Apple Hires ARM's Lead CPU Architect Amid Rumors of ARM-Based Macs as Early as 2020

[ElectricPotato]

Mac evolution, next step:
Macbook Intel -> Macbook ARM
Macbook Pro Intel i series-> Macbook Pro Xeon starting at $4000
Macbook Air -> no updates for a year then discontinued
iMac Intel -> iMac ARM
iMac Pro Xeon -> Updated, same-ish pricing
Mac Pro Xeon -> Updated, same-ish pricing

 

[rlhamil]

Except at the low end where it doesn't matter, ARM-based Macs would eliminate the probably nontrivial class of virtualization users, not just people using Windows VMs on Macs, but people using Docker etc on Macs. There will always be stuff that people want to run, that's just too much of a PITA to port.

Macs, while perceived by some as more or less a walled garden, at least compared to Linux or perhaps even Windows, are far less so than e.g. a ChromeBook, where you don't get enough control for it to matter much what the CPU is.

And mixed CPU architectures on nominally a single platform (e.g. Macs) means "universal" binaries again, a measure not entirely without downsides, both to developers (slower compiles, at least a few more problems) and users (at least unless they can be optionally automatically stripped for the architecture they're installed on; e.g. a system preference to strip on installation, and ways to do that without breaking code signing etc).

Just about everything in the last 40 years or so except for mobile devices (constrained by power and heat limitations) with non x86 or x86-64 compatible CPUs has been a niche product. I don't necessarily like that - I'd prefer a more robustly diverse ecosystem - but that's how it is.

 

[firewood]

What would be the purpose of an A-series Macbook?

Same or better performance in a MacBook than a Core i5, but running cooler, allowing significantly longer battery life (without any weight increase).
Next-gen neural engine(s) for all those new ML/DNN apps.
More secure (e.g. less built-in speculative execution malware exploits than an Intel i5).
Can run Apple developed apps, including Xcode, and almost any macOS (or iOS via catalyst) app built with Xcode tools. Most Linux pro-apps (recompiled) as well.

 

[can$rules]

If they would only hire some good keyboard designers... Sigh!

 

[k27]

Thing is having worked at NeXT and Apple and seen HP-PA RISC, SUN, Moto, x86, PowerPC we always knew the future was x86. We were Quad-FAT binaries at NeXT but at Apple Steve went PowerPC on the assurances from IBM they would keep up with development and performance of x86. They didn't so he moved to x86.

AMD is ahead of Intel now. Apple can just customize their OEM motherboards around Zen2+ and smile all the way to the bank with higher margins and lower overall system prices.

AMD also has an advantage when it comes to security.

https://www.anandtech.com/show/14525/amd-zen-2-microarchitecture-analysis-ryzen-3000-and-epyc-rome/3

Sadly, this topic is often ignored or forgotten (also Apple's A-series concerns this topic.). People also confuse mitigations with fixes.

 

[firewood]

Just about everything in the last 40 years or so except for mobile devices (constrained by power and heat limitations) with non x86 or x86-64 compatible CPUs has been a niche product.

IIRC, Raspberry Pi (ARM CPU) outsells Dell in consumer PCs. Hardly a niche.

 

[fourthtunz]

Apple owned a part of the PowerPC alliance. Did that drive Apple Mac prices down? Dream on. HAHA!

Oh, I know that it didn't. I agree.
Just kind of tired of some kid that lives in his moms basement telling us that he can build a cheaper faster hackintosh, like none of us are aware that you can do this.
I'm just not interested in it thats all.
And the people that have done it often have to screw around to keep them working. nope

 

[gnasher729]

ARM stands for Advanced RISC Machine. RISC never handles more complex code very efficiently because it can't perform every instruction natively. We been through the CISC vs RISC discussions many times through the years and RISC is very good at applications where you limit what you do, but do it very fast. CISC based computer processing is what the business world utilizes. Going to ARM would be a another disaster.

WTH. It doesn't matter what ARM stands for. The A12 processor is an absolute monster. It has massive caches - massive compared to most Intel processors. It is a 7 way superscalar design, Intel is 4 way. It has a throughput of 3 fused multiply-add operations with a latency of 3 cycles, that's 50% better than Intel. "Can't perform every instruction natively" is pure nonsense. I've run benchmarks where an iPhone XR beat a quad core 3.2 GHz iMac by a significant amount.

Apple just needs to take the A12 processor, replace 2 Vortex + 4 Tempest cores with 6 Vortex + 4 Tempest cores, put it into a MacBook Pro, and I'll gladly give up my own MacBook Pro for that.

 

[09872738]

Microsoft saved them in 1997.

No they didn't. When Microsoft made a (rather small) 150 million investment this was little more than a gesture - and they settled the copyright infringement dispute, so this was not really a deed out of sheer generosity or to save Apple. It was a move to settle the dispute, mostly.

Besides, quote (https://en.wikipedia.org/wiki/History_of_Apple_Inc.#Microsoft_deal): "The day before the announcement Apple had a market cap of $2.46 billion,[67] and had ended its previous quarter with quarterly revenues of US$1.7 billion and cash reserves of US$1.2 billion,[68]making the US$150 million amount of the investment largely symbolic".

Would you please stop spreading this simply wrong myth? Thank you

 

[gnasher729]

Arm based macs with an intel co-processor for those that need it, would that work?

Having two totally different processors in one machine would be an absolute nightmare. Even if Apple could make it work without problems, it would be a waste of money to pay for two processors, and a waste of space (literally) to use up the space for two CPUs in your computer.
[doublepost=1561590468][/doublepost]

No they didn't. When Microsoft made a (rather small) 150 million investment this was little more than gesture - and they settled the copyright infringement dispute, so this was not really a deed out of sheer generosity.

It seems Microsoft hired the same company that had done lots of work on QuickTime for Apple - and that company took a lot of code owned by Apple and delivered it to Microsoft. I'm 99% sure this happened without the knowledge of Microsoft, but Apple still had them by the balls.

 

[09872738]

It seems Microsoft hired the same company that had done lots of work on QuickTime for Apple - and that company took a lot of code owned by Apple and delivered it to Microsoft. I'm 99% sure this happened without the knowledge of Microsoft, but Apple still had them by the balls.

I believe the dispute mentioned here is on Windows itself: Apple claimed Windows is just a rip-off of their operating system and pressed for charges. With mentioned deal, Windows finally was legit - what was not necessarily a sure thing back then.

 

[Wonder Warthog]

Because it is a profoundly stupid idea that will invoke a huge amount of headaches, and achieve nothing other than declining sales. Moreover, Apple has tried it before (remember PowerPC) and unless it is going to be 10 or 20 times faster, no-one will notice the speed difference, and there will just be a bag of mess.

I remember PowerPC. You mean the chips that were using far less power and totally buried intel Pentiums in performance? They only started sucking when IBM couldn’t/wouldn’t build newer processors for desktop machines and were only interested in their supercomputing monster chips. Then the G4 had to last a few years longer than anticipated, and they were falling back in performance, and couldn’t compete anymore.

That’s when Steve Jones pulled the plug and transitioned to intel who at the time had the better roadmap for desktop and laptop computers. Now intel has been sucking for too many years, Maybe it’s again time to move on. Luckily Apple is now big enough to produce the CPUs themselves and don’t need to rely on 3rd parties to **** them over again and again.

Having said that I, too, think they should consider AMD. They already have a great relationship with them for GPUs. Just the recent simultaneous announcement of the new Mac Pro and AMDs latest 7nm high end GPU shows that they’re playing well together.

And for all the naysayers: the latest Apple chips in the iPhone X are already outperforming the low end MacBook Pros considerably for single core performance. They do that ”as is”! What do yo think the performance gain would be if you took that tiny super low-power smartphone chip, gave it ample cooling, sized it up and let it suck desktop level power? I could imagine there potentially is a chip that could bury all intels Apple has running in their Macs, except maybe the Mac Pro Xenons.

 

[cmaier]

ARM stands for Advanced RISC Machine. RISC never handles more complex code very efficiently because it can't perform every instruction natively. We been through the CISC vs RISC discussions many times through the years and RISC is very good at applications where you limit what you do, but do it very fast. CISC based computer processing is what the business world utilizes. Going to ARM would be a another disaster. Apple at this point is trying to further optimize and expand its computer based solutions (either CISC or RISC (ARM) not migrate to a less efficient RISC solution that costs business their software investment. Thats not to say that you can't use ARM paired with CISC to enhance computers by removing certain processing from CISC for application specific enhancement.

I’ve designed PowerPC, SPARC and MIPS RISC cpus and x86-64 CISC cpus. “It can’t perform every instruction natively” is utter nonsense. Every CISC machine uses a microcode decoder and ROMs to convert complex instructions into RISC-like instructions. No processor is handling “add the contents of memory location X to the contents of memory location Y and put the results in memory location Z,” or any other similarly complicated instruction, “natively.”

 

[len5]

"AMD is ahead of Intel now."

Are they further ahead enough for Apple to risk shifting to AMD processors? AMD has a history of taking a quick lead but also quickly losing the lead. That's what happened with Athlon64

Until AMD comes up with something in their CPUs to compete against Intel's Quicksync Apple will never use an AMD CPU in a Mac. Intel Macs running FCX blows the doors off any current AMD CPU running Premier.

 

[Realityck]

The A12 processor is an absolute monster. It has massive caches - massive compared to most Intel processors. It is a 7 way superscalar design, Intel is 4 way. It has a throughput of 3 fused multiply-add operations with a latency of 3 cycles, that's 50% better than Intel. "Can't perform every instruction natively" is pure nonsense. I've run benchmarks where an iPhone XR beat a quad core 3.2 GHz iMac by a significant amount

You know of any real PC games that are A12 based?

 

[PlayfullyEvil]

Having used a PowerBook G4 and later moving onto MacBook in the mid 2000s, I have no doubts that Apple can transition from Intel x86/64 to ARM. But whether they make that transition seamlessly is the question.

Provided they have the same development team that allowed the seamless transition from PPC-to-Intel, then all should be well.

This hire is a good one; regardless of whether it's for just the iPhone/iPad chip development or the rumoured move to ARM-based Macs.

 

[Val-kyrie]

On the other hand, maybe it has nothing to do with someone leaving last year and everything to do with future plans. Who knows... really?

Maybe it has something to do both with the need to fill a vacancy in leadership and with future plans. Those often go hand-in-hand.

So will an ARM based Mac be unable to run standard, compiled for Intel applications? Where does this leave the current lineup, and even the new Mac Pro? Is buying a Mac now buying into a machine that's going to have a limited life due to the move to ARM?

Based on the PPC to Intel transition, I would not expect support for macOS updates on Intel machines more than 1 year beyond the transition to ARM, with the exception of security updates.

That's what I'm wondering. They're going to come out with a $4000+ 16" Intel Mac this fall then quickly follow it up with an ARM based Mac? What about the people who spend $12,000 on the Mac Pro?

Or is ARM only coming to the lower-end Macs?

Apple released a revised PPC Mac Pro in October 2005 and replaced it with an Intel Mac Pro 10 months later in August 2006.


The rumored 16” MBP could be Apple’s last gesture to the Pro audience before transitioning to ARM or it could be the first ARM-based MAC.

A new form factor for a new architecture makes sense but it is just as possible Apple designed the 16” MBP with its ARM CPUs in mind and will use it first for a last Intel CPU before transitioning to ARM with the same form factor.

I am processor agnostic. However, i do favor RISC vs CISC generally speaking. If they can demonstrate their ARM based platform, performing "heavy-duty-lifting"* at the same speed(or better than) intel's offering, at roughly the same price, i'll be interested.

They've made transitions like this in the past, so i hope they can follow the "recipe". (Emulation layer for the transition, etc).

If this is their plan, I suspect they will start with consumer-centric Macs first.

*heavy-duty-lifting would include number crunching or video encoding(HEVC) IMHO.

PS>> They better be able to pair their CPUs with off-the-shelf GPUs.

I would not count on pairing off-the-shelf CPUs with any ARM-based Mac. The transition will likely be a push toward greater restrictions, much like the GPUs during the PPC era which required special firmware,... but I would like to be wrong about this.

 

[mschmalenbach]

"Hey prosumer, have a look at our computer that costs at least TWICE AS MUCH as a similarly spec machine anywhere else. ...

When you compare like for like it's not a 2x cost difference. Or even a 2x price difference.

I was PC for years. I bought a MacBook for my business because I just needed the damn thing to work and not spend ages booting, or doing security updates. Or crashing so much.

Comparing like for like in terms of components/quality, the Mac was a bit more pricey to buy, but not hugely.

But in terms of COST of operating overall? The MacBook was waaaaayyyyy cheaper - all that time saved by not having to wait for the thing to do it's own thing.... work not lost due to crashes etc...

I still have a i7 quad core HP machine for messing about with development of some s/w and hardware stuff (I'm an engineer by training & original trade, still dabble...)

But for 'business' it's been Mac for me for quite a while... from bottom of the line stuff to the very top of the line lap tops. Not once have I regretted it. Some of my team use PCs for their video editing etc - at least 2x a week their experiences remind me why I made the jump...

YMMV.

But right now I'm not sure you could pay me to use a PC.

 

[IG88]

I doubt too many 12 inc MacBook users require boot camp. My guess is we will see an ARM MacBook in 2020 and perhaps MacBook Air. The MacBook Pros will remain Intel based for a bit longer.

That's been my thinking ever since ARM Mac discussions began. The 12" MacBook is the perfect target. People that buy that aren't BootCamping it in any significant % that it matters. Apple will stay with Intel for the Pro machines.

 

[sirozha]

I'd really love to know what percentage of Mac users run Windows or Linux, either through Bootcamp or using a virtual machine, and what percentage would be lost as customers if that wasn't possible.

There wouldn't be many customers lost to Wintels because Apple will not completely abandon the x86_64 platform for years to come. Maybe in 5 years the ARM architecture will be able to complete with x86_64 at the higher end but not yet. The higher-end Macs that cost north of $3,000 will have to continue to use the x86_64 processor architecture, which means that those who need to run Windows or Linux for business application would still be able to buy Macs capable of either running Windows or Linux natively or in virtual machines. I'd say that no more than 20% of Mac users (most likely no more than 10%) have to run non-MacOS applications on the Macs. The rest would be just fine with ARM-based Macs.

My personal opinion is that for what I do, I don't want to pay more than $2,000 for a laptop. This means that I would have to either bite the bullet and pay at least $3,000 for a new Mac or buy used x86_64 Macs for a few years. What happens then is not clear. I may have just to switch to Wintel at that point or maybe there will be a way to virtualize Windows/Linux on the ARM architecture then.

 

[Realityck]

I’ve designed PowerPC, SPARC and MIPS RISC cpus and x86-64 CISC cpus. “It can’t perform every instruction natively” is utter nonsense. Every CISC machine uses a microcode decoder and ROMs to convert complex instructions into RISC-like instructions. No processor is handling “add the contents of memory location X to the contents of memory location Y and put the results in memory location Z,” or any other similarly complicated instruction, “natively.”

Your right, should have conveyed it differently. The term natively was the wrong word to utilize. I was trying to show the difference between RISC instruction set requires one to write more efficient software with fewer instructions, versus CISC which is using complex instructions.

 

[sirozha]

I still believe that when Apple releases the ARM-based Macs, it will be a hybrid iOS/macOS device that will run the touch UI in the tablet mode and the mouse pointer UI in the docked mode. I don't think that MacBook Pros, iMac Pros, or Mac Pros will be running on the ARM architecture any time soon; at least not for the first 5 years. The only devices that will be running macOS on the ARM architecture will be either only hybrid iOS/macOS devices or, perhaps, iMacs (non-Pro ones) and Mac Minis will also be ARM-based (without having the tablet mode in them).

 

[R3k]

Yeah, more to do with the existing ARM based products id say. The newly announced Mac Pro

I believe the Mac will have ARM and Intel both inside the machine for maximum capability. The T2 and bridgeOS on the MacBook is already a good example of a hybrid OS.

From the looks of it, the best method is to have ARM and Intel on notebooks first, this will allow the macOS and all the catalyst apps run on ARM. Any other apps that isn't supported by the ARM processor will run on the Intel chip. Why do this? Because on the mobile side, ARM's speed is near intel's speed, which means that have macOS apps running on ARM will not be slower than running on intel, but at the same time more efficient.

As the ARM speed improves, I expect ARM will eventually be included on iMacs as well.

This method is the best method as it is invisible to the end user, doesn't matter if you are a novice macOS user, or a professional user. At the same time, it will not divide up the Mac into 2 platforms (ARM Mac and Intel Mac), which definitely confuses the buyers.

But I expect the price to rise again for the notebooks, especially for the coprocessors.

Sounds like a thermal nightmare. I have doubt.

 

[JimmyHook]

You know of any real PC games that are A12 based?

The App Store has a MASSIVE list.

 

[smulji]

Until AMD comes up with something in their CPUs to compete against Intel's Quicksync Apple will never use an AMD CPU in a Mac. Intel Macs running FCX blows the doors off any current AMD CPU running Premier.

A better comparison would be Intel Macs running FCPX vs AMD Macs running FCPX
[doublepost=1561599628][/doublepost]

I still believe that when Apple releases the ARM-based Macs, it will be a hybrid iOS/macOS device that will run the touch UI in the tablet mode and the mouse pointer UI in the docked mode. I don't think that MacBook Pros, iMac Pros, or Mac Pros will be running on the ARM architecture any time soon; at least not for the first 5 years. The only devices that will be running macOS on the ARM architecture will be either only hybrid iOS/macOS devices or, perhaps, iMacs (non-Pro ones) and Mac Minis will also be ARM-based (without having the tablet mode in them).

"I still believe that when Apple releases the ARM-based Macs, it will be a hybrid iOS/macOS device that will run the touch UI in the tablet mode and the mouse pointer UI in the docked mode."

I wouldn't doubt this at all