>2016
>not polishing your cpu to a mirror finish
For what purpose
>>52498142
How many different grains of sandpaper do you need for such an endeavor?
>>52498166
Better heat conduction?
>>52498142
But won't the fact that the inner side of the integrated heat spreader (which makes contact with the CPU die itself) wasn't polished make the exercise mostly pointless?
if there was any benefit to this, why wouldn't cpu manufacturers do this themselves and sell the mirror cpus at a higher price?
>>52498166
some people say it lowers heat but for all the trouble its just 1-4 degree different, theres no real purpose to doing it other than just for being a special little snowflake on /g/
>>52498169
probably up to 8000 grit
https://www.youtube.com/watch?v=Y_tXF-pKn4Y
These guys have been polishing stones to a mirror finish since forever.
>>52498166
When you have a very polished surface (for metals) it means that the surface is incredibly flat.
When the surface of the CPU IHS and the surface of the heatsink are both at mirror finish, then it will increase surface area contact, thereby increasing metal-metal contact, increasing thermal conductivity and finally improve heat transfer through the heatsink.
You still need paste but its effect will be less pronounced because the surface will be doing a lot of the work.
>>52498201
because running individual IHSes on grindstones would significantly decrease production volume, they'd have to pay for the tooling, and in transport the finish is likely to get fucked up without better packaging.
>>52498142
It's pointless unless you replace this thermal paste between die and heat spreader
How is this better than delidding?
>>52498241
>>52498179
that's what thermal paste is for
>>52498336
metal-metal is obviously better than metal-paste-metal is obviously better than metal-air-metal
>>52498142
I am to figure this does nothing?
even if you also polish the heat sink part you are going to need some paste to create continuity
>>52498336
You probably also believe that you have to change your thermal paste often.
I bought all the right grit years ago for my q6600 and never ended up doing it. Plan to do it 4770k after I get to delidding it but somehow I doubt I'll get to that either. I never go over 70 during full load at 4.5ghz anyway. It would be more for the science of it all.
>>52498276
for delidding, your milage may vary.
It's entirely possible for Intel to literally solder the die to the IHS.
http://www.tomshardware.com/news/Intel-Core-i-74960X,23270.html
Soldering the CPU directly to your heatsink is obviously better than polishing or using paste because solder has the excellent function of being identical to liquid while melting.
It fills in all the gaps, and becomes a metal-metal-metal sandwich for heat to transfer directly from the CPU to the heatsink.
Thus your heat conductivity is limited to the lowest specific heat value in the sandwich. In this case it's probably the die itself. Don't quote me on that.
In other cases Intel/AMD might put thermal paste in between the die and the IHS, and so you might find an improvement by removing the lid.
It should be obvious but there's basically no chance of you as an end user of soldering your CPU to your heatsink without something like a BGA rework station, which do something similar.
>>52498460
This may be of interest to some people.
Thermal conductivity of silicon semiconductors:
http://www.electronics-cooling.com/1998/05/the-thermal-conductivity-of-silicon/
Thermal conductivity of copper
http://www.engineeringtoolbox.com/thermal-conductivity-d_429.html
It seems that the process of heat transfer is rather limited by the die itself.
That still doesn't mean you shouldn't use metal heat sinks. Since metals have high thermal conductivity (it varies internally versus at the surface, which is why you see a lot of copper heatsinks plated with aluminum), they can sink heat into the air faster, as to not allow heat to build up close to the CPU.
>>52498241
im pretty sure that if something is flat it has less surface area, heat sinks are solid blocks of aluminum, there think ones with spacing to help dissipate heat
>>52498650
If something is flat, there is more -surface area contact-.
The object certainly has -less overall surface area-, but more -surface area is in contact if both are flat-.
Have some reading comprehension.
Wait, can someone explain why metal+metal contact is better than metal+paste+metal?
What's the actual purpose of the paste if putting the heatsink on top directly is more effective? It glues them together or what?
>>52498745
it fills the little holes in the surface. when the metal is polished to the point where it's like a mirror, there are no holes to fill
>>52498745
paste is used to fill very small gaps
you don't need paste if you have 2 planar surfaces
>>52498784
>>52498786
By that logic the best way to assemble without polishing like an autist would be to spread the paste evenly over the processor, then press the heatsink on top.
Kinda like that one MSI video where they smear thermal paste all over the processor, but obviously not doing it in that retarded way with a card, just wipe off the excess when you press the sink down.
Yet the correct method is "hurr put a tiny pea sized drop in the center durr". Doesn't make sense to me...
>>52498858
>Doesn't make sense to me
The centre is the most important.
It will spread out over the CPU as the heatsink is pushed onto it, and there will be no air bubbles.
The thermal paste will expand slightly as it heats up, so you don't want to place it too close to the edge.
>>52498858
You can see videos of people preparing thermal paste with Plexiglas sandwiches.
I believe that pasty faggot Logan did quite a few.
Deal is, if you spread it by hand, you're guaranteed to introduce air holes because you're not a machine and there's zero chance of you being able to apply paste at a mirror-finish level of flatness.
There's also the fact that just using a sphere of thermal paste is simply easier to apply and doesn't introduce air holes because you allow nature to do all the work.
There isn't a correct way to apply thermal paste, because manufacturers produce differently shaped heatsinks and differently shaped CPU dies.
Consumer Intel CPU dies don't fill much more than a third or even a fourth of the IHS, but high-end Xeons fill the entire IHS.
A "pea sized dot" would not work for high-end Xeons because you'd end up not covering a core.
>>52498939
If you don't understand why "air bubbles are bad", this thread from /diy/ explains how home insulation works
https://4ch.be/diy/thread/920917/#920917
And yes, the concepts of home insulation also apply at a microscopic level.
>>52498225
>1-4 degree different
And thats only if you do nothing but stress test 24/7
>not liquid cooling the CUP directly
>>52499163
I'm surprised the pump doesn't burst through the channels.
>>52498199
there's a gap an a thermal material between the die and the spreader anyway. unless your cpu heatspreader is soldered it's mostly an exercise in futility for a few degrees
>>52499163
you know an intel engineer somewhere has a 20 core xeon at 6ghz cooled like that
>>52498347
I hope you're not considering using this without thermal paste... You would still need thermal paste regardless of how smooth the top of the CPU lid is. It'd still be metal-paste-metal either way.
>2016
>using a penny instead of a heatsink
>>52501029
well... copper dos make the best heatsinks from what I heard
>>52501208
Depends on the penny, some are mostly copper and some are ~5% copper
>>52498347
it would still be metal tim die no matter what. if you're going to waste your time doing this then you better be deliding
>>52498935
do both. or a combination of both, rather. put a very thin layer over the surface, as smooth as possible. then put a smaller than usual amount in the middle. that way you get good coverage and if the right amount is used, very little spill.
>>52501029
>>52501225
you guys still have pennies?
>>52498142
>>52501603
Yup
>>52501603
3===D
>>52501029
but it's a nickel
>>52501981
correction the IHS is made of nickel but the coin is a penny.
Why don't they just make CPUs with the heatsink built right into it for faggots like OP to jack off over the minor heat conduction benefits?
>>52502062
More like why can't they go back to CPUs without IHSes?
>>52499840
if you use a CPU without paste, it's gonna become paste of a different sort before long, hahaha
>>52503202
jesus christ, anon! why would you want to remove your cpu and ejaculate in the socket?
>>52498166
less heat transfer... Keeps the heat inside the cpu.
>>52498179
it would be WORSE heat conduction, because if it's smooth then there is less surface.
>>52504716
nah, if you polish both the CPU and heatsink, they are closer to eachother and the heat escapes more efficiently
>>52498347
there is no such thing as metal-metal
either there is a thin layer of air inbetween, which isolates and heats up and fries your computer, or you ACTUALLY WELD THE CPU AND HEATSINK, then there is only metal conduction
>>52498142
nigger, this only works if you polish the heatsink as well
>>52498142
The heat spreader is not your cpu
You should only polish your IHS if you can polish it completely flat with some kind of machine.
Don't do it like
>>52505038
and polish it so you turn it into a concave bulge so the actual contact is reduced significantly.
If you can't polish it perfectly flat don't polish it at all.
>>52504682
some people just wanna be happy in weird ways, i guess?
