Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
....I'd like to see non-Apple OS X software run from that first day or first week... stuff like the Adobe Suite, MS Office, Pixelmator, etc down to smaller, popular apps like Skype, Handbrake and so on...I just don't think so...

Those don't all run properly even after a minor OS X upgrade. Sometimes it takes months to get developers to fix apps and plugins -- and that on the exact same CPU.

Anybody who has done major software development knows it's more than a simple recompile. A major commercial app requires extensive stress testing, compatibility testing, performance testing and regression testing. You can't just recompile and ship it. During that testing you often find little glitches involving unintended platform dependencies, exception handling, compiler optimization anomalies, etc. Major apps often undergo profile-based optimization after the final linking phase, sometimes called basic block optimization. This is highly tuned for the processor family's cache behavior.

If you actively support different compiled binaries of a major product on different CPUs, these issues get ironed out. But -- if your major app has existed for several years on a single OS & CPU platform, dependencies creep in. The first attempt to recompile and get it running on a different CPU (even if the underlying OS is the same) is often a big job.
 
Unless the ARM CPU core gets a HUGE leap up in performance, the MacBook Air, MacBook Pro, iMac and Mac Pro models will continue to use Intel CPU's. Especially now that the Broadwell CPU's have gotten very efficient in power consumption.
 
And I just can't get how this vision of ARM-based Airs coexisting with Intel-based Macs (otherwise) is not an explicit shot of the dreaded "fragmentation". There would almost certainly be OS X software that would not run on the ARM-based machines. It would be "For OS X" but "on Intel-based Macs Only". Maybe vice-versa too? Can we actually picture Apple doing that?

Yes.

Because they already do. armv7 iOS apps don't run on armv6 devices (such as the iPhone 3G). arm64 apps don't run on the iPhone 5C and older. The iOS App Store on the device shows you only the compatible apps.

If they can do it on the iOS app store, and sell Billions, they can easily do the same on the Mac App Store.
 
Good to hear. When Apple drops Intel, I will most likely have to drop Macs. That would break compatibility with so many OS X apps (and developers would take a great amount of time to recode their apps, assuming they did at all) and it would break Windows compatibility. As much as I like OS X, I occasionally need to use Windows.

Yes and no. It would break compatibility with Windows, but if Windows moves to ARM as as been speculated, you're going to have to deal with that breakage anyway.

As for compatibility with OS X apps, for 99% of apps out there, it would be as simple as checking an additional CPU architecture and recompiling. Here are a few reasons why an Intel to ARM transition won't be like the PPC to Intel transition:

  • Almost everybody builds with Xcode these days, so there's no risk of the nightmare scenario we had with the Intel transition, where Motorola bought CodeWarrior and sold off their x86 toolchain a few months before the transition, forcing most developers to change IDEs while porting their code.
  • Apple uses a little-endian ARM architecture, so you don't have a difference in byte order.
  • Fewer apps use Intel assembly language code directly, relying instead on Accelerate framework and other similar libraries that already work cross-platform. (Photoshop and audio plug-ins are the exceptions that prove the rule, of course.)


With that said, I'm pretty sure the pundits are still wrong. Although I like the ARM architecture in terms of power-per-watt, and there's a decent case to be made for using them in low-end laptops (e.g. the MacBook Air), in terms of absolute speed, they don't pass muster. They would need to make the A8X four times faster if they wanted it to compete with the speed of the Retina MacBook Pro, much less the Mac Pro.

And then, there's the evil voice in my head that says Apple needs to create a mechanism for transparently migrating application state between x86-64 and ARM64 versions of an application in real time, allowing for ultra-low-power operation, kicking in the real CPU when it is needed, then cycling back, much like the way they swap between the integrated and dedicated GPUs... but this would require a very different compiler design that builds binaries for both architectures at the same time, ensuring that all the executable code is at the same offsets, that all the data structures are the same size, etc. (Yes, this would be nuts.)


So if all these iOS apps are compiled for three different binarily-incompatible CPUs, how come there's only a single app download from the iTunes app store? How does the one compatible version (out of three) find its way to my iPad?

It doesn't. Apple uses multi-architecture binaries that contains code for all supported architectures in a single file, so your 64-bit iPad gets both the code that it runs and the code that older hardware runs. Your iPad then chooses which slice to run based on its capabilities.


The answer is the iOS CPUs are all binary compatible, and there is a single executable download. Saying iOS runs in three different types of CPUs is like saying OS X runs on different Intel CPUs. The microarchitectural features may differ, but they are share essentially the same compatible binary instruction set.

Not true. The first iOS devices used 32-bit CPUs. Current iOS devices use 64-bit CPUs. So most store apps have both a 32-bit ARM slice and a 64-bit ARM slice. And when you compile for the simulator, you're compiling i386 code. There are your three architectures. (And you can also add an armv6 slice for older iOS hardware, but I'd probably call that a microarchitecture difference.)



This actually illustrates why it would be difficult. The xcode iOS simulator runs iOS apps on OS X *far* slower than the native iOS CPU. This is despite the Mac CPU being much faster than any mobile device.

Yes and no. The simulator can be slower for graphics-heavy tasks involving OpenGL because of all the extra abstraction layers (you're essentially running a window server on top of another window server). However, for general computation, the simulator is much, much faster than any iOS device, because the CPU is so much faster.
 
Unless the ARM CPU core gets a HUGE leap up in performance, the MacBook Air, MacBook Pro, iMac and Mac Pro models will continue to use Intel CPU's. Especially now that the Broadwell CPU's have gotten very efficient in power consumption.

Single threaded performance from Intel only increases about 5% on average. ARM is increasing leaps and bounds every year, and by 10 years if not earlier, should reach parity with today's Intels chips. There will always be a need for Intel, but increasingly, ARM will suffice.
 
The day Apple buys NX from Siemens and makes it available to all their users, conceivably at some kind of consumer friendly price point where people will actually buy it rather than the $12,000 per seat it costs now, and thus takes responsibility of providing Mac users with an a functional CAD package that runs natively on its own OS regardless of whatever platform it adopts, then great, switch to ARM.

Until then, outright ditching Intel, instead of just adding a line of low cost toys for the kids, would pull the plug on nearly every Professional user in the design and engineering and architecture fields of every industry. AEC software for the MAC is still BARELY adequate. Better not yank the rug out from under it without something solid in place, or you'll have burned that bridge for the last time.
 
The answer is the iOS CPUs are all binary compatible, ...

Nope.

Apple has shipped 4 different ARM CPUs with 4 different instruction sets: armv6, armv7, armv7s, and arm64. The newer 32-bit ISAs are incompatible with the older CPUs (such as in the iPhone 3G and 1st gen iPod Touch). And the 32-bit CPUs can't run the completely different arm64 ISA.

Most app customers don't even notice, because Xcode is good at auto-magically building and packing application executables for completely different CPUs into one iOS app download. But that's also why some simple apps seem so huge to download from the App store. You are really usually downloading a bundle of multiple iOS apps for different CPUs.

Smart developers know that Xcode is doing this, and fully test their multiple-apps-within-one-app on multiple iOS devices. Clueless developers don't, and this in one reason why some apps crash that the developer "claimed" they tested (on a completely different ARM CPU than the customer was using).
 
Unless the ARM CPU core gets a HUGE leap up in performance...

Not needed.

Most potential bottom-end MBA customers would take slightly lower performance if the MBA was priced more competitively and had better battery life. An ARM CPU might allow a slightly lower price on the bottom-end Macs, enough to entice significantly more marginal potential buyers. Especially in the edu market.

This won't affect the complainers here much at all. Apple doesn't need to use an ARM CPU in their Business class laptops. Enterprises (and the fanboys here) don't usually buy cheapo (for Apple) bottom-end MacBooks.
 
Apple could have done this years ago on Intel chips. They could make it available now on Intel Macs. There's an emulator, no simulator:rolleyes: that already runs iOS software on OS X for the programmers coding iOS apps.

Nope. The iOS Simulator is slower at many things, more expensive (total system cost), and uses a ton more power than an actual ARM device. As well as running the wrong instruction set for even just fully testing iOS apps.
 
Nope.

Apple has shipped 4 different ARM CPUs with 4 different instruction sets: armv6, armv7, armv7s, and arm64. The newer 32-bit ISAs are incompatible with the older CPUs (such as in the iPhone 3G and 1st gen iPod Touch). And the 32-bit CPUs can't run the completely different arm64 ISA.

OK, thanks. I was thinking about this quote ARM architecture reference: "The processor implements a binary compatible subset of the instruction set and features provided by other Cortex-M profile processors. You can move software, including system level software, from the Cortex-M3 processor to other Cortex-M profile processors."

You are right it's possible to target incompatible features in specific ARM CPUs, and if the developer chooses to do that he'd need to build a universal binary.

Besides this I don't see how the entire Mac lineup could move to ARM. The top Mac Pro has a 12-core Xeon, and the next one will almost certainly have an 18-core Xeon E5-2687W v3. There is no ARM CPU remotely close to that, which in turn implies any Mac move to ARM would produce a fragmented product line.
 
68k > PowerPC > Intel > ARM

to be honest there wasn't that much software for macs until they moved to Intel. at least nothing that was dependent on a thousand open and closed source libraries. at the time most of the software written was entirely written in house. there was no real sharing of libraries, because there wasn't much of an internet either. things were much smaller and simpler in those days. i'm not saying it isn't possible, just that it is not an apples to apples comparison anymore. it would take tremendously more effort and it would alienate the heaps of people Macs attracted by moving to Intel. supporting windows in some reasonably efficient way is just something you have to do to be a compelling option on the desktop for most people.
 
Ah

ARM Macs better come soon because we need cheaper MacBook Airs that are around $500-$700. Broadwell-U CPUs from Intel are around $250-$400, and that's not going to cut it for a $500 Macbook Air price point. The CPUs need to be around $50-$100 to hit that price. We could also use a $300 Mac Mini as a small office server, in which an ARM cpu would work well.

intel should be scared. Apple is already at speed parity for the low-end CPUs, and by the time the A10x rolls around, they'll be as fast as Intel's top CPUs.

Apple's ARM A8x cpus are as in the same ballpark speeds as any 10-watt Broadwell CPU. And, given a higher power budget, they could easily go 2-3x faster, putting them in ludicrous speeds. A fully configured 10-core version for a Mac Mini server could even be 10x faster, and wouldn't cost any more than the A8x if they took out the GPU.

Just ship ARM, and have all the vendors just recompile their OS X apps for ARM. No one needs Windows compatibility (I don't know any Mac owner that uses Windows) and the .25% of Mac users that do can go buy an Intel Mac.

Apple typically does not play the low end of the market. They typically play the high end of the market. $500 computers? You're dreaming. This is Apple we're talking about, not Charlie's discount box shop.
 
ARM Macs better come soon because we need cheaper MacBook Airs that are around $500-$700. Broadwell-U CPUs from Intel are around $250-$400, and that's not going to cut it for a $500 Macbook Air price point. The CPUs need to be around $50-$100 to hit that price. We could also use a $300 Mac Mini as a small office server, in which an ARM cpu would work well.

intel should be scared. Apple is already at speed parity for the low-end CPUs, and by the time the A10x rolls around, they'll be as fast as Intel's top CPUs.

Apple's ARM A8x cpus are as in the same ballpark speeds as any 10-watt Broadwell CPU. And, given a higher power budget, they could easily go 2-3x faster, putting them in ludicrous speeds. A fully configured 10-core version for a Mac Mini server could even be 10x faster, and wouldn't cost any more than the A8x if they took out the GPU.

Just ship ARM, and have all the vendors just recompile their OS X apps for ARM. No one needs Windows compatibility (I don't know any Mac owner that uses Windows) and the .25% of Mac users that do can go buy an Intel Mac.

Pretty much everything you have said here is bollocks...
 
Apple typically does not play the low end of the market. They typically play the high end of the market. $500 computers? You're dreaming. This is Apple we're talking about, not Charlie's discount box shop.

The pundits all said something very similar just before the ARM iPad announcement. Wintel tablets were $1000 plus, so the iPad must be priced around $999 or more, because Apple simply does not play the low end of the market.

Surprise!!! $499 instead of $999.

Really put the hurt on all the cheap netbook vendors the following year.
 
Pretty much everything you have said here is bollocks...

I am not so sure about Macs on ARM: there are plenty of questions lingering, the biggest is a need to move a relatively small OS X app library to a completely new CPU, which may be lethal for it and have a very negative impact on growing Mac sales. Not so worried about Windows compatibility, but Windows on ARM will be a new joke from MS.

However, I think that evolving iPad line to a more functional, effective level of subnotebooks is fully possible with with the adaptation of following things: precision input (cursor and trackpad), more user-friendly file structure (already iCloud drive), better multitasking (wasn't it rumored)? Also, sensitivity levels would be good.

Different from MS is that Apple leverages in this case its already existing mobile OS to become a professional mobile OS, while MS tried to make a desktop OS into a mobile and so far failed.

The advantages of iOS moving to Pro line are numerous: battery efficiency is inherent, all iOS apps are staying compatible, moreover, iOS gets multitasking, better input precision, newer opportunities to grow even for existing hardware (though Apple should think of newer iPads). The trackpad can support TouchID, if needed.

In this case, I think that the rumored12 inch device is not Mac on ARM. its iPad Pro on ARM, of course, and it doesn't have hinges. iPad with Keyboard on hinges contradicts iPad concept so its no no. Instead, the very hollow keyboard with trackpad connect to iPad Pro via magnets in two positions: either as a normal keyboard or as a stand. It also provides closed shell for transportation and storage.

So, there is no keyboard built in literally for iPad Pro, but it does gain a new keyboard cover.
 
There is a part of me that says no way in hell would they switch.

Another part says maybe it it's not likely for a good 2-3 years more. ARM needs to get more powerful, especially for anything more than a light use machine like a MacBook Air.

I love the idea of having one underlying system, which might make things flow more continuous between styles. Developers might even be moved to bundle apps so you buy once and get apps for computer, iPad and iPhone rather than now where it's two sets

----------

ARM Macs better come soon because we need cheaper MacBook Airs that are around $500-$700.

You think Apple would drop prices, especially that much. That's so cute
 
The pundits all said something very similar just before the ARM iPad announcement. Wintel tablets were $1000 plus, so the iPad must be priced around $999 or more, because Apple simply does not play the low end of the market.

Surprise!!! $499 instead of $999.

Really put the hurt on all the cheap netbook vendors the following year.

Yeah, but the components of the iPad were also much cheaper. They didn't use full-fledged PC parts like intel processors, so Apple didn't need to market it at PC-like pricing.

Everything we have heard about the redesigned MBA so far still point to it being PC in both form and essence. So I don't see it being so much cheaper. I think Apple aims to further differentiate the air from the pro, as well as appeal to users who particularly prize mobility and don't mind dealing with all the tradeoffs.
 
Single threaded performance from Intel only increases about 5% on average. ARM is increasing leaps and bounds every year, and by 10 years if not earlier, should reach parity with today's Intels chips. There will always be a need for Intel, but increasingly, ARM will suffice.

When you start so low, it is easier to have huge growth.
 
Wine still worked with PPC, with Qemu. Also, VMs will still work for those OSs. Xen and KVM, for instance, have already been ported to ARM. If you work with 'free software', then you have the least to worry about. Go look at the Debian ports. Already most of the open-source software out there has been compiled for ARM, even chromium works on ARM and has worked on the raspberry pi for a long time, for instance. So, you'll be able to use all the open-source software you want.

I don't want to have to emulate the Intel architecture for Wine. Debian isn't bad on ARM, but ARM is a "tier 2" architecture for FreeBSD, which is bothersome. For both OSs, I'll bet there will be something I run into that won't work on ARM.

----------

When you start so low, it is easier to have huge growth.

Yes, exactly. I don't care how quickly the chip inside my computer is being improved upon if it's still weak.

----------

Yes and no. It would break compatibility with Windows, but if Windows moves to ARM as as been speculated, you're going to have to deal with that breakage anyway.

There are far too many "Wintel" programs out there. Nobody really cares about running Windows, just the programs that we're forced to use Windows to run, so Windows itself being ported isn't that big of a deal (except that more new things would be made for Windows ARM).
 
It better be strong or my love for Macs won't be.

I agree I'd never buy an ARM based Mac. But would Apple abandon their Mac Pro line? I can't imagine am MP running ARM.

Without the ability to run a virtual mine the Mac would loose most of it's value for me. I'd just go with IOS for mobile and Linux real work.
 
However my favorite line in all such posts is:

presumably "no one" because...

further supported(?) by a quantitive reference...

A survey of just a few friends is not necessarily representative of the whole world. Picking a percentage out of the air does not make it fact.

haha :)
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.