7
Parzi
4y

So, the future of Apple is ARM, huh? So people are gonna have to maintain, like, SEVEN toolchains for Mac? What, we've got one or two for Mac OS <10.0, 2 PPC OSX chains that are version-dependent, 2 x86 version-dependent x86, Catalina x86 and now Catalina ARM too?

(These are to memory, if i'm wrong please correct me.)

Comments
  • 4
    People actually care about apple PPC toolchains? It's been ages.

    Also, iOS already runs on arm, they'll probably just be adapting that.
  • 2
    @RememberMe people still care about mac os classic and some people can't afford to upgrade, fuck you mean
  • 3
    @Parzi and what fraction of apple users is that?
  • 2
    @RememberMe considering the price of new Macs and how necessary warranties are? (Plus, collectors and enthusiasts exist, so)
  • 4
    @Parzi so basically...a negligible fraction. I don't think anyone would develop new software for PPC anyway, it'd just cost too much for no gain.

    Enthusiasts, sure. But as an enthusiast you're not like a regular customer, so companies won't make their plans around you.

    Also anyone running that old a machine now would find it basically useless for anything but basic document work, and that software's already been written.
  • 2
    nah, as soon as tge first ARM line of apple comes out, all the other will get their planned obsolescence killswitch activated, so there will be only one toolchain.
  • 1
    Xcode a.k.a apple's gift to the developer world will take care of everything!
    Welcome to universal(tm) apps!

    Sheesh. Again.
    And ppc is long dead.
  • 1
    See apple actually don’t use arm cpus they use arm isa and develop their own cores
    Now interesting part here is how they extend the Ax chip architecture for something high performance
    My bet is going multicore I don’t think that a cisc machine can be beaten in single core performance but doing 5nm multicore boy !! That’s where it’s at

    Also they should definitely be extending the iOS toolchain because any other decision doesn’t make sense
  • 2
    @hardfault see, it's still ARM if it's based on ARM... ARMv5 isn't more or less ARM than ARMv7 because of modifications. Phones have extended ARM for all sorts of shit too but it's not heralded as some x86 killer until Apple does it on desktop too. No RPi's gonna kill x86 dammit
  • 0
    @dontbeevil i mean... 2020-2021. https://macrumors.com/2020/06/...
  • 2
    The performance drop with ARM wouldn't be as big as it seems at first glance because Apple's overpriced and shoddy hardware is usually outdated by several years anyway, and the cooling solutions are so bad (at least with the laptops) that they throttle down thermally.

    Another "benefit" is that this move would probably get Apple rid of Hackintoshs because they'd manufacture the CPU themselves. So they could sell more of their shitty Macs to people who just want to develop for iOS.

    Sure, the "top of the line" Macs might suffer somewhat, but Apple has been fucking these customers for a long time now (graphics and audio creatives) so that most of them have already moved to Windows.
  • 3
    @Fast-Nop hardly a performance loss for the everyday customer, I think. The chips inside the iPad Pros are hilariously fast, within spitting distance of x86 laptop chips (eg. https://extremetech.com/mobile/...) (sure it gets beaten in SGEMM and stuff probably due to wider units and special instructions on the intel but that's still nothing short of impressive especially when you look at the a12x's power draw). iPad Pro can already do some pretty heavy stuff, it's just held back by its form factor for truly everyday part use. Apple's been killing it with ARM chip performance lately (Qualcomm seems to have made some progress with 865, let's see).

    A mac would have even more thermal budget so they'd push it even more. Not gonna lie, I might spring for one if they actually do release it and it doesn't cost an arm (hehe) and a leg.

    Also it might provide a good enough push for devs to port desktop software to ARM, which would in turn help MS and Qualcomm's ARM efforts. So it's a win all around pretty much.
  • 1
    @RememberMe The link you gave is about laptop.

    The Core i7-8559U scores 8803 at CpuBenchmark. Now I won't take AMD's Threadripper for desktop because while being ridiculously fast, it's also ridiculously expensive at some $3500.

    But for the $790 of the laptop Core-i7, you could also buy a desktop AMD Ryzen 9 3950X ($700) which rocks 39304 at CpuBenchmark. Sure, you won't get that from Apple because Apple's x86 is always lagging behind.
  • 1
    @RememberMe But in general, I'd agree that when power is a concern, ARM and x86 are converging. ARM from below in terms of performance, x86 from above with regard to power consumption.

    Already back then with the x86 Android tablets, the Intel Atoms were not worse than ARM, technologically (power and performance). It was just that they weren't better either, and incompatible to the market leader in that domain (i.e. ARM), a position that Intel wasn't used to have and failed.
  • 2
    @Fast-Nop desktop Ryzen 9 is hardly comparable, we're talking about thin and light devices. Not to mention it's fairly overkill for ye olde user, and a system built around the Ryzen 9 is going to be big, heavy, and significantly expensive. For the power draw it has and the form factors it's built for, A12X and A12Z's performance is absolutely nuts.

    If you want to see ARM match up in desktop/server performance, check out the Cavium (Marvell) Thunder X2.

    https://anandtech.com/show/12694/...

    Keep in mind this is from 2017-2018, they probably have a new chip in the works. If a smallish house like Cavium can do this three years ago, I'm sure Apple engineers would have no issues at all if they decide to take ARM to high(er) performance.
  • 2
    @RememberMe So the conclusion would be that at least the mobile Macs could be as fast with ARM once the SW vendors make native binaries - but at less power consumption?

    Because, Apple isn't just any ARM, and Apple's A13 is actually quite impressive, as you linked. Also because Intel's 10nm is a neverending story while Apple/TSMC is at 7nm (which is about equal to Intel's 10nm I think, but it works).
  • 3
    @Fast-Nop yup, check out the news from WWDC, they're doing pretty much what was expected, plus they already have software for the arm Macs.
  • 2
    @RememberMe i saw the whole thing
    They are running A12z on mac pro !!!

    I am just thinking all the open source tool we have to recompile, will the st-link work on ARM mac or matlab or lab view ?

    I am so fucked
  • 3
    @hardfault if it's written in something portable like C/C++/Java/Python etc., I don't see why not. I run a lot of stuff on RPi 4.

    Matlab and LabVIEW might be harder given that they're not OSS, but eh, dunno, let's see.
  • 2
    @dontbeevil except Apple has a locked ecosystem and very high performance ARM chips, both of which MS doesn't really have right now. With MS you always have a choice. So if there's one company that can make this transition successfully it's Apple, and even they might fail.

    Also MS relies a lot more on legacy software, while Apple has the massive bank of iOS and iPadOS software to draw from, so a lot of work is already done.
  • 3
    @RememberMe it’s about waiting on the vendors to support
    Also recompilation is step one, all the optimizations that was done for intel chip now have to done for ARM

    Rpi is not a good example, it’s a piping hot mess. For a professional computer software even thing will be needed to be fine tuned
  • 3
    @hardfault right, but it's usable, and that's without any mainstream ARM device that's not a locked thing like a phone or tablet.
  • 0
    @dontbeevil code execution tests require taking EVERYTHING into account, such as exact timings, hardware and bus type differences, drivers, OS, background shit, etc. because modern OSes don't fork your process to bare-metal anymore.
Add Comment