Official OcUK Skylake-X & Kabylake-X Review thread

Simon Belmont · 19 Jun 2017 at 18:43

gavinh87 said:
You are right. I'm talking nonsense. I should get a 7900x and delid it before putting it under an extravagant water loop just to keep intel happy.
The fact that people are defending this is hilarious.

8 Pack is just doing damage control for OCUK. It is there forums so it's understandable

8 Pack · 19 Jun 2017 at 18:44

No need to defend this, it will sell well.

It performs great and OC is high that's my goal.

gavinh87 · 19 Jun 2017 at 18:46

Simon Belmont said:
8 Pack is just doing damage control for OCUK. It is there forums so it's understandable

Rubbing his hands together with the amount of x299 chips that will need delidding.

Simon Belmont · 19 Jun 2017 at 18:46

8 Pack said:
No need to defend this, it will sell well.

It performs great and OC is high that's my goal.

It will sell good for a few days then after that....It's AMD turn

stewski · 19 Jun 2017 at 18:46

Personally I found/find the Anandtech review disappointing.

Clearly in the HEDT market taking a $320 ryzen 1700 and getting 1800x performance is virtually a given and the "recommendations" at the end of the article seem cringingly bad/biased at this point in time.

Simon Belmont · 19 Jun 2017 at 18:48

gavinh87 said:
Rubbing his hands together with the amount of x299 chips that will need delidding.

I thought delidding was pointless? It would only see couple C drop, not worth it

gavinh87 · 19 Jun 2017 at 18:51

Simon Belmont said:
I thought delidding was pointless? It would only see couple C drop, not worth it

Your right, doesn't mean they won't appear on the store though. Delidded and binned chips for the x299 will be here.

Doobedoo · 19 Jun 2017 at 18:54

kitfit1 said:
I think you may be missing the point Doobedoo. If you read the posts i linked to above, the conclusion was that de-lidding just to put "better" Tim on is a complete waste of time because it makes little to no difference in temps. The reason it makes no difference is because of the gap between the core and the IHS.

Perhaps I don't know what I'm talking about which is why I'm asking questions rather than telling people I know what I'm talking about.

Why would Intel design and release a product like that, yet alone an enthusiast product selling for a lot of money? Why does a Kabylake respond well to a tim change but not these?

Seems like a strange decision to move away from solder especially on your high end products.

zornyan · 19 Jun 2017 at 19:02

Doobedoo said:
Perhaps I don't know what I'm talking about which is why I'm asking questions rather than telling people I know what I'm talking about.

Why would Intel design and release a product like that, yet alone an enthusiast product selling for a lot of money? Why does a Kabylake respond well to a tim change but not these?

Seems like a strange decision to move away from solder especially on your high end products.

kabylake had a gap issue, the TIM itself isn't bad, even der8auer said multiple times the actual quality is fine.

skylake for example was pretty good temp.wise yeah? as was devil's canyon etc.

the issue Intel have been trying to sort out (such as devil's canyon relaunch of haswell) is the adhesive they use to attach the heatspreader.

in kabylake this adhesive was too thick causing too much of a gap, hence the big temps drops delidding because that thick glue is removed.

take skylake x, 10c drop from delidding right?

der8auer got a 4 and 7c from from delidding 6950x and 5960x and replacing with liquid metal.

so if Intel had soldered these and got a 6-7c temp reduction, it wouldn't effect things greatly.

10 cores at 4.7ghz with avx512 is gonna be hard to cool, regardless.

Chuk_Chuk · 19 Jun 2017 at 19:02

Impressive performance, but it seems that intel has pushed them for all they have. Judging by how power consumption runs away like crazy.
Also what was happening on Toms hardware. Seems like they had more problems than everyone else.

these are engineering samples right? Aren't these supposed to be the best that intel has?

Zeed · 19 Jun 2017 at 19:05

8 Pack said:
Exactly the issue extreme care not to kill the Cpu needs to be exercised.

We delided around 50 cpu in one week at Asus hq from Kabylake X to skylake X all fine. Drops Temps 20c+ with conductonaut liquid metal.

Asked Gibbo but not heard back. Will OCUK sell Fixed cpus like You do with 7700k?? Asking for mate that is interested in top tier ones for hes workstation

zornyan · 19 Jun 2017 at 19:07

Lord_ZED said:
Asked Gibbo but not heard back. Will OCUK sell Fixed cpus like You do with 7700k?? Asking for mate that is interested in top tier ones for hes workstation

yeah interested in a 7820x if you guys have the delidded ones first, not even worried about binned items if they'll be too long

kitfit1 · 19 Jun 2017 at 19:09

Doobedoo said:
Perhaps I don't know what I'm talking about which is why I'm asking questions rather than telling people I know what I'm talking about.

Why would Intel design and release a product like that, yet alone an enthusiast product selling for a lot of money? Why does a Kabylake respond well to a tim change but not these?

Seems like a strange decision to move away from solder especially on your high end products.

Intel have to my knowledge never given a reason for moving from solder to Tim. A very good reason though might be that the solder they used melted between 80c and 90c. Over the years Intel cpu's have got hotter when clocking, not cooler. My guess is that that is the reason. The last Intel cpu i de-lidded was and i7 975 Gulftown. It was soldered and come off very easily once the mastic was cut and the cpu heated up.

Doobedoo · 19 Jun 2017 at 19:10

zornyan said:
kabylakr had a gap issue, the TIM itself isn't bad, even der8auer said multiple times the actual quality is fine.

skylake for example was pretty good temp.wise yeah? as was devil's canyon etc.

the issue Intel have been trying to sort out (such as devil's canyon relaunch of haswell) is the adhesive they use to attach the heatspreader.

in kabylake this adhesive was too thick causing too much of a gap, hence the big temps drops delidding because that thick glue is removed.

So it's a manufacturing fault then. No? Some may even go as far as saying they are broken.

In all seriousness though why would the glue which is only a silicone sealant or gasket in effect, cause the ihs to not sit down? If they were clamped down with enough fore, which wouldn't need to be a lot, the silicone would be forced outwards. Not saying your wrong it just doesn't make sense but would still validate my original comment, just de-lid remove faulty silicone, re-tim and re-lid.

Doobedoo · 19 Jun 2017 at 19:11

kitfit1 said:
Intel have to my knowledge never given a reason for moving from solder to Tim. A very good reason though might be that the solder they used melted between 80c and 90c. Over the years Intel cpu's have got hotter when clocking, not cooler. My guess is that that is the reason. The last Intel cpu i de-lidded was and i7 975 Gulftown. It was soldered and come off very easily once the mastic was cut and the cpu heated up.

Yeah that makes sense. Seems Intel have hit a wall then.

Chuk_Chuk · 19 Jun 2017 at 19:19

kitfit1 said:
Intel have to my knowledge never given a reason for moving from solder to Tim. A very good reason though might be that the solder they used melted between 80c and 90c. Over the years Intel cpu's have got hotter when clocking, not cooler. My guess is that that is the reason. The last Intel cpu i de-lidded was and i7 975 Gulftown. It was soldered and come off very easily once the mastic was cut and the cpu heated up.

The main material in the solder is Indium which has a melting point around 150c. Potenetially when mixed with the other materials its melting point will drop.

Another reason I heard for the use of TIM (apart from being cheapskates), is that soldering isn't 100% successful and some CPUs do die after being soldered. Due to there large die's, Intel can't afford to have the few cpus they have scavenged die off during soldering

kitfit1 · 19 Jun 2017 at 19:20

Doobedoo said:
Yeah that makes sense. Seems Intel have hit a wall then.

I think they may well be heading very rapidly towards a thermal wall, yes. Die shrinks and better manufacturing techniques should bring lower temps over a period of years, not an increase as seems to be the case with Intel.

gavinh87 · 19 Jun 2017 at 19:24

Chuk_Chuk said:
The main material in the solder is Indium which has a melting point around 150c. Potenetially when mixed with the other materials its melting point will drop.

Another reason I heard for the use of TIM (apart from being cheapskates), is that soldering isn't 100% successful and some CPUs do die after being soldered. Due to there large die's, Intel can't afford to have the few cpus they have scavenged die off during soldering

That sounds about right. These are pretty much Intel's best CPU they can make at the moment and the yeilds won't be very high. If soldering is killing them it's counter productive and I bet those wafers aren't quick to produce.
That said, delidded doesn't drop the temperature by that much according to that de8reabufbredjdjd guy.
It would appear they've pushed these so high that power consumption and temperature have took a big hit.

kitfit1 · 19 Jun 2017 at 19:25

Chuk_Chuk said:
The main material in the solder is Indium which has a melting point around 150c. Potenetially when mixed with the other materials its melting point will drop.

Must be mixed with some very odd materials then, because the Gulftown i mentioned came off using a hairdrier

KompuKare · 19 Jun 2017 at 19:37

stewski said:
Personally I found/find the Anandtech review disappointing.

Clearly in the HEDT market taking a $320 ryzen 1700 and getting 1800x performance is virtually a given and the "recommendations" at the end of the article seem cringingly bad/biased at this point in time.

Yes, barely read Anandtech reviews anymore. While their 'corporate-friendly' wording in reviews is not new it has gotten worse. And since they no longer do the in-depth dive-ins of architectures like they used to (possible because Intel no longer talk much about the internal workings), there's not much point reading their reviews. Their choices of benchmarks are a bit strange too. But they are not the only ones, despite Intel pushing HEDT to gamers with deep pockets, it's almost like Intel's review guide said not to bench games.

Even Tom's has done a better review which has happened a few times recently. That Computerbase and Hardware.fr didn't get any review samples. The European press tends to be less kind to Intel's corporate review policy while plenty of American press tends to lap up everything Intel says.

Hardware.fr:

Unfortunately, we are not able to offer you today a test of these processors. The manufacturer has preferred to favor the American press at the expense of the international press, a situation that tends to repeat itself for Intel for some time and that we can only deplore.

http://www.hardware.fr/news/15159/pre-commandes-skylake-x-kaby-lake-x-ouvertes.html

The embargo on tests for Intel's first Core-X processors based on Skylake-X and Kaby Lake-X has ended. But we do not have a test on ComputerBase. The reason: Intel has not been able to provide any samples so far. And an alternatives organized by the editorial team is stuck in customs.

https://www.computerbase.de/2017-06/intel-core-i9-7900x-nicht-im-test/