The author gives Apple's previous two CPU transitions as an indicator for a successful ARM transition. But he's kind of ignoring the fact that in those two prior transitions, the CPU being moved to was significantly faster than the CPU being migrated from. This then allowed the use of emulation software ("Rosetta") to let users continue using "legacy" software without too much of a performance penalty. ARM is not significantly faster than Intel chips, so how will Apple handle this transitional period in which users need/want to continue using legacy software?
That's not completely true. First, Intel's CPU were not significantly faster than the PowerPCs they replaced. They were fast enough to allow Rosetta to work without users noticing much of change in performance. Also, Rosetta was not an emulator it was a translator. It's a relatively small performance hit (10%) to translate one set of instructions to another. On today's computers that's fairly insignificant as most people rarely "red line" their computers.
Intel's benefit was having a lower power envelop over PPC. Apple simply couldn't put faster PPC chips into their notebooks because it would create too much heat. That's inline with the performance per watt for ARM-based chips designed by Apple today and why Intel's attempts at mobile chips failed miserably.
Finally, and I think importantly, moving the Mac to ARM would mean that Apple has to develop a much wider range of "A"-series CPUs. I just can't see Apple producing:
an "A17" for the iPhone
an "A17X" for the iPad
an "A17XX" for the MacBook (Pro)
an "A17XXX" for the iMac
an "A17XXXX" for the Mac Pro
I'm always confused by people assuming that the chips will all be A-series despite a massive number of changes that will need to be made for traditional PC use and Apple already having many other custom silicon options with different letters. So why assume they will all get additional letter X's because Apple did this for the iPad, which is much similar in terms of how it works with iPhone than a Mac.
Using the Mac for cloud development, this could become an issue for me. All major cloud providers are using an Intel architecture.
If Apple would move the whole Mac product line away from Intel there is absolutely no reason to keep a Mac as a development machine. A Dell XPS with Ubuntu will not only have the price advantage (the current situation), but also the ease of use and speed advantage.
My use case is of course only limited to a few percent of the market, but it could anyhow jeopardize the market share of Macs
Seriously if you are a developer you should already know that the architecture of the processor in the cloud is not a big deal. Most of those machines run Linux anyways and do so with specific feature (software) support on those systems. I really don’t see an issue for cloud developers as long as you still have MacOS on the box and the freedom to install your favorite (cloud providers) software.
Apparently you have no clue about cloud developing. Currently I develop locally, launching a set of Docker containers on a my development machine. The containers can run at almost native speed because the Mac has an Intel architecture. If Apple moves fully away from Intel, it is not sure I can run those Docker containers on the new architecture, and even if Apple provides the necessary tools to eventually run these containers, they will run much slower and not hassle free. Like I said, A Dell XPS with Ubuntu will have the price advantage, the ease of use and speed advantage.
It could be that Apple's goal (or vision) for the Mac is to focus on two of its biggest markets: professional content creators and developers who develop for Apple's platforms.
Using the Mac for cloud development, this could become an issue for me. All major cloud providers are using an Intel architecture.
If Apple would move the whole Mac product line away from Intel there is absolutely no reason to keep a Mac as a development machine. A Dell XPS with Ubuntu will not only have the price advantage (the current situation), but also the ease of use and speed advantage.
My use case is of course only limited to a few percent of the market, but it could anyhow jeopardize the market share of Macs
I don't get this argument: just because you're developing (code/debug/run rinse, repeat) on an ARM based Mac doesn't mean you're deploying to an ARM based cloud. I haven't personally deployed anything to the cloud, but my company does. We develop our applications on our desktops/laptops, then there's a separate build process that puts our app(s) into a container(s) (which is x86/Linux) and those get deployed into a public cloud or into a customer's private cloud.
While a tiny percentage of extreme power users may be able to benefit from a minimal performance boost by switching from Intel to ARM processors, a larger percentage of us that use some Windows-only software (Microsoft Project to name one) and thus need Bootcamp will be left in the cold. I have owned many Macs (currently a top-spec 2017 MacBook Pro 15"), and I reluctantly will stop buying Macs if Apple halts the Windows compatibility feature. The current Macs are fantastically powerful in the higher models, and a switch to ARM is not necessary.
While a tiny percentage of extreme power users may be able to benefit from a minimal performance boost by switching from Intel to ARM processors, a larger percentage of us that use some Windows-only software (Microsoft Project to name one) and thus need Bootcamp will be left in the cold. I have owned many Macs (currently a top-spec 2017 MacBook Pro 15"), and I reluctantly will stop buying Macs if Apple halts the Windows compatibility feature. The current Macs are fantastically powerful in the higher models, and a switch to ARM is not necessary.
You really believe that most people buy Macs to use with Bootcamp than would want faster and less expensive Macs with more features and lower power envelope? Well, no one can say you don't have a unique take.
While a tiny percentage of extreme power users may be able to benefit from a minimal performance boost by switching from Intel to ARM processors, a larger percentage of us that use some Windows-only software (Microsoft Project to name one) and thus need Bootcamp will be left in the cold. I have owned many Macs (currently a top-spec 2017 MacBook Pro 15"), and I reluctantly will stop buying Macs if Apple halts the Windows compatibility feature. The current Macs are fantastically powerful in the higher models, and a switch to ARM is not necessary.
While Bootcamp usage is high in the AppleInsider reading population, it is decidedly not with the vast majority of the Apple user base.
So, it's more like "while a tiny percentage of extreme power users may be left in the cold from a shift to ARM, the larger percentage that don't or have used the iPhone as a gateway to the the Mac will benefit from the performance boost."
All off a sudden we’re getting articles about this because one analyst states it’s happening. As Jon Gruber just said on his page, in an article about this:
”Also worth noting: Ming-Chi Kuo is often wrong, especially about products other than iPhones and iPads. We could be writing this same stuff a year from now and Macs could remain on Intel until the end of the platform. But I do think they’re moving to ARM, sooner rather than later.”
I agree with Jon. It’s likely to happen at some point. Sooner rather than later though, could mean three or four years.
Sooner, is this year.
That’s not a definition of sooner. Now, you’re saying it will arrive in 2020, not even 2021. Dream on.
The technology for ARM to do the extensive multi-tasking that Intel chips provide does not exist today. Arm chips can only now do very small multi-tasking operations. The technology for ARM is still several years away. The main advantage for Apple is Temperature, always on, non comparability with the open architecture that Intel has, (IE Kill Hackintosh) and MORE PROFIT. The most important for them being the last two. Because we know even with the bargain basement chip that ARM is Apple will not drop their pricing. I am old enough to remember when Mac had a completely closed architecture and sales was probably 1% of what it is today. Moving to Intel allowed them to grow to where they are today. ARM will be the death of Mac. Not Apply but the Mac for sure.
Current Arm chips are equal or better in performance compared to Intel x86 chips, single or multi (hardware) threading. ARM will be Macs biggest breakthrough.
Edit: fixed threading
Are you for real, or are you a troll? Not all trolls are against Apple here.
If Apple simply wants to turn up the heat under Intel's slow butt, why not bring some of the high end AMD processors into the Mac platform?
At some level, the bulk of customers that Apple really wants to be selling computers to really do not, and should not, care one little bit about what is under the hood of their shiny new Mac. All they need to know is that it runs every software product they need it to run, helps them complete their daily tasks, helps them solve the problems they need to solve, and does it all very quickly and efficiently.
The logical transition from personal computers being hot rod hobbyist toys to general purpose computational appliances seems to be taking forever. Perhaps, like automobiles, it’ll be over one hundred years into the fray and still unable to escape the realm of toydom. Who needs an 800 HP automobile for a daily commute in a stream of traffic that averages 37 mph? Who needs a 32 core CPU for a personal computer that spends 85% of its cycles animating the screensaver?
I’m not advocating for minimalist designs, or complaining about overkill, I’m just saying that the fascination around how the tools that are supposed to simplify and benefit our lives are being constructed should not be a higher priority for Apple than making sure the tools that are making actually solve the problems that we need them to solve. The essence of being a “Mac” should be based on what the computer does for us, not how it does what it does. If moving from Intel to ARM moves the needle in favor of things that customers value, the decision should be a no brainer.
While I’m personally uncomfortable about AMD, I agree that if Apple wants to put a fire under Intel, beginning a move to AMD would be a much better idea than moving to ARM. Apple could srug its shoulder and say to Intel; “Sorry, but you’re just not keeping up”. The problem is we’ve seen this before, just before Intel shot ahead.
The technology for ARM to do the extensive multi-tasking that Intel chips provide does not exist today. Arm chips can only now do very small multi-tasking operations. The technology for ARM is still several years away. The main advantage for Apple is Temperature, always on, non comparability with the open architecture that Intel has, (IE Kill Hackintosh) and MORE PROFIT. The most important for them being the last two. Because we know even with the bargain basement chip that ARM is Apple will not drop their pricing. I am old enough to remember when Mac had a completely closed architecture and sales was probably 1% of what it is today. Moving to Intel allowed them to grow to where they are today. ARM will be the death of Mac. Not Apply but the Mac for sure.
Current Arm chips are equal or better in performance compared to Intel x86 chips, single or multi (hardware) threading. ARM will be Macs biggest breakthrough.
Edit: fixed threading
That’s not entirely true. They’re better in some things, but not as good in others. You can’t just look at basic tests and assume the rest.
The author gives Apple's previous two CPU transitions as an indicator for a successful ARM transition. But he's kind of ignoring the fact that in those two prior transitions, the CPU being moved to was significantly faster than the CPU being migrated from. This then allowed the use of emulation software ("Rosetta") to let users continue using "legacy" software without too much of a performance penalty. ARM is not significantly faster than Intel chips, so how will Apple handle this transitional period in which users need/want to continue using legacy software?
That's not completely true. First, Intel's CPU were not significantly faster than the PowerPCs they replaced. They were fast enough to allow Rosetta to work without users noticing much of change in performance. Also, Rosetta was not an emulator it was a translator. It's a relatively small performance hit (10%) to translate one set of instructions to another. On today's computers that's fairly insignificant as most people rarely "red line" their computers.
By the time Apple replaced the G5 with Yonah, despite that Yonah required a temporary step back to 32 bit, x86 had gone way past the PPC. There was no turning back. The fact that at that point laptop sales were at least 50% of computer sales. IBM still haddn’t come out with a mobile G5, which left Apple using that dual G4 configuration in its laptops, which still barely kept up, and was an expensive, hot, and complex “solution” to a problem they shouldn’t have had
I'll ask again, "How is Intel making Thunderbolt royalty-free for other chip makers an issue for Apple?" Everything you wrote backs up my reply to jkdsteve.
There is one little problem for Apple if they do move to ARM. How is Thunderbolt going to work with ARM? That’s an engineering problem Apple is going to have to work out, and it’s a real one.
Finally, and I think importantly, moving the Mac to ARM would mean that Apple has to develop a much wider range of "A"-series CPUs. I just can't see Apple producing:
an "A17" for the iPhone
an "A17X" for the iPad
an "A17XX" for the MacBook (Pro)
an "A17XXX" for the iMac
an "A17XXXX" for the Mac Pro
I'm always confused by people assuming that the chips will all be A-series despite a massive number of changes that will need to be made for traditional PC use and Apple already having many other custom silicon options with different letters. So why assume they will all get additional letter X's because Apple did this for the iPad, which is much similar in terms of how it works with iPhone than a Mac.
I agree. But the A probably just designates that it’s an Apple chip. I’ve been trying to come up with a new naming convention, and while I’ve come up with a few, if Apple does do this, I guess we’ll just have to wait. So,maybe ANxxx, standing for Apple notebookxxx. Who knows?
I'll ask again, "How is Intel making Thunderbolt royalty-free for other chip makers an issue for Apple?" Everything you wrote backs up my reply to jkdsteve.
There is one little problem for Apple if they do move to ARM. How is Thunderbolt going to work with ARM? That’s an engineering problem Apple is going to have to work out, and it’s a real one.
Again, why do you think USB4 is a problem for Apple?
While a tiny percentage of extreme power users may be able to benefit from a minimal performance boost by switching from Intel to ARM processors, a larger percentage of us that use some Windows-only software (Microsoft Project to name one) and thus need Bootcamp will be left in the cold. I have owned many Macs (currently a top-spec 2017 MacBook Pro 15"), and I reluctantly will stop buying Macs if Apple halts the Windows compatibility feature. The current Macs are fantastically powerful in the higher models, and a switch to ARM is not necessary.
While Bootcamp usage is high in the AppleInsider reading population, it is decidedly not with the vast majority of the Apple user base.
So, it's more like "while a tiny percentage of extreme power users may be left in the cold from a shift to ARM, the larger percentage that don't or have used the iPhone as a gateway to the the Mac will benefit from the performance boost."
It used to be a much higher percentage as more Windows users were switching over. But the usage seems to have dropped. But a lot of Mac gamers, such as my daughter and here friends, still use Bootcamp for that, as we still can’t keep up with pc gaming. I wonder how big that crowd is? They can’t use Parallels for that. I’ve tried.
While a tiny percentage of extreme power users may be able to benefit from a minimal performance boost by switching from Intel to ARM processors, a larger percentage of us that use some Windows-only software (Microsoft Project to name one) and thus need Bootcamp will be left in the cold. I have owned many Macs (currently a top-spec 2017 MacBook Pro 15"), and I reluctantly will stop buying Macs if Apple halts the Windows compatibility feature. The current Macs are fantastically powerful in the higher models, and a switch to ARM is not necessary.
While Bootcamp usage is high in the AppleInsider reading population, it is decidedly not with the vast majority of the Apple user base.
So, it's more like "while a tiny percentage of extreme power users may be left in the cold from a shift to ARM, the larger percentage that don't or have used the iPhone as a gateway to the the Mac will benefit from the performance boost."
It used to be a much higher percentage as more Windows users were switching over. But the usage seems to have dropped. But a lot of Mac gamers, such as my daughter and here friends, still use Bootcamp for that, as we still can’t keep up with pc gaming. I wonder how big that crowd is? They can’t use Parallels for that. I’ve tried.
1) Do you think that Apple cares about a few Bootcamp users that are Windows gamers that they would hinder their entire line by keeping them slow, hot, limited, and expensive with Intel chips compared to what they already offer in something like an iPad (which I don’t think will be the chip they use for any low-end Mac).
2) Are these Windows gamers using 12" MacBooks without a discreet GPU, which is the type of machine this move to ARM seems most imminent and would very easily trounce what Intel has been able to offer vendors?
While a tiny percentage of extreme power users may be able to benefit from a minimal performance boost by switching from Intel to ARM processors, a larger percentage of us that use some Windows-only software (Microsoft Project to name one) and thus need Bootcamp will be left in the cold. I have owned many Macs (currently a top-spec 2017 MacBook Pro 15"), and I reluctantly will stop buying Macs if Apple halts the Windows compatibility feature. The current Macs are fantastically powerful in the higher models, and a switch to ARM is not necessary.
While Bootcamp usage is high in the AppleInsider reading population, it is decidedly not with the vast majority of the Apple user base.
So, it's more like "while a tiny percentage of extreme power users may be left in the cold from a shift to ARM, the larger percentage that don't or have used the iPhone as a gateway to the the Mac will benefit from the performance boost."
It used to be a much higher percentage as more Windows users were switching over. But the usage seems to have dropped. But a lot of Mac gamers, such as my daughter and here friends, still use Bootcamp for that, as we still can’t keep up with pc gaming. I wonder how big that crowd is? They can’t use Parallels for that. I’ve tried.
1) Do you think that Apple cares about a few Bootcamp users that are Windows gamers that they would hinder their entire line by keeping them slow, hot, limited, and expensive with Intel chips compared to what they already offer in something like an iPad (which I don’t think will be the chip they use for any low-end Mac).
2) Are these Windows gamers using 12" MacBooks without a discreet GPU, which is the type of machine this move to ARM seems most imminent and would very easily trounce what Intel has been able to offer vendors?
When Tim Cook released the first iPad Pro, he said;
"iPad Pro is the clearest expression of our vision of the future of personal computing". If he truly believes this, then what purpose does a consumer-based 12" ARM MB serve? It seems to me that both of these devices would be targeting the same market which is redundant.
While a tiny percentage of extreme power users may be able to benefit from a minimal performance boost by switching from Intel to ARM processors, a larger percentage of us that use some Windows-only software (Microsoft Project to name one) and thus need Bootcamp will be left in the cold. I have owned many Macs (currently a top-spec 2017 MacBook Pro 15"), and I reluctantly will stop buying Macs if Apple halts the Windows compatibility feature. The current Macs are fantastically powerful in the higher models, and a switch to ARM is not necessary.
While Bootcamp usage is high in the AppleInsider reading population, it is decidedly not with the vast majority of the Apple user base.
So, it's more like "while a tiny percentage of extreme power users may be left in the cold from a shift to ARM, the larger percentage that don't or have used the iPhone as a gateway to the the Mac will benefit from the performance boost."
It used to be a much higher percentage as more Windows users were switching over. But the usage seems to have dropped. But a lot of Mac gamers, such as my daughter and here friends, still use Bootcamp for that, as we still can’t keep up with pc gaming. I wonder how big that crowd is? They can’t use Parallels for that. I’ve tried.
The service data that I've been collecting for years suggests that it is about 2% and declining right now, even including enterprise.
sure, for software actively for sale, its developers have an economic incentive to move it to ARM, but what about truly legacy stuff or even open-source applications with little community support? Developers are a big group, enthusiastic Mac user group. We use all sorts of open source software - be it IDEs, web servers, compilers, Java VMs, virtualization software, etc. Can you imagine how slow an emulated VirtualBox would be trying to, itself, emulate an x86 version of Windows?
1. If the Software is actually Open Source, then the Devs. should be able to simply recompile it for ARM, and go on.
2. Windows 10 already does x86 Translation (not Emulation, even though MS calls it that). Similar to how Apple handled the 68k -> PPC Transition, Window 10 ARM has a "Just In Time" (actually "First-Run") Translator (they call it a Compiler), that turns the x86 Object Code and Translates it to ARM. Then, that code Runs at native ARM speeds. And since the Translation is only done on First-Run, only the first time an Application is Launched will be somewhat slower to start-up. After that, Windows 10 ARM will simply Run the already-translated version:
Comments
So, it's more like "while a tiny percentage of extreme power users may be left in the cold from a shift to ARM, the larger percentage that don't or have used the iPhone as a gateway to the the Mac will benefit from the performance boost."
Are you for real, or are you a troll? Not all trolls are against Apple here.
By the time Apple replaced the G5 with Yonah, despite that Yonah required a temporary step back to 32 bit, x86 had gone way past the PPC. There was no turning back. The fact that at that point laptop sales were at least 50% of computer sales. IBM still haddn’t come out with a mobile G5, which left Apple using that dual G4 configuration in its laptops, which still barely kept up, and was an expensive, hot, and complex “solution” to a problem they shouldn’t have had
There is one little problem for Apple if they do move to ARM. How is Thunderbolt going to work with ARM? That’s an engineering problem Apple is going to have to work out, and it’s a real one.
It used to be a much higher percentage as more Windows users were switching over. But the usage seems to have dropped. But a lot of Mac gamers, such as my daughter and here friends, still use Bootcamp for that, as we still can’t keep up with pc gaming. I wonder how big that crowd is? They can’t use Parallels for that. I’ve tried.
2) Are these Windows gamers using 12" MacBooks without a discreet GPU, which is the type of machine this move to ARM seems most imminent and would very easily trounce what Intel has been able to offer vendors?
"iPad Pro is the clearest expression of our vision of the future of personal computing". If he truly believes this, then what purpose does a consumer-based 12" ARM MB serve? It seems to me that both of these devices would be targeting the same market which is redundant.
2. Windows 10 already does x86 Translation (not Emulation, even though MS calls it that). Similar to how Apple handled the 68k -> PPC Transition, Window 10 ARM has a "Just In Time" (actually "First-Run") Translator (they call it a Compiler), that turns the x86 Object Code and Translates it to ARM. Then, that code Runs at native ARM speeds. And since the Translation is only done on First-Run, only the first time an Application is Launched will be somewhat slower to start-up. After that, Windows 10 ARM will simply Run the already-translated version:
https://docs.microsoft.com/en-us/windows/uwp/porting/apps-on-arm-x86-emulation
What doesn't work now (and maybe never), is x86-64 Translation. So no 64-bit Windows Applications... yet.