7600K Naked Die Cooling Temperature Follow-Up @ [H]

Interesting results. I wonder if there was only a 3C advantage because of the waterblock's material. What was that one made of? You have more heat in a smaller area which is harder to dissipate. How about using a 1/8" copper block to spread the heat?

Actually, now that I think about it, you may want to go 1/4" because the 1/8" may deform too much and chip the corners.

BTW, I'm at work and didn't watch the whole video so I apoligize if you already brought this up.


BP
 
First off, thanks for taking the risks so we don't have to.

While the temperature improvements might be disappointing, I think that generally speaking these results are fantastic - we can get within 3C of direct-die without carving up our motherboards - great news, right?

Any plans to lap the IHS and see how much the temperature gap narrows?

(Also, your link to the 7700k de/relid is broken)
 
ButtonPuncher said:
Interesting results. I wonder if there was only a 3C advantage because of the waterblock's material. What was that one made of? You have more heat in a smaller area which is harder to dissipate. How about using a 1/8" copper block to spread the heat?

Actually, now that I think about it, you may want to go 1/4" because the 1/8" may deform too much and chip the corners.

BTW, I'm at work and didn't watch the whole video so I apoligize if you already brought this up.
Click to expand...

It is quite possible that we are just at the "limits" of our ambient temperature cooling abilities given the surface of the die area we are dealing with.
 
Nicepants42 said:
First off, thanks for taking the risks so we don't have to.

While the temperature improvements might be disappointing, I think that generally speaking these results are fantastic - we can get within 3C of direct-die without carving up our motherboards - great news, right?

Any plans to lap the IHS and see how much the temperature gap narrows?
Click to expand...
Those IHS are already so incredibly flat, I doubt you are going to eek much out of those with lapping. Some pictures of IHS mating are here. http://www.hardocp.com/article/2017/02/02/intel_kaby_lake_i57600k_cpu_delid_relid_temp_results/
 
Going back to relid the 6700K just now and am using Phoby metal TIM instead of the CoolLabs Liq Ultra.....This Phobya is like liquid mercury, where the CLU is like a paste. The Phobya literally rolled off my paper towel when I tipped it sideways...

I am not trusting this stuff under the IHS. Laid down a little RTV to cover the exposed contacts under the IHS.

20170207_144821_001.jpg 20170207_145125.jpg
 
In a case like mine where I don't have access to that kind of TIM, would something like Arctic Silver 5 or similar still make a difference? Still worth the delidding/relidding?
 
Wow, that was surprising. I - too - was expecting much better results without the IHS.

Really good information though. Now I can save myself the trouble and not do this on my next buiild!
 
So there it is. Delid/relid TIM upgrade worth it. Delid not so much. Really appreciate all these articles. Saves everyone who has interest in a high powered & QUIET Kaby Lake system a ton of time in testing these scenarios. Long time reader, and I hardly post, but these articles really just save us so much trouble. Thanks Kyle and company.
 
What Load Line Calibration and CPU Current Capability settings are people using?
 
ButtonPuncher said:
Interesting results. I wonder if there was only a 3C advantage because of the waterblock's material. What was that one made of? You have more heat in a smaller area which is harder to dissipate. How about using a 1/8" copper block to spread the heat?
Click to expand...
Wouldn't that just add an extra layer of material for the heat to transfer through, making it less efficient?
 
Patiently waiting for the upcoming CPU goo review!!

Charting Texas weather.... good one!! That is an exercise in futility.
 
I will get it worked out, but when it is 73 at 4pm, and 33 12 hours later it makes stability for temp testing really difficult. And I will likely be testing for for or five days at a minimum.
 
The presence of an IHS is a problem if it's not mounted correctly. In the pictures of de-lidded Kaby Lakes I've seen there seems to have been way too much distance between the CPU die and the IHS, creating a gap that had to be filled with TIM. Ideally there should be no such gap, and the TIM should be visible only like fogging on the IHS and CPU surfaces.
If the tested CPU was well mated from the start there shouldn't be much improvement from de-lidding it.

That said:
1. After de-lidding the CPU could be better mated to the socket by using a shim (stiff frame) where the IHS is typically glued to the PCB.
2. If using water cooling on a de-lidded CPU the obvious choice would be to skip the water block and use direct die cooling. (A dome glued to the PCB with inlet and outlet for the water that flow in direct contact with the CPU die.)
 
I don't think there would be a great deal of difference. The IHS is already solid copper with nickel plating.
 
Naked Die sounds like a weird Sharon Stone / Bruce Willis / Leslie Nielsen movie from the 90s.

And I'd watch it.
 
You must log in or register to reply here.
Back
Top