Delidding an Intel Core i7 6700K - Is it worth it!?
Vložit
- čas přidán 3. 03. 2016
- Does replacing the TIM under your IHS make a big difference to CPU temps? DISCLAIMER: This DEFINITELY voids your warranty. We do not recommend trying this at home.
Tesoro giveaway: linustechtips.com/main/topic/...
TunnelBear message: TunnelBear is the easy-to-use VPN app for mobile and desktop. Visit tunnelbear.com/LTT to try it free and save 10% when you sign up for unlimited TunnelBear data.
Pricing & discussion: linustechtips.com/main/topic/...
Support us: linustechtips.com/main/topic/7...
Join our community forum: bit.ly/ZkLvE7
/ linustech
/ linustech
Intro Screen Music Credit:
Title: Laszlo - Supernova
Video Link: • [Electro] - Laszlo - S...
iTunes Download Link: itunes.apple.com/us/album/sup...
Artist Link: / laszlomusic
Outro Screen Music Credit: Approaching Nirvana - Sugar High / approachingnirvana
Sound effects provided by www.freesfx.co.uk/sfx/ - Věda a technologie
I'm surprised with how resilient the processor is after taking so much abuse. Whenever I get a new processor, I handle it as if it's some kind of ancient relic from the Mayan empire that's been buried for two thousand years.
lest be honest any cpu, gpu ect get's into linus's hands on and it's a review (even at CES) he basically goes mad scientist on it and if it survives it is probably related to superman
CPUs aren't that delicate. Since these are solid state devices and have no moving parts, you could do all sorts of horribly abusive things to a CPU and as long as no surface mounted components are lost and everything's clean and can make correct contact with the pins and cooler with TIM, it'll run fine.
I personally wouldn't bother delidding, though. I think the only people it really benefits are enthusiasts going for a record, and doesn't really serve a benefit for people actually planning to use the computer.
Well said XD
The1rust me 2 man rofl
+ Yaro Kasear well thats all changed now with Kaby. Average performance improvement is 15 to 20 deg C.
I honestly did not know that the big part on top of the CPU was just a heat spreader. I thought the CPU was that whole thing
same
+PlatinumGamer Its even smaller then that. Its covered behind the second "lid".
+PlatinumGamer the CPU itself is silicon :) and the IHS is nickel plated copper :)
+PlatinumGamer One can see who's young in here. 15 years ago every desktop processor was sold without this plate.
deXter2
I'm just 17, and just recently got into computers ;)
The way he's handling the CPU O.o... I thought people say to handle that thing with care and stuff, but Linus is man handling that thing and it still works?!
LGA CPUs are pretty damn resilient, it's almost impossible to damage them by accident, you'd have to actively try to destroy one.
Antares 8001
Ah I learn something everyday
I think when most people say that they aren't referring to LGA (land grid array) footprints or pinouts. Intel, i'm sure as you know, uses contact points on the bottom of the CPU with pogo pins in the socket, effectively opposite of AMD, and older Intel CPUs, which if any of said pins are bent(and they would've been when he put the CPU pin side down on the table and was pressing all over it with a chisel) the CPU is effectively rendered useless unless you can bend them all perfectly back in their place so their line up with the socket, without breaking one off. Effectively the only thing you could do to an LGA is either break off one of the capacitors on the bottom, which is actually pretty difficult unless they weren't sufficiently soldered to begin with, kill them with ESD, or mangle the die once the IHS is removed I guess. You could technically also destroy one of the pads on the bottom or completely remove it, but to do that you'd have to be really careless.
+Eric K No he doesn't lmao
StopTurtle Yeah he does lmao
The CPU is so small! It's hard to believe that there is also an integrated GPU in it too. Now that I know the actual CPU is protected in a ton of casing, I suppose that I don't have to worry too much about handling CPUs, as long as I don't bend the pins or leave unwanted skin oils
You mean bending the pins that dont exist since the Pentium 4 anymore? ^^
Chuckiele
Erm, AMD?
Dante Inouye Talking about AMD under a delidding video seems off topic.
*****
AMD has ihs on the cpus
Dante Inouye But they are soldered. Unless you wanna kill the cpu, you shouldnt remove it and you wont gain better temperatures by doing so either.
I'm going to sleep. When I wake up and check this video again, I had better be the top comment.
jerk
Can't be top comment if you don't use code KEEM
+spartan092 i beg people to don't like above comment, like this one instead ;)
+nathan gauci sounds like somebody didn't use code KEEM
+spartan092 you should change your bio.
Did anyone else get nervous when Linus was waving the surgeons knife around and explaining stuff?
+chumppi I hope your surgeon doesn't use one of those..
+LinusTechTips my surgeon use butter knife :D
+Fredy Rommy how are you alive?
+Fredy Rommy That's the NHS for you...
+chumppi your telling me I kept saying to myself "no cut away from your hand not towards" I have many scars on my hands from knives by using them like Linus lol
wtf,why should this even be possible?One geek with scalpel and "cheap" thermal paste can make cpu perform better??wtf Intel engineers are doing??They can design complex cpu,but can't improve that simple thing that even i can improve with my scalpel and some paste??That's a bad engineering right there!
Ikr. FFS
They used to weld the chip. These days, they just use crap paste to reduce production cost.
dongle seon AMD CPUS still do that
Because they are so hot that they can't work properly without welding, I guess.
Wrong, AMD cpus max out at 65c and 55c for the higher end cpus.
I remember the old days when everyone would blindly follow the loudest proponents of overclocking, even for a 2 degree difference. it's nice to hear you say, "Overall, this isn't worth the risk."
Ben A how young ARE you that you can't read between the lines?
I know that's his opinion. and after 20+ years doing some of my own crazy shit to computers I feel I am qualified to agree that this mod isn't worth it. people are so lucky to be able to BUY water-cooling sets.
Meh. I Am thru explaining myself to trolls...
..... Nice use of dots in your reply though. you are to be commended.
Ben A 5 syllables. wow.
I miss your dots.....
Ben A the really long strings of dots.......
so, telling some to think for themself when they are remembering the days before windows isn't trolling then what is it? general dickishness?
and lets get something else out in the open here. people DO blindly follow others. being a Dick only pushes them to keep following way longer than they need to. how about just saying I disagree?
Ben A not sure about the doing something longer, but that Authoritative Fallacy is pretty fancy. What qualifies you to disagree with my ORIGINAL complement? what experiences do you have. building RAM drives for an 80286 system to play Test Drive at work? oh wait. that's me. tearing down DAT drives and rebuilding? hmmm...
so far, all you have mentioned is big words.
LinusTechTips "because we shit money"
+GamerGab007 Nah, they get this stuff as gifts. These companies sponsor his channel. I'm sure he buys some of it, but most he doesn't have to.
Yes, yes, I know that, I meant it as a joke. A great example, I believe, is "7 Gamers 1 PC"
+El Diablo I just pictured him straddling 2 different toilets, and shitting on the floor between them..
they're actually not rich and they don't keep all that stuff.
+Awwesomemoment this^. plus they are running an office, which has maintenance costs and bills to be paid plus linus, luke and the editors have houses and families to support etc.
People don't use their profit to by sailboats, they use it to go to the supermarket.
>no overclocking
>no fixed rpm cooler
>no most popular tim replacement (liquid metal)
>"NO SIGNIFICANT DIFFERENCE GUYZ"
GJ Linus (nope, not this time, sorry)
+Тёма Бондарев ¨linus is a noobzor
+Тёма Бондарев they might have also done it so tech-newbies don't go around breaking cpu's
+Тёма Бондарев Agreed. I saw a definite difference in my delidded 3770k.
I have never done it back when I had my Ivy bridge cpu, but I do remember it running at 60 degrees while watercooled and not overclocked, so I know why you did it
I accidentally read the title pf the video as "diddling the Intel Core i7 " 😂
Wageesh Arya Same!
Wageesh Arya same
same
It would still apply lol
Ehhhh close enough 😂
There are a lot of OCD enthusiasts who would go to the ends of the earth to get that extra fraction of performance.
Admiral Ackbar honestly i would NEVER do this. i love my hardware to much to put it's lifespan on the line. And gees a couple of degrees? just get a fn water cooler. Easier, cheaper, far better boost and works better for overclocking.
Alex V4 just not true, if the tim is bad a water cooler wont help that much, thats the whole problem.
20°C is "only a fraction"?!?
xXYannuschXx anything is a fraction if you put /1 after it
xX_ScetchUp 9R0_Xx nice
omg. Never realized there is another layer of thermal paste inside. One learns everyday I guess.
same
On some cpus, there's no thermal paste, the ihs is welded to the top
u can like comments in smartphones now :D
u can like comments on phones niw :D
+Björne ッ nice! :)
This public service message is brought to you by Intel.
I'm still going to do it! I'll lap it then polish it with 1 micron polish than 0.5 micron polish :)
I would think Intel would be happy for you to break your chips... so you can keep buying new ones.
+nicklong27 and also they won't have to cover your warranty
Nicholas Gill Exactly
Just remember to buy a performance Tuning Protection Plan from Intel.
Going to revisit this with Kabylake?
Not necessary since it's probably going to be the same result with Intel's cpus these days.
From what I've seen at least one guy who had a 5GHz OC was at ~100 degrees C then same OC dellided went down to 66 degrees C.
I'm not saying it'd be redundant to do it, but more so a bit redundant to make the same video on the matter when it's virtually going to be the same thing shown in terms of the procedure. The results vary, but I'm talking about making the same video on something that will probably prove the same result for him.
If true, that is likely because of a manufacturing defect specific to that individual part.
Skylake = kaby lake.
The 180 flip you did here is the funniest thing of yours I have seen. Good job! Love your work! ;)
what about doing a test without the IHS at all?
+conenubi701 Yeah I was hoping he would do both
agreed.
it wont reach or it might crack the cpu. Jaysz2cents did a video on this already.
+jlrockafella I think some MSI boards have a "delid guard" that helps prevent something like that from happening.
+conenubi701 Straight to die cooling would involve removing the latch mechanism that holds the CPU down because the bottom of your cooler would most likely make contact with the latch mechanism, thus the cooler would never make direct contact with the die. Even after removing the latch mechanism you would still need to somehow shorten the mounting screws of your cooler so that there would be sufficient pressure on the die to spread the TIM.
LinusTechTips "Hey intel you mind sending me a i7-6700k so i can cut it open for a video?
Intel: SURE BRO NO PROBLEM
Me " hey intel you mind sending me a cpu? I would like to make a video.
Intel "no fuck you. Buy the cpu urself"
Because you aren't famous.
MrTechlion ooo someone's a little butthurt... :D
+Blooperinos LOL no. I'm not famous either. I'm just pointing it out.
He's not being butthurt.
Lmao you're just mad because your comment is cringe as fuck
Hey names Eric been a long time fan of the channel, i was looking to get a friend about to head to college a keyboard to go with his new budget pc. It would be an amazing gift, thanks happy new year!
5:45 WHOA LINUS TURNED A 6700K TO A PENTIUM 2? DAMN
Of course the 6th gen cpus wont have much difference when delidding. He should have done this with a 3570k or 4670k, the tim they used for 3rd/4th gen was fucking nasty. My 3570k was hitting 95-100c with a hyper 212 and adequate ventilation (no it werent to do with cooler mount, i remounted several times) at stock speeds running intel burn test. I then delidded it and applied some mx2 to the die and reseated the cooler etc. I ran intel burn test again and my highest core was at 80c, so i got a good 20c drop in temps. Also the idle temps went from around 55c to 37c.
The tim that was on the cpu die had similar texture to the sealant and i had to rub quite hard with an alcohol wipe to get it off. But again, this isnt a problem with skylake cpus so this video is quite misleading because it sways people away from delidding but it doesnt show you the scenarios where it is really beneficial.
Dont get me wrong, i love linustechtips, its just I dont like the way this video sways people away from delidding when it shows results from a cpu that doesnt have the problem that the cpus people actually delid do.
Well even Skylake 6700K regularly gets a ~10 degree Celsius drop from delidding if you do it PROPERLY by putting on Cool Laboratory Liquid Pro/Ultra or Thermal Grizzly Conductonaut etc.. (only $5 for the conductonaut!) liquid metal pastes instead of the no-name junk paste he used here thats no better than what intel had on the chip already really. And you're supposed to SPREAD the paste on the die not just GLOB it on lol. Linus literally had NO clue what he was doing, a lot of people are accusing him of being biased or even being told by Intel to make a delid video that purposely looks bad to keep people from trying it so they won't get as many RMA's (because intel actually doesn't really check their returned RMA CPUs for being delidded usually; only to find out later after they refunded you already lol)
+TheScaleless my point is the skylakes dont need delidding because temps are usually fine whereas if he did it to a 3rd or 4th gen i5 or i7 he would have seen massive results. I didnt even use any special tim just some standard mx2 which i used the line method on the die and ihs and my temps dropped around 20c on average
+MRsjovealek then you are a very lucky person who good a good one
MRsjovealek
i think you'll find its a majority problem rather than minority
MRsjovealek why do you think so many delid their 3rd/4th gen cpus? because the thermal paste under the IHS is complete shit and it hardens over time making it really bad at conducting heat leading to higher temperatures. Also I've heard reports of peoples IHS's not making full contact with the cpu die due to excessive glue holding it down and the actual IHS being slighting convex/concave. If you havent had any problems then good for you, just because you havent heard of the problems it doesnt mean they arent there :)
I'm afraid to say what I'm thinking because people will say mean things to me, which really hurts my feelings.
ehem
Wimp. Man up you pansy.
+SunPower Guru People will say mean thing to you regardless.
+SunPower Guru I don't have feelings cant hurt what I don't have.
+john cavender it's not always about you, it's about me.
OK, I have trust issues. Perhaps I could hint about what I was thinking and see how it goes...
Sharp knives, surgical precision, ability to adjust a chair.
Be nice to me, I bruise easily!
I delidded my 3770k and got at least 15°C less everywhere (and depending on load and ambient temps even more than 15, up to 20). As long as you're using a delidding tool (I used Rockit 88) and know what you're doing, I recommend delidding.
People please dont take this video seriously. You're supposed to do this if you want 5 ghz.
Not for stock clocks, on stock cooler
you could probably boast a pretty medium overclock, even with stock cooler after delid. And atleast keep full turbo speed at max load. Which a cheap cooler can't accomplish. But noone in their right mind would choose a stock cooler for any K version cpu. They don't even come with a stock cooler.
I got 5GHz without delidding. WITH i got 20°C cooler and 5.2GHz
Delidding without OVERCLOCK.... Why the... FACEPALMMMMM!!!!!
2:21
+fatbloke Don't say it - test it! )))
2:21
+fatbloke higher heat output makes the temperature difference bigger.
+kuugeli Exactly !!!
Dont try this with the Haswell-E series of CPU's (5820k, 5930k, 5960X ) Their TIM is soldered to the IHS and if you try to remove it you'll destroy your silicone.
Silicon
Silly cone
Not true actually. There are delidding videos for Broadwell-E already that show the IHS can be successfully removed and the remaining solder scraped clean. Whether that is even worth doing, I'm not sure but it can be done.
Dunno why anyone would want to lap a soldered on IHS, the amount of thermal loss through the soldered on IHS is so insignificant that you'd be overclocking to the point that a small change in room temperature would be the difference between a working or thermal shutoff CPU.
yeah, also don't do it on your P4 CPUs. You'll rip out quiet a bit of glas fiber
These are the old LTT videos that I love and these are the reason that I subscribed to this channel.
6700k??? no i would only do this if
1. the cpu is too hot with a good cooler
2. near the end of it's life
3. the warranty is already over
4. if that cpu module is not selling well on Ebay, Craigsist, ect
5. if i have money to spend on a new cpu
do you agree
it's just i would not do it to a 6700k i would do it with something like a pentium for OC and TBH i have a i5 6600 in my pc and i'm not broke just cheep and i don't care about the price i just want a really good value and no, not everyone can buy top of the line parts like a 6950X CPU or a titan X (pascal) and i don't care about universal heath care right now because we are on Linus tech tips not reddit
thanks for the complement
*POKE*
I don't think points 1-4 matter. for some of the richer people, risking the price of a CPU for fiew degrees of temperature and the joy of living beyond the bleeding edge of available consumer electronics is a small price to pay.
and to them, I say "may I have the stock heatsinks that come with your CPUs, I want to sell them to the recycling center."
Yes 6700K has good cooling, this was before the awful Kaby and Coffee lake. Watch the delid video for the 7900X, it needed delid just to avoid throttling.
How can I download more cores to my i7 after delidding it? Do I have to connect my CPU to my DVD drive in order to allow the lasers to do it?
I have 2 cores but i want 4 cores, will it take more time for the lasers to download those 2 cores?? Also, i want a mechanical keyboard, but i don't have any hands.
+Incytech XD
i died xDDDD
GEEZ XDDDD
let me guess you have half core of a brain
+gelpogi123 let me guess you have completely missed the joke
You've lost a lot of weight since the South Park boys defeated you in WoW!
+LCdrDerrick whuaah?
+LCdrDerrick lol, same gear
+Winicius Benevides There's an episode of South Park where the kids get addicted to World of Warcraft, but this one obese guy manages to spam-kill them over and over again, causing them to grind levels and eventually defeat the fat guy.
fucking savage!
What part?
I like the music you chose for this, it really screams "I'm doing something really stupid just to see what happens".
I first did this back on Ivy Bridge with a watercooling setup up direct to die with Cool labs liquid metal. It took 20C off my overclock so I was hugely pleased at the time!
Ive done it on my 4790k but not had the same success, more like 10-15C.
It's not going to make much difference in this scenario - stock voltages with the stock Intel cooler but start putting more watts through that combination and you'd see more than 3C I'd bet.
5 degrees at stock speeds with a badly done job is significant. To quote a message from a user named Ed Turvey - "This public service message is brought to you by Intel." and Linus' videos certainly seem to have a huge Intel advertiser's slant. With a well done job, an overclocked processor and a watercooling setup pulling away significant heat you can expect 10 degrees. If a 5 degrees difference is significant enough for everyone to advertise good thermal paste over cheap stock thermal paste then 10 degrees is certainly significant enough for a reasonable number of people to do with highly overclocked setups. Its not for everyone, but 10 degrees will allow for higher speeds and a good improvement to processor lifespan, Linus' badly done attempt with 5 degrees will do the same to a lesser degree yet still more than noticeable.
And ultimately this corporate PSA slant to everything Linus does for Intel is why I've largely stopped taking Linus seriously. Hes entertaining but I make sure to take nothing away from the videos and look elsewhere if I need information and suggest others do the same.
You're overthinking it, it's just more content to put out and everyone has their preferences.
Are you taking linus that serious? :s
nah, dude hiding identity so
Why in the hell would you ever want to overclock any CPU anymore? Stock rates of nearly 4Ghz per core, and there are many of those cores. I run an early I7 Quad at its stock 3.6Ghz and it is rare that anything uses it to full potential.
Rot Bear my 2600k is oc for 4 years on 4.0ghz :') cpu based games op
It wasn't Haswell that got it started, it was Ivy Bridge and the bird shit that Intel used for compound. Tualatin and Northwood have the same crap as does all the budget dual cores from the c2d era. So glad that I moved on to Xeons enjoying my super low temps.
As long as within design spec, there simply is no problem, doesn't matter if it's running hot or not. The engineers at Intel are amongst the most educated when it comes to integrated circuit engineering, I trust their knowledge before a bunch of plebians on the internet when it comes to operating temperatures and circuit lifespan. The amount of cooling from a heatsink outweighs the thermal loss from using cheaper thermal paste between the die and IHS, the stock heatsink is designed to maintain a die temperature below the maximum Tjunction temperature at stock clock, and enthusiests wanting to overclock will always buy a better aftermarket heatsink which can cool beyond the stock clock. Overclocking is not condoned by Intel, it can reduce the lifespan of a CPU and they did not design it to operate beyond the stock clock, it provides such overclocking capability as a courtesy. They explicitly do not allow overclocking Xeons because they understand that such a thing is risky to the stability and longevity of a system, and there is no place for that in a professional datacenter. If you want to push boundaries, buy an extreme or "k" suffix CPU. Less than 10% of customers that buy the standard models of Intel CPU end up leaving it overclocked, so spending significantly more money for better thermal paste on millions of CPU's is quite plainly silly.
+GOAE7777 Well said.
Basically everyone that has done some over clocking and knows the shenanigans Intel pulled on their CPU's since Ivy Bridge, disagrees with you that there is no problem. Intel moved away from fluxless solder and used crap paste, so that more profit goes into their pockets! The K and extreme versions of Intel CPU's use the same crap, negatively affecting overall temperatures and overclocking. Higher temperatures DO affect your and everyone's CPU life span!
It isn't just as simple as you saying that as long as it's within design specifications there is no problem. Yes, there IS a problem and that is why there are people that DO pay attention to the consequences of moving away from fluxless solder and have written articles about exactly that! It is why there are delidding videos on the internet, including this one even though it's a bad video I cannot advise to follow.
"The amount of cooling from a heatsink outweighs the thermal loss from using cheaper thermal paste between the die and IHS..."
Dafuq? No. If you have a shitty paste AND it is making sloppy contact between CPU and the heat spreader due to a certain thick adhesive compound, a large heatsink or watercooling won't mean jack shit, because temperatures are going to be HIGH. For common people that do not clean their heatsinks often enough, dust is going to clog up it and potentially fry their CPU thanks to the already high temperatures!!!
Seriously. I recommend you to read some in-depth articles about this. If Intel had a real competitor in the industry (sorry to say AMD is not a good one), Intel wouldn't dare to pull this shit off.
I mean, if we're talking about spending significantly more money for better thermal paste, they could just do that on the high end desktop consumer CPUs.
The ones that only people who intend to over clock them would buy.
But, it's whatever.
The practice of changing out that thermal paste seems a bit... silly.
OC'd with water cooling, my CPU doesn't go above 60 degrees Celsius under my normal everyday load.
I'm kinda fine with that.
You know, I wonder if actual bird shit would be a good thermal compound. Linus should test that for an episode of HOLY $H!T
literally
I saw a massive temperature drop on my 4670k from delidding. It's very important to remove every trace of the glue to ensure that it's seated properly over the cpu.
I actually did this before as well. I had an old P4 HT box I made for strictly media playback. Music, movies, & so on. But, the case I was using was actually handmade and I modded the heatsink with a thick piece of copper and aluminum coils that looked like a stil for making moonshine lol. And then added a 120mm fan directly on top of the coils with a funnel keeping it 'kinda" sealed for the intake. Then another small fan for the exhaust connected to the funnel as well. It wasn't like a HOLY SHIT! moment in cooling. But there was definitely a noticeable improvement in cooling that the stock cooling system couldn't do. Also, those old P4 chips had a pretty thick heatsink and the plastic securing the fan was Jacked up too. Long story short, the only benefit of the mod in my experience, was to incorporate a custom design for cooling the CPU for a small case. Good video bro, definitely brought back memories..
Was picking my nose when that intro popped up lol
yuck.
Me too, I was so deep up in there, yo.
+chiken a rouge booger tickled the inside. I had no choice
+MrAttheMALL No one cares about you Sterben, or laughing coffin. Kappa123
me too.
Was that a consensual delidding Linus?
If you are soft try the Pentium G4400 first ;)
This stuff is cool; learned a lot! Thanks Linus :)
Even with water cooling(h100) my 4770k(stock speeds) was idling around 55°. Load jumped up to 85°+. My fan was constantly running loud.
After delid & lapping:
35° - idle
67 - load (turbo boost)
Hey, Linus!
Saw the Vid with replacing the TIM of the 6700. Guess you miss something.
Was the CPU Fan @ Auto in BIOS? If so, the board lowers the RPM of the Fan to keep the CPU at the "normal default" temperature.All in all: Lower noise, same temps or same noise and cooler.There were some decent hardware magazines doing the same with fixed fan speed which results in an approximately 15 to 20 degrees CPU temperature sink.Keep up making vids, really enjoying them!
Greets from Germany,
Daniel
It would've been much easier to delid with a vice--takes less than 5 seconds once it's in place. It would be less entertaining though.
As for the difference in temps, it's due to the gap caused by the black adhesive around the die and not the excellent thermal compound that Intel uses. Even with liquid metal, temps would be similar if not worse with the black adhesive left on.
+Wooden Marker Vice method not recommended for Skylake due to thinner substrate. As for the black adhesive, editor might have removed it (I didn't watch the video) but I scraped it off.
Ahh, I didn't think about that for skylake.
I did notice that the black adhesive was taken off but what I was referring to was the reason why temps are better with delidding. Temps are better due to the gap reduction when removing the black adhesive and not because of Intel's TIM which is actually excellent.
+Wooden Marker Nice theroy, now it is your turn Linus to test it! Also you should test a high end cooler which might make the diffrence gap even higher than 3 degrees!
+Fergesslich One of the reasons why the difference in temps isn't much is probably that the stock cooler doesn't apply much pressure which might help lessen the original gap after delidding. Most normal thermal pastes also face pump-out issues when between the die and ihs and don't fare as well as the thicker material that intel uses or liquid metal.
+Wooden Marker I think your right there mate. I went from thermal throttling (100deg) to a steady 72deg when stressing with aida64. This was using an EK supremacy block and with a 4.8ghz overclock dialled in. I think when your pushing the limits of the chip it's definitely worth it.
Man Linus, I can't get enough of you saying "video description! I stay and watch the sponsor and get subscribed part each time to hear you say that!
This was a damn good video, Linus!
test seema flawed. you have to use the absolute best cooler you can get to isolate that variable!
***** thats like testing the flow of two liquids in 2 very narrow bottles... At the end they both will barely flow because of the bottleneck
agreed, a good cooler and overclocked will demonstrate the real difference. It's usually pretty substantial, I can vouch for the 20c claims.
that and you know that's likely their cherry picked engineering sample CPU.
No other reason to see such a polished finish on any normal, retail CPU. Baseline with their sample CPU and testing with it to keep hardware tests legit.
Now, did they dial back their voltages since the delid? I don't it...
Intel should offer K series processers with a pre delid option. Ie the IHS and processor are separate bits in the package.
I had removed the lid on one of my old AMD Phenom builds way back when and the biggest pain in the ass was modifying the brackets so the heatsink made contact with the die. And for safety I added 4 solid foam dots on the corners of the chip PCB like the old old old AMD cpus to help protect the die from chipping off a corner.
It doesn't bring much benefit with Skylake anymore but used to be a significant improvement for Haswell CPUs and contrary to what Linus says at 2:14 it matters how good the cooling is because the improvement in temperature difference you could gain with this was up to 10...15 °C for decently overclocked CPUs that would not even run with the stock cooler. Less overclocking produces less heat and thus makes the improvement less significant. Usually it allowed 100...300 MHz higher clock rates for Haswell CPUs.
it does bring massive improvement to skylake, with 20C being the average drop, he saw a fraction of the because he used a stock cooler meant for i3's and below, its a thermal bottleneck
ya slicing your hand would suck :P i did that onces when taking 2 frozen burger apart with a knife :P hello hospital and 9 stitches. hehe
rip
+Finn Andersen he was delidding, i was cringing...
+Prashank Abhishek ya i was doing the same. Just a litle slip and maybe a finger tip gone ..
+Finn Andersen hahaha yeah i cut myselfe so many times while delidding CPUs :,D
+Finn Andersen How was the burger?
it is an improvement, first of all
second of all, you only used a normal heatsink paste but you yourself said that there are more advanced solutions like that "liquid metal" thing you talked about
so i guess from what you said you can improve the temps even more.
i get the risk and all, but you just proved that it works....
+yakir11114 He didn't even do it correctly either (or at least not 100% efficiently for the stuff he had). Using a better compound, and taking the time to prep the compound and surface I would estimate to give from 6 to 8 degrees of difference using the stress test he used. That's a quite significant gain. An 8+ degree shift in ambient temperature over the Summer for non-air conditioned rooms is common.
Like the you. Explain your and the y's. Rock on dude. Great Job.
well i did this on my 4670k and had a good drop in temps (11c underload), the TIM had gotten crusty under there. That skylake CPU looks like it still had some life in it;s TIM so im not surprised
I did this to my 4770k and let's just... I got 4.5ghz instead of 4.4... 😎 totally,,, not even Worth it unless you are just stupid like me. my heart almost stopped when it went flying off the vice onto the ground.... but it survived. and instead of 78° on water it is at like 59° "also bare die mounted". totally not worth it :D
***** haswell got really really hot lol...
***** oh yea man. It was ridiculous. That's why the released the 4790k. That was just a 4770k with no so terrible heat output. That's how bad it was. Good architecture, bad everything else
Wait, so it went from 78 degrees to 59 degrees while increasing the clock speed? If that's not a great improvment I don't know what is...
Gustaf Anderzén oh yea it was for sure. But it was a very dramatic and time consuming process. It became an obsession of mine for a long time. But I got my extract .1 GHz 😂
It actually depends on the condition of the thermal paste. If the CPU is relatively new and haven't gone through prolonged heavy loads, the stock thermal paste works fine. But in my case after 1 and a half years of use and occasional 115watt renders at 4.6ghz @ 1.295v in the 4790k (overclocked from the stock 88 watts) the TIM starts to degrade. My idle temps hovered at 68-75 Degrees, and under full load reached 85 -90 degrees (room temp is 30 degrees). I had to reduce my OC and compromise or lose the cpu but i need to render faster so the scenario doesn't work for me.
Yesterday I delidded my 4790k, and the results were astonishing. My temps improved by 20 degrees and the cpu can take pretty high and stable overclocks (up to 4.8 ghz now). I decided to stay with 4.7 ghz @1.30v and just render at 4.6ghz @ 1.262 (down to 109 watts). Now my max temps are just 72 degrees, idle at 48. Good enough for me, it's as if my CPU was reborn, and even better (Just like Gandalf the Grey becoming Gandalf the White hehe).
Serious question, why does Linus only use Intel CPUs in all rigs? Is it a brand deal?
1: they're better
2: he has actually done amd rigs in the past, but amd hasn't come out with any new cpus worth purchasing since then.
(this coming from an fx-8350 owner)
There are literally 0 high performance amd cpus. Bulldozer was complete shit for amd.
avanasear FX9370 myself.
Sponge Boob The FX 8/9 000's series?
+Dr. ViperKevin He does AMD systems often, its just they have no solutions for many of his projects at this time.
i know its not the same. but i did it with a 4770k. but i used metallic thermal paste (coollaboratory ultra). i also used a thin layer of liquid electrical tape to re seal the IHS. i was also able to have my friend 3d print me a delidder. you place the chip in and place in vice, squeez a bit, and it pops off.
my CPU doesn't OC very well. i set it to 4.4ghz when on 1 or 2 cores and 4.2Ghz when on 3 or 4 cores. i think its at like 41-45c under load. and 30c when idle. on a h100i. i cant remember what it was before.
Improving the heat transfer near the source of heat will improve the performance throught the whole cooling Sink setup (heat resistance of cooling elements is in series, so lowering resistance on source will help you transfer more heat to next part of the cooling chain and if it is overkill cooler you will see more improvement than with stock cooler.
Linus, you mentioned at the beginning that you're only going to change the thermal paste but that there are other methods that people do which you won't cover. But then in the end you make a generalization about how de-lidding isn't worth it. With all due respect, but your conclusion means nothing because none will go through de-lidding their CPU to just put a fresh paste and put it back on. No, people that de-lid their CPUs will actually mount the cooler directly to it, sand down both surfaces, etc etc. With such things, you either go all the way or you don't at all. I think your conclusion is irrelevant and doesn't help the people that are trying to decide whether to do it or not. That being said, I love it that you did such a video. I expect more of this :)
+SaladFingers Your last but one sentence ruined it all :|
+SaladFingers It's worthless unless you're an extreme overclocker that want to run the overclock 24/7. I can reach 49 * 101.5 with my 6700k, what would I achieve more by delidding? Nothing, a couple of degrees at most. Most people don't run their max overclock on a daily basis.
some people are getting 10-30 degrees C on kaby lake 7700K after delidding, adding liquid metal and regluing it ...
this seems like a half-assed attempt?
getting 10-30 degrees lower*
@@nioxic77 7700k Stock with Intel Turbo Boost, from 89c under Aida64 FPU Stress test to 67c after de lidding + liquid metal, so yeah
i thought it was late for you to do so, only after realising this was march 2017..i did sign up for the entry tho haha
You could make a video about undervolting, which was the only thing that helped with my very hot i7-4710hq processor!
Awesome channel btw
I wonder if today is the day Linus finally notices me.
Maybe, one day.
+PhantomTech huzzah.
+LinusTechTips :3
+LinusTechTips hahahaha :D nice one
+LinusTechTips
My GPU is now overclocked.
If you're catching my drift.
❤️💕💕❤️
Use liquid metal
#herpderplinus
That stuff works brilliantly. But it can be tricky to apply. I prefer the gallium-like sheets that don't have to be painted on. Just pop the sheet between cpu and cooler and toast the hell out of the CPU till the temps reach 98C and the metal melts and temps plummet.
Instructions not clear, we think the processor was melted but my mother and I barely escaped the steel mill with our lives
3:29 R.I.P Intel i7 6700k 2015-2016...
Emrldy why would that kill it?!
Start of the process
awesome cable management :)
KillingLynxNL ikr
Looks like Intel just bought LinusTechTips channel. I don't recognize this channel now...
+Diego Cortez Intel would not endorse something that would void the warranty and likely break their chips.
people's stupidity never ceases to amaze me.
+Diego Cortez hes a tech channel? what do you want him to mess with and overclock the softness of puppies?
+Karol Jampolski that is
Not how marketing works
Is that the proper way to put heatsync paste on? I always thought you had to put a very thin layer over any surface, not a blob
pistolschlapper Everyone has their own way of doing it. I've seen the 'rice grain' method used before, as well as a 'blob' , as well as a thin 'sheet'. I personally always use the thin sheet method.
thanks for replying. i just remember back to the days of the celeron 300a. used to over clock them to crazy speeds for a few weeks till they died. i think i was buying them for 30 dollars or so. never seen quake world run so good with my voodoo2. i miss those days :)
should have said, i always noticed the thin sheet method delivered better temps. i used to blob but an electrical engineer introduced me to it. always thought that we was the way...
When I get my muse back, I'll make an autocad model of what came to my mind just now. In my mind, the most efficient way to cool a cpu, is how we used to do it back in the athlon xp era - applying the radiator directly on the chip die. This ALWAYS yields best results. So my idea is for a heatsink unit for a water bloc, that can be secured on top of the die via the mooring plate. That way we'll be 100% sure that it's not sitting at an angle and that the CPU's pcb is making solid contact throughout the entire socket.
I didn't breath while you cut the adhesive. That's some scary shit mate.
"Holy Crapola that is tight".
MMMM you don't say Linus....
Linus Sex Tips
This test was massively flawed.
First, Linus, there is a very good reason people use and recommend Liquid Ultra for application between the die and the IHS. When you use Liquid Ultra normally, between the IHS or heatsink, the difference between it and other good compounds is negligible. But when it's applied to the die, the difference is HUGE. Double digit drops, in C, compared to other non-metallic compounds. (I've found this from the enthusiast laptop community which I used to be a part of, using Liquid Ultra between the die and heatsink on high end laptops - where there is no IHS - easily reduced temperatures by 10-15C compared to any other compound. I'm not exaggerating.)
Second, you need to be very careful about what compound you use between the die and IHS. It looks like you used MX-4 there, would I be correct? If you'd continued to test it, you'd find that a few days later, you'd be running 20C HOTTER than before. Many compounds, such as MX-4, cannot withstand the extremely high thermal gradient that can exist between the die and IHS (or, on a laptop, the die and heatsink). Normally, the IHS (and the thermal interface between IHS and die) would massively smooth that thermal gradient. Again, this is why Liquid Ultra is recommended - it withstands those thermal gradients AND performs far better with those high gradients than with any other type of paste.
Third, testing it with only one cooler, and at only one clock/voltage setting. A thermal interface that is adequate at 80W may be a limiting factor at 120W. Similarly, if you had a proper cooler with multiple heatpipes that can draw heat away far faster, the thermal interface may become far more limiting.
People delidded Haswell because it produced a 10-20C drop on many CPUs, and that's no exaggeration. Because they did it properly, with Liquid Ultra, with a proper cooler, in an overclocked setting.
Linus I'm gonna do my first build in a few months and was wondering if you could do a video on static and how much it actually affects your equipment. You just seem to go all out when building your rigs so I get the impression it's not as much of an issue as it's made out to be.
My FX 8350 sits at a comfortable 50-60c on load with my Cooler Master 212 EVO, and sits anywhere between 9-20c while browsing the web. Gotta love the little things
what about the stories of i7 7700K with 30 degrees improvement
I disagree slightly....
I did this on my i7 3770k as my core temperatures were atrocious with a pretty decent OC and being cooled by a Noctua DH-14 my temps were averaging around 70c. Once delidded I cleaned up all the glue/thermal paste and applied Coolaboratory Liquid Metal between the CPU itself and the IHS and used GC Extreme for the CPU cooler. On the same OC my temps went down to an average 55c, also the other big improvement was the temps across all 4 cores are very consistent now with each other as before the procedure it was very inconsistent.
I think the difference will be relatively bigger if you use some serious cooling setup instead of stock cooler. Combine it with properly lapped and flat base of water cooling, using coollaboratory paste without IHS (if you can mount it properly) an will get results that are worth it.
cool video. thought it was funny seeing the alcohol bottle that said "linus' alcohol".
Linus, you should use the liquid metal thermal solution that you've reviewed on NCIX Tech Tips instead of whatever you've used this time...
Oh i am early! Damn. Back to bed...
Exactly:)
Lucky....
Good one
By me in Germany, you can buy on the site caseking a thing to remove the heat spreader kind of uncomplicated.
This video was put online on my birthday!!!
it's been 4 years and the 6700k I got has awful temps, easily reaching 95c on a hyper TX3 with a modest overclock. I expect that the thermal paste inside has degraded some.
tl;dr "it depends."
Vandoeun Long tl;dw
it would be useful to see this test with different coolers - a decent air cooler and then an aio water cooler perhaps; as well as an overclock.
the problem wasn't the thermal compound between the chip and ihs, it was the glue that they used to seal it that caused a slight gap... clean the adhesive and the thermal compound and use CLU and you will see a good 15-20C depending on how badly the adhesive was applied
I think that if you decide to delid the CPU, then you should go all the way - get better TIM ("that liquid metal stuff" :P), sand IHS (which is not perfectly flat out-of-the-box) and use proper amount of TIM with a aftermarket cooler that has great performance, not that box cooler crap. I know, that Linus only looked for delta between delid and standard CPU, but when you are testing tweak that is on the more of the extreme part of the spectrum, you should also prepare whole environment. I bet that if you would use good setup mentioned earlier and then check delid vs standard, then the results would vary more than 2-3C. I delided my i7-3770k, sanded/polished IHS, replace TIM between die and IHS to Liquid Pro, used good quality TIM between IHS and aftermarket cooler and I've got 12-13C drop in temps. Granted, Ivy Bridge had one of the worst TIMs and space between die and IHS, that's why temps went that much lower. It was corrected in future generations, so you won't see this high drop anymore. It lets me to run this CPU 24/7 4.4GHz with temps lower than standard clock non-delid (still tested with aftermarket cooler and good TIM).
+CoreyPL This is the EXACT review I was looking for by Linus. Thank you and Fail by Linus.
+Tanzu15 Basically you sand the metal part with the name of the CPU. Before doing that there are plenty of other factors to consider: 1. It will void your warranty; 2. If done incorrectly it will damage your CPU; 3. Works best with good aftermarket cooler with flat rectangle\square shape. You must also check if your CPU really needs it - take a metal ruler or something similar that won't bend, put the edge on top of the IHS (metal part) and shine a light through it. If you see more light bleeding out in the centre or edges that means IHS is not perfectly flat. Be sure not to use too low grit sand paper because you can sand too much of IHS and damage it. Remember that after sanding you will expose pure copper on the IHS which will oxidize on the parts that will not be covered by your aftermarket cooler and you will probably need to clean\polish the IHS to get rid of that. That are some basic rules to sand IHS. Be sure to take all the pros and cons under consideration, because once you do the first move on the sanding paper there is no turning back. There are also a lot of sanding tutorials on CZcams - check them before doing anything and if you decide to go with it, prepare all the materials needed beforehand.
Why is Linus still such a noob with PC hardware, lol
He really is.
+Nerdburger85 I'm curious why you think that...
+Yuannan Lin He's never cared about trying to keep the components "safe"...
I understand that is true but I was stating that Linus just doesn't care. He's never gone through any steps to make sure the parts don't get damaged.
yeah he have the money for buy more. like PAST VIDEOS XD
when i used one of those knives 2 weeks ago i sliced my thumb because the knive part broke and it still hasnt healed + my nerves dont work anymore in the top of my thumb... always wear gloves with those kind of knives since its very dangerous when applying force on those knives
When you see that literally any video you can think of, this guy has made already, then you know he deserves these subscribers.
I wonder what would happen if you were to just use mayonnaise instead of thermal compound? Could you post a video of this experiment? I am curious
Racist... why not ketchup
+Savvy Seagull Thicker, whiter..
+CarHub He should just do a video to see what food item is the best alternative to thermal grease. Mayonnaise, ketchup, mustard, miracle whip etc.
+FubarMike There have been a couple sites that have done this over the years. 3 of the best alternatives are 1. Toothpaste 2. Vegemite 3. Peanut Butter
I have actually tried toothpaste myself and it does work slightly better than Arctic Silver 5 for about an hour under heavy load and the temperature will start creeping up as it dries out. You can use toothpaste in a pinch but expect to be reapplying it a lot if you're using the computer heavily.
+FubarMike this isn't techrex haha
Well the reason people did this back on haswell was because the chip would get so hot with such poor quality thermal paste that they could gain 10celcius less heat, many videos on youtube showing this in fact.
I delidded my 7700k and used liquid metal (konductonaut). Temps were hovering around 75-80c and dropped to 52c under 100% load using EK's fluid gaming loop. It was even worse using an H100i v2 (would consistently rise to 90+)
My 8700k on an AIO was able to hit 5.5Ghz with liquid metal after delid with the same temps as a much lower volt 5Ghz OC. If you like to overclock or and build a very small PC and need to pinch cooling performance I think delid is great now with a delid kit. I've bricked 1 of 12 cpu delids and that was prior to having a delid kit. I've also had a 7700k delid hit the same 5.5Ghz with temps being ok enough to validate on an AIO.
I've also noticed a large benefit to delid of older devils canyon cpus as the paste wasnt even hard it was baked.
Without Delidding i had my G3258 with 4.6GHz on 85C. But now after Delidding and making Liquid Metal Between Heatspreader and CPU AND between Heatspeader and a AIO 120mm Watercooling i just have 66C on 4.8Ghz
+OSTutorials
Amazing results for an 120mm AIO indeed. Which liquid metal did you use?
www.amazon.de/gp/product/B001PE5XAC/ref=as_li_qf_sp_asin_il_tl?ie=UTF8&camp=1638&creative=6742&creativeASIN=B001PE5XAC&linkCode=as2&tag=hardw03-21
+Rygar allways liquid pro! I got almost -30c difference when I delided my old 3570k @ 4.7 (custom cooler - d5 , monsta 240 , ek supremacy block) final temp was 50\55ºc prime95
yeah thats make big differents
Yeah Liquid Pro is awesome. I was using it on my old lapped i7 920. Is it worth to go for Ultra? It's supposedly a bit easier and safer to apply.
LINUS you're conclusion is very one sided! I have a i7 3770k (Ivy Bridge was the first to use paste, not Haswell) and an H80i water cooler and I've delidded using MX-4 and Liquid ultra paste. The MX-4 dropped temps 6c at 4.7GHz while the Liquid metal dropped temps by 23c!!! This made a HUGE difference and is VERY much worth it! Liquid metal is SIGNIFICANTLY better with high voltage and better cooling, stock cooler bottlenecks the entire test to boot.
+Marshal Beard I got a massive 20C lower temperatures on my i5-4670K with 0,1V higher voltage. Made my 4,5GHz possible on a Hyper 212+ as it needed 1,375V.
+Marshal Beard your*
Done it using the very helpful "Delid Die Mate" with my 3770k and worked out perfectly in less than 5 minutes with minimal risk of damaging something (a child could do this). 10 - 20 C general improvement under load -> same temps with +1 GHz OC compared to stock before delidding.
Would love to see you testing it again and porperly using a decent air cooler and liquid metal with 7700k (OC)
Just did it on my old 4790k using Kryonaut. I am seeing 20C improvement on a CM 212 hyper, and now it can finally stay away from 100C.
With the auto OC profile on the motherboard I got 20% higher Cinebench scores
2 years of watching stuff on this channel and that G502 on the desk is the first time i have ever seen something i own lol
"I do not recommend you do this at home... so here is how you do it!"
He has to say that in case someone fucks up their CPU so they can't sue him.
Not worth the risk, only a couple of degrees to lose, and voiding the warranty are all the reasons people use to say against Overclocking. I don't think there's risk at all, the cpu under the hood is well protected. I think it's going to be a new trend.
I'd rather have a reliable older computer than a golden pinned paper weight.
Dunno about the couple of degrees.
Here is some prime95 outcomes when you do it correctly with real stuff and not with "something you have laying around"
www.norpparuutu.net/artikkelit/skylake_6700k_korkkaus/prime95_lammot.PNG
sami heiskanen oh cool thanks. Looks like a 2 degrees difference?
***** more like 20-28 degrees celcius. Sorry i forgot to mention that all of those measurments are in celcius :)
sami heiskanen Wow that's awesome. Looks like there's hope yet.
what is the best thermal paste to use now? Is it still arctic silver? I've got some of that mx stuff that needs no cure time as well as some arctic silver but i like seeing my temps improving afterwards like arctic silver does. Unless there's something similar but better. I've seen gold and diamond pastes that must be shit because i never hear of them being recommended. I tried the nano stuff before as well.
wow keren abis om, nice LTT