Apple Plans to Ditch Intel and Use Custom Mac Chips Starting in 2020

[oldmacs]

. If they are able to handle the transition of those older x86 apps, it would solve your problem. I wouldn't imagine Apple is that dumb just to completely negate all of those apps. Of course it wouldn't be every app, but likely 64 bit created or maintained using tools this year through 2020, something like that.

And how will they negate the loss of Windows Applications? What about incredibly useful x86 software which isn't being updated? Why should I loose access to that for a pointless architecture switch?

How on earth are they going to manage the transition of these Apps? Through battery and performance sucking emulation, thus negating any supposed benefits of ARM?

You mention iOS limitations, what limitations do you mean? And what if they removed some of those limitations? Again, we aren't talking about putting iOS on Macs, we are talking about making macOS work on ARM. Same OS.

Limitations in what Apps have access to. Limitations in the hardware that iOS devices run on. Removing those limitations would negate all the benefits you say you like about the iPad.

I don't think it would be dumbing down the Mac at all. This would be better for developers as they could create 1 app that works across iPhone's, iPad's, and Macs. All with a different UX for that given device. It would also allow Apple to introduce new technologies in their chips and bring those features to all 3 devices as well

Oh yeah that would be fantastic. Apps for the Mac Pro being limited by the processing power of an iPhone. We've already been down this path with iWork, when it was dumbed down on the Mac to achieve better cross functionality with the iOS versions. One app across all devices would be pretty inefficient and bloated, with a lack of optimisation.

New technologies such as???? There are already ways of doing this without switching toARM.

 

[Heineken]

There is very little logic in unifying the two platforms to that extent, because that means dumbing down the Mac. iOS apps have inherent limitations. What makes sense is the continued development of iOS for the mainstream, while preserving Macs and improving on them for those that need it. And that would not be a big win for many people. Again it would be the loss of the ability to run almost every desktop application out there.
[doublepost=1522938707][/doublepost]

Apple doesn't even bother updating Macs when the processors are available. They're held hostage to their own laziness.

Compiling for ARM is dumbing down? Well that's a first. Were do you get this idea that Mac apps would suddenly turn into iOS apps? It's a CPU architecture change not OS change.

 

[oldmacs]

Well that's a first. Were do you get this idea that Mac apps would suddenly turn into iOS apps?

If the core of the application is going to be targeted at MacOS and iOS, isn't it going to have to abide by the restrictions of iOS for it to be properly integrated?

 

[Heineken]

If the core of the application is going to be targeted at MacOS and iOS, isn't it going to have to abide by the restrictions of iOS for it to be properly integrated?

I have no idea where are you getting this from.

 

[DNichter]

And how will they negate the loss of Windows Applications? What about incredibly useful x86 software which isn't being updated? Why should I loose access to that for a pointless architecture switch?

How on earth are they going to manage the transition of these Apps? Through battery and performance sucking emulation, thus negating any supposed benefits of ARM?



Limitations in what Apps have access to. Limitations in the hardware that iOS devices run on. Removing those limitations would negate all the benefits you say you like about the iPad.



Oh yeah that would be fantastic. Apps for the Mac Pro being limited by the processing power of an iPhone. We've already been down this path with iWork, when it was dumbed down on the Mac to achieve better cross functionality with the iOS versions. One app across all devices would be pretty inefficient and bloated.

Okay, I will try to work through these questions. You potentially wouldn't have to lose Windows applications. Windows is running on ARM today at limited performance. Maybe 2 years from now it is working much better. If the developer wants to bring that software to this new platform, they will update. If not, they would continue to support on Windows. You wouldn't have to lose access, there would still be a platform supporting it on some level. If not, maybe there is another app that does the same thing. That is up to Apple to manage the transition. If they can maintain proper battery life and performance, than it should be seamless. Again, we are talking about macOS on ARM, not iOS on a Mac. Very different situations. Let's consider the OS stays the same with the same features, it just may need apps to be updated. Lastly, nobody is saying take an iPhone chip and put it in a Mac. This would be a completely different chip made for a laptop/desktop. As of now, we have no idea what Apple is capable of given the proper space/power. I think they have very smart people though in their chip design team and I am sure they can bring something worthwhile to the table.

All in all, your only worry is apps. Which is completely valid, but apps will either need to be updated for ARM or alternatives will need to be found/created. Otherwise, you may need to move to Windows if you are unable to find an alternative. I think the small amount of people in that last bucket is not reason enough to stop Apple from making this transition.

 

[laz232]

It seems that the most important thing for most people on Macrumors forum is Windoze? But i get it, i know the pain of Missing games and getting some games through Steam after 2-3 yrs.

I'm waiting for these to come to Mac (/sarcasm):

https://www.zuken.com/en/products/pcb-design/cr-5000
https://www.keysight.com/en/pc-1297113/advanced-design-system-ads?cc=NO&lc=eng
https://www.altium.com/
https://www.cst.com/
https://www.autodesk.com/products/3ds-max/overview
https://www.zuken.com/en/products/pcb-design/cr-5000
MS Office Suite (hate it, use LO in my business internally, but my clients use it, and Office for Mac 2011 soured me on Office inter-compatibility)

not to mention RDP into work machines.

And before anyone asks - I prefer macOS for terminal, Python, general computer usage, web browsing, ssh etc. - and then I run real (pro) sw in a VM. With current Intel Macbooks I can easily integrate into Windows and Linux customer's systems. With an ARM Macbook, Apple will have lost me as a customer.

 

[DNichter]

If the core of the application is going to be targeted at MacOS and iOS, isn't it going to have to abide by the restrictions of iOS for it to be properly integrated?

I wouldn't think so. I think the goal is to create a singular platform for developers to create 1 app that can be applied to different devices. One for touch on the smaller screen (iPhone), one for touch on the larger screen with keyboard support (iPad, maybe mouse support eventually), and one for large screens with point and click (Mac). I think Apple could create tools to do this.

 

[ipponrg]

I wouldn't think so. I think the goal is to create a singular platform for developers to create 1 app that can be applied to different devices. One for touch on the smaller screen (iPhone), one for touch on the larger screen with keyboard support (iPad, maybe mouse support eventually), and one for large screens with point and click (Mac). I think Apple could create tools to do this.

So what you're saying is ... let the compiler do it? I'm surprised they haven't thought about this yet!!

 

[Heineken]

I wouldn't think so. I think the goal is to create a singular platform for developers to create 1 app that can be applied to different devices. One for touch on the smaller screen (iPhone), one for touch on the larger screen with keyboard support (iPad, maybe mouse support eventually), and one for large screens with point and click (Mac). I think Apple could create tools to do this.

Apple already does this. There are apps that have iPhone specific and iPad specific UI. Going with ARM would simplify development by adding a possible 3rd target (Mac). A developer then would be able to have one project and compile for 3 targets if he wanted to.

But that doesn't mean that you would get some dumbed down Mac app based on iOS app. Think of it like for example GarageBand for iPhone, iPad and Mac.

 

[DNichter]

So what you're saying is ... let the compiler do it? I'm surprised they haven't thought about this yet!!

I am not a developer so I apologize if confused, but if there is a compiler to handle that task, what difference does it make if the app being created is for x86 or for ARM?
[doublepost=1522941247][/doublepost]

Apple already does this. There are apps that have iPhone specific and iPad specific UI. Going with ARM would simplify development by adding a possible 3rd target (Mac). A developer then would be able to have one project and compile for 3 targets if he wanted to.

But that doesn't mean that you would get some dumbed down Mac app based on iOS app. Think of it like for example GarageBand for iPhone, iPad and Mac.

Okay thanks for clarification, that makes plenty of sense.

 

[CodeJoy]

Starting in 2020, a keyboard overhaul for mobile devices is expected with a non-zero probability. According to people familiar with the matter, letters C, M, F, W, Y, G, P, B, V, K, Q, J, X and Z may be dropped from the on-screen keyboard due to not being frequently used by the majority of users. Digits will be completely discontinued because everyone hates math. Instead, the missing keys will be replaced by emojis, which most users find more productive and easier to use. External keyboards will continue to function as long as they comply with the new keyboard layout, though they will carry a price premium and a performance penalty for reasons. Users with work related tasks that require the use of any of the deprecated letters are advised to finish those tasks before 2020 or switch to alternative devices. It is unclear at this time if the same letters will be removed for all languages, or whether the new emojis will be the same for all regions.

Productivity apps will remain available if anyone can figure out how to use them. But who cares?

New wristbands will be released to soften the blow.

Embrace the change.

 

[cram501]

I wouldn't think so. I think the goal is to create a singular platform for developers to create 1 app that can be applied to different devices. One for touch on the smaller screen (iPhone), one for touch on the larger screen with keyboard support (iPad, maybe mouse support eventually), and one for large screens with point and click (Mac). I think Apple could create tools to do this.

Just writing applications for multiple mobile devices with different resolutions on different operating system versions gets complicated enough. Adding another device, like a Mac, with much broader set of capabilities, I would view as a nightmare for all but the simplest applications. Every device, every resolution, every operating system version, every differing capability make it more and more complicated. What you usually end up with is a lowest common denominator of features for all devices. For most applications, I wouldn't want a limited application built for a mobile device on my Mac.

 

[0924487]

Windows supports ARM nowadays. What's the big deal as ARM Windows can run desktop x86 apps. Everybody is blowing this out of proportions as the underlying CPU simply doesn't matter like it did back in 2005. Apple now makes the most power efficient chips in the world. The x86 is fast but it is the modern day equivalent to the PowerPC. Also, why give Intel all the $$$ for buying their chips when Apple has its own more than capable chips. Maybe if users need x86 then Apple could load it on a sort of CPU daughterboard to allow those who need it to use it.

Windows can run on ARM, but x86 apps cannot run on ARM windows lol. You must meant ARM. versions of x86 level apps

 

[DNichter]

Just writing applications for multiple mobile devices with different resolutions on different operating system versions gets complicated enough. Adding another device, like a Mac, with much broader set of capabilities, I would view as a nightmare for all but the simplest applications. Every device, every resolution, every operating system version, every differing capability make it more and more complicated. What you usually end up with is a lowest common denominator of features for all devices. For most applications, I wouldn't want a limited application built for a mobile device on my Mac.

But lets say it worked the other way around. You create an app, has all of the features that you want for the Mac version. Then it is compiled, executed, whatever you call it for the iPad, which maybe cuts out a couple features that it can't support, and then the iPhone, which does the same thing. Would that be a better solution? I keep hearing dumbed down, but I am not sure it has to be that way. There are full featured iOS apps as well, it just comes down to the capabilities Apple allows and developers willing to create the apps.

 

[obitow]

I think you're trolling now because it's 110% clear you're being stubborn, naive, and exemplifying your inexperience in this domain.

Have you heard of Virtualbox, Parallels, or VMWare? If not, Google it and learn something new.

You misunderstood me. I didn't say Windows. I said Windows machine, or a PC other than Apple's to be more precise.

EDIT: In fact, I wasn't clear, my bad. A Mac can be a Windows machine.

PS. Sorry if my comments looked trollish. English is not my native language and I might have used expressions that looked like I was trolling. I'd highly appreciate if you could help me improve.

 

[cram501]

But lets say it worked the other way around. You create an app, has all of the features that you want for the Mac version. Then it is compiled, executed, whatever you call it for the iPad, which maybe cuts out a couple features that it can't support, and then the iPhone, which does the same thing. Would that be a better solution? I keep hearing dumbed down, but I am not sure it has to be that way. There are full featured iOS apps as well, it just comes down to the capabilities Apple allows and developers willing to create the apps.

It doesn't have to be that way but it's that path of least resistance. Universal applications are fine for simple applications but tend to start stumbling when a complicated feature sets is/are added to support every permutation of device, screen size, operating system, capabilities, etc.... IMO, once an application starts getting complicated, needs optimizations, etc... it's easier to write a separate application. A simple one for the mobile devices and a more comprehensive one for pc's.

But this discussion really has nothing to do with the architecture change. Thats a development environment issue.

 

[DNichter]

It doesn't have to be that way but it's that path of least resistance. Universal applications are fine for simple applications but tend to start stumbling when a complicated feature sets is/are added to support every permutation of device, screen size, operating system, capabilities, etc.... IMO, once an application starts getting complicated, needs optimizations, etc... it's easier to write a separate application. A simple one for the mobile devices and a more comprehensive one for pc's.

But this discussion really has nothing to do with the architecture change. Thats a development environment issue.

Understandable. I was only focusing on the development side because I think that is the biggest sticking point for people and I wanted to know more about it. Thanks.

 

[dogslobber]

Windows can run on ARM, but x86 apps cannot run on ARM windows lol. You must meant ARM. versions of x86 level apps

No I don't mean that at all.

 

[techwhiz]

For a person who claims to know so much about cpus - you don’t seem to know much about the a11.

It has 2 high performance cores. The other 4 cores are high efficiency and are much slower. So you cannot just divide by 6. Just look at the single core result to see what the high performance core is capable of.

Look at the iPad Pro using the last generation a10x. It is approx 9500x1.29/3.

In any case, Apple won’t just plonk an a11 in a Mac and call it a day. Mobile processor max out at less than 3w compared to 45w in the MacBook Pro and 100w + in the iMac

"Clearly that's not how we evaluate performance but even normalizing for speed; the multicore performance of the"
You didn't read my post.

But I'll say what I have said before.
The only obstacle of Apple making a processor to compete with Intel for Laptop/Desktop/Server applications is time, money, fab and people resources.

1. Money? So much cash they need to figure out creative ways to spend it. They pay really good dividends to the shareholders too. So that's not a problem.

2. Time? I don't know at what point they started or will start but two years isn't enough time for a family of processors and the story said 2020.

3. Fab? They haven't bought one and to date Global Foundries, TSMC, Samsung or anyone else has announced wafer agreements that Apple would need for this kind of effort. Global Foundries is an AMD partner. Intel uses their own FAB. So I'm still looking for Fab. capacity and they haven't spent the tens of billions to build a fab.

4. People? Can they hire them? Sure. I just haven't seen the massive hiring they need for silicon types like me in the last couple of years or even now. They all need to expand the teams in SEG (Silicon Engineering Group). My friends that are like me doing chips haven't seen massive hiring or gotten any calls.

So my doubt about this story is not based on questions of ARM performance because both Qualcomm and Cavium have server class ARMv8 implementations and there isn't any question if Apple can do it also.

My doubt is whether there is real ROI to spend the billions to compete on performance with Intel. Trying to compete with Intel has driven multiple companies out of the market and to the brink of bankruptcy. That won't happen to Apple, but it could be a misstep that costs time.

Intel has 100K employees and a whole lot of them focused on x86, memory, high speed SERDES, PCIe root complex and all the stuff they need in their chips. Intel owns its fabs and can do port wafers early in the process to work on power optimization and other issues.

I doubt the veracity of Bloomberg and their predictions. This is the same group that said this two years ago and also claimed that by now, Apple would be using it's own modems.

 

[cmaier]

"Clearly that's not how we evaluate performance but even normalizing for speed; the multicore performance of the"
You didn't read my post.

But I'll say what I have said before.
The only obstacle of Apple making a processor to compete with Intel for Laptop/Desktop/Server applications is time, money, fab and people resources.

1. Money? So much cash they need to figure out creative ways to spend it. They pay really good dividends to the shareholders too. So that's not a problem.

2. Time? I don't know at what point they started or will start but two years isn't enough time for a family of processors and the story said 2020.

3. Fab? They haven't bought one and to date Global Foundries, TSMC, Samsung or anyone else has announced wafer agreements that Apple would need for this kind of effort. Global Foundries is an AMD partner. Intel uses their own FAB. So I'm still looking for Fab. capacity and they haven't spent the tens of billions to build a fab.

4. People? Can they hire them? Sure. I just haven't seen the massive hiring they need for silicon types like me in the last couple of years or even now. They all need to expand the teams in SEG (Silicon Engineering Group). My friends that are like me doing chips haven't seen massive hiring or gotten any calls.

So my doubt about this story is not based on questions of ARM performance because both Qualcomm and Cavium have server class ARMv8 implementations and there isn't any question if Apple can do it also.

My doubt is whether there is real ROI to spend the billions to compete on performance with Intel. Trying to compete with Intel has driven multiple companies out of the market and to the brink of bankruptcy. That won't happen to Apple, but it could be a misstep that costs time.

Intel has 100K employees and a whole lot of them focused on x86, memory, high speed SERDES, PCIe root complex and all the stuff they need in their chips. Intel owns its fabs and can do port wafers early in the process to work on power optimization and other issues.

I doubt the veracity of Bloomberg and their predictions. This is the same group that said this two years ago and also claimed that by now, Apple would be using it's own modems.

Again, they dont need to hire anyone new. They have more than enough of the right people with the right experience to get it done. They have most of the design team that designed chips like AMD’s opteron. They have much of the design team that did DEC Alpha, Exponential x704, etc. Many of the folks that work in that group have previously designed UltraSparcs, x86’s, PowerPC desktop processors, etc. It took 20 experienced physical/logic/circuit designers and another 50 or so layout and verification folks to do entire x86 processors at AMD. They have many many times that number already there.

They also don’t need massive fab capacity - mac shipments are a tiny percentage of iOS shipments. Even if the processors are 4x the die area, they would need to increase their wafer orders from TSMC by only a tiny percentage.

 

[dannys1]

You've got it the wrong way around. Intel can't make a CPU as fast as ARM at low power. Intel uses a lot of silicon which uses more power to implement complicated designs to squeeze out the best performance from a complicated legacy architecture. Intel's advantage has mostly been in having advanced fabrication processes, but it is losing that advantage rapidly.

The reason no-one makes high-performance ARMs is becuase users of that arch are mostly interested in low power usage and don't want to invest in the substantial R&D required to implement high performance features. Unfortunatley for Intel Apple has been doing exactly that with their low power ARMs and are now taking the next step to implement them in laptops with somewhat higher performance and power usage.

We will likely never see them in desktops though becuase Apple makes relatively little money there and it's not worth the effort. Intel chips will probably stay there.

Seems today Bloomberg have said they Apple won't ditch x86...so it might not be ARM but rather their custom version of an AMD or something. I prefer that option to be honest. The best of both worlds. Compatibility going forth and Apple being able to differ themselves from PC's just using the same Intel chips.

 

[jblagden]

If it runs well and keeps Linux and Windows compatibility, I am all for it. It feels like the Intel professor cycle has been a bottle neck for Macbook releases in the past several years.

I know that Debian has ARM ports for a few different types of ARM processors (ARM64, ARMel, ARMHF), but I don't know about other distributions.

 

[Sydde]

Just writing applications for multiple mobile devices with different resolutions on different operating system versions gets complicated enough. Adding another device, like a Mac, with much broader set of capabilities, I would view as a nightmare for all but the simplest applications. Every device, every resolution, every operating system version, every differing capability make it more and more complicated. What you usually end up with is a lowest common denominator of features for all devices. For most applications, I wouldn't want a limited application built for a mobile device on my Mac.

Well, most traditional Mac apps that do stuff are not bound by screen resolution. Content is often displayed in a resizable window, so the Mac app must be flexible in dealing with potentially changing view constraints. When this ethos is imposed on iOS, the question of varying screen size becomes all but second nature. Translating things like the menu bar and modified clicks might be more challenging, but not that much more.

macOS and iOS have basically the same underlying system in the Foundation Framework. The main difference is in UIKit vs AppKit but it is not a huge difference. I personally do not envision macOS being “dumbed down” to the level of iOS as much as iOS rising toward macOS, though I could be completely off-base on that. There might be a “Pro” option for some iDevice users that brings more mac-like accessibility while keeping the light-duty iDevice users in a more protected environment.

 

[BeforeTheMeds]

I would not actually own a mac until yhey switched to intel. Worked on them, used some, but app suuport and compatibility was horrible. Bootcamping windows was the only way. Bye, bye.

Maybe good news for those that only surf the internet. Maybe.

 

[techwhiz]

Again, they dont need to hire anyone new. They have more than enough of the right people with the right experience to get it done. They have most of the design team that designed chips like AMD’s opteron. They have much of the design team that did DEC Alpha, Exponential x704, etc. Many of the folks that work in that group have previously designed UltraSparcs, x86’s, PowerPC desktop processors, etc. It took 20 experienced physical/logic/circuit designers and another 50 or so layout and verification folks to do entire x86 processors at AMD. They have many many times that number already there.

They also don’t need massive fab capacity - mac shipments are a tiny percentage of iOS shipments. Even if the processors are 4x the die area, they would need to increase their wafer orders from TSMC by only a tiny percentage.

My last processor (recent) job was extensions for fault tolerance.
It was ISO26262 and ASIL (safety Element Out of Context) SEOoC chips.
We had more than 20.

We will agree to disagree on whether they currently have enough people to do a family of processors.
I think they don't have enough. I know people in SEG in both the processor and graphics groups.
The'll also need support chips, but they needed that for PPC.
I know people that left Apple after they got rid of PPC that are back in SEG working on chips again.
But lets face it; that number really isn't important.
We both agree it's doable.


Also my issue isn't the number of wafers, it will be how fast they can get them.
But then, you can always pay for "hot lots" and "skew lots" for characterization.
Yes, mobil shipments will dwarf the number of desktop orders they have.
Once again, we agree that in some form they can do it.

It really boils down to ROI.
Is there return on investment doing a family of laptop/desktop processors?
They have to keep their performance equivalent to Intel or it will be considered a fools errand.
Once again, I don't have doubts that they can do it.
Apple has very, very deep pockets.
If you are a key component in a device, Apple is trying to figure out how to design you out the day you get the design win. Ask Portal Player and Imagination Tech.

I just don't think Bloomberg knows what they are talking about.