Rumors of Apple going pure ARM across their product stack, why on Earth did they leave POWER PC then?

https://www.bloomberg.com/news/articles/2020-04-23/apple-aims-to-sell-macs-with-its-own-chips-starting-in-2021

Apple was literally on PowerPC for the longest time, but after some kerfuffles with IBM, they decided to move Apple’s entire Eco System onto x86 with Intel. Now that Intel is having some production trouble and getting trounced by AMD, now they want to move to ARM and migrate their eco system onto ARM along with do their own CPU development.

Does this make sense for them in the long term? They literally went from RISC -> CISC -> RISC again. WTF? I feel sorry for their dev’s.

At least Apple is back on RISC where they belong, but this seems like quite a big middle finger to their devs who have to go through a ton of headaches AGAIN.

  1. They left powerPC because there was nothing that they could put in a laptop, and that was 15 years ago (times change). It was nothing to do with abandoning RISC, it was the simple fact that intel did an amazing job with the original Core series power management, and Motorola/IBM built the G5 for servers, and gave no fucks for the laptop market. It simply would not fit in the required thermal envelope. It was pure pragmatism; PPC would not do the job required.

  2. Back then they were almost broke and not one of the most valuable companies in the world. This time around they have the funds to go full in-house CPU design and finally become the master of their own destiny and not beholden to ANY cpu manufacturer.

Expect them to go in-house GPU as well eventually, for exactly the same reasons. Because shifting from one manufacturer to another just puts them at the mercy of somebody else’s supply chain, product stack (which is designed to sell to the larger commodity market), etc.

It’s nothing to do with RISC -> CISC -> RISC ideology. It’s getting the products tailored to the exact specifications they want, WHEN they want them.

Right now they’re stuck with whatever intel shits out, which currently isn’t good. “Switch to AMD!” you say? Well if they do that, they’re in the same boat when AMD inevitably have a bad run. Just like they were with Motorola back in the 68k days, and just like they were with IBM/Motorola in the PPC days. Furthermore, there’s no room for them to differentiate themselves from the mainstream PC market if they’re shipping somebody else’s commodity hardware.

No, fuck that, this is the move they need to make, from a purely company risk mitigation perspective.

Apple can and will ship much better stuff than we’re getting at the moment.

As far as their devs go… you can guarantee they have had macOS running internally on ARM since the original iPhone. iOS was a branch of OS X and back in the day OS X was developed in parallel with x86/PPC from the outset. 100% guarantee they’ve been doing the same with ARM for a decade as a contingency plan.

Most stuff isn’t written in assembly these days, they already have ARM code, and they put massive resources into LLVM. Apple have been prepping for this for a decade or more, guaranteed.

14 Likes

If that’s the case, wouldn’t going RISC-V be a more sensible path than ARM?

1 Like

Why?

They already have 90% plus of their products using ARM, already have an ARM based CPU development team and already have heavily customsied-to-their-requirements ARM CPUs, a well tested ARM compiler toolchain, libraries, etc.

RISC-V is cool, but it is, as yet unproven and would result in Apple starting from scratch with any custom stuff they’ve been working on and be totally incompatible with the rest of their product stack. They’d be throwing the past 10 years of in-house CPU/toolchain development out the window.

Using RISC-V would be an entirely ideological choice (“wow, it’s cool!”), not a practical one, and Apple aren’t making these decisions for ideology reasons really. It’s entirely business costs, performance and risk mitigation.

Right now RISC-V is not battle tested, the resources, documentation and experience is far less than for ARM. Its a much larger risk and just not necessary; they already have their own high performance ARM based architecture.

If you think switching everything to ARM will be hell for their developers, it’s nothing compared to switching the Mac (and potentially iOS as well!) to RISC-V.

You say “at least Apple are back on RISC where they belong” like this matters. It doesn’t. What matters is how performant, reliable, customisable and risky (not in the RISC way) the choice is,.

Whether the choice ends up CISC or RISC doesn’t really matter. And intel CISC based parts have been RISC internally (in some way best of both worlds - CISC code density for better cache hits and mostly RISC performance) since the Pentium Pro (at least) anyhow…

edit:
Also… I guarantee you they will maintain the x86 platform, and track intel cpu developments internally, irrespective of what they actually ship. Just in case. They’ve been burned too many times before by reliance on somebody’s parts to not maintain the agility needed to shift as required. I doubt that will change when they’re building their own.

Fortunately NextStep was originally well architected to enable this sort of thing and macOS is essentially NextStep evolved by 30 years…

7 Likes

Also, it’s been 15-20 years since they ditched PPC. So long ago it practically doesn’t matter it ever happened, except for historians.

3 Likes

Well, their mobile SoCs are top of the pack right now, and they could use it for lower end devices and maximize profit. As long as x86 chips are quicker they won’t go full ARM, imo. By starting with lower end laptops, the buyers of that laptop will be testers of the “experiment” and fund Apple’s groundwork dev for Mac OS on ARM. Given the amount of resources Apple has, performance parity with x86 could be possible one day and at that point they can completely eliminate Intel’s involvement on the cpu’s side.

I give it 5 years before their entire lineup is on ARM with the possible exception of the Mac Pro.

Heavy processing will be shifted out to ASICs (like the afterburner card) and/or GPUs. This is why they are pushing compute APIs like Metal, etc. for the heavy lifting work - so that metal itself (for example) can be optimised for whatever hardware is in the box and your code doesn’t care.

Once their first ARM macOS computer is released (probably a new 12" Macbook, to start with the less performance/more battery life sensitive users) the next few years will come like a tidal wave.

Their ARM based processors in the iPad and iPhone are genuinely amazing in terms of performance per watt. If they can scale those up to the 100-200 watt power consumption range (and I’m willing to bet they can), it will cover their entire lineup pretty easily.

1 Like

However, they’ll lose the Windows dual booting feature which might hurt em in some places.

True, but I don’t think even Microsoft want to support Windows (client) any more. They are, but its not the end game. The money isn’t in Windows any more. It’s just a lever for 365 adoption.

They’re pushing people just as hard as anyone else to slim devices that run apps via Azure / 365 because that’s where the money is. Microsoft don’t give a shit if you’re connecting to 365 via a Windows machine, an iPad or a Mac, you’re still paying them for a license.

This is why they are genuinely making efforts to be cross-platform these days. Not because they suddenly became good guys; its because it opens up more of the market to consumption of their cloud services.

Sure, some people won’t like that they can no longer run windows on a Mac, but that’s an edge case of a fringe market and they just don’t matter. Even I as a formerly heavy VMWare Fusion user don’t really care any more.

1 Like

Yeah I agree.
Also, moving to ARM could also open Mac OS to iOS apps, which could be a value add too.

There’s already an API for that, and some of the currently shipping macOS inbuilt apps are ports from iOS (e.g. the news app).

In terms of supporting Windows on Mac/ARM I wouldn’t be at all surprised if it worked purely as a remote cloud login system. Even Microsoft are headed that way as fast as they can with Microsoft365 - no more local accounts and ALL data is hoovered up by the cloud.

1 Like

sitting here being a ppc enthuse ppc was HOT and the gains of ppc vs x86 at the time were better moving to arm allows better hop to their own pure custom risc system

open power is nearly risc-v in instructions and arm is nearing this as well

but still today power is not for the everyday user atm there is nothing again to gain using x86 or power so the gains of arms design is more for their environment allows a soc style system which again locks out others and the dev work might push arm tb3/usb4 things so im happy

Because they are Apple.
They don’t need an excuse.
Because they can.

And as long as people keep throwing money at them …
They don’t care.

1 Like

well their customers want new wisbangs and wislets so they give em a carrot here and ther

when they suck people look to hackintosh which is thriving atm the ecosystem is why people want it

If rumors are true and they start the transition in 2021 I would be shocked if the switch isn’t done by 2023. Remember, PowerPC to Intel was done in 14 months from announcement to completion.

Hackintosh is gonna be dead in a couple years anyway. When Apple makes T2 a requirement, it’s game over.

1 Like

they already by passed it in hackintosh so i doubt… i will see more if they actually make boot or something built in to a t2.5 chip or the like, so im not worried bout hackintosh im more worried bout using stuff on the systems WITH a t2 chip

I don’t think Apple made it a strict requirement in the OS yet. But when that happens, it’s gonna be deep in the system.

1 Like

well i hope new mac pros go power 10 HUE HUE HUE

1 Like

There are higher positions within the company who will not allow that to happen. It would have died years ago if it weren’t for a small group that advocates for its existence and releases the necessary patches to allow it to continue.