13900k de-lid gone wrong?

Next update. I cleaned the die and applied Kryonout thermal paste. The temperature problems under load persist. I can run Cinebench, but the CPU throttles down to around 5GHz. Strangely enough, idle temps actually dropped when using the thermal paste. Maybe because it is thicker.

Looking at the cooler and die afterwards, it looks to me like there is some contact in the middle of the die and little to no contact around the center towards the edges. This would explain the poor temperatures. Also, I noticed that when assembled the little arrow on the cooler points up and right, while in the assembly documentation from EK it shows down and left (in line with the little arrow on the CPU pcb). The rest of the waterblock is oriented correctly, so I am wondering if EK assembled this one incorrectly? I am hitting a wall here. May need to give EK a call next week to sort it out.

Cleaning up the die


Adding Kryonout


The contact pressure imprint on the die and cooler.



Note the arrow. When assembled this points up/right, but on cpu it points down/left. ☹️

Any more suggestions before I call EK?

 

theres little if any pressure at all there.

 

Looks like pressure is only in center. Still shouldn't reach 100C on 180W.
Whats the voltage and LLC you running for 5.8 ?

Edit: On a second look at the block plate paste contact, you're only cooling properly half the P cores and E cores. Also cores edge are not in great contact

 

In this image, unless my optics fail me, it looks like the cpu die is curved a bit.

 

Exactly

Running out of the box settings at the moment. A little high on voltage for my taste (1.4v for 5.8 boost) but this shouldn't be the reason for the high load temps.

Fully agree on the poor contact.

I put a ruler to it and it looks straight. I thinks it's more the tape wrap making it look skewed. But honestly, I wouldnt rule anything out.

 

phone lense does that.

 

can confirm the EK DD cold plate is convex and im pretty sure thats the problem , convex works on IHS, but with direct die you need it ot be as flat as posible

 

So my remark was certainly spot on earlier, this directdie block is advertised as compatible with the gen 13 die but is milled to the requirements of an ihs, attempting to increase mouting pressure could crack the die in the center too.

That or there is a batch of poorly milled blocks, should definitely get in touch with EK here, @dune2

 

I am having the same issue. I am running about 75-80 in cinebench on all cores except two p cores are hitting 100. No overclocking. To be fair, my last cooler was a coolcastle AIO and I was hitting 95 -100 on all cores in cinebench with no overclocking. I have not taken my loop apart yet to check die pressure and flatness. I am able to run games at around 65 c all day long, which is way better than I was getting before, so I don't know if its worth the hassle to take apart, as I might just make things worse.

 

This specific block is not only meant to be compatible with direct die cooling. It is specifically manufactured for direct die cooling. You can see on the pictures of the block that there is an inner rectangle protruding from the block to make direct contact with the die (at least in theory ). In fact, this block will NOT work with the IHS, direct die only.

 

Yes, but the raised area is convex instead of flat so there isn't correct surface pressure on the edges of the die.

 

the block should have a raised area in the center to fit where the IHS normally is.

either your block is not milled correctly or you are not tightening the block enough.

 

It looks pretty straight to me both diagonally, vertically and horizontally.





I also know that EK makes their standard LGA1700 blocks concave to counter IHS warp. This wouldn't make any sense on a direct die block. Maybe it's just too small a gradient for me to detect with the naked eye But whatever the reason, it's not making the contact it should.

 

I doubt tightening too much will solve it.
The contact is there but only in center which means something is not straight.
Even if there was no tightening, when placed on die the block should have uniform contact.

This should not be happening on a straight contact:

 

I don't dare tightening the screws any further. I think the block may not be milled all the way to spec. So annoying.

 

Have you contacted EK?

 

Yes, I contacted them today. They are usually very good on customer service, so I am hopeful. But up to 5 days wait. I will keep you posted.