Intel Rumors - Kaby Lake-X - Skylake-X and Cascade Lake

FrgMstr

Just Plain Mean
Staff member
Joined
May 18, 1997
Messages
55,532
Intel Rumors - Kaby Lake-X - Skylake-X and Cascade Lake

In this episode of How the Rumor Mill Churns, we address some old Intel CPUs, some new Intel CPUs, and hopefully Intel CPUs that we will never see again. End of Life for good products is often disheartening, but when EOL pertains to something that should have never existed, it goes over a lot better.
 
So appears bottom line is that Intel got caught with their pants down and is hopping around trying to get them up while they address the problem. I assume the manager in charge with their strategy got a HUGE bonus.
 
So Cascade lake will now be Q219. Not suprising with all of the Ice Lake delays.
TR4 owners get a TR+ option this year. X299 owners get... solder this year.

"Kaby Lake-X fell well short of sales expectations" - Huge surprise there! :rolleyes:
 
Solder eh? It's definitely not worth it for current i9 owners to switch over. I will be delidding mine and possibly will skip Cascade Lake X as well unless it really brings some tangible performance increase.
 
So WTH happened to all the 10nm parts? *crickets*

With Ryzen breathing down their back with 4.4 GHz OC chips, and better interconnect timing, Intel better start sweating.
 
At least the Skylake-X refreshes will be using solder under the IHS. Possibly Cascade Lake-X too. I agree that the K-SKU CPUs on the mainstream platform should also use solder, but eh. The delid/relid tools we have available today work great (at least all the ones I saw [H] cover).
 
Glad I didn't jump in on Skylake-X like I had been planning for years :p. A decrease in price would also be nice, but I don't expect that to happen. DDR4 prices have been starting to trend downward, so I may have another look when the refresh is released.
 
Last edited:
Chance of 5ghz with that many cores... Slightly above zero.

Unless they are doing it Xeon Platinum style and they'll be $10k

It'd be a real boon for some workstation uses cases (one in particular for me is driving me crazy because it supports avx but doesn't thread well) but I suspect it'll just be bragging rights and you can't actually get your hands on.

And that price wouldn't necessarily be bad btw. Baking a pointcloud at ultra quality in one of our apps can take literally days. *Mumbles about shit apps that don't run on Linux, thread properly and aren't supported in the cloud*
 
I have a Kaby Lake -X and it will do 5 ghz without breaking a sweat. Let's see how long x299 lasts. Maybe Intel will throw us a bone.
 
Wonder what the odds are of Skylake X running on a z170 skylake mobo.... probably less than zero...
 
Knee-jerk, and priced too high relative to Ryzen as well. Intel got worried, but still couldn't resist greed.

More like they were worried they would damage their high profit multi core parts. So they priced them and hobbled them enough that wouldn't happen. It sticks intel in a difficult position.

Any word on Spectre and Meltdown fixes?
 
300W!!!!!!!!!!!! everyone bashed AMD for putting out the 220W 5GHz FX-9590 ... omg 300W?! let the fun begin!!!
 
hell, it'll probably be the new $3000 gaming processor. they figure if nvidia can charge $3000 for their new gpu then they should be able to do the same for cpu. and you better make sure you got another $3000 leftover for cooling!!! :drowning:
 
So WTH happened to all the 10nm parts? *crickets*
With Ryzen breathing down their back with 4.4 GHz OC chips, and better interconnect timing, Intel better start sweating.

Where are you seeing these 4.4Ghz results? There's a user on the AMD discord who was already struggling to break 4.3GHz just on Cinebench (last thing I'd be using for stability testing). I think he settled on 4.25 for something reasonably 3D stable, but I've definitely lowered my expectations for the refresh :p
Looking at the OCN thread, I guess people are fine with predicting 4350Mhz as top bin for SiliconLottery, but the bell curve is likely to put average chips closer to 4.2Ghz or so.
 
4.35 would be fine imo. 4.4 is great but not needed for these chips to be attractive, they already are imo.
 
Where are you seeing these 4.4Ghz results? There's a user on the AMD discord who was already struggling to break 4.3GHz just on Cinebench (last thing I'd be using for stability testing). I think he settled on 4.25 for something reasonably 3D stable, but I've definitely lowered my expectations for the refresh :p
Looking at the OCN thread, I guess people are fine with predicting 4350Mhz as top bin for SiliconLottery, but the bell curve is likely to put average chips closer to 4.2Ghz or so.

If you read the specs, AMD even touts 4.3 as the boost clock. His board might need tweaking with a ROM update or more. But I have my sources which I am not at liberty to speak about. I also saw similar rumblings floating about on various threads.
 
https://www.gamersnexus.net/hwrevie...x-review-game-streaming-cpu-benchmarks-memory
Looks like GN's findings were quite similar to the Discord user's experience after all.
Granted it's not bad to see 4Ghz being a legit low power-use option now.

https://www.computerbase.de/2018-04...ancerating-frametimes-ryzen-bei-gleichem-takt
Improvements in frametimes are there too, not bad.

But yeah, headroom was my only real complaint and looks like I'll still be waiting just a bit longer.

Anand hit 4.35GHz
 
Going through the rest of the reviews, this is what I observed:

TR - 4.275Ghz, but with the disclaimer that it took more voltage than they'd be comfortable for 24/7 use.

PCPer - 4.2Ghz overclock on *all cores*, despite the 4.3Ghz turbo frequency for single core only.

TT - 4.25Ghz on 2700X, 4.275Ghz on 2600X.

Tom's - 4.2Ghz all-core frequency at 1.3785vcore. They did not want to exceed AMD's recommended 1.40v while trying to hit 4.3Ghz.

TPU - 4.2Ghz all core at 1.4V.

HH - 4.25Ghz all core at 1.4v

Techspot - 4.1Ghz for 2600X, 4.2Ghz on 2700X, noted with difficulty, requiring 1.4v to pass 1hr Blender stress test. I must say the 4.1Ghz for the 2600X is particularly ominous, but hey, maybe just bad luck.

Computerbase - 4.3Ghz for 2700X at 1.4v, 4Ghz at 1.28v. 4.4Ghz single core at 1.475v. 4.2Ghz for 2600x at 1.4v. 4.1Ghz for 2700 at 1.35v. 4.2Ghz for 2600 at 1.4v.

TechGage - 4.12Ghz for 2700X at 1.308v.

4.35Ghz is certainly not the norm.
 
Going through the rest of the reviews, this is what I observed:

TR - 4.275Ghz, but with the disclaimer that it took more voltage than they'd be comfortable for 24/7 use.

PCPer - 4.2Ghz overclock on *all cores*, despite the 4.3Ghz turbo frequency for single core only.

TT - 4.25Ghz on 2700X, 4.275Ghz on 2600X.

Tom's - 4.2Ghz all-core frequency at 1.3785vcore. They did not want to exceed AMD's recommended 1.40v while trying to hit 4.3Ghz.

TPU - 4.2Ghz all core at 1.4V.

HH - 4.25Ghz all core at 1.4v

Techspot - 4.1Ghz for 2600X, 4.2Ghz on 2700X, noted with difficulty, requiring 1.4v to pass 1hr Blender stress test. I must say the 4.1Ghz for the 2600X is particularly ominous, but hey, maybe just bad luck.

Computerbase - 4.3Ghz for 2700X at 1.4v, 4Ghz at 1.28v. 4.4Ghz single core at 1.475v. 4.2Ghz for 2600x at 1.4v. 4.1Ghz for 2700 at 1.35v. 4.2Ghz for 2600 at 1.4v.

TechGage - 4.12Ghz for 2700X at 1.308v.

4.35Ghz is certainly not the norm.

I'm not disagreeing with you. 4.2 seems the reasonable average for what's out there now.

But remember it's early. Reviewers had at most a couple days to play and they had to do stock benchmarks first. There might be another 100 MHz to be had. But I have seen screen shots of it running 4.4 all cores with Prime 95. I would post it, but then my source might get into trouble for posting early benchmarks. Might need some MB firmware tweeking as well to deal with current surge. (Load line calibration issues)
 
It would be nice to finally get a solid 15-20% architectural boost.
I know many people play games that are ultimately GPU bound, but all my games are CPU bound.
Paradox Interactive games (EU4, CK2, Stellaris, HOI4) CIV6, SWTOR and turn based strategy games in general.
 
About time Intel replaced the booger paste with some real solder. But I have to wonder if it's just another plot to delay real architectural enhancements. Intel needs to sustain some improvements year over year, and so this is the low hanging fruit they can pluck off the MHz tree. They've been coasting since Sandy Bridge, pretty much (which was a fucking amazing advancement but still - 7 years ago!).

OTOH if the rumored 8 core mainstream part is really on the way and all it's cracked up to be, THAT would be nice. 6 core Coffee Lake was good. And 8 core version would be killer. AMD would have some serious work to do.
 
Other sites are reporting that Intel made if official today them killing off Kaby Lake-X. May 7th EOL.
 
Sad to hear this. I was hoping my Kaby Lake -X would last longer. It's a decent overclocker, too.
 
These "lake terms" is truly confusing lol. I wonder how they came up with these names
 
These "lake terms" is truly confusing lol. I wonder how they came up with these names
I think they reference actual lakes. No idea what the basis of the decision was.

They had a run of -field names, too.

Smithfield, Kentsfield, Yorkfield, Lynnfield, Clarksfield, Bloomfield...

There were also a whole bunch of -dale names, but I fail to remember them all.
 
I think they reference actual lakes. No idea what the basis of the decision was.

They had a run of -field names, too.

Smithfield, Kentsfield, Yorkfield, Lynnfield, Clarksfield, Bloomfield...

There were also a whole bunch of -dale names, but I fail to remember them all.

CPU code names appear to be cosmic ass pulls. Some dude was looking at a lake one day and wham, we get a bunch of lakes. Or maybe they were looking at a bulldozer, or a horse. Whatever. If it wasn't likely to result in lawsuits, we'd probably have ones named after celebrities by now.
 
  • Like
Reactions: TMCM
like this
Back
Top