Linux works well for most jobs, but to get it to work great requires kernel tuning, and that's never easy. ByteDance has a proposal to make this easier.
  • qazEnglish
    11•7 months ago

    In another case, ByteDance was able to optimize HTTP network latency on an NGINX server by optimizing the tuning of 16 kernel sysctl parameters. In its best scenario, the ML tuning gave the NGNIX network performance a 12% boost over expert manual tuning. Again, that’s a significant improvement.

    Looks very promising

    • PropaGandalfEnglish
      6•7 months ago

      I’m sure that AI will be used to optimize software, hardware and configurations like this somewhere in the future.

      • Cosmonaut_CollinEnglish
        5•7 months ago

        Near future really. Supposedly the algorithms for chatgpt4 are going to be many levels above chatgpt3. I would imagine coding algorithms will follow suit as openai keeps updating their stuff.

        • Cosmonaut_CollinEnglish
          3•7 months ago

          I don’t want to remove my post, but I just learned about the openai situation, and I have low expectations for chatgpt4 now. Microsoft on the other hand may have some good stuff coming out. Hopefully it will be available to the public like chatgpt3, but my expectations on that are low considering the monthly subscription model Microsoft has adopted for the office suite.

        • @RealBot@lemmy.worldEnglish
          1•7 months ago

          True, but algorithms they used in this article are just a “level” higher than primitive statistics methods and even that is not mainstream in kernel parameter (and similar things) optimisation. So it might be some time before we see more advanced methods used for this stuff.