Intel Kaby Lake i7-7700K CPU De-Lid & Re-Lid Results @ [H]

FrgMstr

Just Plain Mean
Staff member
Joined
May 18, 1997
Messages
55,532
Intel Kaby Lake i7-7700K CPU De-Lid & Re-Lid Results - Is there any reason at all to take a razor blade, garage workbench vice, or any sort of other tool to your brand new Kaby Lake CPU? Of course there is. That reason being that the Thermal Interface Material used by Intel on its desktop CPUs is pretty much crap...at least to us enthusiasts.
 
[H] never disappoints! Im surprised with a 23 degree difference in temperature it still wasnt able to hit 5Ghz. Maybe you need more training? (sorry couldn't resist!) :D
Im interested in seeing the full fruits of your labor in future article.

BTW the Discussion link in the article links back to the article itself, not to this forum thread. I had to go to the [H] main page to get to this thread.
 
[H] never disappoints! Im surprised with a 23 degree difference in temperature it still wasnt able to hit 5Ghz. Maybe you need more training? (sorry couldn't resist!) :D
Im interested in seeing the full fruits of your labor in future article.

BTW the Discussion link in the article links back to the article itself, not to this forum thread. I had to go to the [H] main page to get to this thread.

Well, I'll tell you what...I have the exact same situation with my 6700K. No matter what I do it will not be stable above 4.7 GHz. At 4.7 GHz I can throw anything at it and it does just fine; at 4.8 GHz it fails. Depending on voltage it'll boot, but it'll always freeze in any benchmark. Doesn't matter how cool it is, though I admit to not having tried delidding (it's not that important to me). Sometimes you just hit a wall.
 
Never seen anyone use rtv to glue to ihs back on, won't the cpu bracket hold on the ihs or is this done so the cpu is stable while installing and removing the cpu?
Just curious, thanks for the testing, pretty nice decrease in temps for 45 bucks and some time to do it!
 
  • Like
Reactions: Qrash
like this
Nice, thanks for doing this.
Very similar results to my 6700K delid.
About 20C drop, can game 100MHz higher but will fail a long stability test.

ps the link at the end of the article takes you back to the article, not this thread.
 
BTW the Discussion link in the article links back to the article itself, not to this forum thread. I had to go to the [H] main page to get to this thread.
Caching issue on the server, trying to figure that out. Thanks for the heads I up though.

Edit: Should be fixed now.
 
Wow you would think a huge temp drop like that would have more impact. 20c drop is huge. I guess what that tells us is the chips are heat tolerant and if it's not going to hit 5ghz stable lidded, it's not going to be stable delidded either... just cooler?

Has anyone tried ditching the IHS and using a thin piece of copper instead?
 
I have some real disappoint here with my 7700K running hot due to such a simple bit of negligence on Intel's part. The weak spot of a cooling setup shouldn't be a tiny bit of TIM between the CPU due and the heat spreader. The CPU can push 80C and the water going through the radiator isn't even getting warm.
 
Delidding only really helps when you are heat limited. I'd personally do it anyway now that I've had experience and am not worried about damaging/destroying the cpu. Like someone else said, sometimes a cpu will only be pushed so far be it a voltage limitation or something else.
 
Great article. Short but to the point. Results remind me of NVidia Pascal arch in a way. Lowering temps doesn't help OC as much as it used to.

We have also been breaking CPUs since CPUs had lids, well actual before, but that is another story, and one to why CPU have IHS units now days. The safest why we have found is to use a tool specifically designed to remove the IHS.
actual --> actually
one to why CPU --> one reason why CPUs
why --> way

And already mentioned above:
... it likely does not meat yours.
meat --> meet


Guy: I can't wait to meat you! :woot:
Girl: Ugh... <sigh> :punch:
 
The chip is just hitting a speed wall the temperature it's hitting is not the issue. Nice to know a quality paste does make a huge difference in temperature, but that is a lot of money to spend just to make it run cooler not faster.
 
pretty awesome but also wish it was just like this by design... i wonder if other cpus would bernefit from replacing the caps. i bet it would last longer if the cpu was a lower overclock (4.5ghz) and possibly increase the life of the cpu due to lower heat
 
Wtf 25% lower temps and still no 5ghz so disappointed.
 
not sure why everyone is calling this a wash... i would do this just for the temps. with that big of a difference i would bet stock Intel heat sink on delided chip would come pretty close to non delided chip with top of the line air cooler. while this comparison isn't realistic in a real world sense, the point is people pay big bucks to get a couple degree lower temps between the best end air cooling and water cooled setups.
 
And to add to that since my last sentence is slightly confusing. Can you get a higher overclock on this chip with a high end water cooled setup VS a cheap tower cooler like coolermaster hyper 212? l am guessing no, but most people would still go for better cooling just because.
 
Its more like 6%, temps start at absolute zero -273.15C.
But yeah it is still sizeable, bit of a shame.
Hehe, fair deal on the Kelvin. I figured since we were using C I should go from zero. But good point, I should actually go from ambient at the time the temps were taken.
 
Potato? ;)
Potato.png

1.300v VCORE (Adaptive) highest HWiNFO64 has recorded in the past five days was 1.389 VID, but I had it set to 1.320v in the BIOS and backed off slightly by .02v VCORE in Windows. Using a Corsair H115i for cooling.

It pretty much instantly locks when I set it to 5.2Ghz at 1.35v and I don't want to bump it up much higher.

Having four sticks of DDR4-3466 installed seems to have it running hotter by ~3-8C hotter than two stick of DDR4-3000, but my room is also a bit warmer, so I don't know.
 
The chip is just hitting a speed wall the temperature it's hitting is not the issue. Nice to know a quality paste does make a huge difference in temperature, but that is a lot of money to spend just to make it run cooler not faster.

Yea, it is not like ppl would ever spend exorbitant amounts of money just to keep their rigs overclocked whilst running cool and quiet. Geeze, who'd ever want to do that?
 
I see your screenshot at 1.368V and 5GHz, but I was wondering if you were able to achieve 4.9 at a lower VCore than without the new TIM

Also, you said your previous attempts at 5GHz you used 1.35V, does that mean you were able to push more voltage after the de-lid?

Don't you have 3 more processors sitting around ;)
 
I can tell you first hand that Solidworks, Premier and several other programs I've used or my clients use can stress a CPU out for hours upon hours at a time. I've seen 12 - 13 - 15 hour Solidwork renders that pound the CPU.

For a gamer, 2 hours is fine. For a workstation situation, probably not.
 
not sure why everyone is calling this a wash... i would do this just for the temps. with that big of a difference i would bet stock Intel heat sink on delided chip would come pretty close to non delided chip with top of the line air cooler. while this comparison isn't realistic in a real world sense, the point is people pay big bucks to get a couple degree lower temps between the best end air cooling and water cooled setups.
This!

Hell, depending on the need, you might be able to have a fanless or super quiet system after a delidd.
 
Is there any chance that the function of the temperature sensor has been altered in some way by the process? Is the new temperature value an accurate value?
 
I thought the temperature sensors are inside of the die and not on the board - hence the different core temps etc available for reading. So this should not have affected the reliability of the readings unless you broke the die.

Surprising by a large degree (no pun intended) the difference in temperature obtain! WTF is Intel doing having such crap TIM inside their processor!? - hopefully this light will make Intel, for what ever f___ up reason, to do the right thing and make it right.

Then seeing the lack of OC ability is also perplexing as well at the end of the day. Looks like 5ghz is the upper wall for Intel at this time with few exceptions.

I would also try BCLK OCing to see if you can get a better OC. May not apply here but on three separate FX processors the standard fsb would have a ceiling on an OC. Up the fsb, lower the ratio of cpu to fsb, mem speed to fsb etc. brought back stability pushing FX 8120 to 5.2ghz, FX8350 to 4.9ghz and FX 9590 to 5.1ghz.
 
That's a massive decrease in temps. I'm thinking about de-lidding my proc now. I'm not chasing OC, but I'm always chasing lower temps. German don't ever install A/C in homes/apts, so you're stuck dealing with the heat during the summer months. My system hates that, so I had to go watercooling to keep things down.
 
That's a massive decrease in temps. I'm thinking about de-lidding my proc now. I'm not chasing OC, but I'm always chasing lower temps. German don't ever install A/C in homes/apts, so you're stuck dealing with the heat during the summer months. My system hates that, so I had to go watercooling to keep things down.

There's a new delidding tool
comes out at 22.02 at caseking

it's only 30€, though I hate how caseking adds shipping costs; ehh spoiled from Amazon Prime :D


anyway the toughest thing when I delidded my kaby lake was applying the liquid metal

if you're not used to using it (within reason) you better practice just a bit

but then
with enough patience it worked fine
I myself came down from 88 to 68 degrees on an air cooler at 5Ghz
a drop of 20 degrees is something people usually have to buy a watercooling solution for :rolleyes:

5.2 spiked up again in heat and voltage
AiO installed now, needing some time to test :ROFLMAO:
 
I believe so. Kyle's previous OC attempts have always been aiming for "Everything enabled".. That's why he rejected the 5.0 w/o HT enabled OC..

Gotcha - so some chance Kyle's chip could run stably at 5.0 GHz in games (which don't appear to use AVX yet) while being limited to 4.9 GHz for transcoding type applications..
 
Now that would be funny.
But I doubt it, although I do miss the days of the external thermocouple machined into the lid. http://www.hardocp.com/image/MTIyNTM4MDczOTdGT2V4YVR6MzdfMV80X2wuanBn
I think we were the only site to spend the money to do that. :) I remember talking to an Intel engineer about temps on those CPUs. "No real way to get that temp....unless you machined a groove for a thermocouple in there." I thought, well hell, that is pretty easy...and I got out the Dremel. JK, I had a machine shop do it. You could build a jig and do it in the garage though with the Dremel router head though.
 
I think it should be a % decrease compared to ambient. That's the floor in this scenario, without some kind of TEC or phase change.
Yeah, that is the way we do HSF stuff. I will do it that way going forward. I am not going to change the article however. Changing data is never good in this business.

Here is what the graph looks like scaled from ambient.

upload_2017-1-20_9-32-22.png
 
Back
Top