• Re: Hint For GNU/Linux Progrmmers

    From =?UTF-8?Q?St=C3=A9phane?= CARPENTIER@[email protected] to comp.os.linux.misc on Sat Sep 13 13:44:41 2025
    From Newsgroup: comp.os.linux.misc

    Le 12-09-2025, Lester Thorpe <[email protected]> a écrit :
    On Fri, 12 Sep 2025 03:02:09 -0400, c186282 wrote:


    THE best op is to proto, look/think for a few weeks,
    then re-write.

    That will do FAR more than any compiler tweaks.


    Everyone is missing the main point.

    Which one?
    - That you are a fraud? Nope: I know it.
    - That you don't know how to optimize compilation? Nope: I know it.
    - That you can only copy/past code? Nope: I know it.
    - That you are a distro lackey? Nope: I know it.
    - The fact that the more you speak about something, the less you know
    about it? Nope: I know it.
    - That you are a Windows fanboy trying to make Linux users pass like
    morons? Nope: I know it.

    I am referring to optimizing code that is already published
    and available, e.g. the average GNU/Linux package.

    You mean that the guys who wrote and published the code know how to
    compile it? Or do you mean what the people competent enough to write
    code for a great tool are too stupid to be able to know how to compile
    it?

    Do you really understand how your sentence is, at the same time, stupid
    and inconsistent? You explain at the same time they know what they are
    doing and they don't know what they are doing. You just explained you
    need random people to help you find a general way to sort out what
    experts do good and what they don't.

    I have experienced up to 40% performance increase using just compiler options.

    I don't believe that. And your last video proves that it's a lie.

    But finding the best options can at times be difficult

    Agreed. But I don't believe you can find them. And, I believe the distro managers, helped with the people who provided the code, can do it. In
    any case, it would take me hours to find better options than what's
    provided by the distro managers helped by package producers to get
    noticeable results on my own computer. Another way to state it: spending
    hours to win few seconds each moths is a waste of my precious time.
    --
    Si vous avez du temps à perdre :
    https://scarpet42.gitlab.io
    --- Synchronet 3.21a-Linux NewsLink 1.2