Stories
Slash Boxes
Comments

SoylentNews is people

posted by martyb on Friday June 14 2019, @07:10PM   Printer-friendly
from the big-impact-from-very-little-things dept.

China Is Still Multiple Generations Behind In Chip Manufacturing

When it comes to the actual foundries China has within its borders, the picture isn't good for the country. Perhaps the most advanced foundry there is owned by Semiconductor Manufacturing International Corporation (SMIC). A company spokesperson late last year said, "Our 14nm technology will start risk production by 2019, 12nm process development is completed and under customer verification."

Keep in mind how much further along the rest of the world is: TSMC (Taiwan) is already producing high performance AMD CPUs on its 7nm process with low power Apple parts having shipped in 2018, Samsung is readying advanced EUV production lines for NVIDIA's next generation of graphics chips, and Intel is rolling out its 7nm-equivalent this year as well. We even reported yesterday that TSMC is now actively developing its 2nm node!

If China's most advanced foundry is only beginning low-volume 14nm production this year, that would put them about four or five years behind the rest of the world. An eternity in the world of semiconductors.

For now, Huawei is building their world-class and cutting edge SoC, Kirin 980 on TSMC's 7nm process. If they were forced to use SMIC's 14nm process it would force them to regress in both performance and efficiency which would be a death-knell. Currently the Kirin 980 can compete with Qualcomm's Snapdragon 855, but should Huawei be forced to fab its chips within its own countries[sic] borders this wouldn't be the case.

[...] It seems Chinese companies will have to do things the old fashioned way and grit their way through the learning curve with using these chip-production tools. One way around this would be to hire talent away from companies with a mature understanding of the technology, but even this is proving difficult.

For instance a Chinese DRAM company CXMT attempted to hire away a top Samsung engineer who had expertise in his field, but a South Korean court blocked the move. Kim Chi-wook headed the company's DRAM design team and would be a home-run hire for any DRAM company lacking knowledge. The court made no qualms about the fact that the engineer getting hired by CXMT would potentially hurt Samsung's competitive edge. They wrote, "Chinese semiconductor companies are estimated to be three years to 10 years behind in technology gap regarding DRAM designing technique."


Original Submission

 
This discussion has been archived. No new comments can be posted.
Display Options Threshold/Breakthrough Mark All as Read Mark All as Unread
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
  • (Score: 3, Insightful) by HiThere on Friday June 14 2019, @07:54PM (3 children)

    by HiThere (866) Subscriber Badge on Friday June 14 2019, @07:54PM (#855727) Journal

    ISTM, without detailed knowledge, that most of the advantage of electronics can be had within their current levels of expertise. It's true that there are increased advantages with more tightly controlled manufacturing, but the cost rises so much that the advantage is decreased.

    China would probably do better in developing new technologies for the job than in urgently refining their current techniques. Improve things at the most economic rate, but look for different approaches. Photonic computers, spin-based diamond sheets, something. That way they'd be entering at the start, where while the failures are numerous, the profits are immense. And it might obsolete the high-end silicon chips.

    --
    Javascript is what you use to allow unknown third parties to run software you have no idea about on your computer.
    Starting Score:    1  point
    Moderation   +1  
       Insightful=1, Total=1
    Extra 'Insightful' Modifier   0  
    Karma-Bonus Modifier   +1  

    Total Score:   3  
  • (Score: 2) by takyon on Friday June 14 2019, @08:10PM (2 children)

    by takyon (881) <takyonNO@SPAMsoylentnews.org> on Friday June 14 2019, @08:10PM (#855733) Journal

    That probably is a good idea. As silicon scaling continues to slow down, it will be easier to steal the trade secrets and remain competitive. It's the potential 1,000x performance increase from new technologies that has to be found ASAP. The problem is that there are a number of promising technologies and nobody is sure which ones will come out on top.

    --
    [SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
    • (Score: 0) by Anonymous Coward on Friday June 14 2019, @08:38PM

      by Anonymous Coward on Friday June 14 2019, @08:38PM (#855741)

      "The problem is that there are a number of promising technologies and nobody is sure which ones will come out on top."

      In a long-game mindset, try them all, and the State kills of the ones that fail

    • (Score: 3, Interesting) by c0lo on Saturday June 15 2019, @01:29AM

      by c0lo (156) Subscriber Badge on Saturday June 15 2019, @01:29AM (#855843) Journal

      It's the potential 1,000x performance increase from new technologies that has to be found ASAP.

      Let me add to this an immutable fact of nature: the lattice constant of Silicon is 5.431Å [wikipedia.org].
      A 5nm gate = 50 atoms in thickness. Guess what happens when you have too few of them in a gate?

      As silicon scaling continues to slow down, it will be easier to steal the trade secrets and remain competitive.

      Translation: it's already easier to steal trade secrets in electronics and the law of diminishing returns guarantees the situation is not gonna improve.

      --
      https://www.youtube.com/watch?v=aoFiw2jMy-0 https://soylentnews.org/~MichaelDavidCrawford