Nikkei Asian Review, first on Apple rolling their own modem chip:
Two other chip industry executives also echoed Li’s view that Apple will develop its own modem chips or at least boost its related capabilities – a view bolstered by Apple’s poaching top Qualcomm modem chip engineer Esin Terzioglu earlier this year.
But Li added it is was unlikely that Apple could quickly roll out such components within two years. Modem chips have a very high threshold to develop and need to fulfill requirements of different operators worldwide.
A veteran chip industry executive estimates that it would require more than a minimum one thousand engineers to work on such a project.
As far as Macs go:
Core processor chips for the MacBook range is another area Apple is trying to venture into.
Two industry sources say that Apple is trying to cut its dependence on Intel when it comes to notebook chips and instead build those using ARM architecture, referring to the SoftBank-controlled British chip designer.
“Notebooks are becoming thinner, while consumers are demanding better mobility and longer battery life. That gives ARM’s architecture, which is known for its power efficiency, a very good opportunity,” a chip industry executive said.
And:
Apple also aims to design its own chips that could integrate touch, fingerprint and display driver functions, sources say.
“Apple has hired engineers from Taiwan’s No. 1 display-driver chip designers Novatek and panel makers of AU Optronics as it wants to control next-generation display technology and some related key components,” said a Taiwanese chip industry manager.
Apple switching Macs to ARM chips has long been a topic of speculation. And there are significant technical hurdles to overcome. Interesting, but take with a grain of salt.
That would be so cool.
nopunintended
I have no doubt, none, that this will happen. I have always thought it would, I see no reason why Apple wouldn’t.
On an infinite timescale…
With the performance benchmarks using the new Bionic chip, it’ll happen fairly soon I would guess.
A natural next step in their evolution.
Feels more like tea leaf-reading than an actual report. It makes sense from a technical side, but they’d have to devote an awful lot of money and talent towards a product that is, what, 3rd or 4th on the list of revenue and attention?
All of your software needs to be recompiled. ALL of it. Nothing you have will work. Doing that for the move to Intel made sense because it was moving to a more powerful chip that was continuing to evolve.
What does a move to ARM really get me as a Mac user anyway? Battery? They moved the MacBook to a low power Intel chip and threw away the longer battery life by making it razor thin. Power? Perhaps – if they feel the A series will continue to evolve as fast as it has been that’s a consideration.
Apple has gone through recompile transitions multiple times in the past. That alone wouldn’t stop them if it makes strategic sense.
If the A-series CPUs keep improving this quickly, before long it’ll be 2-3x faster than the Intel M CPUs while using less battery power. For thin-and-light that would certainly be compelling from Apple’s perspective… if the product line is important enough to them (as NB implies elsewhere on this comment thread).
Much lower cost, for one thing. Those Intel CPUs are expensive.
They’ve done it once for the Mac. The move to Intel. Not multiple times (I’m ignore the 6502 days here).
It’s not just them. It’s every single developer. It’s a huge investment for the ecosystem to make.
Battery – again, if they were concerned a lot about battery life they didn’t show it in the MacBook which removed battery for thinness. That machine would have a killer battery life if it was in the Air case.
Also, ALL Macs need to make this move, not just the MacBook. The MBP. The iMac. The iMac Pro and Mac Pro. Otherwise you’re asking your 3rd party developers to develop and test two versions of their software (and you have to do it too).
The best reason to do this would be if they felt that in some years their A series will significantly outstrip the Intel CPUs in performance.
They’ve been hampered by Intel’s development cycles, so if they could get equal or better performance out of a chip they control the development of that could be a big win.
Not impossible, and possibly compelling someday.
But until Apple has an ARM chip on the horizon that can do Mac Pro/iMac Pro* level work, it’d mean fragmenting the Mac platform and either endless incompatibilities or fat binaries.
Now, the latter sure ain’t impossible, and Apple’s literally done it before.
But that’s why I’m, like you, not real sanguine on it happening anytime soon.
(* Yeah, the A11 can sort of compete with a MacBook Pro, in some contexts and benchmarks, I hear.
But that’s not the same realm, and I’ll take those benchmarks with a grain of salt until I see someone actually run non-mobile software on one.
Someday, though, it seems very likely to happen.)
They are already spending a lot of this doing the custom development for their chips, so the marginal cost of creating an ARM Mac probably isn’t as great as we might imagine. They are also likely seeing the convergence of the iPhone/iPad/Mac ecosystems overtime, so getting ahead of that by bringing the Mac in line with those other platforms makes sense.
Compilation is a non-issue, though, as a developer.
If you’re not doing inline assembler (and if you are, well, you can cope – almost nobody does) it should be as simple as XCode making a fat binary or a separate ARM binary in parallel to the x64 one.
(Nothing you have will work? Well, that assumes that you’re using a bunch of software that is both not-Apple and not-maintained-by-its-developers.
Which is a big hurdle for some people, yes!
But not as many as enthusiasts like to think.
Plus, again, Apple’s done that one before, twice, during the mc68k-PPC transition and then the PPC-Intel transition, with transparent emulation, and surprisingly small performance hits.)
“Interesting, but take with a grain of salt.”
A BIG grain of salt.
Twice.
The mc68k-PPC transition also had an emulator built in and fat binaries.
Note that the transparent emulation layer means that if your un-maintained software isn’t super-performance-critical it just keeps running…
(I remember this because I went through both transitions, as a long-time Mac guy…)
Good god… I’d utterly forgotten about that transition. (Hangs head…)
Oh, it’s not the actual compilation. It’s making the code work, debugging it, etc. That is, it’s not the mostly mechanical step of telling Xcode to compile the program, it’s the initial debugging (which could be onerous or simple) and it’s the ongoing testing (you now need to test your product on both architectures).
The Apple A11 chip is already as fast as an Intel core i5, and Apple under-clocks their chips for thermal reasons. No one but Apple knows how fast the chips really are. I’ve been following Apple for 30 years and I wouldn’t be surprised if they put two A12 chips in a Mac.
Yep. And they DO like to control things.
I think by now most of us developers know what not to do, but it doesn’t mean we can skip the testing. 🙂
If Apple announces this, by the time they were to ship it virtually every software package would run natively on the new architecture. Except for Microsoft Office.
Which is why I am certain if they can bring it in house they will.
Yeah. Again, it’s not that its undoable but that it imposes a burden, some parts of which are ongoing, on developers. The Mac is far more popular now that when we last did this too which will stretch out the number of legacy Intel Macs.
That main reasons I can see for them to do this are 1) Control and 2) The A series starting to significantly and consistently outperform Intel not just in lower power but in raw CPU power.
Exactly what sigivald said.
And it’s not just about battery. It’s the performance-per-watt. Apple wants to make fast and thin and long-lasting machines. They’ve already accomplished just that with the iPad Pro.
Don’t forget they also have experience with multi-platform binaries (the fat binaries from the aforementioned PPC transition). There’s indeed a question if they want to stick with that as a long-term strategy or not, though.
If they switch chips, will they need extra salsa?
This is only useful to me if it includes good x86 emulation. Being able to run Windows or Linux x86 in a VM is much more useful to me than incremental power/performance gains.
I think if Apple can pull off this transition, they should. Because the results will be a real feather in their hat.
Isn’t the veteran chip industry executive who is saying that it would
require more than a minimum one thousand engineers to work on such a
modem project the same guy that said Apple wan’t going to walk right in and design a successful smartphone? I’m just saying you can’t rely on some top executive who might be an Apple competitor to always be right. Apple could have been planning this stuff for a longer time than this executive realizes.
Jeez, a minimum of 1000 engineers. I had no idea modem chips were that complicated. Does Apple also have that many engineers working on the A-series processor? I always figured it only took a handful of really talented engineers to come up with great processor designs.
It certainly is possible for Apple to have some portable notebooks running ARM processors if the have the power of Intel i5 processors and still retain low-power capabilities. At least it’s something hopeful to think about if Apple can cut costs and still get performance. Contrary to all the future growth negativity, maybe Apple does have a future after all.
I suspect that Apple may move towards using something like Bytecode for Macs in order to solve 2 problems: 1)Put more Mac apps in the Mac App Store. 2)Generate Intel and ARM binaries from the same app bundle.
Apple has said this year that they are not building ARM based Macs. Apple doesn’t lie, they may not be forthcoming with information, but they don’t lie.
I do expect that Apple will continue to integrate ARM chips into the MacBook for a variety of functions.
After reading all the comments on this post, I’m shocked at how entirely civil this discussion is, with good arguments on both sides, no name calling, no snide remarks, and no insults. What sort of internet is this?
Hmmm. I take offense to that. 🙂
Aficionados of “ARM efficiency” should note that there is a HONKING GREAT FAN in the A10X based AppleTV box.
https://d3nevzfk7ii3be.cloudfront.net/igi/ZMNKdBGGBPeXrfSG.huge
And that fan is being managed at gigaflops speeds by thee A10X chip. Just try and top that Amazon Fire.
Yes, a fan that is so quiet, those who hadn’t opened the box and tried to listen for it, were unable to detect whether or not there was a fan in the box.
This is a good conversation. However, there is one HUGE part of the discussion that no one seems to be thinking about. The things that have been mentioned so far are important:
• Speed
• Energy efficiency
• Control of the development cycle
Those are all very important and worthwhile considerations in why Apple might switch to in-house processors for the Mac lineup. However, where I see the true advantage is the chip customizations that Apple has been adding over the years to their A-series in iOS devices. These customizations have allowed for innovations on iOS devices and the A11 Bionic chip includes an exponential increase in custom silicon and the innovations to go along with it.
• M11 motion coprocessor
• allows for tracking of steps and listening for Hey Siri without impacting battery life.
• Dual core ISP Neural engine
• 600 billion operations per second
• allows for Face ID
• Allows for HDR processing instantaneously.
• Allows for Portrait mode
• Perform effects on live video
• Identify objects and adjust the camera on the fly for both photos and video
• Custom controller
• Allows for fast switching between four low-power cores and two high-power cores.
• Also allows for all six cores to be used simultaneously if needed.
• GPU
• Custom, in-house, 3-core GPU
• Graphics processing
• Machine learning
• 30% faster than the previous model, which was the top of the line in mobile graphics processing.
• SSD storage controller
• Allows for much faster SSDs
• Hardware video encoder designed in-house
• Cuts the size of 4K video to less than half of what it used to be
There’s no compelling evidence to suggest that Apple would deprecate the Intel architecture on desktops. If this turned out to be the mobile architecture, or a variant of it, they could easily keep both side-by-side for years.
iOS: ARM
MacBook: ARM
iMac/Mac Pro: Intel
It wouldn’t have to be a Macintosh: it could be a Granny Smith or some other name. Apple has handled multiple CPU product lines in the past (Apple II/GS + Macintosh); adding an A-processor based light laptop line would increase diversity.
A new name is a thought I didn’t have. I dunno if they’d do that, but it’s worth considering at least!
That comment is beneath you, Jim.
They’d just need extra spicy salsa. 🙂
It keeps the riffraff out. We tend to ignore those that become rude.
It’s fantastic! Sorry, I HAD to.
I’d certainly like to see it. But can only imagine the technical challenges. I hope you’re right.
I’m sure its just the most peachy fan in the whole world 🙂
I wonder if its shifting 10 Watts of energy or 20 Watts ??
Even if they migrated every model, developers STILL have to develop and test two versions, since legacy Macs will need support for years to come. I also imagine they would not do this without some support for emulating Intel CPUs.
It’s been a month and I noticed this comment only today, is it too late to throw some insults around? 🙂