• SuperIce@lemmy.world
      link
      fedilink
      arrow-up
      42
      ·
      1 year ago

      Ones that are built into the CPU’s hardware but you have to pay extra to unlock like a DLC for your CPU. That’s not a joke, Intel’s new Xeons are like that.

    • SkyeStarfall
      link
      fedilink
      arrow-up
      16
      ·
      1 year ago

      Honestly, it’s probably more worthwhile to rent a GPU on the cloud rather than buying one for this purpose. Unless you are either working with AI or doing something serious with it, it’s likely the better (and more cost effective) option either way. Hardware for training is fairly cheap.

      • foxblood@lemmy.world
        link
        fedilink
        arrow-up
        1
        ·
        1 year ago

        Yeah. I’m currently using google colab but it is sometimes a bit of a hassle.

        I’m currently planning to resume using it and maybe I will get a cheap second hand card by pure luck.

    • henfredemars@infosec.pub
      link
      fedilink
      arrow-up
      4
      ·
      1 year ago

      I bought Nvidia for the AI packages that support their GPUs, but I’m not happy about it. I would have much preferred to go with AMD.

  • JATtho@sopuli.xyz
    link
    fedilink
    arrow-up
    5
    ·
    1 year ago

    On 1st gen ryzen cpu (until rzen3), don’t have branch recording with “perf record -b” So no way to make your program +15% faster with profile data/PGO/FDO without effort. 😞

  • Björn Tantau@feddit.de
    link
    fedilink
    arrow-up
    4
    ·
    1 year ago

    Why are you an AMD fan? It’s been years (probably decades) since they’ve been the struggling underdog.

    Just buy what fits your budget and needs.

  • ItsMeForRealNow@lemmy.world
    link
    fedilink
    arrow-up
    2
    ·
    1 year ago

    I bought an expensive custom built PC with a threadripper CPU on it and they changed socket type for the next generation. And then they didn’t release any, anymore. Just sucks to invest.