• [object Object]@lemmy.ca
    link
    fedilink
    arrow-up
    1
    ·
    48 minutes ago

    A bit expensive to get test drivers, learned that from amd.

    Lots of ports, seems pretty good, the AI features are fucking stupid though “you can run openclaw on it” okay… you can run openclaw on most anything, what models can it run on low memory and a low power NOU? Realistically you can do fine embedding, but you won’t get realtime VTT or TTS or usable agents with openclaw.

      • gravitas_deficiency@sh.itjust.works
        link
        fedilink
        English
        arrow-up
        1
        ·
        2 hours ago

        Like, I get the value proposition of more open hardware… but Apple is absolutely killing it in terms of low to mid tier homelab ML-oriented boxes. They straight up work better, and require near zero custom configuration - and zero kernel hacking.