Ordered a Silicon Lottery Kaby Lake i7-7700K from HIDevolution

Did you try the settings i suggested? At this point might as well give them a shot. For 4.7Ghz (-35 for core) and Cache at 4.3Ghz (-160)

 

I already swapped my old CPU back. I almost am gonna have a heart attack from how many times I reposted. This CPU reminds me of my first 6700K in my first P870DM. Give it any loaday and the temps sky rocket to 95C within seconds. De ja vu

 

That's because your voltage is skyrocketing along with it.
Looking at my screenshot , you can see the voltage is relatively low thus keeping the temps lower as well, but if it's a sauna where you are, then there is not a lot you can do in the end.

 

Needs tweaking ma man

 

and here's my Silicon Lottery 6700K @ 4.6 GHz / stock voltage with no undervolt

 

Holy cow! Something is wrong man, no way you should be hitting 89 at only 4.6. I think your 6700 was delid'd right? Could something be wrong with your heatsink fitment?

 

He's missing heat sink screws.

 

that's only in a stress test.

I've ordered a new CPU Heatsink and new vapor chamber heatsink just now as something is not right, but that 7700K was flawed for sure

and yes the Silicon Lottery 6700K is factory delidded

 

that too, the cats ate them

but those bottom heatsink screws that I'm missing, how important are they if the CPU 4 screws are already tightened? how do they affect the main vapor chamber heatsink?

 

From what I have seen. Pretty important. Because they clamp down on the CPU heatsink/Fin array.

 

Lol. Reminds me a when I lost a screw inside a laptop I was working on repasting. I was sure it was inside the laptop somewhere, and eventually started tearing it apart. Finally I found it wedged under a heatsink.

 

You cpu is running crazy hot, it must need those screws assuming the paste job and contact is good. Those temps are 25C or so hotter than they should be.

Were the fans spinning?

 

yes sir, max blast

before I lose the screws, max temps were in the 82C in a stress test that is, much lower in day to day usage

 

I know it's warmer where you are, but wow that is crazy. Mine idles in the mid 20s with max fan, and under that same stress test you just did my peak temp was 64. I'm no expert but I still think something bigger is wrong if before you were hitting 82.

 

Yep. that is also pretty high.

 

no that's how it was, it's fine, I'm in Dubai dude not in the north pole

 

probably those missing screws

 

OK, I just assumed we all live in air conditioned homes that are fairly close in temps. Unless our houses are 20C+ apart in temps inside lol

 

I made mention of that before about why the high temps. Reference "Sauna" in a prior post.

Nope. I do not have those screws in either. I only have the 8 around the GPU's and that's it.

 

OK that makes me feel better.

Who ate your screws do you have cats?

 

No pets at the moment. Although I do have some random cat trying to camp out in my yard and in my garage.

Nah, I have it that way so i can swap back and forth between parts.
And when you are testing the cpu. You don't need to keep repasting gpu till you are 100 percent sure the cpu is right. Heck, I haven't repasted my gpus since i got it back from Sager after they swapped the screen. I just lift off do what i need to cpu and then set the vapor chamber right back on.

 

@Phoenix - you gave up too quickly, bro. Get some more IC Diamond, put the 7700K back in. Be patient and try harder. I bet there is nothing wrong with your 7700K. You tried what worked with 6700K... which is a totally normal thing to do first. I did that as well. No point in reinventing the wheel if you don't need to. I get that. But, it didn't work and I had to spend time figuring things out all over again, which is what I expected would need to happy. People can say 7700K and 6700K are the same thing all they want to, but they are not. Take it slow and easy, take notes (or pictures) as you go to avoid rework and repeating mistakes. I'm still not done experimenting.

As far as temps go, I'm not sure the fit of the heat sink is exactly the same as 6700K because I'm having to use IC Diamond as well. That suggests the heat sink may not be as tight as is needs to be for CLU to work on top of the IHS.

And, there is more that one way to skin a cat. This screen shot is using Brother @bloodhawk suggestion. These temps are after 3 back-to-back runs on Cinebench. They are totally stable settings for me as well. Hotter than I would like, but not dangerous. And, I'm going to fix the too hot part when I have more time to tinker with it. Until then, I'll use it as is... not a problem. Remember, the OEMs and Intel are saying anything less than 100°C is "normal" for the turdbooks, and this ain't anywhere near that hot.

 

Exactly. Those are kinda "safe" offsets. I am running 4.8Ghz on mine @ -55 (Cache 4.3Ghz @ - 190)
This is 1 hour RealBench and multiple Cinebench run stable. Max temps were about 85-88C.
Obviously the CPU only cooling of the DM3 will be superior.

Also @Phoenix does you heatsink have a big thermal pad right in between the GPU and CPU heatsink? ( The area where the BIG section clamps down on the CPU heatsink).

 

well this CPU bro is overheating at stock clocks, I requested HIDevolution to send me a courier to my house to collect it and ship it back to them. They are in turn in contact with Silicon Lottery to see what is up or if they are swaying away from what they previously did which is delidding it and applying CLU between the CPU Dye and the IHS

Ill get another one soon. hopefully you can send me a full screenshot galore of your new BIOS settings.

 

It really does sound like you have a heat-sink contact issue. There is no-way the temperatures should ramp up that fast unless there is no "mass" for the heat to dissipate to.

If you had a temperature probe it might be interesting to see what the actual heat-sink temperature is. At a 100 c CPU temperature that heat sink should be sitting at 70-90 c if there is even a semblance of contact between the items.

If the heat sink is as cool as a cucumber and the CPU is roasting... well, you are not transferring heat due to some mechanical issue that is obstructing the ability to remove heat from the chip.

That you tried to go back to the old CPU and it sucks is an indication that there is something wrong. I would not even to try and tweak settings until I resolved the issue with those missing screws and any sort of fitment/alignment problems.

You are fighting too many problems at the same time, while you are attempting to tune. You must have a stable configuration before you tune anything at all. That is a fundamental of any laboratory exercise, you must minimize the variables that are skewing your results.

 

I bet anything the Heatsink isn't making good contact and the CPU is fine. I've had temps like that and did around 10 repastes before I could get it right.
I think the thing that worked for me was actually cleaning off all the glue on the IHS/CPU after the delid. Then sandpapering off a miniscule amount on the bottom of the IHS.
That's at least half the battle. The other half is the HS fit on top.

 

You can make the temperatures worse if you use an incorrect technique in delidding the IHS from the chip package. Jamming a razor blade between a glued down IHS and the edge of the carrier PCB can actually gouge up the IHS bottom and make it sit higher off of the chip die when you are done. Also as TBoneSan said, you must have every speck of adhesive removed from the bottom of the IHS and from the PCB surface before you glue it back down.

After you put something like LM between the die package and the inside surface of the IHS and then set it back down that gap has to be at the absolute minimum. That is determined by the cleanliness of the PCB and the edge at the bottom of the IHS. Also if you use an adhesive that is too viscous (thick) you can make it difficult for the IHS to get fully down against the carrier PCB. You should have something that slightly compresses the IHS downwards while the adhesive cures.

Once I remove the glue I have a machinists flat (a big steel disk that has a mirror finish and is perfectly flat) and a garnet based lapping compound to polish the bottom edge of the IHS and to make it slightly shorter so the clearance between the iHS and die package is gone. This removes less than a millimeter of IHS thickness but gives me assurances that it is flat and as low down as I can make it. The entire lapping/flattening process takes less than two or three minutes and I use a micrometer to measure IHS thickness before and after lapping.

I use a rubberized super-glue. It is black in color and is not as brittle as normal super-glue. I first encountered it way back in the 1990's when we were trying to get fiberglass to stick to aluminum as part of a circuit board assembly. Normal super glue was too brittle and would crack off. The black rubberized stuff really worked great on dissimilar items. Also if you are very neat in the application it looks just like the normal adhesive that came with the CPU and was between the IHS and the PCB that makes up the edges of the CPU.

I use one of the commercial de-lidding jigs that has a compression clamp that aligns the IHS to the package and after gluing I leave it in place for an hour or so until the glue completely cures. I have done a few CPU's other than the 7700K and it has been successful every time. I do not need to worry about messing up a CPU with a razor blade.

As you know, if you gouge up a CPU with a razor blade or knife while delidding you are immediately out the ~ $350 that you spent on the chip. Buying the right tool for delidding is a great investment, if you are successful maybe you can charge your friends $20-30 to delid and LM their CPU's as well, recouping your investement.

 

Well said. That's very much what I've found to be true. So much so I took a few snaps of my last delid ( have a peek below). It's critical the IHS gets in close and firm to the die and the bottom get a little shave. The only difference I've done is I don't glue back on the IHS because I wouldn't be able to get the kind of precision you've mentioned.

http://imgur.com/a/BWHWI?

 

the delid was done by Silicon Lottery. anyway, this CPU is going back to them soon

 

Good grief, what a hassle to get the best performance! So, how often based on daily use, but not necessarily all gaming does this CPU need to be repasted over the course of its lifetime....3-5???? years?

 

you don't have to repaste usually

when you apply Liquid Ultra, i can last for a year

With IC Diamond, it can last up to 3-6 months

The reason I had to repaste like 6 months was purely for a lemon of a CPU, I was thinking my paste job was bad and tried diff. methods/diff. pastes so don't worry about that, it isn't the norm

 

Yeah it can have its moments.
It's also the kind of thing that's a hassle to some, but a labor of love to others.
It's definitely worth the effort and perseverance IMO.
I was testing things out running Project Cars at 5ghz last night, with auto fans chilling out with temps in the high 80's. It's so ridiculous I love it.
I really enjoy it when something pays off. 3 months ago I was sitting in the 90s at 4.4.

 

@Q937 explain what you discovered. Please.

 

My caliper was being kinda wonky, but as far as I could tell, it looked like the retention tabs on the IHS sit a tiny little bit closer to the top than on the 6700K. I measured something around 0.05mm difference, which was enough to make my 6700K barely tall enough for a generous amount of CLU to work well, but the 7700K low enough to require a shim.

EDIT:
Redid the measurements and got similar results. Used my non-delidded 7700K this time because the delidded one is installed in my laptop.

6700K delidded:
4.28mm from bottom of substrate to top of IHS
- 2.45mm from bottom of substrate to top of retention tab
= 1.83mm from top of retention tab to top of IHS

7700K floating:
4.26mm from bottom of substrate to top of IHS
- 2.49mm from bottom of substrate to top of retention tab
= 1.77mm from top of retention tab to top of IHS