• Fedegenerate@lemmynsfw.com
    link
    fedilink
    English
    arrow-up
    3
    arrow-down
    3
    ·
    edit-2
    7 days ago

    If Intel/AMD provides an LLM capable card, by the time I get around to making a big boy server, then that’s what I’ll get. Ideally Intel for the sweet AV1 encoding and Quick sync. Then again, if the card runs LLMs, then transcodes won’t be a problem. Cuda looks nice, but fuck Nvidia, I’ll go without.

    Don’t look at me, I’m an Android gamer these days, and my home lab runs on integrated Intel graphics.

    Maybe one day I’ll build a gaming rig, but mine’s so old now (gtx970) that I’m pretty much starting from scratch. I can’t justify the expense to make a rig from nothing when my Retroid pocket is >£300 and I can play most everything PS2/Gamecube and before.