No announcement yet.

Apple M2 vs. AMD Rembrandt vs. Intel Alder Lake Linux Benchmarks

  • Filter
  • Time
  • Show
Clear All
new posts

  • Originally posted by mdedetrich View Post
    anyone with a uni level degree or a doctorate in statistics will instantly tell you how massively wrong you are.
    I know the US education system is kind of a joke, but I doubt even they really don't get as far as long division until college.

    > Do a course in statistics, then come back.

    oops. So what's your excuse going to be when I tell you I already have? "It only counts if it's a PhD from Harvard", maybe?

    It's funny how "a ridiculous number of installs" was something you considered proof at the start of this. Now that it's come back to bite you in the ass it's suddenly just a random number. The only reason I'm still in this thread *isn't* to get you to admit defeat, it's because it's literally comedic watching you back away from your own words while simultaneously rewriting the script on the fly. It's been a nice relaxing way to wrap up the nights and crack a smile at the end of the day.

    Sadly, I'm out of time now and busy for the rest of the week, so I'll put you out of your misery. Homebrew's homepage doesn't work, but the stats pages for the time periods do. Note that these are package downloads, so they overestimate the activity, possibly substantially: e.g. 24M installs of Chrome in a year doesn't mean 24M Chrome *users*, it means slightly over 2M. (Unless Chrome is gaining market share, which it isn't).
    Over the past year, the top packages are:
    #1 iterm2 362,542 3.17%
    #2 visual-studio-code 360,739 3.15%
    #3 docker 288,850 2.53%
    #4 google-chrome 287,401 2.51%
    #5 adoptopenjdk8 237,162 2.07%
    #6 firefox 215,795 1.89%
    After that you're in the <200K range.

    It's worth reminding anyone who stumbles across this later that Homebrew doesn't include any Apple tools. For all we know, there are eight million Xcode users writing apps for IOS, and those wouldn't show up here at all. Other than that, I'll leave people to draw their own conclusions, since if I do it we'll just have to sit through another cycle of arguments on par with the Republican denial of climate change.

    Download stats for the package manager itself - the "ridiculous number of installs" - are either unavailable or obviously wrong (like, 40M+/mo), so there doesn't seem to be any backing for that claim at all. I mean, unless you need a doctorate to work out that it's unlikely there would be more unique installs of something than there are devices to install it to. I guess we know who actually needs to "Do a course in statistics, then come back". :P

    > If it's ~1M, that's in 1 in 27, and we're both terrible at this.

    As it turned out even that wasn't particularly close, but it's in the right ballpark, especially since this is ignoring at least some (and likely a majority of) IOS devs. That aside though, being very generous with rounding, roughly 1.4% of Macbooks are used by developers. Note that this does include Airs, but since the discussion was about the M1 - and mdedetrich objected strongly to the idea that anyone was buying an M1 for the sake of better battery life rather than better performance - that's apparently the group we're using.
    It's worth pointing out that the Homebrew stats also include *older* Macbooks running on Intel silicon, and Macs in general, so this potentially *over*-estimates the number of developers on M1 machines by a large amount, but it seems reasonable to do so since we're looking at installs over the past year, and can expect that any developer on an older Macbook already has everything installed.

    Only barely related, but that's an impressively high ratio of Firefox installs to Chrome ones. Mac users are obviously a lot better about testing browser compatibility than most, though obviously they also have a lot more incentive to.

    All told, 1 in 1000 is probably a better guess than I thought for "people actually compiling code etc" (rather than just editing Lamda for microservices, running JS on web pages, and so on) in situations where an x86 laptop was also an option; but again, the Homebrew numbers don't take IOS development into account other than for incidental cases where the same developer is also responsible for the backend.

    It's crazy just how obsessed some Apple users are. "The M1 is a really good chip" isn't enough for them: it has to be "M1 is magic, and 140% of developers use it: not only is there literally nobody using x86 hardware any more, M1 went back in time and converted developers from before it was released". It really is cult-level stuff: start with a lie, and when you get called on it change the lie to something else but still claim the original was true anyway, then go all out attacking the infidel to steer the conversation onto them and away from anything that might risk someone else questioning the lie.
    I'm certainly not expecting anything as trivial as a few numbers to persuade anyone that deep in the cult, but they might satisfy the curiosity of those of us who live in reality.

    I look forward to the inevitable "rebuttal", which will doubtless be as devoid of anything but more lies and excuses as the previous ones. I won't have time to reply to it, but don't let that stop you.


    • Originally posted by arQon View Post

      I look forward to the inevitable "rebuttal", which will doubtless be as devoid of anything but more lies and excuses as the previous ones. I won't have time to reply to it, but don't let that stop you.
      Yeah its full "I don't like this figure because its too high so I will ignore it because it doesn't suit my argument" without providing any real evidence of your reason. You can prove anything you want by pulling your own figures out of your ass, you want a trophy?

      See you later, also pro tip don't waste your time with a wall of text in your reply. There are more productive ways to spend your time rather than being unable to deal with being wrong once in a while.
      Last edited by mdedetrich; 19 August 2022, 03:14 AM.