South Korea’s president has described the global semiconductor industry as “a field where all-out national warfare is underway” as he announced a $19 billion to diversify the nation’s silicon sector.

  • IHeartBadCode@kbin.social
    link
    fedilink
    arrow-up
    1
    ·
    6 个月前

    Okay look y’all. RISC V is a lot of different things. I think when people say RISC V they’re hoping some shit over ARM for SoC. That’s likely to not happen as RISC V just doesn’t have major advantages over ARM to justify the costly swap.

    But RISC V as the chip that glues a bunch of controllers is already underway. Chiplet designs that handle IO between the CPU and your various disks are already big things that RISC V dominates in.

    So will SK produce RISC V, yes. Will that stop producing ARM SoCs? Absolutely not. RISC V is an open ISA, but an SoC stack requires a lot of things which still need to be licensed, so RISC V isn’t as “open” as some would like it to be, at least in SoC. So it’s not THAT much cheaper than just sticking with ARM.

    In China where IP is like “who dis?” RISC V is getting big because it’s open and China likes getting access to that kind of knowledge. But ARM SoC in China still has a lot of inertia, but that steam can run out a lot easier there than say South Korea because of the “who cares about IP?”. So if any of you are pulling your Milky-V out and saying “NUH UH!”, that’s why.

    And don’t get me wrong, I love RISC V conceptually. I’ve got a LicheePi here that pulls PDFs beautifully and prints them off on my Brother printer. I love what China is doing with RISC V and yeah, I can see why the US has beef. But most SoCs that come out of South Korea are ARM and there’s very little reason (*it’s not THAT much cheaper) for them to move off of what they already know.

    But they are going to be producing the eff out of some RISC V chiplets. If you don’t need everything a SoC requires, RISC V will get the job done.