drm/xe: Don't hand out the flat CCS storage as usable VRAM · torvalds/linux@818bebe
github.com
get_flat_ccs_offset() reads the base of the flat CCS storage from the hardware, scales it by the number of enabled L3 nodes, and rounds the result up to 128K. Everything below that offset is then ...
  • Mwa@thelemmy.clubEnglish
    2·
    17 hours ago

    But isn’t it as bad as running a graphically intensive AAA game on your computer

    • deadcream@sopuli.xyz
      11·
      17 hours ago

      That’s still just moving the impact from data centers to end users’ machines.

      Gamers (especially those that own 5090) are a tiny minority compared to llm users.

        • deadcream@sopuli.xyz
          11·
          12 hours ago

          This will just waste even more energy since cpus are less efficient at this kind of task.

          LLMs require insane amounts of compute, period. The hardware to do that needs energy and emits heat, there is no way around.

            • deadcream@sopuli.xyz
              11·
              9 hours ago

              It will still run necessary calculations much slower than a gpu, so you will need to run it much longer for the same result, which is not only inconvenient but will likely use more power in the end anyway.