My takeaway from "The Workshop" videos so far: Nothing really matters. Everything is the same. Choice is an illusion. Life has no meaning.

7:52 ... there I just saved you 10 minutes.

1:57 Dot 3:02 Line 3:46 X 4:33 Spiral 5:16 Too Little 5:38 Too Much 6:55 Spread Have a great day ! (We are not gonna talk about this comment being 3 years late)

I tried the pee method but I is hard to just hit the CPU.

what do you apply before using thermal Paste ? . . . . thermal copy . . . . . i am sorry

I just did my first build. Being affraid of puting on to much, i did to little. Idle 42, Load 75. After this video, I took off the heatsink, cleaning, new (and more) thermal paste: Idle 35, Load 65. Thanks LTT!

i always put on thermal paste then get paranoid if i put too much or too little or it went on the side

I think there´s something wrong with the thermal compound I bought... ...it tastes really weird.

I hope so, I have been having to clean my credit cards for years!

I'd really love an updated version of this with new lab conditions to see if there is minor variance in any of the methods

If you ever encounter a "stuck" heat-sink, do NOT try to pull and twist the heat-sink off. This can bend the CPU pins and - in the case of CPU's that have no heatspreader such as the AMD Athlon XP and Duron - damage the exposed core logic. Instead power up the computer for a few minutes if possible, then power down fully. See if the heat-sink can gently be removed. In some instances the CPU heat generated from by the CPU will helps loosen up the thermal paste/thermal pad. I had this very problem with an old Pentium 4 that was left for years.. I couldn't access the CPU lever as it was obscured by the CPU heat-sink. The heat-sink was stuck rock solid. Turned the machine on for 10 minutes, I then turned the machine off, and found the heat-sink shifted MUCH easier, to the point where I could very gently wiggle it until it finally gave up the ghost and separated from the CPU. The culprit was a very old thermal pad that should have been replaced years ago; however the CPU itself survived to fight another day :) Note: Be mindful of running stress tests (which rapidly heat up the CPU) with old CPU's like the AMD Athlon and early Duron. Said products have poor thermal overload which will be compounded by any air gaps between the CPU and heat-sink. Under no circumstance run either of these CPU's with no heat-sink. Better to go into the BIOS and keep a close eye on the CPU temp.. if it rises very quickly towards its threshold then power the system down as quickly as possible then see if the heat-sink budges. Modern CPU's from both AMD and Intel do not have this problem; they will most likely power or under-clock if they sense they are getting too hot.

so conclusion is that the more thermal paste u apply, the more ambient temperature will rise! :)

YES! I've been wondering about this for a while. Excellent summary Luke! You've done some real myth busting here. This is super useful to know for other applications where heatsinks are used too.

How the fuck are you wearing only a t-shirt at 17c???!!! That's at LEAST chilly.

Just upgraded pretty much everything on my PC over the last month or so, and every time I enter a question here you guys got a video about it leaving me with no questions, great and helpful channel :)

The question that I didn't see anything mentioned about and was thinking this whole time since the temps are basically the same - what is the cooler's fan RPM to maintain that temp? That would indicate if one amount/application method was more/less thermally conductive than another amount. Or did you use a fixed fan RPM in your test?

what about using mustard instead of thermal paste.

i made mine in a swastika form. never gets above -3 (c)

You forgot the method where you take your CPU out for an evening of romance.

I have always applied thermal paste by applying it in the middle of the processor, then using a flat blade completely covering the entire chip/thin layer. I never have excess squirt out and it's always been fine. I prefer this over the dot method because there is zero chance of too much be squished out anywhere.

I have always done the Pea dot for my builds, this is very interesting results!

I think the test would be better if you used an air cooler and possibly tried a cycle without thermal past as a control

so if I use non conductive thermal paste I basically can't go wrong as long as I put enough?

You put a grain of rice sized portion on the cpu, then you use a rubber glove or plastic sandwich bag over your finger and spread it evenly over the entire surface before mounting the heat sink / block. The paste is only really supposed to fill in all the micro gaps in the surface. You want the plate to make contact with as much metal surface as it can. It's old school but still the best way. 🤓

TL;DR: As long as you use enough, you're fine.

While I still prefer the full even spread, it becomes a lot easier when you realise that the metal square on top of the CPU is a heatsink to the core and that the core is only in a certain spot on the CPU anyway... with negligible difference I'll just do the full spread to avoid issues if I don't know where the core is.

I think ideally you want nearly full coverage just not overflowing and also the thickness of the paste brand affects how much it spreads

One important information is how fast the fan/watercooler is running - maybe your MB is making things go to 47 degrees regardless of how good the application method is.

I do the Mona Lisa method.

I think The Rock called it: "Let's look at the various ways to pa-" "IT DOESN'T MATTER HOW YOU PASTE IT!"

The classic method during that brief period where CPUs had a bare die was to use a random expired credit card or library card or similar and spread it like you spread butter. The idea was that it was more important that every square mm was covered right when you turned it on rather than having the least amount of air bubbles or whatever.

On thick paste (Arctic Silver 5) I use the double thin method, take a small amount, apply it to the heat sink with a circular motion, which will create a round circle, but it needs to be very thin.(almost see through thin) Then I use the same method for the cpu. When the two very thin coats join, you end up with one perfect layer and it reduces air gaps, it also reduces paste from leaking out. In fact, when I tear down my systems for cleaning every year, I never find any paste that has leaked out and my temps are always super low (below 40c), even overclocked.

i would like to see this done with a much less efficient cooler. might have a bit more room for slightly different temperatures.

the piano music really helps take the whole experiment to another level.

This is helpful. I've got a rigid tubing custom loop and reseated and redid the thermal compound like 3 times before I realized that my thermal paste application, no matter how I did it resulted in the same performance.

"It doesn't spread" he says, as he uses a card with paper over it. You know the kind of paper you use to wipe things up. You ever seen a bricklayer put his trowel in a pillowcase? Hey, if you're afraid you'll get your credit card dirty, you could always wipe it after. It's a piece of plastic.

Canadians... 17°C and Luke is wearing a t-shirt and no signs of discomfort at all!

I get my best results with an application of thermal paste in the shape of the Greek letter theta, written in a firm yet slightly effeminate hand with just a hint of sassy boldness. I would like to see some videos about how to adjust chairs please!

I use the whole tube... My temps stay at 10C, and my Ambient temps are 33C

this video is old, but for people wondering, on the arctic silver guide (specifically for Ryzen CPUs) the best way to apply it is a 1mm wide line across the cores’ length, instead of having the processor completely covered. this is to have better contact *only* directly where it needs it, as to prevent loss of cooling fyi, i think some stress tests might be wrong in videos, because of the “break-in” time of the paste. Arctic Silver 5 has a break-in time of 200 hours, and the results change several degrees when comparing directly after application and letting it cure.

IMO The real test would be thermal paste performance one month after being applied.

shouldve overclocked it so it was running around 70c. if you idle at 20c and load at 47c your delta is 27c. if you idle at 20c and load at 70c your delta is 50c. much easier to see differences that way. small changes in thermal conductivity will be amplified

My temps were running high on my first build with the pea method. So I used a thin layer over the whole thing with a card (not bleeding over the edge--cleaned edges with Q-tips, TP, and soft cloths), and a pea method on that. Went down significantly.

I use about half a pea in the middle and let the pressure of the heatsink do the work. This is for single-die chips, and the paste spreads out far enough to cover the area where the chip is, without getting messy, and being thin enough to let the heatsink contact the heatspreader, so that the thermal paste is ONLY filling in the microscopic gaps.

Why do you use such a high end water cooler for a test like this?? You should have used a stock cooler for higher temperatures where the thermal paste has a larger impact on the cooling of the CPU.

You're wearing a t-shirt and it's 16c? You'd love south-east Australian weather!

I've never got full coverage at the corners using middle dot. So now I go for one big dot middle and smaller dots near each corner. Does the trick. Is not going air bubble anyway, when paste meets paste it just combines.

I would have liked to have seen individual core temp for this to see if more even coverage to the edges would help with the issues that some cpu have with single cores getting to high temps( i7 7700k as an example)

THEME from "The Workshop" thus far: Paste application doesn't matter all that much. And cable management doesn't matter all that much. Love your videos Luke! Keep 'em coming! :)

I put a "T" on my old computer and it made it great again!

No sense video for me, try every test pushing the processor first and then check temps. Idle cpu don't need too much cooling power

I'm curious why you used a liquid cooler for the test. I feel like with a worse performing air cooler, it would matter a lot more how well the paste is performing. and since the temps would be higher in general, there would be more room for tiny imperfections to change the results.

dont you feel cold staying in a 16.8C room with only a tshirt on?

Everybody knows you're supposed to use the whole tube of it.

It would really help to have better shots of the AMOUNT of paste that is applied so we can use it as a reference. All the shots were far from the CPU and tough to gauge the amount.

this thumbnail gets weirder the more you look at it

You should do a video comparing different types of thermal paste. Such as the standard cheap silicone white stuff verses the more expensive silver type. I heard to just get the cheap thermal paste because it will work the same.

is there any diference in using thermal paste instead of thermal pads on the gpu's VRAM?

For the past 25 years of building dozens of PCs, I've always used a razor blade to spread the paste evenly across the entire die while ensuring no missed spots, and no heavy spots. It takes some practice to make it perfectly even. I've seen bigger differences with the type of paste used than the method of application.

Me before replacing cpu: “Don’t mess up! Don’t mess up!” Me after replacing cpu: “I messed up...”

luke, just because nobody else would - you should have done a test with NO paste too.

How about a video where you put pre-applied thermal compound vs using your own?

9:05 Dude that T-SHIRT is absolutely fire

I think the main concern for me is air bubbles/pockets in the paste. I used the the almost too much method with a card to smooth it out evenly on multiple computers. The reason? one of my laptop computers that ran forever without any heat problems had WAY too much thermal paste applied by the manufacturer. It was a laptop and it had the thermal paste all over the place even on outside circuits and I'm not sure it did any harm. It is still running strong today. Granted I didn't apply as much as they did but pretty damn close. I also think there is some type of clear silicone material covering and protecting the circuits around the cpu on that laptop. It is a worn and torn workhorse laptop that I should replace with a newer one so any risks involved with too much thermal paste is not a issue for me. The time to take it apart again could be better spent elsewhere. I also drilled a damn hole in the back of the outer plastic case right above the cpu to let heat escape because there is no vent in that area. They really need to make better cooling solutions for laptops.

17c room temp? how are you not freezing, i need atleast 23c

How about different types of thermal paste? Would that make difference?

This is flash great comparison. "Sooooo.....don't use too much.....or too little." I'd like to see a revisit series of these where you test a variety of pastes with a shorter variety of methods.

Pausing at 1:50 I used to have a piece of paper and a block of wood that had a similar size surface area as the given CPU. I would then apply an amount of paste on the paper, in a dot, of which I'd be guessing how much I'd need. Then I'd press the wooden block on it, too see how much it spread to gauge if I used too much or too little. (Yes I usually buy two tubes. xD ) With that, I'd have a fair approximation of knowing of how much I needed. And I use the single dot method with minimal 'wrinkling' to avoid air-bubbles.

I need Luke's fucking shirt that he is wearing during the CrunchyRoll advertisement.

I think the test should be done with a stock cooler, or better a passive one in order to get a more spread out results.

You didn't really stress test the CPU by running heavy processing. That's when it matters.

Would love to see this retested on current cpus and to see each individual core temp. It may make a slight difference if not all is covered completely

Dayum Luke, Back at it again with the thermal paste

Latest Workshop videos: does it matter? - No

one thing you could try is setting the cooler down at an angle like closing a book instead of trying to place it straight down when doing the spread method. This will basically roll the air bubbles out the side. I have also read that when the CPU heats up to max the air will be pushed out when the paste sets...

Seems like the single dot often misses the corners, and single line often misses the sides. I'd rather like to see, just once, someone try a "2-line" version [ | | ], or a "dice" (5 smaller dots) [ : · : ] version, to see if either gets better coverage over the whole chip surface than just a single dot or line alone, but without leaking over the edge. Like would the two lines spread out and merge in the center, or would it potentially miss the middle with an air pocket? Would the "dice" version hit all 4 quadrants, and make sure not to miss the middle (without creating quirky air pockets)?

KZfaq has a new poll feature which would be easier to use than strawpoll especially on mobile. Could you consider using it?

Ehi, less IS more... less is MORE APPLICATIONS.

Jesus... all of these applications methods and NOBODY bothers to link to the manufacturer's instructions. I've got a Cooler Master 212 hyper on the way, and after reading the manufacturer's instructions, they say to use their own (high quality,) paste. I had ordered Artic Silver5, and THEY say to use it in different ways for different processor core configurations. i.e.: With my i5 6600, a quad core, they recommend a line 1mm wide the goes from top to bottom without hitting the edge, because the quad cores are centered like a tall rectangle, so since the cores align in this way, the majority of the heat is generated down the middle from top to bottom. AS5 also recommends "tinting" the surfaces -applying a very thin layer all over both surfaces, and then a line that gets smeared into place by the cooler base. They also advise that there is a 200 hour break-in period to achieve peak cooling fromt heir thermal grease. Even still... Cooler Master recommends NEVER using your fingers to spread TG, because it leaves behind contaminants that WILL degrade performance of the compound. Also, non-lint cloths. No qtips, no paper towels.

I always used a method I learned from the Arctic Silver website. Finger in plastic bag for the bottom of the cooler (and wipe off excess) and then spreading on the CPU with a plastic card. Glad to see the pea method works as good, will save me the heartache next time.

I am curious: does thermal paste ever need to be re-applied?

next vid do different thermal pastes including stock

I just cleaned and re-applied yesterday. On removal the cooloer and CPU they seemed to have a gelled amount of dried paste. I used a BB size drop and spread it thin and even with a Q-tip, The result was a drop from 72C to 53C. I am liking the q-tip method.

When you use a water block to test for thermal changes, it's not going to show, as clicks take forever to heat up. I appreciate the video to show coverage.

Temps change when the paste dries out a bit.

Hypotheoretically, adding "Too Much" or "More" thermal paste might be the safer bet for the long-term. Because the paste will last longer before drying out to failure. #TeamTooMuch

I have a small (3/32" wide), long-handle plastic spatula made for the purpose. I cover the entire CPU top AND the cooler contact area. If the cooler allows (most do) I push it on, then lift and rotate the cooler 90 degrees and push again, removing any excess. Thermal engineers will tell you that total contact area AND paste thickness both are important -- max area and little thickness. Your tests suggest little difference, but I shall continue to maximize the contact area.

Skip the video(or go to 7:56). Read this comment that summarises the whole video instead. Thank me later. Temperatures: 1:57 Dot - 47C 3:02 Line - 47C 3:46 X - 47C 4:33 Spiral - 47C 5:16 Too Little - 54C 5:38 Too Much - 47C 6:55 Spread - 47C

the important thing is that the thermal paste has to cover the cpu surface

I've always used the spread method and never had any issue. I remember several years ago when some big tech channels endorsed the pea/line method and said the spread method was not suitable, despite not including temps or a comparison such as in this video. Honestly do what works for you, there is no magical application method, just invest in a good paste, (I like Arctic Silver 5), apply it and have fun gaming 😎.

Pretty sure the amount of functional paste remains the same per unit area for a given mounting pressure. Too little thermal paste simply doesn't cover the IHS enough, beyond that it's always going to be the same. The only difference being how much ends up being pressed out and around the sides of the cold plate and IHS. Therefore, best off to go with slightly too much rather than slightly too little. Assuming you're not using a paste which is particularly electrically conductive.

What really impressed me is that you were wearing only a T-shirt with an ambient temperature of 17ºC (that's 63ºF for those into Barbarian measuring units). I live in a subtropical area and anything below 18ºC is enough to make people wear at least a light sweater here. With 10ºC (50ºF), they're wearing heavy wools and thick parkas. Most, of course, will die without ever having seen snow except in photos and movies.

Should try this with the 220w FX processors that crank out more heat.

Draw a phallus with thermal paste. Gives 69 times better cooling.

I love you guys on LTT. but the biggest issue I found with this test was for a thermal test you were using a aio and not an air cooler, you only ran Aida for 10 minutes for each test, not long enough for the aio liquid temps to equalize. for a short test time a air cooler would result in better metrics, metal heats up faster than water. keep making great videos guys!

I put the paste in the middle, put on the cooler and slightly circle it left and right to make the weight to spread the paste evenly.

Try an air cooler instead of a liquid cooler next time.

18:17 Love that shirt!

8:14 "That's what makes these things fun". Luke thinking inside **kill me**

Your "more" was the same because it was still in range of a safe amount. The "more is bad" is when you add even more than you did and squish it out over the sides. Effectively there can't be "too much" in between the CPU and the heatsink because they squish together by force, it's going to squeeze out any excess. It's this squeezed out part that becomes bad/dangerous/causes odd issues or heating. So yea, adding too much is pointless because it's either not actually too much or you've overflowed off the CPU. So long as it's not preventing the heatsink from being as tight as possible the danger is in overflow. Your conclusion is pretty much what anyone with some logic would expect; though it's nice to see it tested. The spread method is at best the same (but messier and takes longer) as simply the dot. Sticking the heat sink on is going to do the spreading for you. At worst if you screw it up it could create air bubbles.