Intel CEO Responds to Rumors of ARM-Based Macs, Says Relationship With Apple Is 'Strong'

[LordVic]

Wine still worked with PPC, with Qemu. Also, VMs will still work for those OSs. Xen and KVM, for instance, have already been ported to ARM. If you work with 'free software', then you have the least to worry about. Go look at the Debian ports. Already most of the open-source software out there has been compiled for ARM, even chromium works on ARM and has worked on the raspberry pi for a long time, for instance. So, you'll be able to use all the open-source software you want.

WINE = Wine is Not an Emulator. (the literal translation of the name)

it does not work like you think it works. you cannot use wine, without some other emulation going on in the background if you're not on an x86 architecture. http://en.wikipedia.org/wiki/Wine_(software)

Emulation via VM's also requires the CPU to emulate or support it at a hardware level, or else you are going to have overhead of the emulation layer, which, in current technology, arm is unable to provide sufficient power to do in any meaningful way. While you can emulate x86 on other platform, the performance overhead would make it entirely innneficient to do so, and honestly, quite painful.

Chromium s not an emulated program. Chromium is Chrome, that has been ported to use linux. the linux flavour MUST be ported to run on that other architecture itself. it is not emulated, but actually changed to handle the different instruction set. You cannot just take linux and install it to Arm, you MUST use an arm code base version of Linux.

You will only be able to use whatever software that has been specifically written for the Arm architecture. you can't just take your standard LInux binaries and run them on Arm.

 

[b34g]

Rest assured you are exactly right, for the reasons you state and a lot more. Proof of this is Apple's own PPC to x86 transition. The software ecosystem consists of a lot more than a few mainstream apps. It took *years* before all the PPC apps and utilities that customers relied on were ported.

No computer company in history with a significant base of binary-compatible software has ever changed to non-compatible CPU easily or overnight.

However history is littered with failed attempts where they *thought* it would be easy -- "just recompile". A notable example was the FHP computer documented in Tracy Kidder's Pulitzer Prize-winning book "Soul of a New Machine": https://en.wikipedia.org/wiki/The_Soul_of_a_New_Machine

For a time Microsoft supported Windows NT and various apps on three different CPU architectures: x86, MIPS and DEC Alpha. It was only possible to port a small number of internal apps and even that was a huge job. And that was using a code base specifically optimized for porting.

A more recent example is Intel's own Itanium CPU. A key reason it failed was apps could not just be recompiled.

Even more recently, a key reason Microsoft's Windows RT tablets failed is due to binary incompatibility with x86, which in turn limited available apps. If it was as simple as "just recompile", WinRT tablets would have had every possible app from day 1.

So there's a lot more involved than "just recompile".

With software that is not locked into a specific CPU instruction set -- e.g, Google apps like Gmail, Sheets, Docs, etc, of course you can change CPUs but this software is only a small % of the overall desktop and laptop marketplace.

Here is the difference now in modern times. Apple has a large amount of IOS apps, which they could easily make work on an OS X arm version. And in addition, web apps are much more popular now than in the past era.

 

[casperes1996]

Why not core-m?

I don't buy all these ARM rumours. I find it much more likely they'd use aCore-M design. The Core-M series would allow for fanless designs as well, with a flexible top, would run existing software without any dynamic translation or anything like that (which is a big one since it could harm both performance and battery life for potential ARM products), wouldn't put this great Intel relationship at risk, and lastly, Core-M's GPU component is extremely good. Whilst you could certainly fit something respectable on an ARM based chip, such as a high end Imagination chip, it still wouldn't rival Intel's offering. This would also have benefits in compute as the Broadwell series of GPU parts support OpenCL 2.0.

 

[AppleFan360]

Basically, if Apple switches to ARM, it will kill the Mac. End of story.

 

[b34g]

WINE = Wine is Not an Emulator. (the literal translation of the name)

it does not work like you think it works. you cannot use wine, without some other emulation going on in the background if you're not on an x86 architecture. http://en.wikipedia.org/wiki/Wine_(software)

Emulation via VM's also requires the CPU to emulate or support it at a hardware level, or else you are going to have overhead of the emulation layer, which, in current technology, arm is unable to provide sufficient power to do in any meaningful way. While you can emulate x86 on other platform, the performance overhead would make it entirely innneficient to do so, and honestly, quite painful.

Chromium s not an emulated program. Chromium is Chrome, that has been ported to use linux. the linux flavour MUST be ported to run on that other architecture itself. it is not emulated, but actually changed to handle the different instruction set. You cannot just take linux and install it to Arm, you MUST use an arm code base version of Linux.

You will only be able to use whatever software that has been specifically written for the Arm architecture. you can't just take your standard LInux binaries and run them on Arm.

This is why I said, that using wine in conjunction with qemu is a trick that was always well-known, dating back to the PPC era. Qemu is an emulator, that for instance, google uses for their android arm emulator. As for all the linux software, again, various distros, like Debian support ARM, and have already compiled vast swathes of software for ARM. Look at the Debian packages yourself. Even Chromium has been compiled for ARM and is used in the raspberry pi as a package, for instance. They compile fine.

 

[PowerMac G4 MDD]

If Apple did this, it would be catastrophic. Their move to Intel was wise, as Intel became better, but they completely kicked-out their PPC users, not to mention leaving-behind some of the greatest Macs ever created. It would be so annoying to have it happen again, and have Intel Macs go down the drain. The ONLY benefit I see would be ARM processors in their smaller computers, such as their MacBook Air, which REALLY needs to be seeing a price drop. I am starting to not like Apple's current tactics, but if THIS happens, I am going to hate them. And of course, it seems as though Apple has an even larger mentality of "never looking back", so I am sure they would ditch their Intel Macs and think nobody would care... I am 100% sure that they have NO idea of the upheld fan-base behind PowerPC Macs. Talk about not knowing anything about your loyal users, Apple!

 

[thekev]

Intel is also doing US no favours by their CPU choices in all their hardware. The Mac Airs are excused for their CPU choice due to power and heat restrictions.

But in the new Mac Mini, and even the lowest end the newer lowest end iMac, intel has decided to use the ULV intel part. The ULV (ultra low voltage) components typically cost more for each watt of performance than the similarly priced Low Voltage (LV) and standard CPU.

I assume you mean Apple here.

this is mostly noticable in latest mac Mini iteration where they opted to use the same intel ULV chip as the Mac Mini at 17w than the previous quad core 25w part, or even the larger faster quade core desktop 85w part. for that same $200, intel could be providing i3/i5 quadcores at 3.0ghz or faster.

That is true. I don't think that would improve by going to ARM. I mean you would see a drop in performance rather than an increase. Apple has allowed that at times on major shifts, although there are sometimes offsets. I would be surprised if they split the Mac lineup between two totally different architectures.

Apple could easily provide cheaper products that give better performance than some of their newest ones, but are opting not too, likely in order to maximize profits by reducing the diversity of their internal components and build requirements.

That has always been the case.

 

[joema2]

...iOS runs on three different types of CPUs and, indeed, apps these days are typically compiled for all three and you really wouldn't notice the difference....

So if all these iOS apps are compiled for three different binarily-incompatible CPUs, how come there's only a single app download from the iTunes app store? How does the one compatible version (out of three) find its way to my iPad?

The answer is the iOS CPUs are all binary compatible, and there is a single executable download. Saying iOS runs in three different types of CPUs is like saying OS X runs on different Intel CPUs. The microarchitectural features may differ, but they are share essentially the same compatible binary instruction set.

 

[canucksfan88]

honestly if that happens. I'm out.

I will hold onto my Intel mac for as long as i can, and then switch back to Windows.

 

[b34g]

One of the reasons for Apples incredible surge in PC sales over the last decade was the switch to Intel. Especially in the BYOD era of computing.

if suddenly Apple computers were no longer compatible with 95% of the worlds enterprise architecture and would no longer be supported by virtually every corporation due to lack of Windows / Linux capability (either via parallels or bootcamp), then Apple would lose any market for enterprise that they might currently be in.

this would happen because Apple has no real enterprise back end solutions when it comes to.. well, Enterprise. its a realm they've stayed far away from. Especially since they canned the xserves.

thats why I dont get at all why this rumour keeps popping up every 6 months or so. for apple it makes absolutely Zero sense at all. none. it would be a financial noose.

----------



while AMD is doing an Arm core in their CPU's, it's not being done in a way that can be used as such. It's more an attempt to bring about a certain low power security and performance metric. It's not going to be "either or". it will be "x86, that borrows some arm routines". and I believe is being aimed more at parallel server workloads. details are a bit sketchy (unless they've done a better announcement since). and isn't likely to be seen till 2016

Apple doesn't focus on selling volume, they focus on maintaining high margins, if ARM can give them higher margins, while not sacrificing usability as Apple determines (as they have with the iphone/ipad, etc. despite critics in the past calling them 'non-practical toys' with 'average specs', etc.), then it is not a big jump for them. Remember, Apple also killed the xserve, and the old serious mac pro, despite there being a customer base for those products. It did result in a shrink for mac demand as a consequence in those markets, but then again, Apple looks at margins, not at volume.

 

[rekhyt]

Sounds like someone's in denial.

 

[76ShovelHead]

If apple were to go this route they'd be better off partnering and/or purchasing AMD to strengthen AMD's efforts in hybrid x86+ARM cores (which would def. ease a transition to ARM with keeping the compatibility of current Apps). With their expertise in developing strong ARM processors and AMD's ability to create SOCs with actual graphics Apple could really push the power/portability envelope. AMD struggles because they don't have the resources Intel has to develop truly cutting edge chips, Apple could deliver these resources ten-fold.

And with such a small marketshare, AMD could easily become exclusive to Apple products.

 

[LordVic]

Apple doesn't focus on selling volume, they focus on maintaining high margins, if ARM can give them higher margins, while not sacrificing usability as Apple determines (as they have with the iphone/ipad, etc. despite critics in the past calling them 'non-practical toys' with 'average specs', etc.), then it is not a big jump for them. Remember, Apple also killed the xserve, and the old serious mac pro, despite there being a customer base for those products. It did result in a shrink for mac demand as a consequence in those markets, but then again, Apple looks at margins, not at volume.

Margins though can only make you so much money. 45% margin is useless if you're not selling anything.

the bean counters would have to determine if loss of revenue potential can be made up by increased margin.

 

[mi7chy]

While it's possible we might see an ARM based Mac addition but to replace Intel is a foolish rumor considering Intel closed the power efficiency and fanless gap with Core M. Personally, I've shifted my ARM tablet purchases to all Core M as of late mainly due to the fact that x86-64 has the greatest selection of professional software. Apps are fun but to do real work you need x86-64 software.

 

[b34g]

this is not the reason for it.

Intel already has chips (and more in pipeline) that are dirt cheap and offer similar price to performance that Arm in a desktop would. Apple has chosen to not go after this market and therefore doesn't bother with those chips.

if it were for a cheaper retail product, there's no reason why Apple doesn't move towards the Atom CPU's that are already out in the wild. They offer similar price, performance and power usage than the rival Arm chips, but remain within the x86 architecture.

the only reason for a shift to Arm for Apple would be pipeline control and removal of intel as a vendor. They would retain control of their chips like they do now for the ipads and iPhones. That is it.

And what was the state of Apple PC's last time they were their own architecture, incompatible with Windows and the rest of the PC world? thats right, < 2% worldwide marketshare of Personal computers and near bankruptcy in the 90's.

Exlusivity isn't the way to go here. Move to Arm would be a disaster. Intel knows that. Apple knows that. For some reason the pundits and forumers dont

The Atom is a joke, it has pathetic performance per watt compared to ARM, which is why it is only found in super-cheap laptops, and only in subsidized smartphones/tablets, by intel, yet despite all this, it still suffers from mediocre battery life complaints and low market share compared to ARM. As ARM maintains superior performance per watt, it will be easier for ARM to scale up, rather than for Intel to scale down.

----------

Margins though can only make you so much money. 45% margin is useless if you're not selling anything.

the bean counters would have to determine if loss of revenue potential can be made up by increased margin.

Why do you assume Apple would not sell anything? They sold products with powerpc before too.

 

[LordVic]

The Atom is a joke, it has pathetic performance per watt compared to ARM, which is why it is only found in super-cheap laptops, and only in subsidized smartphones/tablets, by intel, yet despite all this, it still suffers from mediocre battery life complaints and low market share compared to ARM. As ARM maintains superior performance per watt, it will be easier for ARM to scale up, rather than for Intel to scale down.

are you talking bout the brand new arms or the outgoing generation (considering that intel is actually still selling both as current products).

the new upcoming ones based on Broadwell are proving to be fantastic in the price to power ration. will try and find you some benchies a bit later.

I do think arm will scale up well. But then you run into the problem that it will require more power to do so, so many of the powersaving features would be lost.

the K1 from nvidia is a great example of this. Yes, it scaled the A15 cores up to offer performance that far exceeds that of intel's Atom's. But the cost is power consumption and heat. Nvidia has already stated the K1 isn't intended for the lower power usages that Arm was previously benefiting from and intended for different use.

 

[scottsjack]

Don't worry Intel. My next computer will have an Intel CPU. Whether Apple wants to participate or not is of course up to them. I need Windows and user upgrade-ability so that is what I will buy.

I prefer to work on a Mac but I see glued-together computers with tiny keyboards in Apple's future.

 

[LordVic]

Why do you assume Apple would not sell anything? They sold products with powerpc before too.

Except that Intel over the life of PPC machines Apple saw dwindling revenues, market share and by the end of the PPC's life, Apple saw less than a 2% market share of total computer install base in the world.

so yeah, they sold some products... but they almost went bankrupt doing it


it wasn't Apple's computer line that saved the company. it was cash injections and then a return of Steve Jobs, the iPod and i-devices insane market growth that resurrect a company most people had left for dead. They moved to Intel to ressurect their dead computer line which had seen itself nearly eliminated from all modern usages around the world.

 

[nutjob]

But why do that? when intel offers chips that can already do that at the same cost as the Arm chips. Even AMD has the same.

The move to Arm would be for pipeline control. and pipeline control only.

That's patently untrue, but you keep repeating that without supporting evidence. Intel costs much more than ARM. Maybe you're talking about the mobile chips on which Intel is losing $1000 on each $1 it makes. Short of that Intel costs much more than ARM.

 

[mattoligy]

It would be a good idea for Apple to just use ARM in the new super slim MacBook Airs and keep the rest of the line Intel! A version of OS X coded for ARM just for the super mobile MacBook line would allow iOS apps to run on said OS X and also create a transition period for software companies to port their applications to ARM for use on the MacBook Air Line and those that don't want to wait, just get a MacBook Pro... Everyone would be happy... Intel and ARM should coexist in the MacBook lineup to create diversity!

It is the step toward a "hybrid" in the future... I know Apple say they won't do that but they say plenty of crap...

 

[LordVic]

That's patently untrue, but you keep repeating that without supporting evidence. Intel costs much more than ARM. Maybe you're talking about the mobile chips on which Intel is losing $1000 on each $1 it makes. Short of that Intel costs much more than ARM.

yeah, I'm not really putting consideration into intel's profitability on those chips. I'm only talking about it from the consumer / Apple focus. if intel loses money on each chip, thats their choice

 

[b34g]

are you talking bout the brand new arms or the outgoing generation (considering that intel is actually still selling both as current products).

the new upcoming ones based on Broadwell are proving to be fantastic in the price to power ration. will try and find you some benchies a bit later.

I do think arm will scale up well. But then you run into the problem that it will require more power to do so, so many of the powersaving features would be lost.

the K1 from nvidia is a great example of this. Yes, it scaled the A15 cores up to offer performance that far exceeds that of intel's Atom's. But the cost is power consumption and heat. Nvidia has already stated the K1 isn't intended for the lower power usages that Arm was previously benefiting from and intended for different use.

Intel has a superior manufacturing process to ARM Fabs like TSMC or Samsung, however it does not have a superior architecture. What Intel does is make their chips on 1-gen ahead fab technology, and then compares it to ARM chips made with older fab technology, yet, even then ARM still outperforms Intel, even when Intel had a shot to have the best performance, which was in late 2013, Apple blew the market away with the A7. So, on the same technological process, ARM remains superior to Intel, and even 1 gen behind, Intel still can't catch up. It's inferior x86 technology.

----------

Except that Intel over the life of PPC machines Apple saw dwindling revenues, market share and by the end of the PPC's life, Apple saw less than a 2% market share of total computer install base in the world.

so yeah, they sold some products... but they almost went bankrupt doing it


it wasn't Apple's computer line that saved the company. it was cash injections and then a return of Steve Jobs, the iPod and i-devices insane market growth that resurrect a company most people had left for dead. They moved to Intel to ressurect their dead computer line which had seen itself nearly eliminated from all modern usages around the world.

Actually, Apple's rise started in the early 00s, with Steve Jobs. So, Apple had already recovered before it transitioned to Intel. What made Apple huge, was the iphone, starting in 2007.

 

[joema2]

Here is the difference now in modern times. Apple has a large amount of IOS apps, which they could easily make work on an OS X arm version. And in addition, web apps are much more popular now than in the past era.

If the Mac has an ARM CPU, it loses all binary compatibility with the gigantic existing base of OS X x86 apps. Who would want a Mac that could only run mostly iOS apps?

Web apps are not sufficiently popular to make the transition to an ARM-only Mac feasible. If you rely on that, why not get a ChromeBook?

An ARM MacBook would essentially be an Apple ChromeBook. It would not be very compelling -- even if it could run iOS apps.

OTOH it is conceivable Apple is experimenting with merging iOS and OS X. One way to do that is use both ARM and x86 CPUs in a single machine. A super-thin MacBook Air or a iPad Pro that could *natively* run iOS apps and OS X x86 apps would be interesting. It would require a lot of software engineering but it could be done much quicker than porting 10 years of OS X x86 apps to ARM.

 

[LordVic]

Intel has a superior manufacturing process to ARM Fabs like TSMC or Samsung, however it does not have a superior architecture. What Intel does is make their chips on 1-gen ahead fab technology, and then compares it to ARM chips made with older fab technology, yet, even then ARM still outperforms Intel, even when Intel had a shot to have the best performance, which was in late 2013, Apple blew the market away with the A7. So, on the same technological process, ARM remains superior to Intel, and even 1 gen behind, Intel still can't catch up. It's inferior x86 technology.

But regardless of the fab and what not. it comes down to what is available right now to the use for performance.

I'm well aware of the defficiencies of x86. not trying to sound like its the best thing ever. We've seen intel seriously goof up before (Pentium4's were a disaster).

I'm also aware that this might change dramatically within the next, 2, 5 or 10 years.

But we can only go by what is currently out, and what is in the immediate pipeline. And based on that. What I stated still holds true.

p.s. I am actually enjoying this discussion. I think it's rather informative and there's no ill will or anger behidn it.

 

[b34g]

If the Mac has an ARM CPU, it loses all binary compatibility with the gigantic existing base of OS X x86 apps. Who would want a Mac that could only run mostly iOS apps?

Web apps are not sufficiently popular to make the transition to an ARM-only Mac feasible. If you rely on that, why not get a ChromeBook?

An ARM MacBook would essentially be an Apple ChromeBook. It would not be very compelling -- even if it could run iOS apps.

OTOH it is conceivable Apple is experimenting with merging iOS and OS X. One way to do that is use both ARM and x86 CPUs in a single machine. A super-thin MacBook Air or a iPad Pro that could *natively* run iOS apps and OS X x86 apps would be interesting. It would require a lot of software engineering but it could be done much quicker than porting 10 years of OS X x86 apps to ARM.

A Chromebook doesn't have a full-OS, Apple would still port all their apps, and in addition, open-source apps like VLC, or Transmission which already work on arm linux, would be quickly ported too. Java apps would also work. Apple would of course push developers through the mac app store to port their apps to ARM too.