Jump to content

Delidding a Core i7 6700K

5 hours ago, Progressor said:

I agree with Linus that the risk of delidding is not worth it for the general user or/and if your cpu is a newer one like that one he tested. For people who want to get every bit of performance out of their cpu it's worth it. It is also worth it if you're using an older cpu like 3770/4770 because the thermal paste is is gone completely bad after that many years. A lot of people on forums are reporting 20C temperature drops after delidding a 3770 so definitely worth it. I hoped Linus would do a test without the IHS(idk if the socket and cooler would have allowed it), it would been really interesting to see the results. 

 

Now regarding the cpu he used. Man, I can't believe that Intel would pull a shit like this. How can Intel put thermal paste on a high end processor, one in fact that is supposed to be overclocked by people who want to get the best out of their cpu.. not only did they put thermal paste but they put a shitty one. What were they thinking!? It's fine to do this on a i3 or lower bracket i5 but on a i7? Seriously? $400 for a cpu and you couldn't bothered to solder the ihs or at least spend a few more cents and put a high end thermal paste? This is bullshit, fuck Intel for doing this. This is what happens when there is no competition from AMD.

I was actually thinking that Intel swapped over to solder with Skylake instead of sticking with thermal compound under the IHS. I really would like to know what prompted them to switch from solder to thermal compound, since I know they used solder with Pentium 4s -- though according to one source, Intel started using TIM with the Core 2 Duos, releasing some that were soldered and some that were not. AMD also had a mixed record according to that source, but the FX line is soldered, whereas the Phenoms and Athlons may not be, or whether it is may be dependent on model. I'm not sure about the APUs.

 

Another source provides some explanation as to why Intel may have changed away from using solder entirely. That source describes someone who delidded a Skylake to replace it with solder, with results similar to metallic TIMs.

 

And I'm with you on the reason to delid. Doing this with a brand new, or new-ish, processor is obviously not going to see significant improvement unless you replace it with a metalic TIM. Plus the central idea with delidding is to leave the lid off. EK sells a mounting kit for their water blocks specifically for delidded processors that sets the block just that little bit lower so the block makes good contact with the die without crushing it. Putting the lid back on kind of defeats the purpose of delidding it, but without "relidding" the CPU, the cooling solution may not make good contact, or you risk crushing the die in the attempt.

Wife's build: Amethyst - Ryzen 9 3900X, 32GB G.Skill Ripjaws V DDR4-3200, ASUS Prime X570-P, EVGA RTX 3080 FTW3 12GB, Corsair Obsidian 750D, Corsair RM1000 (yellow label)

My build: Mira - Ryzen 7 3700X, 32GB EVGA DDR4-3200, ASUS Prime X470-PRO, EVGA RTX 3070 XC3, beQuiet Dark Base 900, EVGA 1000 G6

Link to comment
Share on other sites

Link to post
Share on other sites

I've delidded over 100 6700Ks, and I've never seen anything less than a 10C drop across the cores (closer to 15C when pushing the voltage.)

 

If you're going to delid, a liquid metal like CLU is a must and you need to clean as much of Intel's sealant off as you can to narrow the gap between the die and the IHS.

Link to comment
Share on other sites

Link to post
Share on other sites

I like the fact that the group actually took the time to do this because i always wonder what "todays" CPUs look like with the IHS. The last i looked it was a pentium 4 lolz

If cons are the opposite of pros, what is the opposite of progress? Think about it.

Link to comment
Share on other sites

Link to post
Share on other sites

I'd be willing to bet the delid was only to satiate someone's curiosity on what the cpu looked like underneath.

My eyes see the past…

My camera lens sees the present…

Link to comment
Share on other sites

Link to post
Share on other sites

@LinusTech, since you bothered deliding that CPU, you should've gone full length and do some testing with Coollaboratory Liquid Pro

here's an example of someone replacing Intel's TIM with this stuff, drops in temp in the ranges of 10deg: http://muropaketti.com/artikkelit/intel-skylaken-lammonlevittajan-irrottaminen-ja-vakiotahnan-vaihto/3/

bench_prime95.pngbench_handbrake.png

 

 

Link to comment
Share on other sites

Link to post
Share on other sites

1 hour ago, Zodiark1593 said:

I'd be willing to bet the delid was only to satiate someone's curiosity on what the cpu looked like underneath.

Or to quote Jeff Foxworthy, every guy is thinking "I'd like a beer and to see something nekid", even if the only thing they'll see nekid is a delidded CPU...

Wife's build: Amethyst - Ryzen 9 3900X, 32GB G.Skill Ripjaws V DDR4-3200, ASUS Prime X570-P, EVGA RTX 3080 FTW3 12GB, Corsair Obsidian 750D, Corsair RM1000 (yellow label)

My build: Mira - Ryzen 7 3700X, 32GB EVGA DDR4-3200, ASUS Prime X470-PRO, EVGA RTX 3070 XC3, beQuiet Dark Base 900, EVGA 1000 G6

Link to comment
Share on other sites

Link to post
Share on other sites

I did this with my 4770k and got a 20 degree drop but i used liquid metal for the new TIM.

Link to comment
Share on other sites

Link to post
Share on other sites

Ok, so a while back Luke said he'd rather I criticize them instead of other LTT members. I'll take that to heart here.

 

This video is terrible. This is probably worse than that one EK review without any comparisons to other coolers. My Skylake thread is trashing this video. Didn't clean off paste properly? Didn't use CLU? Assumed delidding any chip will net similar gains regardless of generation? Problem due to paste only starting with Haswell? What? Don't touch the back of the chip with your skin if possible! What about the vice method? What about SI delid + reseal service? What about bare die? Which chips might benefit from delid, which would not? Were you able to find some information on the failure rates of delidding, if not by method, just in general?

 

How does a professional WITH A TEAM screw this up, allowing random people on forums to take his place? Is your entire team centered around maximizing video quality and editing and NOBODY is around to do any research?

 

One guy got his CPU actually replaced via Intel Tuning Plan after delid.

 

I'm going to blow a fuse. What are you people doing, what are you people doing, you guys are supposed to be the professionals, I'm supposed to be the random pleb that is learning all he needs to know and then some from tech Youtubers...

 

Yeah, I understand that from the Youtuber's point of view, I'm some guy popping up and exploding in the comments over details, but that's what makes information good. Information about delidding is dime a dozen, I can just Google it. It's the details. It needs finesse. Are you guys about quality or do you just want to crank out as many videos as you can in good video quality with dank humor so a million kids will watch you?

In Placebo We Trust - Resident Obnoxious Objective Fangirl (R.O.O.F) - Your Eyes Cannot Hear
Haswell Overclocking Guide | Skylake Overclocking GuideCan my amp power my headphones?

Link to comment
Share on other sites

Link to post
Share on other sites

You should rename the video to "How not to delid an intel cpu in one video."

Spoiler

Chernobyl

AMD FX8350 @ 5GHz | Asus Sabretooth 990FX R2 | 16GB HyperX Savage @1950mhz CL9 | 120GB Kingston SSDNow

EK AMD LTX CSQ | XSPC D5 Dual Bay | Alphacool NexXxoS XT45 240mm & Coolgate Triple HD360

 

Spoiler

Kraken

Intel i5 4670K Bare Die 4.9GHz | ASUS Maximus VII Ranger Z97 | 16GB HyperX Savage 2400MHz | Samsung EVO 250GB

EK Supremecy EVO & EK-MOSFET M7G  | Dual 360mm Rads | Primochill CTR Phase II w/D5 | MSI GTX970 1670MHz/8000MHz

 

Graphic Design Student & Overall Nerd

 

Link to comment
Share on other sites

Link to post
Share on other sites

Really really shoulda used CLU imo.

 

Good vid (from a media stand point), but let down in terms of how informative it was with that more regular TIM use. If your gunna do something less mainstream like de-lidding, you should use the appropriate TIM (CLU). Also as others have pointed out, with a proper OC, proper cooling, proper TIM use and such, u can get way more than 3c difference. usualy 10c+.

CPU: Intel i7 3930k w/OC & EK Supremacy EVO Block | Motherboard: Asus P9x79 Pro  | RAM: G.Skill 4x4 1866 CL9 | PSU: Seasonic Platinum 1000w Corsair RM 750w Gold (2021)|

VDU: Panasonic 42" Plasma | GPU: Gigabyte 1080ti Gaming OC & Barrow Block (RIP)...GTX 980ti | Sound: Asus Xonar D2X - Z5500 -FiiO X3K DAP/DAC - ATH-M50S | Case: Phantek Enthoo Primo White |

Storage: Samsung 850 Pro 1TB SSD + WD Blue 1TB SSD | Cooling: XSPC D5 Photon 270 Res & Pump | 2x XSPC AX240 White Rads | NexXxos Monsta 80x240 Rad P/P | NF-A12x25 fans |

Link to comment
Share on other sites

Link to post
Share on other sites

Another misinforming video. What are you doing @LinusTech

 

Using CLU only drops it by... "a couple degrees"?

 

You're using 5-6 w/mk material. Liquid Ultra is 38.4 w/mk and soldering kits are 80 w/mk. It's not "a couple of degrees" better. I got my 4670K @ 1,386V running 55 degrees running FFT's prime95 27.9. Where it was 90-100 before the delid.

 

http://i.imgur.com/f6MvpyV.jpg (it says 1,35V, but CPU-Z back then didn't properly read the FIVR. only what i set in the BIOS). Cooler was an H105.

 

Using another TIM doesn't improve the effectiveness of the cooler, only the speed at which the heat is transferred. The stock heatsink is NOT a means of testing a potential bottleneck in thermal transferrence or thermal resistance because it's personal thermal resistance is pretty shit. You'd opt for an AIO or watercooling. And test with something that's not only marginally better than Intel's TIM.

 

Really, these video's and the workshop video's are pretty ill-informed and deceptive. And just seems to be more aimed to stop people from positing the questions online or directly. I'd be willing to bet if anyone had the spare time to replicate your tests, they'd arrive at much more conclusive evidence. Like Luke's airflow test. C'mon, noone believes that shit.

 

Just stop.

Link to comment
Share on other sites

Link to post
Share on other sites

I'm glad people have shared their opinion on this video, it's a complete shambles and should be redone but PROPERLY. 

LTT, you are NOT doing it right, no one in their right mind would go through the hassle of delidding their CPU, using a slightly better TIM, using a stock cooler and not even overclocking, seriously, what the fuck?

In spite of your poor testing methods, especially only using a stock cooler, I think 3C difference is really quite significant.

Incipere V5.0

Spoiler

CPU | i7-4790k | GPU | Nvidia GTX Titan X | Motherboard | MSI Z97S SLI Krait Edition | Memory | 2x8GB Kingston HyperX Fury DDR3 1866MHz | PSU | EVGA 650 G2 | Storage | Crucial BX200 240GB + Toshiba 3TB | Case | Cooler Master MasterCase Pro 5 | CPU Cooler | Noctua NH-D15

Parvulus V1.0

Spoiler

CPU | i5-4690k | GPU | Zotac GTX 960 | Motherboard | ASRock Z97M-ITX/ac | Memory | 2x4GB G.Skill Ripjaws X DDR3 1600MHz | PSU | EVGA 650 GS | Storage | Crucial BX200 240GB + WD 1TB Blue 2.5" | Case | Silverstone Sugo SG13

If you want to join a group chat of like-minded techies, gaming, and all things dank, join our Discord group. Message me or get into contact with Galaxy. http://linustechtips.com/main/user/107351-gaiaxy/

Link to comment
Share on other sites

Link to post
Share on other sites

On 3/4/2016 at 8:36 AM, Silicon Lottery said:

I've delidded over 100 6700Ks, and I've never seen anything less than a 10C drop across the cores (closer to 15C when pushing the voltage.)

 

If you're going to delid, a liquid metal like CLU is a must and you need to clean as much of Intel's sealant off as you can to narrow the gap between the die and the IHS.

Oh hey, SI. how r u m8?

 

I'd be willing to bet the delid was only to satiate someone's curiosity on what the cpu looked like underneath.

Then that person needs to try Google search.

In Placebo We Trust - Resident Obnoxious Objective Fangirl (R.O.O.F) - Your Eyes Cannot Hear
Haswell Overclocking Guide | Skylake Overclocking GuideCan my amp power my headphones?

Link to comment
Share on other sites

Link to post
Share on other sites

a better test was done by watercool.de click here to read (it its in german)

 

They used an INTEL I5-6600K @4700 Mhz (Vcore 1,408) on a  ASUS Z170M-PLUS cooled by their HEATKILLER® IV water-cooler.

 

Exchanging the TIM with Arctic MX-2 resulted in a drop of 2,35° C from 45.25°C to 42,9°C. After that they used GELID GC EXTREME which dropped the Temperature to 35,25°C.

Link to comment
Share on other sites

Link to post
Share on other sites

8 hours ago, Opferlamm113 said:

a better test was done by watercool.de click here to read (it its in german)

 

They used an INTEL I5-6600K @4700 Mhz (Vcore 1,408) on a  ASUS Z170M-PLUS cooled by their HEATKILLER® IV water-cooler.

 

Exchanging the TIM with Arctic MX-2 resulted in a drop of 2,35° C from 45.25°C to 42,9°C. After that they used GELID GC EXTREME which dropped the Temperature to 35,25°C.

Yep, just goes to show that the top end compounds make a massive difference, and the IC Diamond was used really just isn't up to the job.

 

Also proves my point that Thermal Grizzly Kryonaut is a good alternative to liquid metal, because Kryonaut is even better than the Gelid GC Extreme used in the article.

System specs
  • Graphics card: Asus GTX 980 Ti (Temp target: 60c, fan speed: slow as hell)
  • CPU: Intel 6700k @ 4.2Ghz
  • CPU Heatsink: ThermalRight Silver Arrow Extreme
  • Motherboard: Asus Maximus Viii Gene
  • Ram: 8GB of DDR4 @ 3000Mhz
  • Headphone source: O2 + Odac 
  • Mic input: Creative X-Fi Titanium HD
  • Case: Fractal Design Arc midi R2
  • Boot Drive: Samsung 840 Pro 128GB 
  • Storage: Seagate SSHD 2TB
  • PSU: Be quiet! Dark Power Pro 550w

Peripherals

  • Monitor: Asus ROG Swift PG278Q
  • Mouse: Razer DeathAdder Chroma (16.5 inch/360)
  • Mouse surface: Mionix Sargas 900
  • Tablet: Wacom Intuos Pen
  • Keyboard: Filco Majestouch Ninja, MX Brown, Ten Keyless 
  • Headphones: AKG K7xx
  • IEMs: BrainWavs S1
Link to comment
Share on other sites

Link to post
Share on other sites

Link to comment
Share on other sites

Link to post
Share on other sites

On 3/4/2016 at 6:22 PM, Dark_wizzie said:

Ok, so a while back Luke said he'd rather I criticize them instead of other LTT members. I'll take that to heart here.

 

This video is terrible. This is probably worse than that one EK review without any comparisons to other coolers. My Skylake thread is trashing this video. Didn't clean off paste properly? Didn't use CLU? Assumed delidding any chip will net similar gains regardless of generation? Problem due to paste only starting with Haswell? What? Don't touch the back of the chip with your skin if possible! What about the vice method? What about SI delid + reseal service? What about bare die? Which chips might benefit from delid, which would not? Were you able to find some information on the failure rates of delidding, if not by method, just in general?

 

How does a professional WITH A TEAM screw this up, allowing random people on forums to take his place? Is your entire team centered around maximizing video quality and editing and NOBODY is around to do any research?

 

One guy got his CPU actually replaced via Intel Tuning Plan after delid.

 

I'm going to blow a fuse. What are you people doing, what are you people doing, you guys are supposed to be the professionals, I'm supposed to be the random pleb that is learning all he needs to know and then some from tech Youtubers...

 

Yeah, I understand that from the Youtuber's point of view, I'm some guy popping up and exploding in the comments over details, but that's what makes information good. Information about delidding is dime a dozen, I can just Google it. It's the details. It needs finesse. Are you guys about quality or do you just want to crank out as many videos as you can in good video quality with dank humor so a million kids will watch you?

He also completely ignored the REAL reason people delid. When the difference in temps between Core 0 and Core 3 are over 10C apart from each other, it means there are more than likely air bubbles on the paste under the IHS. If he paid attention to his own core 0-3 temps, he would have noticed that he went from a 13C difference, down to a 6C difference (with one core still going higher than the others, possibly due to his less than great paste application). That is primarily why people delid. Yes, getting all 4 cores to a much closer temp range is helpful for overclocking, but it's not just done for overclocking. There are entire threads on OCN of people hitting 90C on Prime95 AVX2 load on stock speeds, and needing to delid. 

 

This video was a disaster for multiple reasons, but the whole "overclockers delid for only a few C difference" attitude really irks me.

My (incomplete) memory overclocking guide: 

 

Does memory speed impact gaming performance? Click here to find out!

On 1/2/2017 at 9:32 PM, MageTank said:

Sometimes, we all need a little inspiration.

 

 

 

Link to comment
Share on other sites

Link to post
Share on other sites

Why do these topics actually get made when @LinusTech never bothers to explain himself in these topics?

 

Essentially passive-agressively calling all delidders morons, but shy away from criticism. Does anyone in that place even have anything resembling an engineering degree? Because noone in there seems to understand how science/research is done.

 

Stick to unboxing/reviewing things Linus, please. You have 2,5million subs and as such bear a responsibility. Making false claims and disgingenuous statements only does more harm than good. Because you know there will be hordes of cretins now using this as a club to bash everyone delidding his CPU with. Because they lack the same skills to recognize poor research as you do creating them.\

 

- Lack of proper cooling (stock cooler)

- Lack of proper load testing (aida64 instead of prime95)

- Lack of voltage increase (self evident)

- Lack of proper TIM (run of the mill TIM instead of LCU)

- Improper target audience/scenario. (regular users instead of overclockers).

 

Just so much fail.

 

I mean, it's like testing the influence of CPU in gaming at 4K...oh wait.

Link to comment
Share on other sites

Link to post
Share on other sites

I already commented on Youtube and Twitter, but i think Linus deserves one more shout in here aswell. Terrible, terrible video as people have already said. Lets hope Linus fairly admits that this went completely wrong and makes better video of it.

Link to comment
Share on other sites

Link to post
Share on other sites

Did I even mention the fact hat lowering the T-delta also makes overclocking viable on cooling systems they're not originally designed for? Like the noctua NH-U9S HTPC cooler i'm using with a 4670K

 

Which is specified for low overclocking capability;

http://noctua.at/en/tdp-guide

 

delidding_isnt_pointless.thumb.jpg.09aa4

 

CPU running 1.24V on a cooler which is meant for running chips stock or at mild OC. It's running only 62% fanspeed and i'm not even breaking 70 degrees with Intel XTU.

 

Running AVX prime (27,9); Getting 1.312V on the core

 

delidding_isnt_pointless2.thumb.jpg.8194

 

Still not running the cooler flat-out. And not using dual-fans.

 

Yes, 1.312V running AVX prime at 70-ish degrees on a HTPC cooling unit. I could probably raise the clocks a bit on that voltage, but Haswell scales the most with voltage, so it's the most predominant factor when checking temps.

Link to comment
Share on other sites

Link to post
Share on other sites

3 hours ago, Majestic said:

Why do these topics actually get made when @LinusTech never bothers to explain himself in these topics?

 

Essentially passive-agressively calling all delidders morons, but shy away from criticism. Does anyone in that place even have anything resembling an engineering degree? Because noone in there seems to understand how science/research is done.

 

Stick to unboxing/reviewing things Linus, please. You have 2,5million subs and as such bear a responsibility. Making false claims and disgingenuous statements only does more harm than good. Because you know there will be hordes of cretins now using this as a club to bash everyone delidding his CPU with. Because they lack the same skills to recognize poor research as you do creating them.\

 

- Lack of proper cooling (stock cooler)

- Lack of proper load testing (aida64 instead of prime95)

- Lack of voltage increase (self evident)

- Lack of proper TIM (run of the mill TIM instead of LCU)

- Improper target audience/scenario. (regular users instead of overclockers).

 

Just so much fail.

 

I mean, it's like testing the influence of CPU in gaming at 4K...oh wait.

The "using videos against people" thing is a current problem of mine with one of his older videos about memory overclocking. I can certainly see this video having the same result. Just like the whole "Don't use Prime95, it will blow up your CPU!" debacle, this anti-delid crusade will end badly for the people that actually need to delid out of necessity. Let's face it. Some people roll the dice, and get terrible CPU's with air bubbles in the thermal compound under the IHS. Delids are required to fix this. Someone could say "Just RMA the CPU" but that option is not for everyone. Sometimes people buy the CPU from a third party, or buy an OEM/tray unit that lacks a warranty (like my 6600T) and basically have no other option.

 

With all of their recent mini-series (workshop, holy $***, etc) I hope we see a new mini series, where they go back and re-test videos with less than accurate conclusions. God knows I want to see that memory speed video redone, and this video needs to be redone properly too. While we are at it, maybe we can show people that Prime95 is only dangerous to people with absurd expectations? Come on LTT, i'm giving you a gold mine here. 

My (incomplete) memory overclocking guide: 

 

Does memory speed impact gaming performance? Click here to find out!

On 1/2/2017 at 9:32 PM, MageTank said:

Sometimes, we all need a little inspiration.

 

 

 

Link to comment
Share on other sites

Link to post
Share on other sites

2 minutes ago, MageTank said:

God knows I want to see that memory speed video redone, and this video needs to be redone properly too. While we are at it, maybe we can show people that Prime95 is only dangerous to people with absurd expectations? Come on LTT, i'm giving you a gold mine here. 

Especially if you've seen the digitalfoundry video on the same subject....

 

I'm not sure LTT really cares about objectivity or being accurate. They care about youtube metrics and pandering to their audience. It's why they never fucking respond to this type of criticism. I've yet to see Linus respond to any of my valid criticisms. Like testing CPU's at 4K, the airflow tests, or this.

Link to comment
Share on other sites

Link to post
Share on other sites

Just now, Majestic said:

Especially if you've seen the digitalfoundry video on the same subject....

 

I'm not sure LTT really cares about objectivity or being accurate. They care about youtube metrics and pandering to their audience. It's why they never fucking respond to this type of criticism. I've yet to see Linus respond to any of my valid criticisms. Like testing CPU's at 4K, the airflow tests, or this.

I've seen the video. I also saw the thread on OCN that came out years before the video, and even have done tests myself (in my signature) to see it with my own eyes. It's part of the reason why I have a problem with some of his videos (and to an extent, the people that treat them like gospel). His testing methodology is often flawed, and at times, even he does not know what it is he should be looking for.

 

For example: memory speeds impact on gaming performance is mostly dictated by the amount of CPU overhead, and is often most noticeable when looking at minimum FPS. In Linus' test, he not only failed to monitor minimum FPS, but he used a freaking GTX 660 Ti running games on Ultra 1080p with 4/8x AA, as if the card could even handle those games. Yet people still call me a mad man every single time i recommend people to invest a few extra dollars into faster ram, or when i try to teach people the basics of memory overclocking. That one video has put me so far back that its a never ending chore to get this through to people. If he would just redo the test, with a better test methodology, maybe then people would start paying attention. 

 

Sadly, when it's your word against a man with millions of followers, you are going to be seen as wrong anyways. Luckily, DigitalFoundry has begun to show people my point, and prove that old OCN thread was correct. Not only that, but I've even reached out to more tech channels, and they are running the tests too. So if LTT will not go back and retest, at the very least, we can provide alternative sources for our claims. 

My (incomplete) memory overclocking guide: 

 

Does memory speed impact gaming performance? Click here to find out!

On 1/2/2017 at 9:32 PM, MageTank said:

Sometimes, we all need a little inspiration.

 

 

 

Link to comment
Share on other sites

Link to post
Share on other sites

Yep, and show that AMD has driver overhead, and that CPU matters for gaming, etc. etc. Digital Foundry is a really awesome channel.

They're so objective, that most people who've grown accustomed to people like Linus feeding them pandering lies that they're calling them biased xD

 

It's so ironic and pathetic at the same time.

 

And yes, most people are eager to call out cassandra complex on those who oppose established truths. it's why I'm always labelled a dickhead on this forum.

Link to comment
Share on other sites

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now


×