Intel Mentions 10nm, Briefly

by Ian Cutress on 1/9/2018 11:14 AM EST
Comments Locked

18 Comments

Back to Article

  • Drumsticks - Tuesday, January 9, 2018 - link

    I wonder if we'll ever find out what happened to 10nm. It seems one of Tech's great mysteries along with Vega, at least in the modern world.
  • Drumsticks - Tuesday, January 9, 2018 - link

    Modern technology world that is, i.e. the last 5ish years or so.
  • HStewart - Tuesday, January 9, 2018 - link

    They realize the threat from AMD and QualComm and taking actions to make sure it not issue in the future.

    As for Intel / Vega thing - if you look at previous technical information from Intel on designs. This is just part of existing planned to make the mobile GPU faster connection with Mobile CPU. I believe it also part of planned to make simplified the motherboards designed with discrete GPU in mobile area. Apple was probably heavy behind this - I don't believe it is limited to just AMD GPU, likely Intel based discrete GPU and even possibly NVidia. I not sure we will see this technology in desktops - possibly but it requires major designed changes in desktop layouts.

    Of course like everything else on internet, this is just my opinion.
  • HStewart - Tuesday, January 9, 2018 - link

    I believe it been well documented that Intel 10nm is actually more dense than other 10nm - likely closer to rumor 7nm on the other side.

    Intel is also not stupid - they are completely aware that that AMD was working on Zen before it was released and also likely aware of Qualcomm emulations and are likely taking extra efforts to make sure they have strong hold in the industry once release.

    This is of course my speculation in what I would be doing if I was Intel - larger amount of cores and lower power is what I expected. I also believe that GPU's will be enhanced with experience from Ruju - probably more in aligned with what Apple really wants in the GPU component and not what they got from AMD.
  • lilmoe - Tuesday, January 9, 2018 - link

    Samsung, TSMC and GF 7nm will probably be more dense, and will probably ship around or before Intel's 10nm. Density is what matters more, not the nomenclature. Besides, unless Intel's foundry business starts fabbing in mass for others, it wouldn't really matter which is the "superior" process. What will matter is which CHIP/package offers better performance and/or efficiency for the intended task, and takes better advantage of is respective process.

    Time will tell. They have other security problems to worry about now, and those problems need fundamental changes to the base architecture. They all do, but it looks like Intel's problems go a bit deeper.
  • HStewart - Tuesday, January 9, 2018 - link

    Time will tell, right now we honest don't know about it - what is important is not the size in nm - but how many transistors are in a give chip. I like Intel designed primary because they know how to used the denser technology for where is it is important on CPU and GPU and allow cheaper technology on other areas. I think Intel was very smart when they purchase Altera and of course Raju coming about will have impact.

    I not to worried about security issues, software has bugs - even CPU's which I actually found one on IBM486SLC. I think the main problem with Intel is that they are top dog in PC Market and there are some others that don't like that and will go for the Underdog. I know that there are some especially gamers that want to believe AMD is Intel primary threat, but it actually not ARM is actually it - but that is different market.
  • lilmoe - Tuesday, January 9, 2018 - link

    Intel has had a monopoly for years and have been milking their customers...bad. That's why they're not everyone's favorite. They're charging an arm and a leg in terms of silicon area. They've been cashing in on the same architecture for years, and just started introducing more cores because of Zen. They've been really dependent on their process advances, doing very little in terms of innovation and creativity (as Lisa Su likes to put it, in which I agree).

    Any potential GPU advances is still years off. I wouldn't bet on any of that for the time being. By the time they do deliver, AMD and NVidia aren't going to stay still.

    Zen has proved to be great, and now, the future is all about cores and co-processing. The single threaded advantage Intel has is slowly diminishing, in terms of both hardware AND software.

    There's also Windows on ARM which has yet to show its full potential. I bet a dedicated 8 core (all A75 cores) platform with a competent GPU would give a quad core Intel U-series a good run for its money in performance, using less power while providing many more hardware features, and would make huge dent in Intel's high margins in the consumer business.

    I honestly wouldn't be as bullish as you are when it comes to Intel. I'd definitely start building shorts.
  • TEAMSWITCHER - Tuesday, January 9, 2018 - link

    What monopoly is that? What "milking" are you referring to? As I remember, Intel offered 6 and 8 core CPU's years before Zen. And nearly three years before Zen shipped, Intel had a 6 core CPU for less than $400 ... the Core i7-5820k. Granted, I'm mixing the mainstream and HEDT platforms, but full-featured mainstream and HEDT motherboards were never so different in price -
    certainly not enough to claim that Intel didn't offer mainstream consumers more performance.
  • Alistair - Tuesday, January 9, 2018 - link

    No that is milking. Intel offloaded the price to the platform and motherboard fees to make that CPU appear cheaper than it was.

    No reason they couldn't have released 6 core CPU on their main platform years ago, when the 4 core hit record low area sizes.
  • HStewart - Tuesday, January 9, 2018 - link

    To AMD HEDT platforms is where they can attack Intel, but to Intel HEDT is a minor market.

    But the AMD fans so called false vision of Intel monopoly is good for industry - because what AMD does not have is capital, Intel has that and can crush AMD in seconds. But for the industry, this keeps Intel on it toes. But Intel primary threat is not AMD but ARM and that is why that are focusing in lower power. But the problem with ARM is that it extremely slower that Intel cores - mostly equilivent to Intel's Atom cores - and emulation has not proving itself to give performance - especially when trying to emulate CISC architecture on RISC architecture.
  • HStewart - Tuesday, January 9, 2018 - link

    Intel has a monopoly if you living in a blind world - I would be more concern about company's like Qualcomm and 5G and ARM environments.

    And talking about Windows on ARM, keep in mind it is Windows S for a reason, emulation mode is extremely slow with out APP enhancements to make sure it runs in App Approved windows environment. It basically has a performance of Atom but in Windows .net mode its probably pretty good.

    "Any potential GPU advances is still years off. I wouldn't bet on any of that for the time being. By the time they do deliver, AMD and NVidia aren't going to stay still."

    Do you really think Intel is not going to stay still - if you do, you are living in a fantasy world. Keep in mind Raju left AMD for a reason and he probably was reading the writing on wall. It is funny that AMD fanboy's say Intel Glue itself together - but basic designed of Zen is multiple cpus Glued together to get AMD the temporary claim of more cores.

    "They've been cashing in on the same architecture for years, and just started introducing more cores because of Zen. "

    The same can be said about AMD - just that Zen is release new - but AMD had the ill-fated BullDozer core about the same time as intel start the i series. Intel does call the core - but the iSeries are significantly different than the Core 2 base series.

    Please educate your self on history of both Intel and AMD

    https://en.wikipedia.org/wiki/List_of_Intel_microp...–_Core_microarchitecture
    https://en.wikipedia.org/wiki/List_of_AMD_micropro...

    " They've been really dependent on their process advances, doing very little in terms of innovation and creativity (as Lisa Su likes to put it, in which I agree)."

    Ok anything you say - but Intel has 49 qubit quantum chip - which is next generation of intelligent computers which AMD or ARM can never even think of competing with it. Lisa Su is basically hoping to grab some of market from Intel and loosing engineers like Raju.

    Go ahead and believe your fantasy world - AMD is no where near the thread to Intel as ARM is. Intel is fully aware of that that is why they are putting efforts in mobile and not desktops.
  • tomi832 - Tuesday, January 9, 2018 - link

    So you really think that in like 3-4 years everybody will have an intel Quantom CPU?
    If that's a yes from you than you're the one who should be educated.
    Quantom CPUs are superbly expensive. Not only to manufacture, develop and buy but also to maintain - they need a really low temperature to operate, check LinusTechTips's video where he went to a facility which had a super computer with a Quantom CPU - it took him like 5 minutes to get in because of all the things around it to make it cool. It's like a little chamber/huge fridge, which operates at a way lower point than 0 Celsius.
    Only big companies like Microsoft and Google would maybe buy it...but for the most people (and 100% of the consumers) Qubit CPU are not useful at all and they won't be able to buy/operate it...
    Also I think that you don't really get it but Intel is a serious underdog in Quantom computing...just because they entered doesn't mean they are the best - they are probably the worst right now, that's why they only have samples but the don't really sell them.
  • HStewart - Tuesday, January 9, 2018 - link

    "Besides, unless Intel's foundry business starts fabbing in mass for others, it wouldn't really matter which is the "superior" process"

    Also this is exactly what they did with Intel 8th Gen CPU core integrated with Vega GPU. They work with AMD to make this happen. But it in a package and other GPU likely could be also integrated.
  • GrouchoMarx - Tuesday, January 9, 2018 - link

    @lilmoe,AMD fanboy detected. Even AMD's partners already admitted that their 7nm chips are less dense than Intel's 10nm. Intel's rivals are too much behind. I'll get an Intel 10nm this years while you will get an inferior AMD 7nm on 2020(likely as I don't buy AMD's schedule of 2019. Look how hard it's been for Intel)

    @HStewart, you are a smart guy ;-)
  • tomi832 - Tuesday, January 9, 2018 - link

    BS much?
    First of all - I didn't see GloFo/TSMC/Samsung admit that intel's 10 nm is more sense than their 7 nm, only Intel claimed that like 2-3 years ago when they talked about their "upcoming" 10 nm, which according to many leaks - is truly disappointing, which is why they just release 14++ instead of the 10 nm node - because their 14++ is actually better.
    Now take that, with the fact that GloFo's 14 nm is seriously behind Intel's 14+, yet Ryzen isnt a lot behind Kaby Lake, and the gap between Coffee lake and Zen+ will be even shorter.
    Intel is losing their Node advantage that they had for years. About 7 years ago they claimed that they'll have 10 nm by 2015. It's 2018 now! Intel isn't on schedule - unlike other Foundries. They also claimed that they are way ahead of other companies while actually the 10 nm that is on the market is about as good as Intel's 14+/+. Yet according to Intel, they should have a way bigger advatnage...which just isn't there, it's gone.
    GloFo's 7 nm will probably be about as dense as Intel's 10 nm.
    So - I think that the only fanboy around here is actually you, especially by the way you talk ("inferior AMD 7 nm etc") and say that you won't even consider buying AMD's products...
  • FunBunny2 - Tuesday, January 9, 2018 - link

    makes one wonder whether Heisenberg and the rest of physics is nearly ready to say "enough already!!"
  • jjj - Tuesday, January 9, 2018 - link

    For Intel QS means shipping and they do this all the time, ship QS to claim that they shipped on schedule.
  • bugnguts - Tuesday, January 9, 2018 - link

    Intel's problem is 10nm is too small for the core architecture. It will work and it will consume less power, but it will need to decrease clock speeds because it cannot transfer heat fast enough at this smaller node.
    Haswell(22nm) to Broadwell(14) had this problem where the typical max overclock actually lowered. Intel has worked a lot to push the clocks higher but we are seeing very little increase after Skylake. Even worse Intel must lower clocks on anything above a 4 core. IPC improvements have decreased as this architecture has been improved upon for over 15 years. The move to 10nm will mean lowering clocks in desktops, who wants to downgrade to save a bit of power? Servers, well the problem is even worse since they use high core counts the clocks will certainly be pushed lower. Sure you will consume less power per chip, but you will actually have slower chips, which means you will need more, which in turns negates the lower power draw.
    This all falls in line with what Intel announced quietly today. "We are shipping 10nm Cannon Lake...they are low power dual core..., but we are shipping 10nm.

Log in

Don't have an account? Sign up now