• Airfried@piefed.social
    link
    fedilink
    English
    arrow-up
    17
    arrow-down
    3
    ·
    3 days ago

    AMD does the exact same as Nvidia. They’re just less successful with it. It doesn’t really matter who you buy from in that regard, I think.

    • FlexibleToast@lemmy.world
      link
      fedilink
      English
      arrow-up
      9
      ·
      3 days ago

      AMD doesn’t gate the high vram to be only their top end model. They have generally higher vram across the board.

      • Airfried@piefed.social
        link
        fedilink
        English
        arrow-up
        1
        arrow-down
        2
        ·
        21 hours ago

        Whether that’s important or not is your choice as a consumer. If you use Blender for example, Nvidia gives you more bang for your buck. It just works much better with those cards. That’s simply what the spread sheets say. If you need slightly higher VRAM for different reasons go for something else.

        But that’s completely besides the argument. Whatever you choose doesn’t change the fact that neither of those companies gives a shit about us end consumers. They’re barely making any money from us compared to AI companies. They’re all in on AI.

      • tempest@lemmy.ca
        link
        fedilink
        arrow-up
        2
        ·
        3 days ago

        They did but do their latest cards have an option for 24gb like my 7900xtx?

        I like them for Linux support but I don’t doubt they would start doing the same crap if they could break CUDAs strangle hold on the ML LLM space

        • FlexibleToast@lemmy.world
          link
          fedilink
          English
          arrow-up
          4
          ·
          2 days ago

          I don’t know. I’m still using my 7900XT with 20GB. For my home server I just bought an Intel Arc Pro B70 32GB.

    • alpine_jim@piefed.social
      link
      fedilink
      English
      arrow-up
      4
      ·
      3 days ago

      yeah the main reason AMD isn’t quite as price-gouged is that ROCm isn’t quite as in demand as CUDA for non-gaming uses for GPUs.

      • boonhet@sopuli.xyz
        link
        fedilink
        arrow-up
        3
        ·
        3 days ago

        Lisa Su has also been cozying up to Trump to try and accelerate data center construction.