Experimenting with CachyOS kernel on NixOS via Chaotic's Nyx
discourse.nixos.org
I’ve heard praise about the kernel performance with CachyOS, and was looking to replicate it a bit on NixOS when I learned about Chaotic’s Nyx. After simply swapping to their Link-Time Optimization kernel varent, I’m noticing a ~10% single and ~8% multi core geekbench scores improvement after repeated testing on my 10yo laptop: https://browser.geekbench.com/v6/cpu/compare/14141971?baseline=14145053 Over 5 consecutive trials in performance mode, results are relatively consistent: I’m als...
  • dinckelman@programming.devEnglish
    3·
    2 days ago

    My biggest issue with Nyx, and the reason why I’ve yanked it out of my flake twice now, is because unfortunately this repo has no quality control.

    If the Nix foundation Hydra fails, the whole merge is cancelled. If Nyx’ build system fails, they just write the broken packages into the equivalent of shit_that_broke.json, and still push it.

    Given that I’ve only wanted to use 1 package from the entire repo, it is shocking that it’s both impossible to bisect for debugging, but also increasingly frustrating to get any help with.

    This is especially obvious for kernel packages, where nixos-unstable is a little behind Nyx, so fully expect your builds to error out frequently, if you use applications, reliant on specific kernel functions, like what openrgb/openrazer people experienced a few releases ago

    • ruffsl@programming.devOPEnglish
      1·
      2 days ago

      Appreciate the heads up. Looks like they use merge bots to auto update the package version JSON files for git packages, making for a very large/frequent commit history. Was that what made bisecting imposable?

      I also see they pin the nixpkgs input, but do others normally modify that nixpkgs input to follow their global nixpkgs from their own system flake, or does that invalidate the use of Nyx community cache?

    • ruffsl@programming.devOPEnglish
      2·
      3 days ago

      I wasn’t sure if these results were specific to my older hardware, or more generalizable to newer systems.

      • just_another_person@lemmy.world
        1·
        3 days ago

        Wait…what? If anything, building on your current hardware would yield more realistic results.

        What exactly did you do here, just swap packages?

        • ruffsl@programming.devOPEnglish
          2·
          3 days ago

          I don’t have resources to locally build an optimized kennel for every update for each of my systems, thus my interests in keeping with a community cache.

          I didn’t do much here, just swapped the kernel via config and ran some benchmarks. I posted as I was more curious to hear of what engineering trade offs may be at play, and what experience folks have had in daily driving CatchyOS’s kernel patch sets.

            • ruffsl@programming.devOPEnglish
              31·
              3 days ago

              Compiling the Linux kernel from scratch takes over an hour on this laptop. Given I’m tracking the unstable channel on a rolling distro means doing that several times a week. Ain’t nobody got time for that. Or at least I don’t…

    • Telorand@reddthat.com
      1·
      3 days ago

      “I was so focused on whether I could, I didn’t stop to consider whether I should.” —inventor of the Steam Brick