[Bloomberg] Apple starting process to dump Intel in Macs

[jpiniero]

https://www.bloomberg.com/news/arti...to-move-from-intel-to-own-mac-chips-from-2020

Light on details, but rumor states Apple is gearing up to begin the transition away from Intel in Macs to their own processors with products as early as 2020. This has been rumored for some time that Apple would do this, but this is the first time I've seen any actual confirmation this is in the works.

You could also read this as Apple doesn't think Intel will deliver anything to keep them interested in staying.

 

[ZGR]

The software transition has me worried more than performance itself.

 

[Nothingness]

I doubt this could happen as early as 2020. There would be some obvious hints such that ARM support for MacOS apps under Xcode (and fat binaries that'd embed both x86 and ARM). They can't afford to rely on emulation as they did in the past because there's no hope an ARM CPU will be fast enough by 2020 to overcome the cost of runtime translation (look at Windows on ARM).

Another step might be dual CPU support. Hasn't this already been talked about?

 

[ultimatebob]

Yeah, there has been rumors about having a power saving dual CPU setup in their notebooks for awhile now, where the ARM CPU does basic tasks like text editing and music streaming, and the APU only kicks in for things like gaming and video editing.

Migrating everything off of Intel is a huge step, though. It also means that the resale values of used Apple hardware is going to plummet, as you know that they'll drop Intel CPU support 2 or 3 OS releases after the last Intel powered Mac is produced. The days of getting a Mac and expecting it to be supported for 6+ years are coming to an end, boys and girls.

 

[senseamp]

Are they migrating PC's to ARM or both PC's and Mobile to Risc V to not have to pay ARM licensing fees anymore ?

 

[whm1974]

Are they migrating PC's to ARM or both PC's and Mobile to Risc V to not have to pay ARM licensing fees anymore ?

ARM is more likely as there are no desktop/workstation RISC-V CPUs out right now. Keep in mind however, neither AMD or Intel are standing still.

 

[jpiniero]

I doubt this could happen as early as 2020. There would be some obvious hints such that ARM support for MacOS apps under Xcode (and fat binaries that'd embed both x86 and ARM).

Well, it's been previously rumored that the next version of XCode is going to allow the ability to create a single app that will work with iOS 12 and macOS 10.14. Perhaps that's when you would see the ARM macOS support.

 

[el etro]

Time for AMD and MSFT/NVDA jump out of the x86 boat too. to AMD is better to port the Ryzen IP to ARM....

 

[Ajay]

I think that there is no doubt that Apple is testing this out, just as they ran x86 builds of MacOS in testing during the PPC era. The question of when and if Apple makes a change would depend on actual an perceived performance vs the advantages of a more tightly controlled ecosystem. Apple is the only one who knows how this is going and I doubt they are talking to anyone given their penchant for secrecy. The success of Marzipan might be an indicated - though a weak one since it's still running on x86. I do wonder if Apple was hoping Intel's ATOM lineup would have produced better results and is seeking ways to get platform power even lower given some of the power issues with current MacPros and the delay of 10nm x86 CPUs.

 

[Ajay]

Time for AMD and MSFT/NVDA jump out of the x86 boat too. to AMD is better to port the Ryzen IP to ARM....

LOL. x86 is still a cash cow - neither MSFT or NV are leaving that platform anytime soon.

 

[lopri]

You could also read this as Apple doesn't think Intel will deliver anything to keep them interested in staying.

Alternatively, it could mean that TSMC's (and maybe Samsung's?) roadmap gives them enough confidence. Rumor has it that TSMC's 7nm+ process is a huge leap in P/W.

 

[senseamp]

ARM is more likely as there are no desktop/workstation RISC-V CPUs out right now. Keep in mind however, neither AMD or Intel are standing still.

2020 is enough time for Apple to design their own. Intel and AMD are pretty much standing still compared to Apple. Since 2011, Intel went from Sandy Bridge to Coffee Lake (~2x), while Apple went from A5 to A11 (~10x)

 

[LightningZ71]

As we've been seeing for a while now, there isn't a lot that the top end iPad's aren't capable of doing that the average Mac buyer wants to do. Professional and high end gaming workloads are maybe the only concern at the moment, but, neither of those really represent the bulk of Mac purchases. I think that in 2020, you'll see the lower end MacBook and iMac lines transition to some sort of ARM based platform with the Mac Pro line continuing for another cycle or two to offer an upgrade path to their higher margin lines. As it stands, ARM doesn't currently offer much that can compare to the higher end Xeon lines at the moment, though, with some of the newer server products, that also appears to be in the long term offering.

What's sad is that AMD has the ability to produce a compelling product for the MacBook (both regular and pro) in the near term with the Raven Ridge and later versions of the APU being more than enough for more than 90% of their use cases at likely lower unit cost. Seeing that not explored, as well as the possibility of a new Mac Mini based on it is a bit sad. Its not shocking, however. Apple has been investing in improving their A series processors far faster than the rest of the market, and the only reason to push that hard is if you have bigger plans.

 

[CHADBOGA]

If true, this will be the perfect legacy for Brian Krzanich

 

[DisEnchantment]

Time for AMD and MSFT/NVDA jump out of the x86 boat too. to AMD is better to port the Ryzen IP to ARM....

Why jump out of x86? Both AMD and MSFT have ARM licenses and have had or considered ARM architectures in their products.

MSFT have so many products for the enterprise which are mainly x86 only.
For usual desktop professional workloads x86 is much faster than ARM.
For server ARM is far away for usual scalable cloud hosting, virtualized environments, Compute backends, WebApi gateways etc etc. compared to x86. Check out Phoronix benchmarking of 96 Core ARM vs Power and x86, it is barely faster than a single Ryzen 1800X in most benchmarks.
AMD has more to gain from x86 than from ARM.
One thing x86 should do is start dropping off archaic features which take up silicon space. From what I read there are many.
ARM SoCs are for now seems much better suited for portable devices though.

The problem with x86 is that we had stagnation in a major part of the decade due to no competition. Competition among ARM Licensees is fierce.
If you have ever opened an RFQ to SoC suppliers you can see how fierce the competition is.

For NVDA though they are on the ARM bandwagon since the Tegra 1days, makes total sense for them.
Why they are hanging around in x86 is because of that juicy PCIe bus for both client compute and Enterprise ML.
nvidia supplies ARM SoCs to many OEMs which usually the general gaming public are not aware of because they don't show up in any reports or news being B2B transactions.

With the booming API Hosting and Computes Services ( for IoTs and ML/AI ), Azure is doing well for MSFT (as do AWS etc) and from what I see, x86 is usually the only choice.
What is interesting now is the fight between Intel, AMD and the other ARM SoC makers on a near parity in manufacturing processes.
Also not only the ARM SoCs will gain a leap in performance and efficiency moving to 7 nm, but x86 as well. These few quarters are interesting because lots of designs are transitioning to 7nm/10nm and for x86 a return to competition after years of stagnation.

 

[ultimatebob]

If true, this will be the perfect legacy for Brian Krzanich

I guess you can't really blame him for the failure of the Atom product line in smartphones, but then he certainly didn't help much to improve it since he took over.

If Intel had an awesome low power processor for phones and tablets, Apple probably would have never got into the ARM processor business to begin with.

 

[cytg111]

Damson.
Next thing we know, Apple is going into the foundry business.

 

[NTMBK]

It's not really surprising. Why keep paying Intel margins for stagnant architecture and stagnant process?

 

[paperfist]

I thought overall desktop & laptop sales where in decline or a holding pattern? Seems kinda odd for Apple to be putting resources into moving away from Intel and into their own processors.

 

[moinmoin]

Damson.
Next thing we know, Apple is going into the foundry business.

Not happening, Apple is not manufacturing anything on its own nowadays. But they are supporting their manufacturing partners to some degree.

 

[Qwertilot]

I thought overall desktop & laptop sales where in decline or a holding pattern? Seems kinda odd for Apple to be putting resources into moving away from Intel and into their own processors.

Its the mobile phones (+ ipads etc) that have driven all of the R&D to get them a really, really good CPU architecture and a fair bit of GPU expertise with it. Moving over to it for their laptops/desktop machines as well is basically a direct cost benefit for the disruption caused by the switch vs the non trivial benefits of going all home grown - probably mostly about having more control over everything although they'd likely get much better integrated GPU's out of it too.

They've had ages to port their main operating system & own brand software over properly, so not huge costs there. Its definitely plausible but not automatic either. Myabe a bit problematic for their really high performance machines. We'll see.

 

[William Gaatjes]

I thought overall desktop & laptop sales where in decline or a holding pattern? Seems kinda odd for Apple to be putting resources into moving away from Intel and into their own processors.

Well, there is the idea that the smartphone one day will be powerfull enough to sort of replace complete desktop systems and laptops.
just wireless connecting the smartphone to a Qi charging docking station and continue to work at home or at work.
When 5G works really well, i am sure there will be very good short range alternatives as well( point to point phone to dockinstation wireless connection with > 20 gigabits/sec).
And, the dockingstation can always hold auxillary cpu and gpu calculation power for the heavy calculations.

Apple produces smartphones as well.
If it is not Apple, it will be a Chinese manufacturer who will come up with the concept.

At least, that is my assumption...

 

[NTMBK]

Don't forget that Apple also gets the benefit of no longer giving the PC market a leg up. Apple invented the MacBook Air, Intel repackages it as the Ultrabook and sells it to their competitors. Apple requested that Intel get the TDP and size down low enough to make the fanless MacBook possible, and Intel turns around and sells it to their competitors.

 

[krumme]

Don't forget that Apple also gets the benefit of no longer giving the PC market a leg up. Apple invented the MacBook Air, Intel repackages it as the Ultrabook and sells it to their competitors. Apple requested that Intel get the TDP and size down low enough to make the fanless MacBook possible, and Intel turns around and sells it to their competitors.

Yep. Thats the past.
But Steve Jobs is not there to drag Apple forward. Imo its pretty obvious he is missing with his eye for good ideas and relentless temper and energy to force the nessesary quality.
Apple isnt the force anymore. Its a predictable boring cash cow. Not a product leader.
Heck look at the new Huawei P20 pro. Small A73 cores that is fine and nobody cares for st gb perf beyond that but a innovative stellar camera wrapped in a nice look. That phone should have been Apple.

 

[chrisjames61]

The software transition has me worried more than performance itself.

Worried? Why? It will go off without a hitch. Apple switched from 68K to PPC and it was a seamless transition. The swith from PPC to X86 also went off without a hiccup.