What We've Learned from Delidding Intel CPUs 

You might like our Intel X299 VRM thermal analysis: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-0qYHWAnvXv8.html Or our Bitwit Delid content: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-nBAeyzaRPBs.html

when i resealed mine i just put a tiny dot in each corner of the IHS in order to hold it onto the PCB enough for me to handle it while putting it back in the socket. When i initially delidded i didn't have any LM on me (it wasn't gonna be delivered for a few days and i was too excited to wait). Just delidding, cleaning of intel's sealant, and adding in my own thermal paste (hydronaut) i got a 10C drop in temps. Redoing it later with LM i got another 10C drop, giving me a 20C drop overall. What a great experience and i'd recommend delidding to any and all enthusiasts out there.

"Apply gravity" - GN 2018

No no no this nail polish is for my cpu.

Here's what I learned from delidding my i7-4790K : 1) Like Steve, too much RTV silicone just creates too much of a gap between the IHS and the die, so I run mine without any adhesive. 2) Even without adhesive, the gains can be low depending on whether your pre-delidding temperatures weren't bad and how much power your CPU dissipates. In my case, it's running at stock and the IHS was probably mounted correctly without too much TIM nor adhesive at the factory, so I ended up gaining...3-4°C under load. I decided to also use liquid metal on IHS-waterblock interface for an extra 3-4°C drop, so the total temperature drop is around 6-8°C, as my measurements' accuracy are limited to 1°C with Intel Extreme Tuning Utility. 3) Make sure to apply enough liquid metal so that the surface is at least shiny ; I once tried to use as little as possible, and got no gains because there were just too many air microbubbles between the die and the IHS. 4) If applying liquid metal on the IHS-cooler interface, I'd suggest making a central square of liquid metal and a ring of regular thermal paste outside of it ; like this, you likely won't have any runaway liquid metal when tightening the cooler, as the excess liquid metal will get stuck in the regular thermal paste. 5) Your gains with liquid metal can be limited even if you do everything perfectly ; google "thermal spreading resistance" and read a little about it. Basically, liquid metal will provide great gains if the thermal spreading resistance (of the die-IHS interface) varies a lot with thermal conductivity, which implies a significant ratio between the effective surface area through which there's heat output (it can be fairly smaller than the die's surface area) and the IHS's surface area. Small dies like the 7700K benefit a lot from liquid metal, but if you've got a CPU with a die nearly the size of the IHS (the IHS here doesn't really spread heat, it's more like protecting the die from cracking when mounting the cooler), liquid metal will do about nothing compared to regular paste.

I have no intention of ever de-lidding and yet I watched the whole thing. Bet I'm not the only one either.

Thanks for another quality video! A few things I've learned with lidding, lapping and liquid metal-ing, some already covered in video. - Lapping the IHS on CPU's when delidding can improve temps when also using LM between cooler and IHS - It is possible to lap the bottom of the IHS very slightly to reduce clearance from chip to IHS and further improve temps. Could crack chip, take care, remove little, if no POST, back off cooler. - Using a small amount of silicone to reseal IHS and clamping in the socket to do so minimizes the clearance when resealing - Take great care not to allow dust or other contaminants in to the exposed liquid metal, don't leave it sitting there. - On 6700k, 7700k, 8700k, remove all factory sealant (fingernail works) - When clamping, clamp may attempt to push IHS along the CPU PCB. Beware surface components and potentially compensate by positioning IHS with offset before clamping. - Liquid metal on top of the IHS requires very flat surfaces and careful application - Apply LM to both surfaces to be mated - LM is extremely effective on the GPU chip, particularly on water, protect surrounding components though. - Brush on electrical tape is easy to use and effective at protecting surrounding components - When applied to shunt resistors on nvidia GPUs to increase power limit, the LM eventually permanently discolors the resistor and solder - I wouldn't be inclined to recover liquid metal with the sucker as the chemicals within it may have separated

I haven't bothered resealing my delidded CPU's. If you're just going to leave it in the motherboard for a long time anyway (like I would guess 99% of people will), I don't see the point. The clamping force of the socket retention bracket is more than enough to keep the IHS in place. Add the force of a mounted cooler and it's not going anywhere. Vertical and horizontal, the PC being transported in a car with all the movement and vibrations that entails, and there has never been any issues.

Even though I don’t understand half of what is being explained, I still watch because it is soothing to my soul. It’s a great day when gamers nexus posts a video

Just wanted to tell you I've been fascinated with your channel ever since I was in a live stream you did of Ocing Vega. You discuss everything with a scientific approach of which few or no other tech channels I've seen do. Truly an awesome channel for informational teaching of computers. Thanks!

Steve is right... The BEST way to re-seal.... if you must, is to use a very light corners application and clamp in the socket. Using the delid tool pushing down in the center can cause a slight warping of the IHS that thins the interface while curing but allows the seal interfaces to cure thicker... When you unclamp, the thermal interface actually comes apart and becomes thicker and potentially voids due to squeeze out. USE THE SOCKET to cure sealer. Then the interface will stay the same through insertion/removal (repeated). BTW.... there is no excuse for NOT completely cleaning off old sealer and then triple cleaning the entire CPU and IHS with appropriate solvents to make sure they are clean. THEN, completely clean and reset up your workbench with an eye toward NOT having any contaminant particles around that might get into the thermal interface areas. THEN, you can apply your liquid metal.... then apply the corner sealant (again, if you must) and then socket clamp to cure.

Damn as a beginning OC'er I've been glued to YT for months and THIS VIDEO is what I have been waiting for for all the goods

25:39 mug-O-watt, bottom left :D

Anyone tried polishing the cover inside? In engines, we polish the combustion chambers to mirror finish to keep from hot spot areas from pockets or pits in the metal. Would something like that benefit heat dissipation?

Use conformal coating in place of nail-polish. CC is 1: Made exactly for this purpose. 2: Fluoresces under UV light for the purpose of coverage checking after curing. 3: Is easy to remove; once set it peels off like rubber and cleans up with isopropyl alcohol.

Looser format but golden content as usual 👌

Fantastic intro

It is possible to reseal the CPU without losing the benefits of using liquid metal. I delidded my 8700k today and when I put the IHS back on I put a very small smudge of silicone adhesive to the 4 corners of the IHS. It's enough to hold the IHS secure but not enough to affect the thermal transfer between the die and IHS, at least on my chip. I was able to drop my load temperatures down around 25°c at 4.9GHz and gave me enough thermal headroom to take my chip to 5.0GHz which I couldn't do before. My chip is a dud in terms of overclocking. Hitting 4.9GHz was easy, and it would do it at 1.296v, hitting 5.0GHz requires 1.392v which was way too much for the factory TIM to handle. One thing I noted when I took off the IHS was the factory TIM on my chip was already getting hard and crusty and I have only had the chip for a couple of weeks now. It's probably been sitting on a shelf somewhere for several months, but it was already at the crusty level of over a years time on many other compounds I have used.

IMHO, I think it's best to use "Ultra Black" silicone sealant. The "Ultra" in the name denotes that it is (automotive) Oxygen sensor safe. Basically, it does not emit fumes which can damage sensitive electronic components. Also, plastic razor blades for removing old silicone available at auto parts store. Dorman/help part number 22018. About $3.49 for a pack of 5.

I left my 6600K unsealed. Since it stays locked on my MOBO, it's been a non-issue, either vertical or horizontal.

Done quite a few delids and found some tips to make it look factory. First, the q-tip method sucks, use a small paint brush to apply, you get more control of the spread, you can get them at most drug stores, hobby shops etc and they are cheap and reusable. Second, if you remove all the old adhesive you really dont need LM on both sides, i always just do one side with no problem. Third, use a automotive high temp black rtv, i use permatex ultra black, every auto parts store has it, has a resealable lid also. Use a small screw driver to apply or a tooth pick ( or anything with a small flat tip). Use the relid clamp part of the delid tool, let it set for at least 12hrs. Remove any excess with a qtip and alcohol, To remove the old stuff use plastic razor blades, got them on amazon for a few bucks for 250 of them. Every cpu i have done looks factory stock but run atleast 15c cooler, even on locked i5's like i5-6500.

Id like to see a test done with a glass plate to see if there would actualy be air bubbles with this way of applying.

I just imagine being a 20 something qt in a nailsaloon & some hairy dude walks in with a cameraman, talking about not-electrically conductive nailpolish & how it will shield the components from liquid metal to get better temps on an overclock.

The volume of knowledge I gain from gamer Nexus...thanks guys amazing job.

My 3770k is still going after delid and one LM application. It's been 2yrs no problems. I killed a g3258 during delid a month ago lol. Both of them used razor blade method. Never again will I do that I'll get the tool from now on.

I know this video is old, but another advantage to delidding is for older cpus that may have degraded thermal conductivity between the die and ihs. I'm going through this process now on a 2015 iMac that's been used extensively in a work environment. Over time, the thermals have "naturally" degraded to the point the system idles at 55c and hits 100c almost immediately under load.

Am I the only one getting distracted by their wobbly table?

Fantastically thorough and helpful. I'm running Haswell and don't "need" to delid but almost want to now just because I want to try it out!

The retention device on the motherboard is pretty easy to remove on a motherboard. With my 4770k back in the day I screwed off the retention device, sat the bare cpu without ihs in there, applied coolaboratory liquid pro, and mounted the block right on top of the cpu. Earned a couple extra decrees over using the ihs.

i didn't use nail polish to protect stuff on my PCB, i just bought some kapton tape and used that. Nail polish works great as you said though.

Someone bolt that table to a wall or add some cross struts or something...that wobble is really destracting. Thx for the video...i will probably never delid a CPU but knowledge doesn't hurt ;)

I know it's an older video now, but thank you! I finally got round to doing this. It went really well. I don't have comparitive data because I upgraded the stock intel cooler at the same time to a be quiet dark pro. But temps are halved overall under load and a third better at idle. So I think I spread the Lmetal ok. Obviously the cooler is making a big difference. Was hard to get the metal to spread without having too much, and then getting it spread evenly tended to leave too little. I might take it apart eventually and reapply and compare results.

*Steve* Thank you for producing this detailed instruction.

I use non conductive thermal paste instead of nail polish. Especially due to heat, i think nail polish would crack and discolor and will be almost impossible to remove after that

I love the ironic intros, keep em coming

I delidded and put regular thermal paste (Cooler Master MasterGel Maker) on the die, a nice, thin layer spread out with a spatula. The cpu (6600k) package temp under 0-25 % load is about 3-5 degrees celsius above ambient with a Noctua NH-D12 cooler on the cpu in a Fractal Design Define Nano S case with a Kraken X61 cooler on the intake which cools the gpu (MSI GTX 970) which is like 2 degrees above ambient at idle). Max (peak) temps reached during an hour of 95 % + utilization is ~62 degrees on cpu (1 core) and 43 degrees on gpu. Both cpu and gpu are non-overclocked at these temps, but I'm really happy with the result. I re-glued the cpu IHS with electronics compatible silicon adhesive, just putting a small dot at each corner of the ihs. I have no worries at all about the thermal paste leaking out.

just buy conformal coating it is easy to get from amazon.

I used *Menda Chino #35622* spudger to remove IHS glue, works very good/better than fingernail. Didn't glue it back, just put the cpu first to socket and IHS top of it. No problems whatsoever. Isopropanol removes residues from black glue easily after.

Don't use nail polish and just use the correct stuff called Conformal Coating (can be found on Amazon)

Thank you for this video. I watched a number of other video's, but the additional information this video provided made me feel more comfortable.

For people that have no idea what IHS stands for, it seems fair to occasionally say the full term for some abbreviations. Just so newcomers can understand things a little better.

When it comes to the testing - always save the raw log files. You can always repeat the analysis, but you cannot repeat the measurement. If you're a precision freak, you can plot the raw data against time in something like Excel and get a derivation. Derivation is basically a rate of chage and yo ucan very easily compute it by subtracting every two consecutive number, i.e. for data "60, 68, 70, 70" the derivations would be "65´8-60=8, 70-98=2, 70-70=0". (There will always be one less derivations than the data.) What does this tells you? If the derivation is positive (above 0), it means your CPU is heating up. If it's negative (below 0) it'S cooling down. The only relevant data (the steady testing mentioned in the video) is when the derivation is around 0 (it can bounce around the 0 but it shouldn't be more than one or two degrees). It's a bit of a chore, and it's not required, but it gives a data-based argument for the quality of the testing done, if you're into that.

Hey Steve, the q tips you buy at the dollar store are wound tight as can be, theyre not super absorbant so they work great to apply the liquid metal too. Also.. You'll never believe how cheap they are!

I use a toothpick to get just a small amount of silicon to apply to the IHS and lock it all down in the socket to dry overnight.

if i join your patreon, would you consider getting a sturdier table pls?

You do not need to put liquid metal on the ihs. Just applying to the die does the same job.

I've often wondered that instead of using thermal paste that if using layers of gold flake would be better because gold is soft and highly conductive and gold flake would probably fill in all the little spaces just good as the paste and with a lot more conductivity

Super useful info here. Thanks for sharing your results!

Love the vid, usually I stick with Kryonaut - especially if I'm doing it for someone else. Removing the stock adhesive* really improves temps.

Thanks for taking the time to make a brief how to sand not what to do. Side note. Your link in this video for instructions/notes is page not found.

Saying you need to seal at the start really threw me off. Then later you clarified, I think. Was contradicting information though. Had to google to make sure not sealing was okay, since I'm not gonna be moving or removing it after my delid and LM application.

For mine, I put the CPU (i7 6700K) back into the delid tool to hold it. I then used one small drop of thick CA on each corner of the IHS placing it on top of the CPU and used the delid tool to hold it down for 30-45 min to dry. Not sure if that's the best method, but worked flawlessly for me. I'm sure I could use the tool to pop it back off if needed. Temps dropped 20C afterwards. I should add this was after completely removing all the old sealant with denatured alcohol.

Really helpful, thanks for the video 🐨 think I need to search for a guide from you guys to actually delid the cpu in the first place and then practice on my older cpu.

hell yeah JB Weld RTV silicone , saved my ass too when i changed my valve cover gasket as an extra layer of sealing haha, no more oily spark plugs.

Tech Jesus turning Intel paste into liquid wine, marvelous!

I want to see the difference between paste and liquid when it comes to delidding.

Official quote: "it likes to explode out of the back" - taco bell

Thank you for the post. Helped me through the process of delidding.

I recently put a waterblock on my 1080Ti, going for liquid metal on the die. Because I had so much LM left, I went ahead and yolo delidded my 6600K. I didn't bother resealing it, just dropped the chip into the socket and puth the IHS ontop, clamping it on with the holddown. Without the GPU in the loop and on a 360mm radiator, the CPU is now basically at the water temperature even while stresstesting.

great video, been entertaining the thought of delidding my cpu and you've answered a lot of questions so thanks for that ^^

I left my CPU unsealed, and I seated the IHS wrong. I heard a snap right after I clamped the CPU in the socket, and the IHS had gone at an angle since there was the full pressure on one side of the die. I killed my 8700K. Then I bought some silicone adhesive and made it look like nothing happened. The warranty covered it and I got a brand new 8700K for free.

I got some thermal grizzly liquid metal for xmas and applied it to my gtx 1070 and it is running at around 58c under load with a 22c room temp. Before it was running at around 63c in a 22c room temp. As a bonus my 1070 is stable is 2062mhz (2050 before). I've replaced thermal paste for a lot of machines now and nothing compares to replacing thermal paste with liquid metal that contains gallium but when you try to start the computer for the first time but the power to the house goes out before it posts. Waiting 2 hours hoping I didn't put too much on there and short out the card. TLDR: Great product. Bad electrical grid where I live.

That sigh at 9:57 killed me.. "fuck this shit"

Ive considered doing this to my 6700k thanks for info. Especially nail polish. The thing that holds me back is the price of delid kits along with the LM

OMFG. Ive been watching videos about the dilidding process for hours trying to figure out wtf to seal with and I even had a thought earlier "ya know, probably just jb weld". Ive never used the stuff but my bud s=wears by it.

Single part RTV chemistry is acid based and activated by moisture, acetic acid in most products and this is the sharp smell, but there are a few hydrochloric based (has some advantages where the corrosive nature during cure is not an issue) Both acetic and HCl will evaporate with some time. Other RTV are two part products and may or may not involve acid. Sulfur containing products can interfere with RTV curing.

I have an 8700k delidded and overclocked to 5.1ghz @ 1.39 volts. In realbench I max out at 65 degrees C with ambient temp between 24-25 C w a 280mm Corsair h115i. Without the delid I could barely make 5.0ghz and it would jump over 80 C in realbench Edit. Added room temp

Got a successful delid on my 7700k thanks to your vid. Literally dropped the package 12°C under load

Dude, you need to discover Kon-form. De-lid, clean, mask the die, spray conformal coating on chip. While the coat is wet, remove the mask, add TIM, reseat the IHS, and the Kon-form will act as a seal and glue to hold the IHS on. Sorry, I forgot to mention that the conformal coating protects the components from the electrically conductive liquid metal and from the heat produced by the die.

What I learned: 1) It's ridiculously easy, if you know what you're doing (at least with Coffee Lake) 2) You should watch a tutorial or two, and not be sloppy, but that's it 3) You certainly don't need tools that cost nearly as much as the CPU 4) That being said, use a vice. Don't put anything sharp anywhere near your CPU (Linus' blade method) 5) Delidding helps, even with crap cooler. 6) Proper airflow probably helps more. Clean those filters 7) It helps, but if that couple of degrees make any difference in your setup, you are running it at a load/speed that you probably shouldn't be 8) probably not worth the lost warranty, if your CPU comes from reputable vendor

a common way to tell which way IHS goes and I have seen this on ALL Intel CPU's I have worked with (some AMD too) is that the gold triangle is bottom left of CPU and holding it with triangle in bottom corner the text on IHS is the right way up

Thanks for the tips I delidded my 7700k and OMG I dropped like 20 degrees with thermal grizzly. I now have a very stable 5ghz and temps around the 60s and 70s under load using just a bequiet tower in a push pull config with around 600rpm on all fans in the case

"Hey do you have any nail polish that is not electrically conductive?" "Not for a girlfriend I take it then..."

the package of condutonaut says " use a pinhead sized amount" and its VERY serious its easier to add a little more than it is to remove a small pool that over flowed

Absolutely great tutorial! Thanks for that!

There is one very important point about what silicone to use that I have never seen any de-lidding tutorial bring up. Don't use silicone that uses ammonia as part of its curing process. Its often sold as "bathroom and kitchen grade". this stuff releases ammonia as it cures and ammonia destroys electronics.

Do not forget about the deterioration of Intel's compounds. I have several extra 2500k's and 3570k's. They commonly end up in lab builds, and with minimal difference in overclock ceiling, I'd grab a 2500k for max freq, and a 3570k if I really needed the pcie 3.0 bandwidth. I really needed both for a recent storage accelerator build, so I intended to push a 3570k as far as I could. I was amazed at how the thermal ceiling had dropped over the years. I though maybe it was just that processor, so I tested a couple of others. I found that all of them had dropped about 500mhz. I delidded one just to have a look, and it was visibly apparent just how much the compound under the IHS has contracted and shrunk. So if you're an oldster like me with an accumulation of parts, you very likely have a 5ghz processor waiting to be released after delidding and lapping.

Great video, i like the in-depth approach to this topic! I've just bought i8700k series, but honestly, I'm being pretty skeptic. Not sure if i would took the risk, mainly because the process of delidding looks very sketchy. I think that I'll run it like this for some time, check it's stock performance, and maybe after some time try to dare myself to do this.

Some questions, planning to delid soon: 1. How do you know what size the LM application on the IHS should be? I guess you would just eyeball it but In this video it looks like you put a larger amount than is needed to cover the die. Does that matter? 2.Would you say it's necessary to cover the surrounding components with nail polish, or is that just for peace of mind? Is it ok to use liquid electrical tape? 3. What are your thoughts on scraping off all adhesive and resealing using 4 drops of gel superglue on the corners of the IHS? Thanks in advance and amazing job as always.

nice video!! educational as always! How about a benchmarking guide with test-programs and basic procedure for checking delidding or OC?

I delidded my 7700k a year ago, and I left the default sealant on the substrate, to hold the IHS in the same position, but a year later, when I had to switch coolers, I discovered that the IHS glued itself to the substrate/CPU. It seems to me that the liquid metal soldered the two together. For obvious reasons, I did not try to do a second delid. It is still great for my temps, tho.

The enthusiast side of the industry would be in a total different place without the numbers generated from GN. Thanks for the efforts

Great video Steve.

I like using liquid electrical tape. Seems to work just fine, and goes on thin enough to not cause issues. It has a pretty high temp rating as well.

Rather than "nail polish", consider buying some purpose-made "conformal coating" specifically formulated for coating electronic components and covering PCB material. This stuff that I will link after has a UV reactance that allows checking for full coverage after cure: www.amazon.com/MG-Chemicals-Silicone-Modified-Conformal/dp/B07BSK5S7B That way, you can use a UV light (even one in your case) to check and see if you got coverage to protect from the liquid metal.

Definitely recommend liquid electrical tape rather than nail polish. Liquid electrical tape is meant for this very purpose. You don't risk it being conductive, it won't mess up the substrate or pcb, it will last forever. It's also black so you know if you missed a spot. Still see a capacitor? Add more. It takes overnight to cure but you know the SMDs as safe as can be.

sorry if someone else has already mentioned this but you might try Permatex Black RTV. you can pick it up at any autoparts store and seems to work well

I'd leave a thousand likes if I could. These problems are exactly something I would otherwise spend hours testing for a right solution or just feel uneasy about.

you need to do a colab with some precision machinist with the tooling and the measuring

TOOTH pick is your best friend to apply RTV

Interesting method. Thanks for the info man!

did this on a 8086k and the temps went from 90c to 62c.

steve the tech jesus

This video could have been less than half the length with no loss in information content.

Even if your cpu will be working vertically, you don't *have to* seal it. Unless you put far too much LM the high surface tension, coupled with the oxidation layer on its surface will hold it in place. I have multiple devices with CPUs working vertically under LM with no problems. Including a "bare die" laptop cpus in a micro-sff system.

You can use electronics lacquer, to protect the components, as its purpose is just to protect from shorts the electronic components.

You earned a Subscriber Thank you for the tips !

So much nice info, thank you.

Amazing video. Thank you!