Intel Will Finally Make its Big 10nm Push in 2019

Posted on December 12, 2018 by Paul Thurrott in Hardware with 19 Comments

Intel today announced that it will finally move its mainstream Core and Xeon processors to a 10nm manufacturing process. The new chipset family, dubbed Sunny Cove, will debut in 2019, the firm says.

Details of the plan are still vague, since Intel is disclosing the new roadmap as I write this at its annual Intel Architecture Day event. But given the company’s struggles in moving to a more efficient processor architecture, and the many delays it has experienced in trying to do, this is of course big news.

Sunny Cover won’t actually be Intel’s first 10nm chipset: Its Cannon Lake CPUs, currently limited to a Core i3 model, have been shipping in very low volumes all year. But Sunny Cove marks Intel’s big push to the more efficient chipset design, and it will apply across all of its mainstream chipsets.

The first Core-branded Sunny Cove chips are expected in late 2019, Intel says. These will be followed up by Willow Cove, with its redesigned cache, security features, and transistor optimization, in 2020. And then Intel plans to debut the Golden Cove chipset family in 2021, adding improved single-threaded and machine learning performance, plus 5G-optimized networking. The Xeon versions of Sunny Cove may not debut until 2020.

 

Join the discussion!

BECOME A THURROTT MEMBER:

Don't have a login but want to join the conversation? Become a Thurrott Premium or Basic User to participate

Register
Comments (19)

19 responses to “Intel Will Finally Make its Big 10nm Push in 2019”

  1. Avatar

    will

    So December 2019?


    We should have a running poll on when we expect Intel to ship 10nm :-)

  2. Avatar

    dontbe evil

    Intel announce their 10nm cpu for 2016 no, 2017 no, 2018 no, 2019... kind of, late 2019

    waiting for AMD zen 2 7nm, in spring

  3. Avatar

    wright_is

    They are looking at re-working 10nm completely, with some little.BIG combinations, like ARM, although that could cause problems, as the OS will need to be aware of the cores' capabilities, because the little cores won't have FP units, for example.

    Great for embedded or always on devices though.

    They were also at great pains to say that, while the competitors are all moving from 10nm to 7nm that using transistor size isn't a good method of comparing chips, now that they are still stuck on 14nm for at least another year (chips should start appearing in numbers in 2020).

    Given that the Samsung Exynos chips that power the Samsung Galaxy line of phones will move to 7nm for the Galaxy S10 and Hauwei's Kirin 980 is already on 7nm and the next generation of AMD processors are lined up to be 7nm parts built at the TMSC foundaries, Intel need to get their act together and provide a big leap forward and prove their point that the nm-race isn't important...

  4. Avatar

    John Craig

    Good news all round, but they're still miles behind. The road map indicates they'll still be improving on 10nm in 2020/21, which suggests they'll push 10nm as the Intel mainstream product offering for years beyond 2021. Qualcomm, on the other hand, will likely be in the land of 5nm and beyond by 2021.



  5. Avatar

    mattbg

    I get that it's a tentative measure of sophistication and progress, but why do you care if your processor is a 10nm part if it does what you need it to do?


    I'd consider myself a power user (though not an extreme one). I have an Intel CPU on my desktop and I don't want anything else that is on the market for this purpose, 10nm or not.


    It's not like there has been a lot of progress in this space that many people care about from any company - what is the incentive? Powerful desktop PCs are becoming a niche, though I admit it'd be nice to have more power with longer battery life on mobile (though, there again, I have no issues with my current Ultrabook).


    Anyway, if you believe the latest from Intel, 10nm when it eventually does arrive will be followed very shortly by 7nm as the latter uses a different, less problematic process and has been under development in parallel by a different team. I would not bet any money on this, but there you have it.

    • Avatar

      wright_is

      In reply to mattbg:

      Lower power consumption (= cheaper to run) or more power for the same price (= more performance for the same amount of electricity used). That is the main benefit.

      Electricity in the USA is relatively cheap, but in some other countries the price for KWh is twice as high as the USA average or more.

      For example I was listening to TWiT a while back and they were complaining about the high prices of electricity in California, I then checked what we were paying and wished we had such cheap electricity!

      • Avatar

        mattbg

        In reply to wright_is:

        I get that, and it's very important on mobile, but haven't most of the gains already been had? Idle power consumption has been low for a long time. Sleep/standby is the biggest benefit.


        It seems like we're arguing over benchmark-type statistics (i.e. mine's 5% faster but nobody notices) that don't make a noticeable difference to anyone's individual bottom line at this point.


        It probably makes a big difference at the aggregate level, and in data centers, but these don't seem to be the people making the most noise about 10nm vs 14nm.

  6. Avatar

    Brazbit

    After a decade and a half of development we did finally get a new Duke Nukem game so *shrug* anything is possible I guess.

  7. Avatar

    dcdevito

    Umm..isn't the competition already on 7nm?

    • Avatar

      wright_is

      In reply to dcdevito:

      Technically yes, practically no. The other problem is, Intel dove headlong down a dead-end street with their move to 10nm and had to scrap it - they make the i3 for the Chinese market, but that's as far as they got, the techniques they used weren't good enough for high production yields and more powerful chips.

      So they had to scrap it and start again. Interestingly there is a 7nm team also working on other methods to get them there sometime in the future.

    • Avatar

      evox81

      In reply to dcdevito:

      Yes and no. There is no standardized measurement companies have to use when defining the scale of the transistors on their chips, so they're not directly comparable. Transistor density is actually a more directly-comparable measurement. So, while TSMC and others have chips where the smallest components are technically 7nm, the actual density of transistors in those products are on-par with Intel's transistor density on their 10nm products.

  8. Avatar

    skane2600

    The important thing from a business perspective is how competitive a product is in the market it's designed for. The technical details only matter in that context. For PCs and Macs Intel has remained dominant despite not having 10nm chips. That may be only because of compatibility considerations but those considerations are important to the market.

    • Avatar

      Oreo

      In reply to skane2600:

      Compatibility is not nearly as important now as it was before. x86 and x86-64 have been the juggernauts not just because of Windows, but also because of the ubiquity of developer tools (starting from good compilers and optimized libraries that provide basic functionality). Given that there are many more ARM-based computers now than x86-based computers, this advantage is gone. Apple could switch its Macs to ARM-based CPUs without impacting end users too much. Also on the server side there are lots of projects now, and e. g. a small company known as Amazon has recently brought ARM-based systems on the market whose performance is at least on par with comparable Intel machines at a much lower price point. Seeing as the server market is Intels big bastion of profitability, Intel is under attack from all sides, and has to execute close to perfectly to survive as a company that we would recognize as Intel now.

  9. Avatar

    solomonrex

    The big news was how they were doing it - by mix and matching chips from different nodes. "10nm-based" I think is the key phrase.

    • Avatar

      locust infested orchard inc

      In reply to solomonrex:

      Rather than speculate and give the impression to those who read your comment that Intel's 10nm process is just a marketing spin, playing the numbers game, as is the case notably with Samsung and to a lesser degree TSMC, all Intel's process nodes thus far, are numerically the closest to the distance on an atomic scale.


      Your quoted words of "10nm-based" in reference to Intel's process is pure conjecture on your part, without substance and without merit, unless you can provide genuine evidence to support your seemingly anti-Intel stance.


      The trio of Keller, Koduri, and Murthy at today's Intel Architecture Day made no suggestion that, to quote you "mixing and matching chips from different nodes" were being sought – a fudge, if you will.


      Unless you can provide with something substantial to back-up your claims, may I ask you to refrain from posting, as Donald J. Trump would put it, "fake news".

  10. Avatar

    shameermulji

    "The first Core-branded Sunny Cove chips are expected in late 2019, "


    I know this is a Windows site, but it's probably a good bet that this the processor that will go into the next generation Mac Pro

  11. Avatar

    Elindalyne

    The 3d die stacking with the Foveros stuff seems like it will end up being a huge deal for ultramobile chips.

  12. Avatar

    melinau

    This is starting to sound like the Mhz wars of the 1980's!

    Moving to 7nm process is a bit of a red-herring, outside use-cases (such as portable & mobile) where efficiency & low-power-consumption is vital. But if Intel can reduce the size of my desktop PC and remove all those whirring fans and or water cooling gubbins, whilst maintaining the same or better performance, so-much the better. Reducing my carbon footprint and more importantly that of huge Corporations through more efficient PCs is also worth doing.


    The other opportunity provided by smaller process sizes is the ability to integrate improved features, such-as Machine-learning or AI, and to add hardware-based security, without increasing processor size and TDP.

    I'm currently using AMD CPUs because they are superior for my (non-gaming) workloads. AMD is currently giving Intel a run for their money, hopefully Intel's response on 7nm will be to improve competition and value-for-money

Leave a Reply