Yeah I was being “bet several zeros” conservative. 
I agree, likely within 2 years from start to finish, but I’d put money on it being within 5 from today.
1 Like
W10 has an ARM port, see the new Surface Pro X.
W10 ARM is still new, but if Apple and/or Microsoft cared I bet they could get dual boot functional pretty swiftly.
1 Like
Agreed.
Microsoft has only one thing going for their OS and that is backwards compatibility. Apple has made drastic moves in the past and didn’t care about legacy stuff at all. Nobody expects differently when the time comes. And with that attitude they have remained agile.
My prediction would be this: Microsoft will follow Apple onto ARM and see it as a chance to pull away from X86. And at that point we might see a shift that goes way further than just Apple computers.
2 Likes
I doubt Microsoft could get devs to port important software to ARM from x86. Apple’s ecosystem is small enough that they could get it done over time, but I doubt MS even cares as they could risk alienating users and devs.
1 Like
The world of computing is going ARM. This is like “the year of desktop Linux”, it will never be a 100% takeover, there will be pockets of x86/PowerPC/RISC-V, but it is comming. ARM has won mobile compute, ARM (not x86 or PowerPC) will be the major player in IoT. IoT market will be larger than the mobile market which is larger than laptops which is larger than desktop market. ARM will even make inroads into the server market. Cloud services are poised to
With ARM core production and there for R&D budget to soon to dwarf all other processors combined (if not already), every software and hardware developer/engineer should be executing (not planning) on how to transition to ARM.
Apple has shown disregard for their dev community plenty of times. But I just see this as Apple moving with the winds of change. Not even sure they are that far ahead of the curve here. If your software is not running on ARM or you are not working on making it run on ARM, what industry are you in?
1 Like
The whole architecture thing (being stuck on x86 or whatever) is no longer as important as some may think. Not for Apple anyway… this will be their 4th CPU architecture shift, and they’ve done a pretty great job of it from 68k to ppc and ppc to intel in the past.
Sure, there are niche applications that are natively compiled for x86 only. But for the vast majority of end users, their modern workflow is mostly applications written in freakin javascript and run either in a browser or on a mobile device.
Apple is conveniently placed with macOS in that the architecture has been there since nextstep within the OS to handle multiple architectures in an elegant manner (multi-arch binaries, previously rosetta to do real time translation for the non-framework linked code in an app, run-time linking via a runtime broker rather than direct binary linking, etc.).
Apple/Mac/iOS software is generally very well behaved, normally uses the OS provided Frameworks (which are very comprehensive) to do stuff, etc. So apple can re-write/re-compile their frameworks natively and many apps will “just work” with mostly native code provided by frameworks (e.g., metal, OpenCL, etc.) for the heavy lifting.
The amount of actual compute-intensive code written by a typical Apple developer (i.e., stuff that Apple won’t have compiled natively in the new ARM OS) that isn’t mostly telling the OS frameworks what to do is… not much. It’s mostly just glue-code.
Microsoft not so much… but again it mostly won’t matter, and for Microsoft their OS is big/common enough that the third party developers will do the grunt work to make their new app work properly.
The only thing holding Apple back in my view is confidence that they can get a good enough performance jump beyond x86 with their new CPUs to more than (or at least mostly) off-set the translation during the cutover period so they aren’t burning battery life and losing performance overall.
But if you compare intel’s performance jumps vs. Apple’s over the past decade with their Ax series processors… the writing has been on the wall for 5+ years at this point.
3 Likes
x86 Walled itself in so bad, they can’t cross their own walls anymore.
1 Like
Problem with that is that Windows for ARM is totally useless. The only applications for Windows anyone care about are built for x86 (legacy stuff, games) and even if the ARM processors Microsoft were using weren’t trash tier, the desirable Windows apps won’t run because Microsoft have no CPU migration strategy for legacy applications (Apple does, and have executed successfully multiple times).
People don’t run windows for “windows” compatibility, they do it specifically for legacy x86 windows application compatibility 
Has been the case for some time. ARM is used in so many things even before “IoT” - car ECUs, mobiles, desktop phones, tablets, watches, embedded controllers, etc.
ARM has probably shipped more units than any other cpu vendor for decades at this point, they’re just not always in high profile devices. They’re everywhere though.
2 Likes
So this thread pops up every 6 months or so. Apple is moving towards arm for a unified ecosystem and for power efficiency reasons. They’re choosing arm because it’s a tried and true architecture that they have over a decade experience with.
And that’s if these rumors are actually true. They’ve been saying this since 2015.
Although, the argument is becoming more and more compelling (but then again, so are the amd cpus)
2 Likes
The people the most screwed are people requesting a extremely powerful notebook to power their creative apps and being limited to proprietary accelerator chips. x86 at least has enough power for rendering at the moment on the go, and right now most people still swear by their MacBook Pros for on the go dailies.
But, with a move to ARM, most Cinema RAW companies will have to make accelerator chips again. This is how the Red Rocket was born:
The new Mac Pro has a dedicated ASIC option called the “Afterburner” that’s similar to the RED Rocket for ProRes RAW, (and this is the same dangerous path that people went down for ProRes 422 SSD recorders, since ProRes RAW is not an open standard)
so you bet they will have to make another low power ASIC for ProRes RAW in a ARM MacBook and under pressure from RED, they will have to make a mini-RED Rocket because a BIG.little 2+4 core config is not enough for 4K, 6K and 8K RAW.
1 Like
I think the popularity of Android and iOS has opened the door for other CPU chipsets. Combine that with their ARM engineering and I think they are ready to extend the hardware benefits cough and lockdown methods cough that customers enjoy on their mobile devices to Macs.
Nah, I think physics limitations are opening the door.
You can only throw so much power at a chip before it’s impractical to cool it, and it’s not worth running. You can only get so much performance out of a chip at a certain lithography. We’re getting to the point where paths and gates in the CPU are only like 10 atoms wide. (5nm) and this basically means we either need to find smaller atoms or more efficient CPU architectures, or rethink the way we’re doing CPU architectures entirely.
ARM is incredibly power efficient because it runs a reduced instruction set, it doesn’t need as much power per core. (physically less transistors) This is good because it means you can fit more cores in there. Now, obviously, this means that certain operations take a lot longer because there’s no dedicated instruction (set of transistors on the physical core) to handle this, but that might not be a problem for commodity hardware.
I’m really interested in the future of processors because we are actually getting close to the limitations of the natural world, when it comes to our current methodology.
4 Likes
Same.
I find optical and quantum-effect transistors super interesting.
The one aspect that strikes me as odd is the lack of any news from any semi conductor manufacturer on these. So those technologies stay locked away in the labs of universities round the world…
2 Likes
You might find it interesting then that 5nm encountering quantum tunneling of electrons. Not that it’s controllable or helpful.
It’s slow work. They can’t really keep posting articles saying “yep, this week we failed 58 times, but we’re failing towards a goal at least!”
Which isn’t to say they’re not doing good work, just that it’s slow.
2 Likes
I think you really need to look at the iPad processors of the past 3 years.
The iPad Pro I have from 2017 (10.5") runs a lot of stuff faster than my brand new MacBook Air with i7-1060NG7 in it. And it is fan-less, in less watts, on a competitor 10nm process (TSMC) from 3 years ago.
Swift playgrounds for example. Fan screams away on the MacBook Air to run it. iPad? No fan - runs just as well - or faster.
The days of intel CPUs being faster than ARM for media type things that the typical Apple user cares about are over dude. You can do entry level 4k video editing on a freaking iPad these days.
You can bet apple will be continuing their path of ASIC accelerators in the Mac Pro for those who care about such things.
3 Likes
Just thought actually, if you consider the history of macOS as a derivative of NextStep (which it is), the CPU path has been
Motorola 68030 (NextStep)
486 (NextStep - in like 1993, concurrent with NextStep 68k using “fat binaries” like OS X later)
PPC (and Pentium internally only) - original OS X
Intel Core
ARM (for iOS/tvOS/watchOS/etc.)
ARM (Mac)
Classic macOS had the 680x0 to PPC shift as well, but classic is a different hierarchy to modern macOS entirely. Which is another platform change they managed for that matter - janky single-user 68k code from the 80s running under a PPC architecture on UNIX. Both a CPU and major OS platform shift without breaking application compatibility for years!
Rosetta was that good that I remember running Diablo 1 (a PPC/classic MacOS app) on my first Mac - an intel Mac mini under OS X 10.5, and didn’t even know it was non-native (didn’t even think about it, it “just worked” perfectly until I upgraded to 10.6.
In 10.6, Rosetta the translation platform was made a non-default install and when I tried again I had to manually install Rosetta to make Diablo work on 10.6. That blew my mind, Rosetta was so seamless and performant.
Point being though… they’ve successfully changed over a lot more than you might otherwise think, if you don’t consider the NextStep origins, which still underpin the core of macOS today.
Apple’s biggest problem they’ve had with these shifts IMHO is that if anything, they’re so seamless that end users don’t realise the software they’re using from 5 years ago isn’t native until one day it just breaks (when the translation software is removed) and then they complain.
2 Likes
If I may:
I guess the rumors are true
4 Likes
Prepare yourselves, content creators. Adobe’s gonna take their sweet time porting the pro software suite to ARM. Plus to please pros, they must have acceleration chips for every thing in Creative Cloud.
Really curious about that development. If they can build CPUs that scale up really well, this is gonna be another huge dagger in the back for Intel.
1 Like
I mean its not like they have to be quick, the Mac pro is not exactly old and it has both an Intel CPU and an accelerator. Pros would be mighty pissed if their $50,000+ creation workstation was dumped for arm in a year.
The pro stuff will trundle along on old big x86 while apple work on all the consumer stuff for YouTube and social media consuming with arm while getting the bigger stuff updated in the background so they can switch pros over at a much later date smoothly.
This will not be an overnight sensation.
1 Like