AMD doesn't fuck around with TIM garbage. https://www

File: mpv-shot0017.jpg (78KB, 1280x720px) Image search: [Google]

78KB, 1280x720px

>>59185782

No need for deliding but it still might benefit from direct die cooling or at least sanding/lapping
The latter being more sane and economical

>>59185812

>No need for deliding but it still might benefit from direct die cooling

If has soldered, no.

>>59185856
Direct die is super autism but it still should be more efficient than through the IHS, even soldered.
Lapping is just easier and less destructive.

>>59185898
The difference would be so minimal i'm not sure why anyone would bother. You won't get any more of a stable OC, you'll just lower the temps by a single degree or two, which isn't gonna be the difference between 4.5GHz and 4.6GHz, it'll just run slightly cooler.

>>59185898
>Direct die is super autism but it still should be more efficient than through the IHS, even soldered.

No again. The material used in the solder has an infinitely higher degree of conductivity than any commercially available liquid metal thermal paste.

Your statement would have made sense if you used the same material as the solder.

>>59185782
Isn't the gap caused by it being a MCM?!?

also touching the pins wtf!

>>59185940
From all I've read so far Zen's OC seems to be 100% limited to heat, these chips can't take 100C temps like Intel, I think I read somewhere that TJmax is somewhere around 70C

>>59185959
>Isn't the gap caused by it being a MCM?!?
No, you moron. Its a single die. Those are solder pads.

>also touching the pins wtf!
Doesn't matter if you're properly grounded.

If it's soldered then sanding should be a cheap way to lower temps, if you're overclocking to a notable degree of course

>>59185968
From the looks of it the biggest limiting factor is power delivery, the VRMs on most of the launch motherboards don't seem that great.

>>59185898
>>59185812

Would lapping using a dremel with a felt wheel and very fine grit polishing paste be extremely retarded?
It's what I use for trigger jobs and it gives you a mirror finish without really removing any material down the road.

>>59186016
As expected of jews wanting to cheap out on pennies on a new platform by using low amp chokes, fucking kikes.
Hopefully ASrock saves us.

>>59186019
Are you actually retarded?
The surface of the IHS needs to be perfectly flat for good heat transfer into the cooler.

>>59186019
I'd go by hand on a measured flat surface going from 600 grit to 2000 in 200 increments.

File: 1200px-Belt_sander_bosch.jpg (185KB, 1200x951px) Image search: [Google]

185KB, 1200x951px

>>59186019
I'd use a belt sander, bitch.

File: Heastsink_diagram.jpg (13KB, 500x197px) Image search: [Google]

13KB, 500x197px

>>59186038
bruh what do you think the TIM/thermal paste is for?

>>59186300
Nigger, you want to use as little TIM as possible.
TIM exists so there are no air pockets between the metal surfaces to act an an insulator.
You cannot hold a dremel and perceiving a 100th of a mil valley with your naked eye, but that will negatively impact temps. You have to be an actual, honest to god, sub human retard to think thats a good idea.

Nice to see how it's oriented under the heatspreader. So, for paste application orient it so that the triangle is top right and paste line accross it horizontally is the way to go?

>>59186393
he don't want his heatsink overheat you idiot!

>>59185968
Intel 6700K has a TJmax of 71'c.
Spec sheets are not good indicators of what the parts can run at.

>>59187108
Eh, really? My i7-920 had a 105C one IIRC.
When did it go down so much?

>>59187133
Even if the spec sheet says thats the TJmax, almost everyone i've seen considers ~85-90C to be the true max.

>>59185977
>>also touching the pins wtf!

>finger oils corroding the copper
W T F

>>59187180
>>59187108
I've heard that the 7700k and the 6700k have the same TJmax @ around 100°C, which would explain why they throttle at about 95°C, pretty sure the 71°C value floating about is actually for Tcase.

>>59187358
>>59187358
gold*

>>59187372
Yeah tcase is always a good bit lower usually 15-20s lower than coremax.

File: who else.webm (1MB, 800x450px) Image search: [Google]

1MB, 800x450px

>>59185786

>>59187972
wtf

>>59187358
Paranoid Autist

>>59187358
>gold corroding
ok buddy

>>59186029
anyone fucking around with the asrock taichi for review or was everyone given the same asus motherboard?

>>59185947
removing material between the heat pipes and the cpu will always help cool it. with everything else constant, more material allows for a greater temperature gradient between the heat source (cpu) and the outside world

>>59186019
hand lap that shit till its flat and around 2000 grit,

if you didnt fuck up, it should be damn near mirror finish already, you want to take some green paste and put a buffer on a dremel at that point, i honestly doubt you will make a pit.

>>59188582

My nigga, if you do this using a material other than the same one used in the solder, you will have a lower heat transfer than you would have using the soldered IHS.

The delid process only makes sense in processor that are not solder like Intel does in the mainstream series.

What is difficult to understand it?

>>59188777
I was assuming that the IHS isn't replaced, and so removing it will help. Granted using thermal compound between the die and the cooler may not improve heat transfer, but it might. You have to weigh up using more material, of which some conducts heat better, with using less material. It's not that simple

>>59187358
>copper
it's gold

>gold corroding
stay in school kid.

>>59190241
> not sweating aqua regia

get on my level kid

>>59185812
>it still might benefit from direct die cooling

1. you can't remove the IHS without ripping the die in half. They are soldered together.
2. solder transfers heat better than any other thermal interface you can put between the die and a cooler.

>>59185782
Why don't we just solder heat sinks onto the CPU? Like, with molten tin, or something.

>>59190539
it'd be a bit tricky to install since the heatsink will quickly suck out the heat from the molten metal

>>59190667
Not if you heat the heatsink on a frying pan first.

>>59190667
Just heat up the heatsink with a hair dryer.

>>59185940
The guy who did this in the OP image is a LN2 overclocker, so that minimal difference matters because he's literally trying to achieve world OC records.

>>59188262
>>59190241

https://en.wikipedia.org/wiki/Gold_plating#Electronics

>With direct gold-on-copper plating, the copper atoms tend to diffuse through the gold layer, causing tarnishing of its surface and formation of an oxide and/or sulphide layer.

BTFO'd