Roadmap shows that in 2021, Intel desktop CPUs will remain on 14nm

Discussion in 'HardForum Tech News' started by _mockingbird, Apr 24, 2019.

  1. Gideon

    Gideon 2[H]4U

    Messages:
    2,317
    Joined:
    Apr 13, 2006
    They tried to use lots of masks which is a huge issue and why their yields are in the toilet. With EUV they dont need need all those masks and thus why 7nm will work while 10nm will stay trash.
     
  2. pillagenburn

    pillagenburn Gawd

    Messages:
    1,008
    Joined:
    Oct 3, 2006
    14nm+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-

    ^ see it all cancels out in the end.
     
    schmide, GoldenTiger and Jim Kim like this.
  3. Comixbooks

    Comixbooks Ignore Me

    Messages:
    13,863
    Joined:
    Jun 7, 2008
    You would think motherboards should be getting smaller and the CPU is at it's peak already.
     
  4. merid14

    merid14 [H]Lite

    Messages:
    75
    Joined:
    Jul 8, 2008
    I would agree with you here and add Intel needs to get 7nm done right and they need to get all the speculative exploits out of the hardware!
     
  5. Chimpee

    Chimpee [H]ard|Gawd

    Messages:
    1,457
    Joined:
    Jul 6, 2015
    You cannot get all the speculative exploits out of their hardware, there will always be an exploit someone will find. You can design a chip to mitigate as much as you can but the sad reality that most of the time, there is a delay between exploits found and a hardware fix.
     
  6. merid14

    merid14 [H]Lite

    Messages:
    75
    Joined:
    Jul 8, 2008
    That is true. Do you think it has to do with the popularity of Intel over AMD as to why more haven't been found on that side?
     
  7. Nobu

    Nobu 2[H]4U

    Messages:
    3,279
    Joined:
    Jun 7, 2007
    I doubt it has a huge influence. i mean, sure, there are way more intel than AMD servers, but there are still plenty of AMD, ARM, and other servers that would be prime targets for these exploits.
     
  8. Chimpee

    Chimpee [H]ard|Gawd

    Messages:
    1,457
    Joined:
    Jul 6, 2015
    I doubt it has to do with Intel being with the higher market share as why it always seem to be Intel, more like Speculative Branching CPU design in general as Nobu indicated other CPU brand are affected as well in varying degrees. It just Intel left a bigger hole compare to AMD in their Speculative execution, whether that is intention (which I doubt it is) or not is a whole another debate. This is my opinion, I think as long CPU use Speculative Branching, there will always be some sort of exploits.
     
    N4CR likes this.
  9. Staples

    Staples [H]ardness Supreme

    Messages:
    7,858
    Joined:
    Jul 18, 2001
    But if they are still faster than the competition, I don't see this being a problem.
     
    Snowdog likes this.
  10. Nebell

    Nebell [H]ard|Gawd

    Messages:
    1,582
    Joined:
    Jul 20, 2015
    Eh, I'm on i3-8350k and 1080Ti, happily gaming on a 1440p 50" screen. Even Metro Exodus ran on max details. If Intel/AMD/Nvidia want my money, they will have to do better.
     
  11. merid14

    merid14 [H]Lite

    Messages:
    75
    Joined:
    Jul 8, 2008
    I guess that's the crux of the argument for me. Will Intel at 7nm actually be faster than AMD even when you factor in a potential speculative exploit? Sure on a benchmark after a patch they may loose 100 points overall but in the real world it's actually 1000 points for a specific application. For example, say I purchase an i7 8700 today because i'm running Photoshop for my business and it's faster than a 2700x. But in 3 months an exploit comes out that doesn't hinder AMD and for the most part doesn't hinder Intel except when trying to rasterize an image. Now my workflow is hindered but I've already bought into the platform based on the benchmarks.
     
    N4CR likes this.
  12. juanrga

    juanrga Pro-Intel / Anti-AMD Just FYI

    Messages:
    2,550
    Joined:
    Feb 22, 2017
  13. clockdogg

    clockdogg Gawd

    Messages:
    918
    Joined:
    Dec 12, 2007
    If only we had a trusted [H] source on the Intel inside to give us the [H]ard++++++ truth.
     
  14. juanrga

    juanrga Pro-Intel / Anti-AMD Just FYI

    Messages:
    2,550
    Joined:
    Feb 22, 2017
    I was out of computer world for a time. On return, I asked the SiliconGang for updates on last news/leaks. This was the April 30 update for the Intel 10nm status:

     
  15. schmide

    schmide Limp Gawd

    Messages:
    240
    Joined:
    Jul 22, 2008
    So this person is more of a source than the original?

    I would however, trust him or her for Spider-Man leaks!!!
     
    Last edited: May 6, 2019
  16. juanrga

    juanrga Pro-Intel / Anti-AMD Just FYI

    Messages:
    2,550
    Joined:
    Feb 22, 2017
    You didn't understand anything. He reported the findings of the SiliconGang.
     
  17. Snowdog

    Snowdog [H]ardForum Junkie

    Messages:
    9,442
    Joined:
    Apr 22, 2006
    I am not calling this one fake. But logically your argument doesn't hold water.

    If I was making a fake roadmap, I wouldn't be dumb enough to contradict the publicly available info, and render my fake obvious from the start.

    I would use as much verified info as possible, and extrapolate from there.

    So having a "leaked" roadmap with some verified info, says absolutely nothing about it's veracity.
     
  18. schmide

    schmide Limp Gawd

    Messages:
    240
    Joined:
    Jul 22, 2008
    Huh? OK enlighten me.

    You post on twitter here.

    Spider dude/dudette replied to you.

    So Spider dude/dudette heads this gang? Why should we trust this gang?
     
    N4CR likes this.
  19. Nobu

    Nobu 2[H]4U

    Messages:
    3,279
    Joined:
    Jun 7, 2007
    silicongang appears to be a group of old hats (and some marketing types) who follow a tag on Twitter and masterdebate about tech. I could imagine some credible leaks appearing on there, but it looks like mostly rumor mill lately.
     
    schmide likes this.
  20. juanrga

    juanrga Pro-Intel / Anti-AMD Just FYI

    Messages:
    2,550
    Joined:
    Feb 22, 2017
    @schime you continue without having idea.

    The twitter tag was created after the group. We have even a website. And one of our members has joined Intel a pair of days ago.

    The leaked roadmap was analised by members of the silicongang and debunked as fake.

    In this case I would say one doesn't even need to be familiar with the technology or have contacts in the industry to question the validity of this leaked roadmap, When see a slide says slide "152 of 14" one would start to question its validity.
     
  21. Gideon

    Gideon 2[H]4U

    Messages:
    2,317
    Joined:
    Apr 13, 2006
    Yeah cause you never intentionally mislabel something in a slide to see who leaks the info. It's one of the oldest tricks in the book by large corporations.
     
  22. schmide

    schmide Limp Gawd

    Messages:
    240
    Joined:
    Jul 22, 2008
    This one has the features you claim. https://hardforum.com/threads/skyla...s-and-reviews.1933735/page-36#post-1044186426

    But that isn't this and that one is unlinked and uploaded to [H] unsourced.

    Regardless. My point being silicongang has no more credibility than any other rumor esp when the dude/dudette is only listed as a spiderman fan.

    Even more egregious. You changed the phrasing without citing it.

     
    Last edited: May 7, 2019
    N4CR likes this.
  23. Nobu

    Nobu 2[H]4U

    Messages:
    3,279
    Joined:
    Jun 7, 2007
    afa the wording change, spider-dood corrected himself in a later tweet.
     
    juanrga likes this.
  24. extide

    extide 2[H]4U

    Messages:
    3,435
    Joined:
    Dec 19, 2008
    Honestly I think people make a much bigger deal when it's Intel -- I mean there are guys out there who's primary focus is litterally breaking the security on modern CPU's. For example, check this out -- the VIA Samuel 2 core is COMPLETELY BROKEN -- you can execute Ring 0 code from Ring 3.



    But I bet most of you guys have never even heard of that exploit. I mean sure, you won't ever find one of those in a server, but they exist in a lot of places like POS systems and whatnot, and those potentially have access to credit card information...
     
    Last edited: May 7, 2019
  25. Pantalaimon

    Pantalaimon Limp Gawd

    Messages:
    187
    Joined:
    Aug 17, 2006
    That's a really weird resolution and monitor size combination.
     
  26. juanrga

    juanrga Pro-Intel / Anti-AMD Just FYI

    Messages:
    2,550
    Joined:
    Feb 22, 2017
    We could think that If it was the only mistake on the leaked slides.
     
  27. N4CR

    N4CR 2[H]4U

    Messages:
    3,931
    Joined:
    Oct 17, 2011
    It's probably a Sammy 4k that runs in 120hz 1440p freesync mode.
    Nice handle, loved those books when I was a youngin.
     
  28. blackacidevil

    blackacidevil Gawd

    Messages:
    677
    Joined:
    Oct 1, 2004
    You should go back to wherever that was, and stay there.
     
  29. juanrga

    juanrga Pro-Intel / Anti-AMD Just FYI

    Messages:
    2,550
    Joined:
    Feb 22, 2017
    The OP listed here only two slides of the pack of three leaked. In that other thread I gave a list of link of mistakes on the third slide, including the slide "152 of 14". About the other two slides uploaded here.


    First slide:
    • Intel's SIPP website does not list any Kaby Lake Refresh or Amber Lake Y SKUs as SIPP eligible.

    • Amber Lake Y is only 2 core, not 2 core and 4 core.

    • Coffee Lake R is also not listed as eligible on Intel's SIPP website.

    • Rocket Lake U has "+14nm GFX" coming in Q2 2021, Intel will not have 14nm discrete GPU die, another issue is that Slide 2 mentions Rocket Lake U "+10nm GFX" coming in Q3 2020, why would the 10nm version come first?

    • No dell logo present unlike the other two slides

    Second slide:
    • Kaby Lake G launched on January 7th during CES, not March.

    • Coffee Lake H launched on April 2nd - not March.

    • Amber Lake Y is only 2 core, not 2 core and 4 core.

    • Intel Executives have stated that Lakefield will come after Icelake, this fake roadmap puts it before it [*].

    • There are zero signs of Lakefield launching this quarter like this fake roadmap states, Intel said in Lakefield announcement that products would arrive in H2'19 (Q3 or Q4).

    • Dell logo is fully visible.

    It was a typo, and he corrected himself latter. Instead quoting him (with the typo) and then quoting the correction about Tiger-Lake, I corrected the quote myself.


    REFERENCE:

    [*] In fact Intel has just remarked again yesterday that Lakefield comes after Icelake.

    D6E_094UEAAH3fl?format=jpg.jpg
     
    Last edited: May 9, 2019
  30. Nebell

    Nebell [H]ard|Gawd

    Messages:
    1,582
    Joined:
    Jul 20, 2015
    It's actually 55".
     
  31. Master_shake_

    Master_shake_ [H]ardForum Junkie

    Messages:
    9,700
    Joined:
    Apr 9, 2012
    so at intel it used to be tick tock

    so now is tick tick tick tick tick tick tick tick tick tick?
     
    GhostCow likes this.
  32. Snowdog

    Snowdog [H]ardForum Junkie

    Messages:
    9,442
    Joined:
    Apr 22, 2006
    Which doesn't exactly add veracity to this "leaked" roadmap.

    Though looking at the actual information Intel presented. It does appear that it might be 2021+ before Desktop migrates to 10nm.

    But that isn't necessarily a competitive problem.

    I think Intels big problem now is 14nm clocks so ridiculously high, that 10nm/7nm can't match it. Not just from Intel, but from anyone. I am betting the only CPUs in the next year running 5GHz will be Intel 14nm based.

    Intel Not reaching 5GHz on 10nm means Intel moving their top desktop parts to 10nm would slow them down. AMD doesn't have this issue because they currently top out around 4.3GHz. If they can get to 4.5GHz on 7nm, then its a significant win for them. Intel desktop parts at 4.5GHz would be a significant step backwards.

    IOW, the exact same performance on 10nm/7nm for Intel and AMD, would a be huge step backwards for Intel, and step forward for AMD.

    Intel is moving 10nm first into Mobile where they don't need the high clock speeds. On Paper, Ice Lake looks set to extend Intels mobile lead. New generation Sunny Cove cores, New gen 11 GPU on new 10nm process. Will be competing against a minor Raven Ridge refresh on GF 12nm.

    After that Server parts are the next to get 10nm, where again, you don't need/use top end clock speeds.

    14nm will linger for top end desktop parts, not because 10nm is so bad (it will probably equal TSMC 7nm), but because 14nm is just so good for that application (super high clock speed).

    What does Intel do if it can never get 10nm/7nm to 5GHz?

    It should be interesting to watch this unfold.
     
  33. juanrga

    juanrga Pro-Intel / Anti-AMD Just FYI

    Messages:
    2,550
    Joined:
    Feb 22, 2017
    Are you sure?

    renduchintala-presentation-2-1-jpg.jpg