Could Apple be going through yet another architecture change?

[KawaiiAurora]

Hey everybody,

Today, while I was browsing MR, I came across a front page that said that Apple's probably develops their own chips. I wonder, will they jump to an ARM architecture or keep the same x86 to maintain compatibility?

Will this spark yet another 68k/PPC and PPC/x86 transition period? Perhaps going from one of the most-widespread architectures to something more obscure wouldn't be a good idea but what do you think? Will Intel Macs end up like our trusty old PPC Macs, that is supported by devoted community members?

I think there's one thing we can all agree about: Apple's the company that changed the most CPU architectures in their products lol. 6502->68k->PPC->x86(_64)->??? (ARM, custom x86 or some new concoction from Apple?)

Article source: https://www.macrumors.com/2018/04/02/apple-custom-mac-chips-2020/

 

[AL1630]

If this happens, eventually people will come up with something like what we have here, with a small developer community and users who try to keep their computers relevant, just because of the massive number of Intel Macs in the wild today. Lots of people will want to use their $1k+ macs as long as possible.

 

[Dronecatcher]

Think it's more about saving $ using their own chips - more profit.

 

[Eriamjh1138@DAN]

I predicted this as an extremely likely possibility in 2013 (around IPhone 5 time). There could be dual architecture macs or pure Arm macs. Using Rosetta emulation tech, we could run old apps on ARM.

Intel machines may end up with their own sub-forum like PPC, but they can at least run Windows if they want to stay updated.

Bottom line, Apple is big enough to play with the big boys, now. In the age of the internet, the CPU is not as relevant. It’s the apps.

PS: it’s too bad Apple doesn’t open source Rosetta tech for PPC on... well, anything. Just to keep those apps alive. Someday even an emulated PPC could be faster than it was natively on the fastest G5.

 

[Gamer9430]

I’ve said this many times and I’ll say it again... I simply don’t see ARM working for an actual computer. Sure, the A11 Bionic benchmark the same score as the current low-end MBPs (which is really sad on Intel’s part), but that doesn’t mean the A11 can run several Windows 10 VMs, Photoshop CC 2018, and AAA games. The A11 and other ARM CPUs are designed for mobile usage, to make pretty AR selfies in Snapchat. Also, if Apple doesn’t encode the future ARM CPUs with x86 compatibility, a lot of people are going to leave. PPC adoption was slow because no x86 compatibility, so everything had to be encoded special for PPC and no one could run Windows natively. Now, these days, yes, Windows does run on ARM, but it’s not great. If Apple were to switch to ARM tomorrow, it simply would not work and it would alienate a lot of their clientele. In 2 years, it might be a completely different story. Apple would need to do serious work to turn their current ARM CPUs into full blown Professional Performance class CPUs, needing to double the power for the MacBook and MacBook Pro, iMac and Mac Mini, and triple the power for the Mac Pro and other high end, top-tier products.

 

[Eriamjh1138@DAN]

Apple makes so much money from iPhones and iOS that the ones who switch because of the loss of intel processors will not compare to the money Apple makes and saves from the switch.

Besides, the same thing was said about Intel. Turns out the first Intel macs ran those Apple optimized apps faster than PPC could at slower overall speeds. They even emulated PPC pretty nicely, too.

This has all happened before. It will all happen again.

 

[AL1630]

I wonder how long it'll be before Intel Macs are pretty much only for enthusiasts like us. It took less than 10 years for PPC...

 

[Project Alice]

Hey everybody,

Today, while I was browsing MR, I came across a front page that said that Apple's probably develops their own chips. I wonder, will they jump to an ARM architecture or keep the same x86 to maintain compatibility?

Will this spark yet another 68k/PPC and PPC/x86 transition period? Perhaps going from one of the most-widespread architectures to something more obscure wouldn't be a good idea but what do you think? Will Intel Macs end up like our trusty old PPC Macs, that is supported by devoted community members?

I think there's one thing we can all agree about: Apple's the company that changed the most CPU architectures in their products lol. 6502->68k->PPC->x86(_64)->??? (ARM, custom x86 or some new concoction from Apple?)

Article source: https://www.macrumors.com/2018/04/02/apple-custom-mac-chips-2020/

I would be an advocate for custom x86.. I think making ARM systems are not going to fly with the community. especially for pro users. It would seem like a big step backwards to me. Nobody gives a F*** about uniting iOS and Mac OS. In fact I think its a terrible idea. Did nobody learn from windows 8? Okay maybe the girl that sits in starbucks with her iphone and macbook and drives a 2012 honda civic but cmon. None of us actual people who have been loyal to apple want that.

All the work they did optimizing Mac OS, removing PPC binaries, next removing even 32bit binaries, all to make the OS more fluid and simple with a small foot print. But hey lets use a mobile CPU for everything and add a whole other architecture to the OS and majority of apps to undo all the work we just did optimizing it and have another 2006 fiasco. It works like it is. Build your own CPUs apple, great. But for the love of god just leave it x86.

 

[Kcetech1]

my thought becomes, then will Apple abandon thunderbolt as well? Thunderbolt is basically Intel LightPeak, and they own almost all the IP on it. no intel CPU = no on chip thunderbolt. Apple would then need to buy third party TB interfaces and then add all that circuitry, and headaches to their system. assuming intel even bothers to make the drivers for anything other than x86/x64

as for a custom x86 chip of their own, will not happen. intel has the x86 architecture all locked up, AMD has x64 and licences x64 to intel and will not outside licence, the remaining x86 licencing package that VIA has is so obsolete they would be going back to a Pentium 4.

 

[AL1630]

I could see putting ARM chips in a MacBook or base iMac since it doesn't need much horsepower, but unless Apple rolls out something comparable to an Intel/AMD chip, which is certainly something they might be able to pull off it couldn't take off in a 'Pro' environment (If Apple cares about that market anymore). I think it's most likely that they roll it out in iMac/Mac Pro slowly as it improves, possibly even over a few years.

 

[Eriamjh1138@DAN]

If Apple wants pros to use ARM Processors, then it will have to make professional apps that run on those processors. Or convince others to do so. They know this.

 

[AphoticD]

It looks likely the MacBook will be ARM based within the next year or two. Apple will ensure it is a smooth (enough) process as with PowerPC>i386 in ‘06.

The performance situation is the reverse of the previous cases though, where the Mac was suffering from restricted hardware, which Apple solved by architecture shifting. Apple now have all of the advantages to selectively move to lower cost hardware and still depend on Intel for performance where it’s needed.

The 64-bit app cut-off happening in 10.14 will ensure the next generation of Mac apps will safely compile for both x86_64 and arm64 with all of the currently implemented optimizations in clang/llvm which has been building Swift and Obj-C code to run on both platforms for the past few years already. Apps will simply be “Universal” again and the Rosetta tech (QuickTransit) will resurface for high performance x86_64 translation on arm64.

I imagine it will be a long transition period for Pro level machines, driven by performance over price. Apple have never had a problem charging (gouging) the Pro market for high performance Macs and will continue to do so.

Once ARM outperforms x86 then it will be a simple and safe transition for Apple to quietly cut Intel out of the hardware and the customer/end-user will hardly need to recognise the underpinnings.

 

[0248294]

It'll be interesting to see what happens to Bootcamp on the Mac if they do switch. Will they kill it off, or will Microsoft and Apple work out a deal to get ARM64 Windows 10 running on Macs? ARM64 Windows 10 and Server 2016 already exist, so it'd probably be trivial to get those running on an ARM-based Mac.

 

[weckart]

ARM64 Windows 10 and Server 2016 already exist, so it'd probably be trivial to get those running on an ARM-based Mac.

The thing is, Apple customises its chips almost from the ground up. The only thing Apple's ARM offerings share with other ARM chips is the IP. While I can see applications from Microsoft (a very cut down Office runs on iOS), I just don't see Redmond reinventing the wheel for yet another platform.

 

[AphoticD]

If something like the MacBook 12" houses the next generation architecture, how many owners of these machines would rely on Bootcamp I wonder? Would it be fair to say that the user in need of a dual boot Mac is likely to skip the consumer line?

Just talking from my own experience, I've only used Bootcamp for gaming, which is a rare occurrence. Even so, a consumer level Mac portable isn't going to fulfill many gaming requirements for most.

 

[pl1984]

If this happens, eventually people will come up with something like what we have here, with a small developer community and users who try to keep their computers relevant, just because of the massive number of Intel Macs in the wild today. Lots of people will want to use their $1k+ macs as long as possible.

It's easy to keep an old, Intel based Macintosh relevant: Just install any other x86 OS on it. Sadly the only obsolescence for Intel based Macs comes from trying to run OS X, everything else works fine.

 

[Hastings101]

Little bit different since x86 chips are so powerful and you can install any alternative OS on them. No one will need a "10.14 Fox" kind of thing in 2024 or whenever if you can just install Windows. But yeah, I see a transition as being possible one day.

 

[AphoticD]

Unfortunately for any of the OS X / macOS team who might have an opinion about portability and sustained longevity, Apple's "free" OS model is designed to simply sell hardware. It's in their best interest to make their 5year+ hardware obsolete. They aren't competing for the best operating system to convert more vanilla boxes to macOS.

Given that most of the Macs built in 2009 can still run the current macOS, it's the contrived limitations that drive new sales. This is where a new architecture shift will amplify and boost sales to existing Mac customers. A new arch, means new software which becomes exclusive to the new arch, which drives sales to existing customers who are in fear of missing out.

 

[Eriamjh1138@DAN]

I never bought about the whole old Macs running the latest OS being a sales blocker. It means hardware is stagnating. If ARM opens up the processor wars again, I welcome it.

 

[mmphosis]

The dominant "platform" is currently the cell phone, and they use ARM.

I've abandoned new Apple platforms and the Windows/Office platforms and run Linux on aging, compromised Intel CPUs. I say compromised because Spectre, meltdown, who knows what else and the ten year old backdoor called Intel Management Engine (ME). Apparently, Intel's latest i9 is mediocre: https://news.ycombinator.com/item?id=16743983 Maximum 16Gb RAM?

I am not a "Pro" user, I don't need video editing abilities, and I am not a gamer that requires the latest PC gaming hardware with graphics cards et al.

Going forward, I am looking at:

• a fanless Intel-based mini-PC with way more than 16Gb RAM running Linux, but I decided to delay this as I don't like Intel chips
• a new MacBook with Intel chip, but again I don't like Intel, and new Apple hardware means joining the Apple ecosystem, giving Apple my credit card, paying for proprietary apps, paying to publish iOS software
• wait until 2020 for a MacBook with Apple ARM chip, but again I don't wish to join the new Apple ecosystem
• a Power or some other CPU (not ARM, not Intel) running Linux, these are better but expensive

I like the idea of a MacBook with an Apple ARM chip. I heard of some people currently using a keyboard with an iPad Pro which sounds appealing to me for my limited purposes, but paying the Store to run SSH? Maybe in 2020, Apple should offer an option of high-end Intel chip or Apple ARM chip, but as others have said it may make more economic sense to switch exclusively to their own ARM chips.

I have two old Intel-based MacBooks and they are no longer working. All of the PowerPC Macs happily keep running, albeit a little slow, but they are running.

 

[pl1984]

ARM was used in desktop systems long before smartphones ever existed. It was considered a solid processor for its time. I see no reason why the architecture can't be used in today's desktops.

 

[Nermal]

ARM was used in desktop systems long before smartphones ever existed. It was considered a solid processor for its time. I see no reason why the architecture can't be used in today's desktops.

Indeed; in the late 80s/early 90s you could get ARM-based Unix (RISC iX) workstations. I wonder how fast a modern ARM could run if you effectively turn off all the power management...

 

[G4PPC]

I think it'll be great only if it drives down the price of Intel Macs even by slight amount cause they cost so much as it is. My Fear is that they will eventually dump Intel's completely, and then a few more years down the road some company is going invent the next level 128bit PC.

 

[DearthnVader]

At some point we could see the lowend Air, Mini, iMac, and Macbook go Arm64. However the limitation is the Arm can't emulate x86_64 very well. With the 68k>PPC switch PPC could emulate 68k at near native speed, and with PPC>x86, x86 could emulate PPC at acceptable speeds.

As it stand, Arm just isn't able to emulate x86 at acceptable speeds, but that doesn't mean it won't be able to in a few years if Apple wants to push it in design and optimizations.

Clock for clock Arm is not faster than x86, even with apps that are compiled to run native on Arm. With the switch to PPC/x86, those CPUs were faster with native compiled apps.

 

[mmphosis]

I think it'll be great only if it drives down the price of Intel Macs even by slight amount cause they cost so much as it is. My Fear is that they will eventually dump Intel's completely, and then a few more years down the road some company is going invent the next level 128bit PC.

"There is not a large performance increase by switching to 64 bit."
The Ultimate Comparison Between 32 bit and 64 bit (wisecleaner.com)

I am not sure that 128-bit datapaths are worth it. On the other hand, it sounds like later PowerPC Macs (G4 and up) have AltiVec that can work on 128-bit vector registers for floating point operations. I find the whole 1 4 8 16 24 32 36 64 128 bit thing to get a bit blurry as different architectures can handle (or not handle) varying amounts of bit widths depending on what layer we're talking about, and what the application is.

“Dennard Scaling and Moore’s Law are dead, Now what?”
Fifty (or Sixty) Years of Processor Development…for This? (eejournal.com)

For the price, I am looking at fanless Intel mini-PCs which I can load up with SSD and big amounts of RAM for less than the price of an Intel Mac with the caveat that I am using Linux and not macos. Mac OS X and Linux each have their pros and cons.

setxkbmap -option "" -option ctrl:swap_lalt_lctl​

Yes, I miss Mac OS X, but that's where using my PowerPC Macs come in.