Monday, October 14th 2019

Intel Scraps 10nm for Desktop, Brazen it Out with 14nm Skylake Till 2022?

In a shocking piece of news, Intel has reportedly scrapped plans to launch its 10 nm "Ice Lake" microarchitecture on the client desktop platform. The company will confine its 10 nm microarchitectures, "Ice Lake" and "Tiger Lake" to only the mobile platform, while the desktop platform will see derivatives of "Skylake" hold Intel's fort under the year 2022! Intel gambles that with HyperThreading enabled across the board and increased clock-speeds, it can restore competitiveness with AMD's 7 nm "Zen 2" Ryzen processors with its "Comet Lake" silicon that offers core-counts of up to 10.

"Comet Lake" will be succeeded in 2021 by the 14 nm "Rocket Lake" silicon, which somehow combines a Gen12 iGPU with "Skylake" derived CPU cores, and possibly increased core-counts and clock speeds over "Comet Lake." It's only 2022 that Intel will ship out a truly new microarchitecture on the desktop platform, with "Meteor Lake." This chip will be built on Intel's swanky 7 nm EUV silicon fabrication node, and possibly integrate CPU cores more advanced than even "Willow Cove," possibly "Golden Cove."
The HardwareLuxx article making these explosive revelations attributes the sudden change in Intel's plans to the company not being able to scale clock-speeds of "Ice Lake" high enough to establish product leadership. It feels "Skylake," which has IPC parity with "Zen 2," has enough scalability and clock-speed headroom to stay competitive with AMD at high clock-speeds. The company will augment next-generation uncore (revamped memory controllers, support for PCIe gen 4.0, Gen12 iGPU, etc.), with "Skylake" CPU cores, over time. Other areas where Intel could grow its mainstream desktop silicon is cache rebalancing similar to its HEDT chips, and implementing the Mesh Interconnect to maintain low latencies as core-counts enter two-figures.

Interestingly, 10 nm "Ice Lake" remains on Intel's enterprise roadmap, where the company appears more desperate not to cede market-share to AMD, especially as businesses around the world set their 5G plans rolling, springing a cycle of hardware updates in the data-center. 2020 could see the introduction of Xeon Scalable processors based on 10 nm "Ice Lake" microarchitecture with "Sunny Cove" CPU cores. In 2021, the company will introduce the "Sapphire Rapids" Xeon processor with even more cores and larger I/O connectivity, spearheaded with PCI-Express gen 5.0.

Update Oct 15th: Intel has released a statement, denying these claims, read more here. Source: HardwareLuxx.de
Add your own comment

148 Comments on Intel Scraps 10nm for Desktop, Brazen it Out with 14nm Skylake Till 2022?

#76
Agentbb007
Great now it will be easy to convince myself to keep my 9900k until 2022, saving money yay!
Posted on Reply
#77
Darmok N Jalad
I kinda wondered what the desktop was going to look like when their mobile 10nm lineup had a boost clock deficit to 14nm. Sure, it had an IPC uplift, but they just gave it back on frequency. I don’t think 10nm is clocking well at all, and mobile is the only place they can even try to sell it, and they are doing that with a stronger GPU. The GPU might have even been an afterthought considering how long the 10nm delay is.
Posted on Reply
#78
theoneandonlymrk
Darmok N Jalad
I kinda wondered what the desktop was going to look like when their mobile 10nm lineup had a boost clock deficit to 14nm. Sure, it had an IPC uplift, but they just gave it back on frequency. I don’t think 10nm is clocking well at all, and mobile is the only place they can even try to sell it, and they are doing that with a stronger GPU. The GPU might have even been an afterthought considering how long the 10nm delay is.
10nm is coming to servers with icelake ,28 core parts initially in 2020, but obviously they're high margin and lower frequency anyway ,just to keep Tdp in check with god knows what to follow , probably not more than two generation's then 7nm?.

Though at this point I am not confident this isn't nonsense since so many of intels timelines have been fantastical products of pr.
Posted on Reply
#79
tony359
Tomorrow
Ryzen 4000 on a B350 board? Maybe. A320 - problably not.
As a 320 owner I enjoyed all the people staying that the 320 wouldn't be 3000 compatible because <put a random reason here not backed up by any technical background>. While everything is possible, I think I'll sit back, relax and enjoy! ;)
Posted on Reply
#80
Vya Domus
theoneandonlymrk
10nm is coming to servers with icelake ,28 core parts initially in 2020, but obviously they're high margin and lower frequency anyway ,just to keep Tdp in check with god knows what to follow , probably not more than two generation's then 7nm?.

Though at this point I am not confident this isn't nonsense since so many of intels timelines have been fantastical products of pr.
You know what no one takes into consideration ? Yields. I have a feeling they're still horrible hence these mediocre mobile 10nm mobile parts being the only real thing out there at this moment. More importantly what that would imply is volume production of huge monolithic dies for their Xeons could still be a far dream at this point. They're certainly no near what TSMC has right know, you can be sure of that. I doubt AMD can make something huge in volume on their 7nm node right now as well.

Unless they have something akin to AMD's chiplets they are going to need a hell of an asspull to get anything remotely competitive. The more you look into it the bleaker it gets.
Posted on Reply
#81
Super XP
Vya Domus
You know what no one takes into consideration ? Yields. I have a feeling they're still horrible hence these mediocre mobile 10nm mobile parts being the only real thing out there at this moment. More importantly what that would imply is volume production of huge monolithic dies for their Xeons could still be a far dream at this point. They're certainly no near what TSMC has right know, you can be sure of that. I doubt AMD can make something huge in volume on their 7nm node right now as well.

Unless they have something akin to AMD's chiplets they are going to need a hell of an asspull to get anything remotely competitive. The more you look into it the bleaker it gets.
Hopefully Jim Keller doesn't share trade secrets with Intel, seeing how he was the lead in the of the ZEN micro-architecture design. Anyhow Intel will have a great struggle to keep pace with AMDs current processors and the upcoming ZEN3, ZEN4 and ZEN5 designs.
Posted on Reply
#82
ppn
A wise decision. Next big thing. 10nm is already obsolete, not being released 2016 when it should. 2019 skip 7nm, so in 2022 is time for 5nm. Which is precisely what 7nmEUV really is. Because it is 5.5 times the density of 14 nm.
Posted on Reply
#83
ShrimpBrime
AMD lacks frequency IMO. I miss the 5ghz club.....
Posted on Reply
#84
newtekie1
Semi-Retired Folder
"Skylake derived" ha, that's funny. These cores date back to Sandybridge. Intel's been relying on optimizations and node shrinks to gain performance since then.
Posted on Reply
#85
theoneandonlymrk
ppn
A wise decision. Next big thing. 10nm is already obsolete, not being released 2016 when it should. 2019 skip 7nm, so in 2022 is time for 5nm. Which is precisely what 7nmEUV really is. Because it is 5.5 times the density of 14 nm.
AMD could well be on 5,nm GAA FETs by then ,im seeing a two horse race minimum from here on in./10years minimum.
Posted on Reply
#86
danbert2000
I'm not surprised that they're just going to try to forget 10nm was even a thing. It was a failed node from the start, and they have thrown millions of dollars at it and it still can't support the clocks that they need to match Coffee Lake. I've said this before, but it's a lot like Haswell to Broadwell. They almost didn't release Broadwell desktop because the clocks were lower and the small IPC bump couldn't make up for it. But the customers were tetchy so they did release it, and it was a sidegrade at best. Eventually they got the node up to speed but at that time they could afford one dud generation, and that's from someone with the "dud" gen chip in my PC. Now with an actual competitor out there, they wouldn't dare put out a processor that they couldn't guarantee to be faster at everything. And without a future for 10 nm with 7 nm coming as the actually usable node, it doesn't even make sense to put effort into 10 nm at this point. They'll sell mobile chips on it as long as they are competitive with AMD, which should be a couple years until AMD finally figures out the idle power and voltage issues and irons out all of the mobility kinks, and they they'll write off the rest of the capital expense loss as the biggest failure in modern business history.

I would be very wary to be an Intel shareholder at this point. This 10 nm boondoggle was a decade-long conspiracy of lies and "technically" meeting public commitments. I don't know it there's any possibility of 7 nm news being taken seriously from Intel until they actually show us the goods. Intel squandered a nearly five year fab advantage and I doubt they'll ever have that chance again. No wonder they snapped up Jim Keller, at this point even their IP is out of date. All that's left is inertia and cash. Let's just hope Intel avoids bankruptcy so AMD doesn't turn into the monster they're in the process of slaying.
Posted on Reply
#89
TheGuruStud
Mephis
Intel just denied this rumor to Tom's Hardware, so looks like it wasn't true.

https://www-tomshardware-com.cdn.ampproject.org/v/s/www.tomshardware.com/amp/news/intel-yes-there-will-be-10nm-desktop-cpus?amp_js_v=a2&amp_gsa=1&usqp=mq331AQCKAE%3D#referrer=https%3A%2F%2Fwww.google.com&amp_tf=From%20%251%24s&ampshare=https%3A%2F%2Fwww.tomshardware.com%2Fnews%2Fintel-yes-there-will-be-10nm-desktop-cpus
They also denied 10nm was dead. A vegetable on life support is functionally dead lol
Posted on Reply
#90
Prima.Vera
How about DDR5 or PCI-EX 4.0 ? Any news about that? Currently I see less and less reasons to go for an uber expensive Intel platform...
Posted on Reply
#91
ShurikN
Prima.Vera
How about DDR5 or PCI-EX 4.0 ? Any news about that. Currently I see less and less reasons to go for an uber expensive Intel platform...
Both AMD and Intel plan to intorduce DDR5 in 2021. Servers first of course.
Posted on Reply
#92
Mephis
TheGuruStud
They also denied 10nm was dead. A vegetable on life support is functionally dead lol
Let's see, 10nm is shipping now, there are more designs and chips on the roadmap for laptops, desktops and server, you are right it is dead.

I know you are a fanboy, but even through your red colored glasses you should be able to see that 10nm isn't "dead."
Posted on Reply
#94
UltraThicc
14nm+++++++++++++++++++++++++++++++ when???
Posted on Reply
#95
R0H1T
Mephis
Let's see, 10nm is shipping now, there are more designs and chips on the roadmap for laptops, desktops and server, you are right it is dead.

I know you are a fanboy, but even through your red colored glasses you should be able to see that 10nm isn't "dead."
You do know that ICL on 10nm is pretty much a CC of Broadwell launch, talking about CL ~ well I wouldn't even know where to begin :ohwell:

Yes yes let's not go there, I can show you the ever changing roadmaps of Intel with 10nm projected release date about 3 (4?) years back? Anyone thinking server chips on 10nm will just fly off the fabs need to have a good hard look at what Intel has achieved on 10nm thus far, real hard look & that's with a less aggressive 10nm node than what was originally planned :shadedshu:
Posted on Reply
#96
notb
Super XP
AMD needs to capitalize on this news as much as possible.
I expect triple digit % gains in CPU market share.
You know this is physically not possible, right? :)
Posted on Reply
#97
JalleR
This Starts to smell like P4, with good performance though :D
Posted on Reply
#98
Vayra86
theoneandonlymrk
10nm is coming to servers with icelake ,28 core parts initially in 2020, but obviously they're high margin and lower frequency anyway ,just to keep Tdp in check with god knows what to follow , probably not more than two generation's then 7nm?.

Though at this point I am not confident this isn't nonsense since so many of intels timelines have been fantastical products of pr.
Where is the product? :)

A page or so back we had lots of posts about AM4 and Zen 3 and how it would or wouldn't happen, but we know for sure it'll happen on A socket. With Intel right now, we have not even a remote clue what an Ice Lake enterprise part even looks like. All we really know is that they have something that's not entirely shit for a laptop. Well yay, they can make a quad core on 10nm... :p

And here's the funniest part. Look at the TDP increase they need to not have an utterly abysmal (Atom comes to mind) baseclock and include turbo beyond 4 Ghz.



In other words, this scales like shit. A many core enterprise part will easily surpass 200W and for what? 2.3 base? teehee

But let's try to find some products then.... :) @notb you will love this, too

___
When can I buy Intel's 10th Gen Core CPUs?

Intel hasn't provided a timeline for when its 10th Gen Core chips will be available to buy on their own yet. However, we expect the first batch of laptops to be available with these new processors soon.

Intel says it expects around 35 laptop designs from various manufacturers to debut throughout the rest of 2019. We expect to see an influx of new and refreshed laptops sporting these chips in the holiday season.

___


Aha! OK. China. But that 2.3 Ghz part... hmmm

Anything else? I clicked DuckDuckGo 'more results' about 10 times, this XPS product is all I could find...

https://www.dell.com/en-us/shop/dell-laptops/new-xps-13-2-in-1-laptop/spd/xps-13-7390-2-in-1-laptop


Painful, I say. The best part they can make apparently is the 1.3 Ghz one. We're close to 2020 already. Only 34 laptop models to be released :D

notb
Mobile segment means laptops.
You mean laptop, singular not plural? :p

Mephis
Let's see, 10nm is shipping now, there are more designs and chips on the roadmap for laptops, desktops and server, you are right it is dead.

I know you are a fanboy, but even through your red colored glasses you should be able to see that 10nm isn't "dead."
Come again?

notb
You know this is physically not possible, right? :)
If AMD doubles they'll have 100% growth in share... right?
Posted on Reply
#99
1d10t
Hifihedgehog
Hardly. They may make a superior product for now (I have my eyes set on that tasty 3950X!), but AMD has a mountain of mind share and market share that you cannot believe they have to contend with so you won't see them slowing anytime soon. They are still very much David against Goliath and they only make a tenth of Intel's mega-revenue. We still see potentially hundreds of thousands of loyalists in IT departments across all kinds of industries (school districts, financial institutions, airlines, department stores, factories, shipping services, etc.) dotting the globe who are sticking to Intel because of tradition (It's in our blood and our rich history and it's who we are.), track record (It's battle tested in the field so why change brands now and risk more help desk calls?), and brand image and recognition (Who's AMD? Oh, are they the guys with the hot, slow processors?). You can also thank the clueless Dell, HP, and Lenovo representatives who insist on Intel when people ask for AMD for continue these trains of thought. We may get it, but most people are generally slow to change, absorbing and adapting to it. They have a ways to go--likely three to five years before they absorb a half of the market if they keep at their current clip--before they can safely let off the gas. Until then, they have to fight Intel's marketing and their own stigma.
This statement are right on the heels.For long time, Intel known to be good at building brand awareness, it's hard to get over with it.Their money is well spend toward marketing and "rebate". It's not a secret that OEM and channel partner are giving rebates and offering rewards if you engaging with particular product.
Time will tell, let's see if Fiat Multipla clouded with billions of dollar can run another year :D
Posted on Reply
#100
DeathtoGnomes
phanbuey
If they get a tiny IPC lift (7-8%) but can boost the clockspeeds then they will be on top by a decent margin. If they can lower latency even more then it will be tough for intel to keep up.
i'm more in favor of latency being a priority.
Posted on Reply
Add your own comment