• utopiah@lemmy.ml
    link
    fedilink
    arrow-up
    0
    ·
    13 hours ago

    It’s bad, OK, I’m not trying to remove any blame or criticism for the situation.

    That being said, what actual normal usage for a normal person (not a National Geographic 8K stereo 360 video editor or a developer compiling Firefox daily) needs a lot, say more than 16GB, of RAM?

    Again I’m not trying to find a way to accept a bad situation, only to see which actual usages are in trouble because honestly I don’t see any.

    • BJW@lemmus.org
      link
      fedilink
      English
      arrow-up
      0
      ·
      5 hours ago

      Hosting an accurate AI locally typically takes 24GB of VRAM. Not having sufficient VRAM makes you reliant on remote models as a paid service. You pay many more times for tokens what you could pay for RAM just once for limitless tokens.

      • utopiah@lemmy.ml
        link
        fedilink
        arrow-up
        0
        ·
        2 hours ago

        Same, that’s not a normal usage. Spending a lot of time on Lemmy/Reddit/HackerNews tech community might lead to believe running a local model is normal but very few people do. When they do it’s typically on GPUs or Mac precisely for their unified memory architecture… but it’s not a normal setup even for gamers. Also typically it’s to tinker locally and learn from than actually productivity. I’m repeating here, I’m not saying it’s not an interesting use case for some, just not what most people do.