To anyone contemplating De-Lid

faceman123 · 20 Jun 2018 at 16:15

Hey Everyone, anyone who's on the fence about Deliding - Go for it.

I've dropped 23c ( same ambient ) whilst increasing my OC to 5.1ghz ( able to push to 5.2 but happy with 1.38v for 5.1 temps are now solid 68c-70c Full load drop from the 93c on 5ghz

shout out to Awahaw for lending me his delid kit

Thermal grizzly and Vital black RTV are the way to go.

end of rant.

bakes0310 · 20 Jun 2018 at 16:17

Sounds really worth while, huge difference.

BriT · 20 Jun 2018 at 16:38

Yeah my 8600k dropped 15c. Wasn't that hard to do as well just need to be careful when handling the CPU.

chaparral · 20 Jun 2018 at 17:31

I don't feel like it really worth it to de-lid my cpu :confused:

mine runs 5ghz at around 60c

faceman123 · 20 Jun 2018 at 17:48

chaparral said:
I don't feel like it really worth it to de-lid my cpu

mine runs 5ghz at around 60c

Fair enough was worth it to me, i was 94c before with h115i pro, i can keep 5.2ghz 1.39 at 78c

If you're at 5 with 60, what voltage and you should push that bad boy lol

Destination · 20 Jun 2018 at 21:28

How much cheaper is the manufacturing process for $400-1200 processors for Intel to use toothpaste when they could make them soldered to begin with?

slade · 21 Jun 2018 at 09:07

I was going to upgrade to a 8700K but a guy up the road from me did a de-lid on my 6700K for $30 (20quid).
It was at 4.2ghz at around 35/42c idle, now its at 4.5ghz at 28/32 idle with slightly less volts too. I wish i'd done it years ago.

I'll always get my future cpu's delidded!

Armageus · 21 Jun 2018 at 09:27

Destination said:
How much cheaper is the manufacturing process for $400-1200 processors for Intel to use toothpaste when they could make them soldered to begin with?

At least $2-$5 per CPU depending on size of the core. Although it's not just a case of cost, it's actually a fairly difficult process to get a copper heatspreader to "solder" to a silicon core.
http://overclocking.guide/the-truth-about-cpu-soldering/

EDIT:
The comments from der8auer also go on to explain why intel use the brand of TIM they do as well - largely that although the thermal performance isn't the best, there is no long term degradation (which is essential as you can't expect normal users to replace TIM under the IHS on a regular basis)

passifid · 21 Jun 2018 at 10:12

Armageus said:
At least $2-$5 per CPU depending on size of the core. Although it's not just a case of cost, it's actually a fairly difficult process to get a copper heatspreader to "solder" to a silicon core.
http://overclocking.guide/the-truth-about-cpu-soldering/

EDIT:
The comments from der8auer also go on to explain why intel use the TIM they do as well - largely that although the thermal performance isn't the best, there is no long term degradation (which is essential as you can't expect normal users to replace TIM under the IHS on a regular basis)

It makes sense in theory and there only. Sandy bridge chips are still running with decent thermals what 7years later?

Kelt · 21 Jun 2018 at 10:33

passifid said:
It makes sense in theory and there only. Sandy bridge chips are still running with decent thermals what 7years later?

I believe it's just Intel fudge dished out by paid Intel hirelings.

I'm running one of their Xeon soldered chips...probably would have been running 24/7 at stock in a server for years, then I'm running it over-volted, overclocked to 4.4Ghz every day for a couple of years absolutely fine.

So eight years on from its release, still running perfectly, and I've had zero degradation that is observable in the time I've had it.

passifid · 21 Jun 2018 at 10:59

Kelt said:
I believe it's just Intel fudge dished out by paid Intel hirelings.

I'm running one of their Xeon soldered chips...probably would have been running 24/7 at stock in a server for years, then I'm running it over-volted, overclocked to 4.4Ghz every day for a couple of years absolutely fine.

So eight years on from its release, still running perfectly, and I've had zero degradation that is observable in the time I've had it.

Even if there is degredation to a noticeable amount at the 8700k level it would have to get 10-20 degrees worse before it matches the vegetable fat spread that is used instead

benjii · 21 Jun 2018 at 11:33

Kelt said:
I believe it's just Intel fudge dished out by paid Intel hirelings.

I'm running one of their Xeon soldered chips...probably would have been running 24/7 at stock in a server for years, then I'm running it over-volted, overclocked to 4.4Ghz every day for a couple of years absolutely fine.

So eight years on from its release, still running perfectly, and I've had zero degradation that is observable in the time I've had it.

I know it's a small sample size, but I gotta agree with you there. I've got an i7 920 and a X5670 that are still pushing 4Ghz from launch, great chips.

Skorpion · 21 Jun 2018 at 12:45

Good job, and well done Faceman

Chuk_Chuk · 22 Jun 2018 at 12:54

Destination said:
How much cheaper is the manufacturing process for $400-1200 processors for Intel to use toothpaste when they could make them soldered to begin with?

FYI It's not just cost. There is a chance of killing the cpu during the soldering process.

Armageus · 22 Jun 2018 at 13:20

Kelt said:
I believe it's just Intel fudge dished out by paid Intel hirelings.

To some degree it probably is, but I believe there is some truth in it that as die sizes have got smaller, the risk of solder cracking, and the potentially greater risk of the CPU/IHS being distorted whilst soldering (due to smaller contact patch) is greater.

Ryzen is soldered, but is still a larger die (192mm2) than most recent Intel chips (Sandy 216mm2 -> Ivy 160mm2, Haswell 177mm2, Skylake 122.4mm2)

https://www.anandtech.com/show/9505/skylake-cpu-package-analysis

AiiR · 22 Jun 2018 at 18:11

Intel's HEDT line has a large die area, and that stil0l has toothpaste under the hood.

Still think it's because some Intel exec done the maths and saw dollar signs in savings.

LuckyBenski · 24 Jun 2018 at 00:54

One thing to note is that delidding's main benefit is correcting the imperfect contact between die and IHS. Some CPUs have nice even contact and some don't, resulting in uneven heat dissipation. Ever seen a CPU where two cores are about 10-15 degrees hotter than the other two? Probably uneven IHS contact. So delidding needn't be done on every single CPU, but should be an option once you've got your parts and tested them. There are people who bin chips based on this variation, rather than delidding.

I'm convinced that the TIM/IHS contact can be disturbed through stress too, I've seen chips that seem to have become awful which were then improved by a delid.

Ice Tea · 24 Jun 2018 at 09:24

Seems to be varying opinions online on how often to change Thermal grizzly ranging from 1year to never.

:rolleyes:

LuckyBenski · 24 Jun 2018 at 10:25

Ice Tea said:
Seems to be varying opinions online on how often to change Thermal grizzly ranging from 1year to never.

I've seen that the Coollaboratory product available before Conductonaut seemed to dry out, so I think people still believe that's a risk. The people I've seen using Conductonaut seem to believe it's fine for ages. Personally I glued my IHS back on and will treat it as a stock CPU, only taking it apart to re-paste if it stops performing.

keyser · 28 Jun 2018 at 13:30

LuckyBenski said:
Ever seen a CPU where two cores are about 10-15 degrees hotter than the other two? Probably uneven IHS contact.

That's more like cores next to eachother, like penguins trying to keep warm. Aw