• @brucethemoose@lemmy.world
    link
    fedilink
    English
    0
    edit-2
    3 days ago

    …once the tech has advanced enough that putting one together makes for a substantial boost in what you get for the same price and power envelope.

    It already does.

    Its more a question of economics of scale. Taping out a single custom chip is extremely expensive, like hundreds of millions of dollars before a single chip is sold.

    AMD could make a custom Strix Halo SKU for Valve (think a 6-core X3D CCD, a 32-40CU GPU clocked low for efficiency) for much less. Perhaps something like that (a custom multi-die configuration of Strix Halo’s successor?) is what Valve opted for.

    • MentalEdge
      link
      fedilink
      English
      03 days ago

      Well yeah. But I used the word “substantial” together with the word “price”.

      If the return ain’t worth it, it’s not “substantial” yet, and hence “It already does.” is a false statement.

      • @brucethemoose@lemmy.world
        link
        fedilink
        English
        03 days ago

        Yeah. But what I’m getting at is the economics may hint at what valve is planning.

        Maybe AMD isn’t making a “specialized” monolithic die like Van Gogh? Perhaps Valve is simply customizing blocks of AMD’s existing product (die) stack, which is more financially plausible.

        AFAIK one of the current issues with Strix Halo for a handheld would be high idle power, but maybe the next generation is better in that respect.

        • MentalEdge
          link
          fedilink
          English
          03 days ago

          I think they’ll do whatever makes sense.

          I’m just saying Valve and AMD are on the same page about making it happen when and if it makes sense.

          For now, the first steamdeck is still great.