How often do you refresh your thermal compound?
>>51714340
Never, because thermal paste is nothing that should ever be refreshed.
>>51714340
Why should one do that?
>>51714348
It should be redone if you ever take off the heat sink, but this.
Once a year at the beginning of the year so I have good paste for the whole year. You know how it goes. New year new meme #FreshStart
>>51714348
>>51714360
Doesn't it eventually dry out and stop working as efficiently as newer paste would? Tell me I'm retarded if that's a retarded question.
Every time I change my cooler
>>51714381
After a very long time. Longer than you'll have to worry about if you upgrade your computer relatively frequently.
>>51714381
Paste is only there to get rid of all those irregularities and to have no air between CPU and heatsink. Metal on metal is way better than Metal on themal paste
Every 30,000 miles.
>>51714381
depends on the paste and also doesnt matter in the consumerist gamer era where people get new builds every year or so
Every month, but I have my own original method. I just add a drop of paste wihout removing what was there before.
>>51714348
>>51714360
You're wrong.
Thermal paste is a product that does expire.
Thermal paste can and will dry outside of its container.
It definitely lasts a VERY long time (around 3-5 years), but after that point you may find severe temperature issues on laptops, which can cause the battery to overheat, and the laptop to shut down as a safety precaution. Obviously the problem gets worse as time goes on.
Personally, I replace the thermal paste once every three years.
>>51714360
Well, I was obviously implying that you never need to refresh it when you dont exchange/take off the heatsink but thx for clearing this up for all the newfigs.
Also, pro tip for anyone wondering:
whipe off pre applied thermal paste from any cooler you buy because its most likely shit.
>>51714381
Not a retarded question, people often get confused. Dry paste should be as efficient as long as it makes good contact with both parts. Meaning, if you let your paste dry out and then you take off the cooler and put it back onto the dry paste cooling will most likely be shit.
Thats why you need to reapply it,when you take your heatsink off.
As long as you dont take off your heatsink though, you will be fine 100% of the time.
>>51714403
>metal on metal is way better
Right and wrong. While "metal on metal" can theoretically be better, there is literally no scenario in which you can put 2 metal pieces together without having at least a few nm layer of unwanted molecules inbetween (mostly air), which makes your cooling turn into shit.
This is why you always use some kind of paste when efficient cooling is required in a "metal on metal" scenario.
>>51714419
Ive had preapplied thermal glue on my fathers prebuilt PC for ~15 years and cooling was as good as on day one (after dusting the thing off).
Again, as long as you dont remove your heatsink and/or let a "gap" between heatsink and CPU happen (meaning even wiggling the heatsink to left and right can be a bad Idea) you will be fine.
every ~5 years
>>51714419
that's fucking retarded. if you apply thermal paste PROPERLY you should NEVER have to replace it. shit is not fucking mayo you slather on a sandwich. thermal paste is only there to fill in the MICROSCOPIC GAPS on the surface of the cpu lid. as another poster pointed out - metal and metal is what you want
I replace it ever 1-3 months to keep things nice and cool, I also take the time to clean and dust my heatsink and make use some compressed air to blow dust from the socket.
>>51714366
>>51714418
>>51714437
>>51714455
You are literally retards
>>51714455
You do know thermal paste has a curing period. Replacing it that often is stupid.
>>51714447
I'm aware you fucking dunce.
You think you can seriously do better than some stupid chink putting together laptops?
They have machines for this.
Laptops with pre-applied thermal paste last 4 years tops.
>>51714418
More paste is worse than less paste. What you're doing is decreasing the amount of good metal-metal contact.
Metal-metal is best, metal-paste-metal is acceptable, metal-air-metal is not.
By putting new paste on top of old paste, what you're doing is decreasing metal-metal contact, which reduces your contact efficiency.
Not only that, but old paste eventually dries out, which adds air into the mix.
>>51714496
>Laptops with pre-applied thermal paste last 4 years top
Jesus, its time to stop posting dude. Just stop and go read stuff on the internet before making up bullshit like this.
>metal - metal
There is no such thing as "metal-metal". There is ALWAYS air in the middle which is why we need paste.
Again, stop posting.
>>51714496
>I'm aware you fucking dunce.
>You think you can seriously do better than some stupid chink putting together laptops?
>They have machines for this.
>Laptops with pre-applied thermal paste last 4 years tops.
that's not a problem with the thermal paste einstein. it's your piece of shit acer using sub-standard components and pants on head stupid engineering/thermal design that's causing the solder joints to get fucked. do you even fucking repair laptops? it isn't about replacing thermal paste it's about playing follow the lead and find the weak solder joints roulette.
>>51714496
>You think you can seriously do better than some stupid chink putting together laptops?
Yes I'm 100% confident that I can. Have you even ever replaced the paste on a laptop before?
The quality of thermal paste used for laptops is of shit quality to begin with because of planned obsolescence
>>51714535
>confirmed for never repaired a laptop
Dumbass seriously thinks that cold solder joints are the cause of most computer failures, and not consumable components.
Enjoy reballing your chips and soldering broken traces together to no affect, try something simpler.
>>51714530
Faggot have you ever even seen a picture of a scanning electron microscope?
It's plain that there is such a thing as metal-metal contact, what the fuck do you think friction is?
Please go and take your below highschool education with you.
>>51714573
>there is such a thing as metal-metal contact
Im talking about a usual CPU-heatsink scenario you fucking moron. Because unlike shitposting faggots I like to stay on topic.
>>51714593
please don't touch a computer again as you've got no idea what you're doing.
>>51714419
>You're wrong.
>Thermal paste is a product that does expire.
I have several 15+ year old computers which haven't had their CPUs removed since I put them on. Had the paste expired, the temps would have gone up at some point. Laptops have heat issues because the people putting the paste on them in the factory are poorly trained and glob it on. It has nothing to do with some sort of expiration date.
>>51714606
Are you literally this fucking retarded?
We use thermal paste because "metal on metal" DOES NOT WORK inside a regular fucking PC. Because even if 50% of the metal layers are physically touching, there is still a lot of irregularities and air between the rest of metal pieces
You are srsly posting in a thread about thermal paste and you have literally no fucking clue what its for and how heat works.
Please kill yourself you fucking piece of shit.
Does thermal paste really have an expiration date because this thread is the first I'm ever hearing of such a thing.
>>51714661
No. That's just the idiots talking in this thread. Don't listen to them.
>>51714496
Are you fucking serious?
The shit they put on laptops comes out of a giant fucking tube that still costs less than a tiny thing of AS5 and is thrown on with no care at all.
The BEST fucking thing you can ever do to a laptop is the second you get it, fucking the useless factory thermal paste off and putting actual good shit on.
I did this with my MSI CX70 and the temps when gaming dropped a good 4 degrees.
>>51714653
no need to get so upset, idiot.
do i just fill the entire area with paste then?
>>51714704
is this how i want it
>>51714704
You want to cover the socket first so it's sealed properly.
>>51714704
Apply one small drop to the center and place your heatsink down in one swift movement. Make sure not to slide it around on the cpu before you tie it down.
This is objectively the best method
>>51714722
I once saw a computer restore shop employee do this. I called him out on his bullshit. Only a true fucking retard would come up with that idea.
>>51714724
dony troll people anon!!!!
>>51714729
>hurr one method for all CPUs
Modern intel CPUs work better with a thin line down the center and a small bump in the center, and you ESPECIALLY do NOT want to do that method with a 10+ core CPU, as there are cores right out near the very edges and you need full coverage.
https://www.youtube.com/watch?v=-hNgFNH7zhQ
>>51714770
absolute mad man
>>51714752
>there are cores right out near the very edges
The actual die is much smaller than the top surface of the CPU. There are no cores out near the edge silly.
>>51714690
>don't be upset because I was a fucking retard.
Me being upset because of your fucking stupidity doesn't make you less of a retard, though.
Kill yourself.
>>51714798
calm down kiddo. now is not the time to throw a fit.
>>51714790
This is 8 cores.
Now picture 18 in almost the same size.
>>51714835
Are there Haswell-E CPUs with 18 cores? This picture doesn't help your argument.
I generally repaste my computers on average every year. I only repaste them after the first time if I have to remove the heatsink though. I tinker a lot and I end up swapping out parts more than I like to admit and so I find myself with a pile of parts that used to be, and will be shortly again, my laptop or my desktop.
Unfortunately, I'll never be able to repaste my SP2. Oh well.
Also, I used to repair computers for a living and I swear every single time I cracked one open (1-3 years old on average), it looked like the paste was over applied and it had dried and cracked off. I learned that if you leave the OEM stuff on too long then it will dry up and, I guess because of thermal properties of different materials, crack up and be worse than having no material at all since you now have large air gaps since they were always running super warm.
>>51714853
How the fuck do you not know this?
http://ark.intel.com/products/81061/Intel-Xeon-Processor-E5-2699-v3-45M-Cache-2_30-GHz
And yes, it does help my argument because even the fucking 8 core is using a hell of a lot more than just the 'middle' of the heat spreader.
Using too small of a pea on a big core chip is a very, very bad idea.
>>51714752
It varies with certain heatsinks as well.
If you have one of those retarded heatsinks that aren't perfectly flat then you have to apply paste to the cooler itself, and then apply the cooler to the CPU to ensure proper spreading.
You run the risk of putting too much but it's better than having no spread.
>>51714653
Would be interesting if someday a company releases some kind of cpu/cooler/mobo combo thats just made out of 1 hunk of metal without the need for thermal goop.. Qould be the death of cpu upgradeability tho
>>51714835
I think that this argument is very poor because the people that this would actually be true for is such a small minority in the overall scheme of things.
On most processors, the die is very small and in the center of the heat *spreader*.
![387508135_979[1].jpg](https://i.imgur.com/ilJiH2l.jpg "387508135_979[1].jpg")
>>51714867
Cheapo chink thermal pastes have curing times.
What makes you think expensive ones don't?
I've heard that there are thermal pastes with curing times of up to 2 years, or several days straight of continuous thermal cycling.
>>51714835
Did the CPU make it?
>>51714889
The argument being poor due to applying to a minority doesn't mean it's wrong.
If anything, the information should be as applicable to as wide an audience as possible.
If you get server monkeys applying paste according to /g/ advice, and then subsequently Facebook fails across the United States for a whole two seconds as a result of a few cores failing, then what?
Wish that someone provided more accurate information?
Wish that server monkeys knew more about the intrinsic properties of heat transfer and fluid dynamics in regards to amorphous solids?
Or perhaps just distribute accurate information in the first place?
>>51714872
>How the fuck do you not know this?
It's well out of my price range.
>And yes, it does help my argument because even the fucking 8 core is using a hell of a lot more than just the 'middle' of the heat spreader.
I think it would still be fine for that 8 core CPU, but I agree if they tried to shoehorn 18 in there you would need to do tiny x perhaps.
>>51714897
I guess my point was that oems are generally very liberal with thermal paste and it's a mess to clean up. On top of that, it seems that the paste cooks and becomes less effective. I only ever replace my paste if I have to remove the heatsink. I don't seek out to replace the paste, it just so happens that I do something that I have to remove the heat sink enough that I end up replacing the paste every year or so on my machines.
eg - I fell for the ITX meme so to replace the wifi card in my desktop, I need to be very creative with a screwdriver bit or remove the motherboard and heatsink. The mobo came with a bgn card that I replaced with an abgn card then eventually an ac card when I upgraded to ac.
>>51714436
>there is literally no scenario in which you can put 2 metal pieces together without
>he doesn't solder the heatsink onto his CPU
everyday so the thormal paste wont go bad and my gaming temps stay good
it sure costs many dollars monthly but good gaming temps worth it...
>>51714917
>my racecar only runs on race fuel
>I better put race fuel in my civic too
The way you handle an enthusiast or server processor has nothing to do with a literal facebook machine.
Not that I consider facebook important, but I swear if someone responible for facebook servers fucks up the application of thermal paste so bad that it brings the machine down because of something he read on /g/ then holy fuck would facebook deserve it.
The difference between optimal application of TIM and incorrect application will only be a few degrees.
>>51714986
No fucktard, the difference between applying TIM properly and not is the CPU shutting down due to hitting the limit of it's throttle.
On 10 core+ CPUs you can do this to a single core causing massive problems especially if you have software setup to send a warning for temps on average rather than on hottest core.
This is not the same as running race fuel, it's the difference between having a wee bit of paste go onto the side of the CPU for your lazy ass to clean up or having your CPU burn
>>51714835
That's 12 cores with 4 of them locked down.
>put my pc together around 3years ago
>average cpu temp, 30 C on idle, up to 50 on load
>check temps now
>still same temps
All this shit about changing thermal paste is bullshit
>>51715031
And the 18 cores are probably 20 with 2 of them locked down.
>>51715019
I feel like you have no concept of thermodynamics. Do you know what a heat spreader actually does?
If for some reason only 9/10 cores had the proper cooling then the 10th core is going to use the other 9 cores as the heatsink and dissipate the heat that way. It may result in slightly higher temps, but again, we're only talking about a few degrees here.
PS: I'm talking about suboptimal and optimal applications of thermal interface material, not using a crooked penny as a heatsink.
>>51715031
Ha you're right. He did bring up a good argument though. Once you get 8+ cores you might need to rethink how you apply thermal paste. In the future it might be better for Intel to increase the size of the board to POWER proportions so you don't run into this issue.
>>51715077
And if you have a 12 core with 8 running and one of those isn't close to the others, you're relying on a thin bit of metal to spread all that heat.
That does not work properly or we'd see much more novel CPU cooling solutions.
>>51715121
>you're relying on a thin bit of metal to spread all that heat
You still have that big heatsink touching the processor.
>one of those isn't close to the others
I don't believe that this layout exists but I wouldn't mind being proved wrong.
I still think your forgetting that the TIM is not the only thing cooling the processor. You're not wrong, you're just exaggerating so much and making up hypothetical situations that won't happen.
Again, someone with server-grade or enthusiast-grade hardware will have already done their homework or will be able to fix it before anything breaks. Suboptimal application of TIM is NOT going to cause spontaneous combustion of the CPU. Yes, temps may be worse. Yes, it may throttle down slightly. That's it though.
Grind up some solder into a dust and smear it around the CPU. Overclock until you hit 200°C to melt it and you end up with a nice metal-metal-metal heat transfer. No air gaps here
>>51715201
How to apply Thermal Interface Material - Housefire Edition.
>>51715201
See now i'm fucking wondering if you could de-lid a CPU, weld a heatsink onto the heatspreader and then put it back on and have it work
>>51714340
I never 'refresh' it because by the time that the paste would degrade, years time, i've already upgraded my processor or swapped out the system for a new one. There is no need to ever re-do a thermal paste application unless you're still running an original P4 or something. Or you fucked it up and have obviously high temps.
Never obviously you just leave it for 10 - 20 years and use toothpaste that was what I was told by a nice user on here.
>>51714340
At summer, when the heat is so hight that the computer stops for overhead with just playing a flash game.
>>51715201
Now you say that jokingly, but...
>>51715201
this is what intel (used) to do for a long time, only it was between the die and the IHS.
Of course you couldn't actually heat up the CPU to 200C but I assume there's another way to do it without cooking the chip.
https://www.youtube.com/watch?v=D-pr4cf70qY
>>51715302
Shit, that actually seems like it could work.
How do the sooper dooper liquid nitrogen cooled overclockers do it?
>>51714403
thermal paste IS metal
>>51715032
it really is. the entire purpose of paste is to fill in the air gap between cpu and heatsink. it then freezes into that molded shape. so sure, if for some reason your heatsink or cpu shell somehow develops a new shape chances are you really dont have to worry about it.
and if you still think you should please give me a scientific reason explaining your position
>>51715895
This is a troll thread.
when i change cpu
so never
>>51714419
true, every product have expire time. Its demands on quality of paste, and other shit. But I'm changing it only when do massive cleanwork. At last, 1 time in 3 years
