• MyOpinion@lemmy.today
    link
    fedilink
    English
    arrow-up
    6
    arrow-down
    1
    ·
    6 hours ago

    Jenson will find a way to have sex with another spy and get around it as an independent agent of China.

    • abc@suppo.fi
      link
      fedilink
      English
      arrow-up
      4
      arrow-down
      1
      ·
      5 hours ago

      You think just laziness was preventing them before?

      • rumimevlevi@sh.itjust.works
        link
        fedilink
        English
        arrow-up
        4
        arrow-down
        1
        ·
        5 hours ago

        When you can easily buy the best hardware off the shelf, it doesn’t make financial sense to spend billions trying to reinvent the wheel. Now that the supply is cut off, they have no choice but to pour resources into catching up."

        • 3abas@lemmy.world
          link
          fedilink
          English
          arrow-up
          5
          ·
          3 hours ago

          The Huawei Ascend 950DT reportedly reached 86% inference performance of Nvidia’s H200, designed for better interconnectivity with more than double the bandwidth.

          They lack the most advanced photolithography tech Nvidia has access to, but they haven’t been lazy.

    • NOT_RICK@lemmy.world
      link
      fedilink
      English
      arrow-up
      13
      ·
      9 hours ago

      Huawei is certainly trying, along with other Chinese companies, but they have a handful of significant issues holding back their chips. The big one is lack of EUV machines, and their lack of a domestic supply of high bandwidth memory and the machines needed to make them. Their yields are significantly worse than TSMC’s because they are relying on DUV to make their chips.

      • Nola@piefed.zip
        link
        fedilink
        English
        arrow-up
        2
        ·
        2 hours ago

        Give it a few years.

        Closing the gap in chip manufacturing only become a tier 1 priority in the latest 5 year plan that started this year.

        Generally speaking the target is total self sufficiency in most input components for AI by 2030 and hitting or exceeding certain benchmarks for advanced computing.

        By and large Chine tends to hit and exceed their 5 year plans when it comes to supply side development and production targets