STK series of audio amplifiers chips get pretty hot.
I would encourage you to update this video review / test and include another method which is very very old school. Put a plastic sandwich bag over your index finger and use that to spread a very small amount of thermal paste, even smaller than the size of your pea /dot. The thermal paste is really only to fill in the microscopic pits in the surface of the metal. You want as much of the metal on the heat sink to make contact with the metal of the chip. The paste is to eliminate the micro pores of air which act as insulators trapping the heat in the chip. The best conductor is metal to metal contact. The paste will conduct some heat but is worse at doing so than the metal touching metal.
Seems like a bad idea in the sense of if you screw it up, you probably REALLY screw it up, but it would certainly be interesting to see the results if done well.
Can you include an additional test for the situation in which you accidentally lift up the block after paste application and quickly put it down again? I keep seeing people saying that you have to clean everything and reapply if that occurs but I don't really know if it makes any difference.
I have performed final assembly on printed circuit boards requiring a heat spreader be attached using the Penta dot method. We are required to remove the heat spreader from every fourth unit to check the spread of the thermal paste to ensure the CPU is completely covered. If it looks good, then we reattach the spreader without reapplying the paste.
The concern is about trapping air bubbles. Personally (20+ yr system builder), I clean the edges of both CPU & Cooler, respread to even on both, and then re-attach. Never has there been an issue (and yes, I support many of the same units, so they are monitored daily).
So cool! As techies we can get hung up on our methods and this shows that there is very little difference between methods. Heck the results are probably within the margin of error for measuring temperatures. Thanks for doing this!
Gamers nexus did a video about this a while ago, his results are pretty much in line with yours. And having too much thermal paste is a lot better than not enough. GN also went overkill with a massive blob, as long as it's not electrically conductive a bit getting squeezed out the sides won't hurt anything.
I remember watching it. Everyone should see it. Though considering the comments I received, prompting me to create this video, it seems to not be the case.
I use the spiral method, just cause it covers the IHS and I don't really have to think about it, have been doing it for over 15+ years and it works out fine.
Thanks for this detailed video. Since I don't rebuild my computer often, I always have the apprehension of how to apply paste properly. You have settled that matter for me in a satisfying way, once and for all! Thank you for showing me!
Probably also depends on the paste. Some pastes are more viscous than others and may not spread out as well under the pressure from the heatsink and may actually benefit from the "buttered toast" method.
I always used to use it for Arctic Silver. I seem to remember it was their suggested method, and it was one of the thicker pastes I've encountered. Part of it might be down to the fact that the paste was conductive; if you pre-spread it, then there's less risk of going wrong.
I usually just do some art... I use different colored thermal pastes. Last time I did a Xenomorph picture from Alien Isolation... detailed and everything. When I removed the heatsink 6 months later to redo things... The Alien was still there... only.. he was flicking me off. I ... I don't remember painting that part....
I got my own preference for this, I first spread a really thin amount all over the IHS and then I add a small dot in the center. This ensures 100% coverage and ALSO total absence of micro bubbles of air that might get trapped in the center when using the spread method.
I had this same idea and my then employer said no. It sounds like double positive to me but, was told air can get trapped between "layers" of compound. The reason we fully wipe off old paste.
@@cardinaldriver it really doesnt matter to be honest, for normal use - excluding extreme o/c - you can just apply paste however you feel like, as long as its enough of it. Ive been reusing paste and i didnt notice any difference. If the paste is in good condition, i just slap de cooler back on it and the temps are the same.
Significant impact of bubbles getting trapped is a fallacy. Makes no difference. Gamers Nexus did a great video on paste application and went absolutely nuts with all manner of paste quantity and methods of application. As long as there's a layer on the IHS (preferably thin) all is well. Microbubbles aren't a thing.
@@martinw245 yeah it's probably just my OCD. It's true, the method doesn't really make any significant difference, as long as there's enough of it. I even reseat coolers without changing the paste and didn't see any meaningful difference.
@@HDJess I'm not offended but I've been an OCD sufferer from the age of 8. It's a dreadful disorder that severely impacts our lives. Unless the above applies to you, being a bit fussy isn't OCD. Just saying, I know you didn't mean to offend, but its a bit annoying when people trivialise a very serious disorder that sometimes results in suicide. I wish this "my OCD" for being fussy hadn't become a common expression.
You call yourself tech illiterate but your academic methodology skills are far better than most academic staff. Thank you for these perfectly scientific videos.
Having had extensive thermal engineering experience, it does not surprise me that differences are small. But I continue to believe that lowest temps, though not by much, will come with a uniforn, thin coating.
Very interesting! I was looking exactly for this. Having done thermodynamics courses, we learned the reasons and mathematics for thermal paste between heat sources and heat sink components. It would, in theory, be best to have the smallest layer possible spread out evenly. It was nice to see the results of a real-life test. I appreciate your commitment to this. -- Some explainations Well, we really just want the paste to fill the nanoscopic holes between our two metal planes. If the metal surfaces were perfectly smooth, there would be no paste needed, the paste is an awful heat conductor, it is just there to replace the pockets of air, which are even worse conductors. So the tiniest layer, spread out evenly will most likely work best, because the surface roughness on the processor top and a heatsink are very small, so a very thin layer will be sufficient to fill those holes. And you want it everywhere, so it fills the imperfection wherever it is. Moreover, What really matters, is the type of paste. There are multiple high thermal conductivity pastes, but they are very dangerous to work with, because they are electrically conductive, so watchout!
Thank you. I always says that and everybody is mocking me because it takes much more time to complete the application. But it saves paste too and is overall a better way to apply thicker pastes.
I prefer the "X" method on newer CPUs..... but i do remember the times i would thinly spread the thermal paste over entire IHS on P4 cpus. That method provided so much suction between the cpu and heatsink, the cpu would actually get pulled from its socket because it stuck to the heatsink.
the best method is a thin layer specifically covering the die of the cpu, when you put the heat sink on it should spread almost to the edges of the cpu without seeping through
@Keebine Kobold well it's not the "best" in terms of thermal performance, since any method with enough paste will work fine. but it might be the best in terms of using as little paste as possible, getting a consistent application, all without sacrificing performance. a novice may not realise they didn't put enough paste if they use another method, but if they use the buttered toast method it will be very obvious when there's not enough. for someone proficient, yeah it doesn't matter bc they have an idea of how much to use and they know to ensure even mounting pressure
A lot of enthusiasts are religious about applying thermal paste. I love the fact that you spent the time to prove (IMHO) that they are simply wasting their time. I've been using the 5-dot method for years because it is self-evident that the better the contact between the CPU and cooler ... the lower your CPU temps will be. Thank you!
Seriously I am not a 20 year PC vet. But I have built a few PC’s. My latest Rog strix B550-f-gaming. Ryzen 7 5800x , Asus Dual 3070, PSU -Corsair Rm 850x Gold. Noctua No-d15’s chromax dual fans - 6 Noctua Chromamax black hot swap. That being said. I have applied Every paste except metal‘because’ it scare the bejesus out of me. So to prove the five dot method for at least AMD 5800x. I applied 4 applications already of thermal paste. My second go around with Noctua thermal paste AND thermal grizzly kryonaut extreme. The 5 dot method made it to where I do not have a hot spot!!!!! 5 dot decreased my my Celsius by 15-20%c. Period. And best. Is no Huge hot spot!
Thank you so much I did the spread method and at basic tasks, my fan speed at 70%, cpu was 100C, did the cross method and fan speed 30%, went down to 46C. Cross is the way
I been using a thin spread style since 360 RROD days, you had to apply it directly on the die. Just makes me sleep better at night knowing evey nano meter has some thermal paste crammed in there.
Wish you had done "no paste!". Thanks for calming me about the hardest part I've had in the one PC I built (destroyed first motherboard due to paste issues but that's between you and me). I think the takeaway is: "apply a reasonable amount of paste BUT don't let any get off the CPU cause that could cause issues".
The object of a chosen method of applying the paste is to A) get maximum coverage with minimum thickness and B) not create air pockets by entrapment. Flattening the paste will always have a radial spread, so a dot will always reach the sides before it reaches a corner. Since the paste has a rather heavy viscosity, the more paste applied to a single spot and the farther it has to spread, the more pressure will be required to reach a desired thickness and spread. So the optimum is a pattern that needs to spread the least to get full coverage, with a minimum amount of pile to the paste, without using a pattern that would trap air as it spreads. So: the X and pentadot, so long as the dot at the center is largest on the latter. If the central dot was too small with the pentadot method, four air pockets would be caught at the cardinal directions around the center. The "buttered toast" method manually creates coverage, but will usually trap air from an uneven surface and have a greater final thickness from using the most paste.
Did you actually watch this entire video which quite conclusively shows that it makes no difference at all, and then feel the need to type all this out, insinuating that there is an optimal method? 🙄
@@5stringaustin It's putting a little science behind the empirical observations. He did them all in ways that won't really cause air entrapment, but some of the methods _RISK_ it for the reasons I explained, which does make a difference.
I found it good to fill the gaps or grooves between the heatpipes on the cooler (GPU), then the the butter method on the GPU. Otherwise it probably wont fill the grooves, depending on the cooler if there actually are grooves...
What I always do: - Get some isopropyl alcohol 99.9% - Clean base of cooler and cpu IHS with it - Get some kitchen paper, dip it in the isopropyl and rub my index with it. This makes it super dry and without any oils - Use as little thermal paste as possible and use my index finger to spread it evenly across the whole IHS. - Carefully apply heatsink and tighten each screw slowly. Works every time. Some people are against using your hands/fingers to apply thermal paste because it has dirt and natural oils. Makes sense. But if you use isopropyl you get rid of that issue. The card method usually leaves a much thicker layer of paste.
I've done 100's of custom gaming PC's from $500-$5,000+. A thin even layer spread across the whole ihs keeps temps low and excess paste to clean off the MB later.
And now I'm watching two different videos, both done 10/10 with good tests. And another one says it does matter, which one I trust XD And of course I trust my own experience of building over 1k PC, over years. Yeah, it does matter, because it depends on CPU itself.
I think the take away from this is that the plates are probably several sizes bigger than what they actually need to be so that they have a large enough area to accomodate the different amount of paste being applied to each model. Also the thermal conductivity of the paste is probably large enough for small variations in the cross section area to be negligible.
The CPU and CPU cooler parts that touch are actually not flat. There are grooves and curves on both sides. That's why the thermal paste looks all wrinkled after you separate them after a while. It's best to put a blob in the center so that when it spreads out, it gets into those grooves properly for heat dispersal.
Agreed. I've been doing a pea size blob in the middle going on 22 years now. All my machines, friends, clients', all the same. My son's gaming PC I helped him build 7 years ago, since upgraded, same method, doing fantastic. I say pea size, but actually just a tad bit bigger than a pea but not much more. I'm currently building a new one for me and here I am researching to make sure I'm keeping up with the times. I see no reason to change now. Long live my new i7
I really appreciate your video. I'm building myself a higher end gaming PC and it's sitting on my counter waiting for me to install the cooler. I wasn't sure which method to use so I've been searching for a while. I'm definitely more comfortable, I can go down and actually finish the build
I'd be curious to the longevity of thermal performance among the different methods. I wouldn't be surprised to see the "buttered toast" method retain its performance longer.
Just to add some historical context to see why this such a pretentious topic: CPU's once upon a time didn't have a heat-spreader, and thermal paste used to contain electrically conductive material, and thermal protections weren't as prevalent. With direct-die cooling, it was critical to ensure every bit of the die was covered, but you couldn't use too much as spill over and short out a surface mounted devices (components); basically any extreme (too much paste, too little, improper coverage), could kill your CPU. Even uneven mounting pressure could crack the die, or be a silent killer creating a hot spot. So the "buttered toast" method became ideal as it helped ensure complete coverage, and helped meter the amount of paste used. There was also emphasis in tightening in small increments in a cross pattern. But these days, with a heat spreader mitigating a lot of the dangers and non-conductive paste, it really doesn't matter; if anything, better err on the side of too much paste than too little, but neither will necessarily spell death for the CPU. The dot or grain of rice method gained prevalence with heat spreaders; while the heat spreader protected some of the SMDs, prominent thermal paste was still electrically conductive, so excess could kill the socket. The grain of rice allowed mounting pressure to evenly distribute across the spreader without causing excess. Improper spreading could create air-pockets, so "buttered toast" fell out of favor with heat spreaders as the grain of rice or dot methods were more fool-proof leaving the spreading to the cooler's mounting pressure. AMD recommends the "penta dot" method because of their chiplet design (especially with ThreadRipper/EPYC) means there are multiple dies underneath the (massive) heat spreader; pentadot ensures each heat creating die has some thermal paste coverage; a singular dot may mean one of the out chiplets(/dies) doesn't have direct coverage (e.g. not enough, or might spill over the side).
love how you make the video and use the scientific the method to compare the different ways to apply the thermal paste. However I think that we should also consider the long term repercusion of each method. Maybe the thermal paste gets wasted earlier if you use the dot application rather than the 5 dots or something like that. Again, great video!
It really depends on the type of paste. The paste that's currently supplied with Arctic coolers, for example, is very sticky and virtually impossible to spread "manually" (you end up with half of it sticking to the spreader, and a lot of air bubbles in the middle). A central blob or a cross generally works fine for all types of paste (and will spread itself out without leaving any voids), as long as you use about the right amount.
I work on large inverters that utilize large 1200 amp IGBTs that get bolted to large aluminum heat sinks. If any “method” other than thin coat across a thoroughly cleaned surface is used, that IGBT will overheat and blow its guts in about an hour, depending on load and ambient temp. All of the old thermal paste must be scraped off, then clean both the heatsink and the device with alcohol wipes. I then heat the mating surface of the IGBT with a hair dryer on low til it’s warm enough to not cause the paste to thicken when it’s applied, at least 100F. I squeeze out one long bead in the center running the length of the device. I use a small plastic paddle to spread it out as evenly as I can. Mount the device and press it down firmly and install the four corner screws to secure it. Once the screws make contact, they get tightened down in a cross pattern one quarter turn at a time until snug, then the torque wrench comes in and tighten in same pattern until desired torque rating is achieved. Reassemble the inverter, start it up and let it run for an hour with no load. Shut it down, make it safe, recheck torque, first time always results in about an eighth of a turn to get it there again. Repeat until after test run it clicks the wrench without turning. Then, and only then can it be returned to service with confidence the customer isn’t going to call in an hour or two saying it blew down again.
New sub. Great vid, super impressed. Puget systems & cisco all recommend the 'X' method. Used that since about 2013. Long term (2-4y) is where differences will show in terms of temps. This is where using quality *paste* matters. Using 20% more than you think you need (EXCL liquid metal) is a good idea for business systems.
A topic that when 1,000 UA-cam videos say the same thing MAYBE someone will be convinced. There is a video out there with Steve from Gamer's Nexus and Der8auer that basically says, placement doesn't matter, too little IS BAD, there is never too much (but gets messy), paste doesn't "dry out" and air bubbles will work themselves out or be compressed so small, it doesn't matter. And the grooves are unavoidable when the maker puts heat pipes right against the IHS. Compressing them too much to be perfectly flat will make them not be pipes anymore which defeats the purpose of the whole heat pipe idea
I used to spread it out (and I've never bought thermal paste that didn't come with a spetula, so it made sense), but I switched to the dot because it's less work and makes no difference. However, given that I'm currently reinstalling the slightly bigger brother of your test CPU I might try the penta dot for a change.
The only difference is when you apply too little, causing CPU to overheat. So what's necessary is applying efficiently. If your Cpu is rectangular (Laptop CPUs/ Direct die) line technique is good as it spreads in length and little breadth, less wastage. If your CPU is square , Pea sized Dot/ X /Penta Dot works good as it spreads inside out in circle shape. Spread Method it's really messy as compared to other techniques, not at all useful when u have little paste remaining. Few of it will be wasted on that spatula/spreader. Time Consuming. Thanks For the video !
Thermal paste pro here. It's not the pattern, it's the amount, thicknes of the paste and the surface area. X will not help if area is large. For a super large CPU I would zigzag instead of dots or exes.
Great stuff man! I've been a Systems Engineer for just over 25 years and had to buy some new CPUs which required upgraded heatsinks for a server, and the thermal pads on one of the heatsinks actually had a fingerprint it! Clean it with 90% isopropyl and apply some new paste, right? Well, I've used "buttered toast" forever, I've always been curious as to what the best method is and you have laid that to rest. Thank you, sir!
You asked for it....NO THERMAL PASTE TEST ua-cam.com/video/VAh_p90y_DU/v-deo.html Remember to LIKE and SUBSCRIBE! Thank you everyone for the continued support!
Even though the results are probably alright it doesn't seem like you've controlled for power consumption/boost which casts a huge doubt on these results. Plus we're talking about a pretty low power consuption, you'll have a bigger differential with overclock.
Cool of ya to add a methodology, wish more people would do this!
@adamI'll try it out.
Well, all this may be true for the IHS chips. But in laptops/notebooks with direct die contact partial spread can cause more problems, like not cooling properly some of the cores, or maybe even chipping the silicone with no material in between chip and heatsink...
Apply the paste to the ram not the cpu
You didn't use enough paste. I can clearly still see motherboard. Cover the entire motherboard, including the RAM. It wouldn't be a bad idea to cover the USB ports as well.
That's considered good practice, isn't it?
Yeah, not nearly enough paste XD.
🤣🤣🤣🤣🤣
If your pc isn't just a solid block of thermal paste with components stuck inside you're not using nearly enough bro
@@xav7311 Preach on. The moar the better. Mo Paste, mo better. Paste is Life. Paste is Love. The more you paste, the more you save.
It just works.
Bro i read that in his voice haha
I never thought that Johnny Sins is also a PC guy.
That comment is unholy
😂😂😂
Shut. Just shut
Cursed comment.
I actually watched a scene where he is a Tech Support guy.
As someone who has never done this before, your conclusion has definitely eased my mind. Put some thermal paste on it, make sure it's enough, and everything will be fine. Definitely a load off. Thank you for the video.
Same.
Aparently putting more than enough can be bad as well as putting not enough, so good luck getting a good balance xd today and yesterday I changed my consoles paste and used different methods, somehow I still feel like they’re boiling :c
@@meappeam6917 Actually, adding too much thermal paste is a lot better than little or none, as it covers the same area as acquitte thermal paste application, so it doesn't matter that much. The worst that could happen in most is that it goes off of the IHS, which is not bad. Most extreme, basically impossible, is that it goes off the IHS into the CPU socket contacting the pins, though I've found sources claiming that it is still possible to normally run a PC with non-conductive thermal paste even in the CPU socket (this does not mean to attempt to put thermal paste in your CPU socket).
Though, putting too little can defiantly impact performance significantly (as per seen in the video)
If you feel as if your consoles are still "boiling", make sure you have a good thermal solution (I would recommend Arctic MX-6 or Thermal Grizzly Kryonaught), or perhaps look to get better / higher quality fans that would move more CF/m (Cubic Ft / minute)
I cut PTM7950 from the sheet and do not play with dots, dashes and crosses.
no this is wrong, it just cover the whole surface that cpu back panel so all the excess heat can be conducted
Despite this video being two years old, it’s still comforting. I built my very first pc last week and forgot about all application methods for thermal paste… but I still used thermal paste!! And not just a tiny dot either! My heart is at ease knowing I did it right! 😂
Well done bro, I just built my first rig and I didn’t install the cooler properly and now it overheats. I ordere a new cooler and hope I dont mess it again
@@drakata27... why would you need a new cooler??
@@denks7849 old one was amd stock and i didnt like it
@@drakata27Yes, it's terrible. Also ordered a new one today :)
@@drakata27yeah same, using stock wraith stealth with the ryzen 5 5600 non x.. 95*C on full use…. it‘s so loud. Ordered a pure rock 2 today. Hope it‘s better than the awful loud stock one..
Not being a tech person, I'm very happy to find your video, and want to say: Finally! A thermal paste demo that actually clarifies the most important considerations and helps me understand what I'm doing! Thank you! :)
There are dozens of the same tests, wdym XD
Pfft! Everyone knows the best way to apply thermal paste is to swallow a dollop and then regurgitate it onto the CPU lid. I call it the "Feedin' my baby bird" method.
That's some big brain shit right there.
that's some verge techniques right there
@@Pouria_1664 You need certified tweezers to apply it.
@@aguyandhiscomputer ones that protect your cables from electric overcharge
@@Pouria_1664 Route the cables with confidence.
Funny thing, I've been doing this a long time but I still revisit the topic every couple years. GREAT video. Clear and concise. Good job sir.
It's to see if a new best method was discover
and here i am again XDD
Jayz2cents use the cross, so i do aswell xD
I tried the spred, but bloody hell its messy.
You forgot the bukake method that the Verge used 🤣😂
oh my XD
I think Gamers Nexus tested that some time ago and the result was as long as there is enough it doesn't matter.
lmao i hate that guy
@@420cm what guy? The verga guy? Or nexus dude?
HAHAHA
Gotta admire your commitment in testing those thermal paste combos. I was just curious on how to put a thermal paste properly. Now that I watched your vid, I learned a lot. You exceeded my expectations.
Thanks. I'm glad it helped.
Subscribing just for the amount of effort it must've took to make this video. Mounting/ unmounting applying thermal paste/ removing it with alcohol. 20+ testing 4 to 5 times Sheesh!! 😬 you have my respect 💯
This guy saying "It'll be fine" in that calming voice is very nice to hear as someone who's almost constantly afraid to break the whole computer with my sheer presence
Well I just bent pins on my 2700x removing the stupid wraith prism that had molded to the cpu with its thermal paste preapplication like the Thing making a new creature
@@casedistorted must hurt
Facts over theory, I like it.
Every known fact started with a theory.
@@itsthatYEStoogoodguy aaaand the point of theorys is to be turned in to facts, theorys are useless if they don't get tested and confirmed.
@@xeviusUsagi all theories are Proven. And if it's not, then it's just an hypothesis
This guy is amazing. He also happened to be a computer guy. Nice one, Johnny.
sins
420 likes, not gonna ruin it
Its sins?
Congratulations! The UA-cam algorithms worked for you today! I watched about 6 videos straight and then subscribed. I enjoyed your format.
old school trick, you get a plastic baggy, use it as a glove, apply paste to baggy covered finger tip and dab the paste on the chip evenly with a very light coating. The goal is to fill in valleys and provide good contact and not cake it on to the point where it acts more like insulation. Great video by the way! I guess my method is more like the buttered toast method.
If this guy really is Johny Sins, i starting to doubt that white liquid on CPU was A thermal paste
well he wants you to think that it's thermal paste and not ask any questions and considering what else it could be I am just going to assume that it is thermal paste cause the alternative is just gross
I've always heard buttered toast was the best back in the day. Been doing it that way ever since the Pentium 4.
I thankfully never heard that, and never did it, and never had problems.
Its about long term usage. The less amount of paste will dry out early. So the best is buttered, not because of temperature.
@@RedShark493 the paste doesn't dry out, it's pumped out due to lots of thermal cycles it goes through
i even saw someone use a banana
@@michakrzyzanowski8554 You mean evaporate? How tf do you "pump out" thermal paste?
He deserves a like for removing and placing pump 7+ times
I'm someone else who has not done this and will need to on my new build very soon. Thank you for the experiment! This definitely puts my mind at ease. It seems like there is so much controversy about this topic that makes people nervous that have never done it. When in reality its a simple process and no need to worry so much, just do it!
Let's be clear. The paste is not the primary medium for heat transfer. Metal-to-metal is still far more efficient. The paste (dielectric grease with suspended silver shavings) is used because it is orders of magnitude more efficient than AIR. You want the most/closest metal-to-metal contact for proper heat transfer, and the paste is there to fill in all of the microscopic gaps for the rest. Too little or uneven paste causes the cooler to ride on a "cushion" of paste and leaves air gaps that insulate rather than transfer heat. Too much paste is going to make for less-efficient transfer of heat. You want to have the thinnest possible layer of paste with the most even distribution. So the "buttered toast" method is important, but so is the THICKNESS of paste. Only so much can be pressed out under the spring load of the cooler. This was demonstrated with the "small dot" method. Once the tension of the cooler mount matched the resistance of the paste to compression, it stopped flowing outward, but you could have spread an equal amount of paste across the entirety of the surface and achieved great results.
uber careful spreading has always been my go to. seems logical really.
Depends on the paste thermal conductivity. Thickness of the paste that goes with the viscosity of the paste, what is it made of. Cpu cooler mounting if its screwed one, it will squeeze the paste to the limit, so it will make the thinnest paste spread possible. If its just press fitted one, like stock intel cooler, then hell yeah you are dealing with cooler just touching thermal paste, that is touching cpu. Thermal paste isnt just liquid with silver shavings in it. By the use of it it will become chemically something like the solid pads you can buy, but fused without air bubbles to the cooler and cpu face. So its more to it than you think.
you are thinking to hard on this. Curb the OCD, the pressure and weight of the heatsink/screws squeezes the paste out so its not riding on some crazy 1/4" of goo. Its quite thin its better to put on too much vs too thin. a tiny dot will not cover the square edges so a X or 5 dot is the best 2nd is the credit card spread or maybe tied.
@@Physics072 The truth is that nowadays all CPUs come from the factory with an integral heat spreader on them, so if you're thick and sloppy on the Wal-Mart rig you're fixing for aunt Betty, it will be perfectly fine. If you like to actually push rigs with high computation loads and run exotic watercooling setups like I have for 20+ years, then yes, doing it right pays dividends
I'm an electrical engineer with over 25 years of experience (with power semiconductors). Simply put, use what you can control. In the case of a CPU, you can control what Nick calls (or perhaps someone else) "buttered toast" - this is the only way you can be reasonably certain (Schrödinger's paste?) is the cover the IHS and be certain the when you tighten down the water block/air cooler with *EVEN* pressure. PERIOD!
From someone whom works in the IT space (25 years), I ALWAYS spread the thermal paste over the entire IHS.
I hate Electrical Engineer's ,just kidding I use to wire House's back in the day. so what's a Electrical Engineer doing playing around with Electronics and CPU paste And semi Conductor's , Maybe you like Tesla Coil's for Breakfast , I don t know , probably not . LOL
@@madmad8582 Power semiconductors like SCRs/IGBTs/GTOs/Diodes, etc. Mostly for robotics, AC & DC drives and high current applications like plating. Playing with computers is just a by-product of industrial automation which is what I've done most of my life.
Hi, manufacturer of these materials here, cover the motherfucker. That's what it's made for.
I am a rocket scientist and before that I was a brain surgeon.
I'd say 2 bottles of thermal paste will do, you want it to be as cool as possible, the motherboard wouldn't mind either.
The issue here being long term when the thermal starts to dry up (especially with cheap/crappy thermal pastes).
Penta Dot/Butter Toast method means more contact points so would probably be more reliable in the long term.
Oh, absolutely. Penta dot is my go to for most clients, since it usually lasts almost 6 months more than single dot.
@@KevinJDildonik only the 1% can upgrade their pcs all the time or at least less than twice a decade. Not everyone is an american video editor with disposable income.
@@StarmenRock Bro, anyone with a proper job can upgrade their PC almost all the time, atleast once per month, and still afford all the bills.
@@KevinJDildonik I was gonna say it’s not an expensive or difficult process you probably still have 19/20ths of a tube from when you built the thing just redo it if your temps start creeping that’s always worked for me
@@StarmenRock if you built it, you definitely did not use an entire tube of thermal paste. Just unscrew the cooler clean everything up and reapply. If you have the tube from when you built the computer then it’s literally free to do this
it used to be much more important in the days of the original Athlon and Pentium 2/3 when you'd put the cooler directly in contact with the silicon - the risk of creating an air bubble and then have that area of the chip overheating were high. Ever since most chips started coming with a heat spreader, the risk has gone down.
Oh I remember those little protruded "mirror" thing! Those were a PITA.
Ogs unite 💯💯💯
Not to forget the risk of damaging the edges of the silicon during mounting of the cooler :)
Btw, graphic cards are still bare silicon, but I never had issues with it. Repasted my Gpu yesterday.
@@knightwolf200612 true... somewhat. You'll notice that most GPU dies are surrounded by a spacer to prevent MOST (not all) damage.
Remember the copper shims? Eeek. I had to clean a die with rubbing alcohol before, wouldn't boot. I was so relieved when it worked.
I have always went with the peanut spread method earlier, but have decided to go with the penta dot on my Ryzen 5X 5600X CPU. AIO Thermal Take TH 360 ARGB Liquid cooling is going to be doing a better job than the prism AMD stock cooler. Thank you for this video!
I watched 5 videos about this topic in a row and your video was the only on convincing. A lot of people, a lot of opinions, only one guy approving his methods. Good job 🙂
Thank you
True. took your advise and worked on my G750 cause its been 8 years past since I bought it. I chose Thermal Grizzly Kryonaut, did the spread on both CPU and GPU, cleaned out the fans and their heatsinks, and well running good as new. Temps dropped to 15C, and now I'm happy. Thank you again for a nice video as well with your advise.
As long as you use enough paste to cover the cpu without the paste running down onto the motherboard
it really doesn't matter what method you use. What does matter is the more even the pressure applied
as you secure the cooler to the motherboard, the more even the thermal paste spread will be....
Yeah, too much heatsink paste on the die could be worse than too little. I normally swivel the heatsink back and forth before clamping it down. I think this helps spread out the paste a little better.
They keep the grooves to get direct contact with the heatpipes, rather than adding more distance with an additional flat plate, as the pipes move more heat, faster, than solid metal.
Did you consider filling (buttering/scraping) the bottom of the cooler plate just to fill in the gaps before applying, or using a cooler with a flat plate?
In my earlier years I used to be a machinist, and I kept some of my tools. All of these methods will work, but I prefer to spread a thin layer covering the entire processor, but...Most copper cold plates have lines with minute grooves left over from the machining process. I blue the cold plate of my AIO, then pull my machinist straitedge across it, you will usually find a few low spots. I then start with the medium grit lapping stone, and lap it until all bluing is gone (low spots). I blue it again, pull the straightedge across it, if it's flat and the pockets gone, I lap it with the fine side. I then lap it a couple more times, each time with a finer stone. I then finish with an Arkansas stone that's super smooth (looks like amber marble), this leaves the cold plate super flat and with a mirror finish. Now the contact area between the CPU and cold plate is almost a 100% with no paste used. I spread a very thin layer just in case of any flatness Inconsistencies the processor might have. At idle with a 280 AIO, all my systems have run 2 - 10 degrees (F) over the ambient temperature of the room. IE, 68 degree room temp, CPU at idle (70 - 78 degrees F) About (5) systems with both AMD and Intel processors.
I personally do the corner dots with the center dot. I dont care if its better or not, but it does give me peace of mind
The reassurance that "it'll be fine" made me confident enough to do it!
you underestimate how much I min max my shit in every video game I play and I will absolutely be using the Penta Dot method and I will stab people without warning in its defense
Dumb.
@@xPathettic stop introducing yourself
@@vulnerablerummy That was creative.
"It doesn't matter how you put the thermal paste on there.." - everybody
If you really wanna get min max, you should use thermal epoxy instead of paste
@@pastelcastiel1481 Just weld the cpu to the cooler
My understanding is that the function of thermal compounds is to fill the tiny surface irregularities in the IHS, and NOT to create a thick thermally conductive layer between the cooler heatsink and the CPU. Arctic advises that a rice grain-sized dollop properly applied (to the smaller Intel die surface area) will spread out as CPU heating cycles cause it to migrate into an optimum thin layer.
I'd like to see a test that puts the CPU through 20-30 heating/cooling cycles for each method. I bet the 'spread' would look quite different than after a single stress test.
The spread would also look better if this guy hadn't use a dog shit heat sink. The pea method is the best method.
And i second to that
Yup, the paste is less conductive than the cooler and IHS, so you don't need it to be bathed in it. If it's too thick it works more as insulator. Any excess gets pushed out by the mounting pressure anyway.
I did get some good results by applying paste, then applying the cooler and then removing the cooler and wiping the paste off the cooler and leave what was left on the CPU. This was a few years ago and I don't remember the stats but it was not bad at all. The thing is it's really got down to +-1 degrees either way, is it worth worrying about?
It’s impossible to create a thick layer as the pressure of the cooler will push out any excess. More is almost always better.
Just bought some thermal paste and was wondering about this. Very interesting to see that it makes almost zero difference how it's applied as long as you're using enough of it. Spreading in out across the entire CPU certainly seems like it would be the best way though as it's the most logical, although perhaps you'd want just a little bit more in the center for optimal effect if you're really obsessed with getting that extra less then one degree cooler, lol.
Thinly Spread it first then do dot/pea at the middle works well for me.
@@zxcv97 i've always used the buttered bread method, i'll add the pea in the middle.
i had the pump die on my water cooled computer last week, due to the thin thermaltake cpu block being blocked.
i have some proper coolant in the mail, and a newer cpu block for my old 1155 m/b.
Ok I applaud you for taking the time to do this test. It certainly is a help. Personally I have been building PC since 98'. Been using the Butter Spread since 2003. The only diff is that I put masking tape around the edge of the CPU approx 1/8" overlap, all around, onto the die surface. Then apply a bit of paste onto the die. Spread with a credit card or plastic always keeping the edges of the card on the tape. Simply remove the masking tape when finished. In this way you can have a very uniform result without too much overkill on the paste. I have always had great results. Again, thank you for your time.
This is genius!
"It doesn't really matter"
*Guy who is .5c away from thermal throttling (sweats nervously) XDXDXD
Time to either get a better cooler, get rid of dust, finetune the oc, or any combination of those.
me and my laptop jajajajaja
Funny I had my first pc with 3 screws because one broke, so I still overclocked the cpu from 3.6 to 4.0 Ghz, it would stay 80 95 in games, and in the summer I had to down clock it so it doesnt thermal throttle.
The funniest part is that when I'd hit with my leg the pc case, the fan would fall, because all the screws were loose so this would happen frequently and I never changed the paste.
I used the pc like that for about 3 years and when I bought the new one, I gave the old one to my cousin, so he used it for 2 years more.
That cpu is still working btw.
@@alexanderciordas5307 The ChadPU
I think what he is saying, is he could start over, and run the tests again, and while within .5c for all of them, the "winner" is random.
The only thing I find truly remarkable about this is the number of times you uninstall and reinstall your cooler without praying to god that the computer will turn back on
I'd understand if it was a laptop being constantly taken apart but he's only removing the cooler on this
Mike what... what are you doing when you take your cooler off? What have you seen???
@@phatman808 he's seen enough
@@phatman808 he’s seen enough leave him poor guy
@@younos435 ??
Lord Johnny Sins with his thermal paste tips lmao !
lmaaooo
I see what you did there mate, plenty of women love that thermal paste too !
Love your tests. I remember when I was little I cleaned by PC and took the heatsink off. I replaced it like nothing happened and ran the PC like normal. Didn't even know i did anything wrong.
I’ve been in the electronics industry for 45 years, now retired. I would go with the “buttered toast” every time.
I’ve repaired every household appliances inc audio amps to computers. STK series of audio amplifiers chips get pretty hot.
I just had to reapply some thermal paste for the first time in 4-5 years so thanks for the informative and well made video bro!
Lmao I've still not replaced mine after 8 years. Surprisingly temps are still as good as before.
I would encourage you to update this video review / test and include another method which is very very old school.
Put a plastic sandwich bag over your index finger and use that to spread a very small amount of thermal paste,
even smaller than the size of your pea /dot. The thermal paste is really only to fill in the microscopic pits in the surface
of the metal. You want as much of the metal on the heat sink to make contact with the metal of the chip.
The paste is to eliminate the micro pores of air which act as insulators trapping the heat in the chip.
The best conductor is metal to metal contact. The paste will conduct some heat but is worse at doing so than
the metal touching metal.
Seems like a bad idea in the sense of if you screw it up, you probably REALLY screw it up, but it would certainly be interesting to see the results if done well.
Stumbled across this video right before first build. 5900x idles at 33-39c thanks for the video. Subbed and continued success man.
Awesome! Thank you!
You just crushed thousands of videos regarding thermal past application.
Great content!!
I used X and will always use X. I highly recommend it to you guys.
I do a % sign, a thin diagonal line with a dot in the 2 empty corners, pretty foolproof way to get a good result.
You are to computers what Project Farm is to engines, glues and oils.
How dare you denigrate Project Farm like that!
Thanks ;)
Can you include an additional test for the situation in which you accidentally lift up the block after paste application and quickly put it down again? I keep seeing people saying that you have to clean everything and reapply if that occurs but I don't really know if it makes any difference.
Yes this is a good point. It is something I am curious about. I will give it a test in the near future.
I have performed final assembly on printed circuit boards requiring a heat spreader be attached using the Penta dot method. We are required to remove the heat spreader from every fourth unit to check the spread of the thermal paste to ensure the CPU is completely covered. If it looks good, then we reattach the spreader without reapplying the paste.
The concern is about trapping air bubbles. Personally (20+ yr system builder), I clean the edges of both CPU & Cooler, respread to even on both, and then re-attach. Never has there been an issue (and yes, I support many of the same units, so they are monitored daily).
@@philsigman9088 You mean IHS or heatsink?
@@txtardis7887 Air bubbles are a myth.
So cool! As techies we can get hung up on our methods and this shows that there is very little difference between methods. Heck the results are probably within the margin of error for measuring temperatures. Thanks for doing this!
This video is the answer of my questions...heap on my mind thanks brother your effort is highly commendable...
I'm the "slim smaller rectangle" guy.
The die is well covered and the excess paste is getting pushed to the sides, but not over the edge.
Gamers nexus did a video about this a while ago, his results are pretty much in line with yours. And having too much thermal paste is a lot better than not enough. GN also went overkill with a massive blob, as long as it's not electrically conductive a bit getting squeezed out the sides won't hurt anything.
I remember watching it. Everyone should see it. Though considering the comments I received, prompting me to create this video, it seems to not be the case.
I use the spiral method, just cause it covers the IHS and I don't really have to think about it, have been doing it for over 15+ years and it works out fine.
I could see that working great.
Thanks for this detailed video. Since I don't rebuild my computer often, I always have the apprehension of how to apply paste properly. You have settled that matter for me in a satisfying way, once and for all! Thank you for showing me!
Probably also depends on the paste. Some pastes are more viscous than others and may not spread out as well under the pressure from the heatsink and may actually benefit from the "buttered toast" method.
I always used to use it for Arctic Silver. I seem to remember it was their suggested method, and it was one of the thicker pastes I've encountered. Part of it might be down to the fact that the paste was conductive; if you pre-spread it, then there's less risk of going wrong.
I usually just do some art... I use different colored thermal pastes.
Last time I did a Xenomorph picture from Alien Isolation... detailed and everything.
When I removed the heatsink 6 months later to redo things...
The Alien was still there... only.. he was flicking me off.
I ... I don't remember painting that part....
I got my own preference for this, I first spread a really thin amount all over the IHS and then I add a small dot in the center. This ensures 100% coverage and ALSO total absence of micro bubbles of air that might get trapped in the center when using the spread method.
I had this same idea and my then employer said no. It sounds like double positive to me but, was told air can get trapped between "layers" of compound. The reason we fully wipe off old paste.
@@cardinaldriver it really doesnt matter to be honest, for normal use - excluding extreme o/c - you can just apply paste however you feel like, as long as its enough of it. Ive been reusing paste and i didnt notice any difference. If the paste is in good condition, i just slap de cooler back on it and the temps are the same.
Significant impact of bubbles getting trapped is a fallacy.
Makes no difference. Gamers Nexus did a great video on paste application and went absolutely nuts with all manner of paste quantity and methods of application. As long as there's a layer on the IHS (preferably thin) all is well. Microbubbles aren't a thing.
@@martinw245 yeah it's probably just my OCD. It's true, the method doesn't really make any significant difference, as long as there's enough of it. I even reseat coolers without changing the paste and didn't see any meaningful difference.
@@HDJess
I'm not offended but I've been an OCD sufferer from the age of 8. It's a dreadful disorder that severely impacts our lives.
Unless the above applies to you, being a bit fussy isn't OCD.
Just saying, I know you didn't mean to offend, but its a bit annoying when people trivialise a very serious disorder that sometimes results in suicide.
I wish this "my OCD" for being fussy hadn't become a common expression.
Refreshing to see a nice calm paced to the point video, looking forward to checking out more videos later on :)
You call yourself tech illiterate but your academic methodology skills are far better than most academic staff. Thank you for these perfectly scientific videos.
As a professional post-secondary student of over 6 years, I immensely appreciate his scientific method 👏👏👏
Having had extensive thermal engineering experience, it does not surprise me that differences are small. But I continue to believe that lowest temps, though not by much, will come with a uniforn, thin coating.
Very interesting! I was looking exactly for this. Having done thermodynamics courses, we learned the reasons and mathematics for thermal paste between heat sources and heat sink components. It would, in theory, be best to have the smallest layer possible spread out evenly. It was nice to see the results of a real-life test. I appreciate your commitment to this.
-- Some explainations
Well, we really just want the paste to fill the nanoscopic holes between our two metal planes. If the metal surfaces were perfectly smooth, there would be no paste needed, the paste is an awful heat conductor, it is just there to replace the pockets of air, which are even worse conductors. So the tiniest layer, spread out evenly will most likely work best, because the surface roughness on the processor top and a heatsink are very small, so a very thin layer will be sufficient to fill those holes. And you want it everywhere, so it fills the imperfection wherever it is.
Moreover, What really matters, is the type of paste. There are multiple high thermal conductivity pastes, but they are very dangerous to work with, because they are electrically conductive, so watchout!
Thank you. I always says that and everybody is mocking me because it takes much more time to complete the application. But it saves paste too and is overall a better way to apply thicker pastes.
thats exactly what works best
so you talking about the butter technique, right? And do you mean "thin" layer by saying "small", or "small" as in area of application....
@@serdareski3325 He's talking about the "butter technique," a thin layer across the face of the CPU...
I prefer the "X" method on newer CPUs..... but i do remember the times i would thinly spread the thermal paste over entire IHS on P4 cpus. That method provided so much suction between the cpu and heatsink, the cpu would actually get pulled from its socket because it stuck to the heatsink.
the best method is a thin layer specifically covering the die of the cpu, when you put the heat sink on it should spread almost to the edges of the cpu without seeping through
@Keebine Kobold well it's not the "best" in terms of thermal performance, since any method with enough paste will work fine. but it might be the best in terms of using as little paste as possible, getting a consistent application, all without sacrificing performance.
a novice may not realise they didn't put enough paste if they use another method, but if they use the buttered toast method it will be very obvious when there's not enough. for someone proficient, yeah it doesn't matter bc they have an idea of how much to use and they know to ensure even mounting pressure
A lot of enthusiasts are religious about applying thermal paste. I love the fact that you spent the time to prove (IMHO) that they are simply wasting their time. I've been using the 5-dot method for years because it is self-evident that the better the contact between the CPU and cooler ... the lower your CPU temps will be. Thank you!
Seriously I am not a 20 year PC vet. But I have built a few PC’s. My latest Rog strix B550-f-gaming. Ryzen 7 5800x , Asus Dual 3070, PSU -Corsair Rm 850x Gold. Noctua No-d15’s chromax dual fans - 6 Noctua Chromamax black hot swap. That being said. I have applied Every paste except metal‘because’ it scare the bejesus out of me. So to prove the five dot method for at least AMD 5800x. I applied 4 applications already of thermal paste. My second go around with Noctua thermal paste AND thermal grizzly kryonaut extreme. The 5 dot method made it to where I do not have a hot spot!!!!! 5 dot decreased my my Celsius by 15-20%c. Period. And best. Is no Huge hot spot!
Thank you so much I did the spread method and at basic tasks, my fan speed at 70%, cpu was 100C, did the cross method and fan speed 30%, went down to 46C. Cross is the way
I been using a thin spread style since 360 RROD days, you had to apply it directly on the die. Just makes me sleep better at night knowing evey nano meter has some thermal paste crammed in there.
Wish you had done "no paste!". Thanks for calming me about the hardest part I've had in the one PC I built (destroyed first motherboard due to paste issues but that's between you and me). I think the takeaway is: "apply a reasonable amount of paste BUT don't let any get off the CPU cause that could cause issues".
ua-cam.com/video/VAh_p90y_DU/v-deo.html
Yeah that's a good takeaway.
This is how to do science, with great interpretation of results. Well done!
I know this isn't going to stop the arguments, but at least _you_ know the answer, because you actually did the work to find out. Thank you.
For beginners, I would probably reccomend the spread method, I feel as if that would help people get used to how much they actually need
The object of a chosen method of applying the paste is to A) get maximum coverage with minimum thickness and B) not create air pockets by entrapment.
Flattening the paste will always have a radial spread, so a dot will always reach the sides before it reaches a corner. Since the paste has a rather heavy viscosity, the more paste applied to a single spot and the farther it has to spread, the more pressure will be required to reach a desired thickness and spread. So the optimum is a pattern that needs to spread the least to get full coverage, with a minimum amount of pile to the paste, without using a pattern that would trap air as it spreads. So: the X and pentadot, so long as the dot at the center is largest on the latter. If the central dot was too small with the pentadot method, four air pockets would be caught at the cardinal directions around the center. The "buttered toast" method manually creates coverage, but will usually trap air from an uneven surface and have a greater final thickness from using the most paste.
Did you actually watch this entire video which quite conclusively shows that it makes no difference at all, and then feel the need to type all this out, insinuating that there is an optimal method? 🙄
@@5stringaustin It's putting a little science behind the empirical observations. He did them all in ways that won't really cause air entrapment, but some of the methods _RISK_ it for the reasons I explained, which does make a difference.
I found it good to fill the gaps or grooves between the heatpipes on the cooler (GPU), then the the butter method on the GPU. Otherwise it probably wont fill the grooves, depending on the cooler if there actually are grooves...
What I always do:
- Get some isopropyl alcohol 99.9%
- Clean base of cooler and cpu IHS with it
- Get some kitchen paper, dip it in the isopropyl and rub my index with it. This makes it super dry and without any oils
- Use as little thermal paste as possible and use my index finger to spread it evenly across the whole IHS.
- Carefully apply heatsink and tighten each screw slowly.
Works every time.
Some people are against using your hands/fingers to apply thermal paste because it has dirt and natural oils. Makes sense. But if you use isopropyl you get rid of that issue. The card method usually leaves a much thicker layer of paste.
This is best and most professional video i've seen regarding this topic, well done.
I've done 100's of custom gaming PC's from $500-$5,000+. A thin even layer spread across the whole ihs keeps temps low and excess paste to clean off the MB later.
"Shrodingers thermal paste."
Locked an an eternal quantum state of being enough and needing more.
And now I'm watching two different videos, both done 10/10 with good tests. And another one says it does matter, which one I trust XD
And of course I trust my own experience of building over 1k PC, over years. Yeah, it does matter, because it depends on CPU itself.
Which be the other video
This video made me subscribe to you. Short, useful information, no blah, blah for 30 minutes. Practical.
Why doesn't this channel have more subscribers? Great video.
Thanks. Slowly but surely.
I think the take away from this is that the plates are probably several sizes bigger than what they actually need to be so that they have a large enough area to accomodate the different amount of paste being applied to each model. Also the thermal conductivity of the paste is probably large enough for small variations in the cross section area to be negligible.
2:41 - Always clean up 2mm thermal paste around the cpu because the CPU cooler will push the thermal paste about 1mm more.
I didn't realize that. Thank for the info.
The CPU and CPU cooler parts that touch are actually not flat. There are grooves and curves on both sides. That's why the thermal paste looks all wrinkled after you separate them after a while. It's best to put a blob in the center so that when it spreads out, it gets into those grooves properly for heat dispersal.
👍🏻 agree
the results clearly don't agree with your logic
Agreed. I've been doing a pea size blob in the middle going on 22 years now. All my machines, friends, clients', all the same. My son's gaming PC I helped him build 7 years ago, since upgraded, same method, doing fantastic. I say pea size, but actually just a tad bit bigger than a pea but not much more. I'm currently building a new one for me and here I am researching to make sure I'm keeping up with the times. I see no reason to change now. Long live my new i7
@@keppycs you were able to see the grooves get filled? I missed that one.
@@scotmaciver no, but the results speak for themselves
I really appreciate your video. I'm building myself a higher end gaming PC and it's sitting on my counter waiting for me to install the cooler. I wasn't sure which method to use so I've been searching for a while. I'm definitely more comfortable, I can go down and actually finish the build
My question was for the first test of the big thermal dot and the pea-sized thermal dot at the last test. Thank you for answering it.
I'd be curious to the longevity of thermal performance among the different methods. I wouldn't be surprised to see the "buttered toast" method retain its performance longer.
Just to add some historical context to see why this such a pretentious topic: CPU's once upon a time didn't have a heat-spreader, and thermal paste used to contain electrically conductive material, and thermal protections weren't as prevalent. With direct-die cooling, it was critical to ensure every bit of the die was covered, but you couldn't use too much as spill over and short out a surface mounted devices (components); basically any extreme (too much paste, too little, improper coverage), could kill your CPU. Even uneven mounting pressure could crack the die, or be a silent killer creating a hot spot. So the "buttered toast" method became ideal as it helped ensure complete coverage, and helped meter the amount of paste used. There was also emphasis in tightening in small increments in a cross pattern.
But these days, with a heat spreader mitigating a lot of the dangers and non-conductive paste, it really doesn't matter; if anything, better err on the side of too much paste than too little, but neither will necessarily spell death for the CPU.
The dot or grain of rice method gained prevalence with heat spreaders; while the heat spreader protected some of the SMDs, prominent thermal paste was still electrically conductive, so excess could kill the socket. The grain of rice allowed mounting pressure to evenly distribute across the spreader without causing excess. Improper spreading could create air-pockets, so "buttered toast" fell out of favor with heat spreaders as the grain of rice or dot methods were more fool-proof leaving the spreading to the cooler's mounting pressure. AMD recommends the "penta dot" method because of their chiplet design (especially with ThreadRipper/EPYC) means there are multiple dies underneath the (massive) heat spreader; pentadot ensures each heat creating die has some thermal paste coverage; a singular dot may mean one of the out chiplets(/dies) doesn't have direct coverage (e.g. not enough, or might spill over the side).
love how you make the video and use the scientific the method to compare the different ways to apply the thermal paste. However I think that we should also consider the long term repercusion of each method. Maybe the thermal paste gets wasted earlier if you use the dot application rather than the 5 dots or something like that. Again, great video!
It really depends on the type of paste. The paste that's currently supplied with Arctic coolers, for example, is very sticky and virtually impossible to spread "manually" (you end up with half of it sticking to the spreader, and a lot of air bubbles in the middle). A central blob or a cross generally works fine for all types of paste (and will spread itself out without leaving any voids), as long as you use about the right amount.
I work on large inverters that utilize large 1200 amp IGBTs that get bolted to large aluminum heat sinks. If any “method” other than thin coat across a thoroughly cleaned surface is used, that IGBT will overheat and blow its guts in about an hour, depending on load and ambient temp. All of the old thermal paste must be scraped off, then clean both the heatsink and the device with alcohol wipes. I then heat the mating surface of the IGBT with a hair dryer on low til it’s warm enough to not cause the paste to thicken when it’s applied, at least 100F. I squeeze out one long bead in the center running the length of the device. I use a small plastic paddle to spread it out as evenly as I can. Mount the device and press it down firmly and install the four corner screws to secure it. Once the screws make contact, they get tightened down in a cross pattern one quarter turn at a time until snug, then the torque wrench comes in and tighten in same pattern until desired torque rating is achieved. Reassemble the inverter, start it up and let it run for an hour with no load. Shut it down, make it safe, recheck torque, first time always results in about an eighth of a turn to get it there again. Repeat until after test run it clicks the wrench without turning. Then, and only then can it be returned to service with confidence the customer isn’t going to call in an hour or two saying it blew down again.
TLDR; thin even coat across entire surface is best.
I would have liked to see a test without any thermal paste at all !
New sub. Great vid, super impressed. Puget systems & cisco all recommend the 'X' method. Used that since about 2013. Long term (2-4y) is where differences will show in terms of temps. This is where using quality *paste* matters. Using 20% more than you think you need (EXCL liquid metal) is a good idea for business systems.
I thought it didn't matter too much, thank you for confirming man!
I was just debating with my bid about TP application. I'm a butter toast guy. I can't wait to show him this video.
A topic that when 1,000 UA-cam videos say the same thing MAYBE someone will be convinced. There is a video out there with Steve from Gamer's Nexus and Der8auer that basically says, placement doesn't matter, too little IS BAD, there is never too much (but gets messy), paste doesn't "dry out" and air bubbles will work themselves out or be compressed so small, it doesn't matter. And the grooves are unavoidable when the maker puts heat pipes right against the IHS. Compressing them too much to be perfectly flat will make them not be pipes anymore which defeats the purpose of the whole heat pipe idea
I used to spread it out (and I've never bought thermal paste that didn't come with a spetula, so it made sense), but I switched to the dot because it's less work and makes no difference.
However, given that I'm currently reinstalling the slightly bigger brother of your test CPU I might try the penta dot for a change.
The only difference is when you apply too little, causing CPU to overheat. So what's necessary is applying efficiently.
If your Cpu is rectangular (Laptop CPUs/ Direct die) line technique is good as it spreads in length and little breadth, less wastage.
If your CPU is square , Pea sized Dot/ X /Penta Dot works good as it spreads inside out in circle shape.
Spread Method it's really messy as compared to other techniques, not at all useful when u have little paste remaining. Few of it will be wasted on that spatula/spreader. Time Consuming.
Thanks For the video !
Astronaut, doctor, plumber, pc builder, is there anything he cant do?
Thermal paste pro here.
It's not the pattern, it's the amount, thicknes of the paste and the surface area.
X will not help if area is large.
For a super large CPU I would zigzag instead of dots or exes.
Great stuff man! I've been a Systems Engineer for just over 25 years and had to buy some new CPUs which required upgraded heatsinks for a server, and the thermal pads on one of the heatsinks actually had a fingerprint it! Clean it with 90% isopropyl and apply some new paste, right? Well, I've used "buttered toast" forever, I've always been curious as to what the best method is and you have laid that to rest. Thank you, sir!