Editorial: Apple's move to ARM is possible because most users want power more than compati...

245678

Comments

  • Reply 21 of 154
    “MacARM” is the iPad Pro. There will be no more “Mac” ARM but there will be even more iPad Pro with iPad OS.

    iPad OS + iPad Pro combo is a more advanced concept than MacARM. The future is the iPad. Macs will continue to exist but for backwards compatibility and for industrial legacy applications only. And if Intel does well within this limited role, it will always do.


    The lack of filesystem accessibility on iPadOS or IOS is a huge problem to me and all developers. We work with files all the time. MacOS allows us to do that. Attachment to emails are files. I add files to emails all the time. Makes passing data easy. Why go to the complexity of using DropBox and uploading a file (those darn files won't go away will they)
    IMHO, IOS (in any of its guises) has a long way to go to be a total replacement for MacOS.  I may well be pushing up daisies before that happens.
    My condolences. You expect to dead by the fall, then?

    More seriously... Did you miss everything last week? IOS now offers filesystem access. (And no, that's not enough to totally replace MacOS. But nobody with a clue is seriously considering that in the first place. That unfortunately includes the person you're replying to.)
  • Reply 22 of 154
    sflocalsflocal Posts: 6,092member
    It’s inevitable that Apple will throw Intel under the bus.  Intel just doesn’t have the chips to keep up in the industry anymore.  They’re like IBM.

    that being said, there must be a big enough market for Windows running on Macs to keep companies like VMWare and Parallels busy developing their virtual machine packages.

    I went to Macs because they could run Windows.  I don’t use it much but for certain development tools and software I use it’s imperative.  I would have to buy a Wintel PC if Apple goes this route.

    I’m hoping Apple’s next chips for desktops are not ARM chips but their own home-grown x86-compatible version with Intel’s blessing.
    viclauyyc
  • Reply 23 of 154
    [...]
    I’m also looking forward to a ARM based MacBook, but performance might be a bigger headache than straight speed tests suggest.  Qualcomm and Microsoft have put considerable effort into their ARM offering, but performance has been terrible.  Apple has done a lot of tinkering to optimize ARM and iOS and performance is impressive, but it also isn’t designed to multitask (they do have clever workarounds).  A ARM processor in a MacBook is going need to excel at multitasking.  This might be a much bigger problem/challenge than Apple fans are thinking, it might be a fundamental limitation of ARM that throwing additional cores/memory at the problem doesn’t fix.  We’ll see...
    Your claim is fantastical nonsense. "Not designed to multitask"? IOS runs dozens of simultaneous tasks all the time. The kernel is designed from the ground up for multitasking. Qualcomm's problem was that they were more than two years behind Apple in terms of hardware performance, until the 855 generation, which has "leapfrogged" to only ~1.5 years behind Apple (and is not, AFAIK, in any windows-running devices).

    There is no multitasking issue in ARM's fundamental design. The notion is just silly. There are challenges in a transition to ARM (which Apple is clearly taking steps to mitigate, right now). Your concern, however, is not one of them.
    tmaydanhJWSCjdb8167
  • Reply 24 of 154
    melgrossmelgross Posts: 33,510member
    Soli said:
    melgross said:
    firstly, no, the ‘a serir]es still has a long way to go before it can compete
    1) It's an odd assumption to think that the A-series chip is the most powerful chip Apple can design seeing as how those are designed for handheld devices with no options for fans, very limited batteries, and incredibly small spaces, but I find it even more odd that you assume that it would be an A-series chip and not another chip designation, like all the others Apple has come up with already.

    2) We know that Apple's chip designs can compete because we already have evidence that their chips are outpacing what they're putting into many of their Macs, but you're incorrectly assuming that they only way Apple could come out with an ARM-based Macs on the lower-end (where compatibility for Bootcamp and virtualization won't have to be supported) is for Apple to also best Intel on their upcoming Mac Pro performance.

    if Apple does begin this process, it will be a very difficult one.
    Of course Apple has begun this process. Do you really not see all these changes to their codebase at WWDC as not leading to this eventual transition to offer some ARM-baed Macs?
    I’ve never said it was the most powerful chip they can design. But it’s not as easy to design a complex chip. As performances goes up, so, will cost, power draw and heat. I’m not saying they can’t do it. But even if they do, developers are still going to have to go native. Yes, no? Who knows how that would go?
  • Reply 25 of 154
    Mike WuertheleMike Wuerthele Posts: 6,858administrator
    damk said:
    You don't take into account HUGE community of programmers which use Macs to develop:
    1) Android software
    2) Enterprise software (with java, x86 Linux and Windows servers with x86 databases like Oracle and MS SQL Server).

    This is possible thanks to virtualization technologies like Parallels, VirtualBox, Docker and cross-platform development environments.

    I use Docker virtualization with Linux x86 container and within it Oracle database - I don't see way to migrate with OS/X to non Intel CPU with this software stack. It is 3 levels deep dependency on x86.

    Generally, ARM based Mac would be almost use-less for backend programming (depending on project and used technologies of course).


    Nope, we do account for that, and did here as well. Don't confuse what some AI readers need, with what the vast majority of the market will be fine with. This kind of differentiation is addressed in the article.
    edited June 2019 Solidocno42
  • Reply 26 of 154
    DAalsethDAalseth Posts: 2,783member
    “MacARM” is the iPad Pro. There will be no more “Mac” ARM but there will be even more iPad Pro with iPad OS.

    iPad OS + iPad Pro combo is a more advanced concept than MacARM. The future is the iPad. Macs will continue to exist but for backwards compatibility and for industrial legacy applications only. And if Intel does well within this limited role, it will always do.


    The lack of filesystem accessibility on iPadOS or IOS is a huge problem to me and all developers. We work with files all the time. MacOS allows us to do that. Attachment to emails are files. I add files to emails all the time. Makes passing data easy. Why go to the complexity of using DropBox and uploading a file (those darn files won't go away will they)
    IMHO, IOS (in any of its guises) has a long way to go to be a total replacement for MacOS.  I may well be pushing up daisies before that happens.
    There have been several articles on AI and elsewhere covering the Files App in iOS 13. It goes a long way toward improving file system access. 
  • Reply 27 of 154
    melgross said:
    It’s a nice, fairly long article. It be]rings up a number of things that are correct. But, I still believe moving to Apple’s ARM chips is more difficult that some people think.

    firstly, no, the ‘a serir]es still has a long way to go before it can compete with AND and Intel at the higher levels, with no guarantee that it ever will. That’s all speculation. Are we going to see a 6 core chip? An 8 core chip? These will be needed to compete on the higher laptop level. 16GB RAM? 64GB RAM? Same thing.

    convincing developers, particularly those with very large, high performance software to go native ARM? And, yes, that will be needed.

    if Apple does begin this process, it will be a very difficult one.
    Not because of hardware performance. You're quite mistaken about this. The A series *already* competes quite well with AMD's and Intel's mid/upper-range chips.

    The current iPad Pro already has an 8-core chip, and every current iPhone has a 6-core chip. The fact that four of the cores in each case is a lower-power core is irrelevant - regardless of power, a core still requires the same services from the bus (or mesh, or whatever) servicing it. Also, supporting large amounts of RAM is trivial, not a challenge at all.

    I suggest you re-read (I'm sure you've seen it already) this coverage of the A12 chip: https://www.anandtech.com/show/13392/the-iphone-xs-xs-max-review-unveiling-the-silicon-secrets/4

    Those two are the best articles I've seen covering the A12 generation. Here's a quote from the last paragraph of the second:
    "What is quite astonishing, is just how close Apple’s A11 and A12 are to current desktop CPUs. [...] we see that the A12 outperforms a moderately-clocked Skylake CPU in single-threaded performance. [...] we’re now talking about very small margins until Apple’s mobile SoCs outperform the fastest desktop CPUs in terms of ST performance."

    Moving to ARM isn't trivial. But raw performance isn't going to be a significant issue, unless you're actually trying to run Intel code in emulation. And not even then, unless it's really CPU-intensive.
    SoliJWSCdocno42
  • Reply 28 of 154
    dick applebaumdick applebaum Posts: 12,527member
    larrya said:
    The question I have is, if you're kissing off 35% of your customers, where do the replacements come from?  I doubt you can grow the base back by adding a bunch of ported iOS apps.  I will probably be one of them because, although I stopped running Parallels a long time ago, I need the option for work stuff.

    Interesting...... We have been a Windows-Free household for years, until recently:

    1. My daughter's new job requires proprietary Windows software
    2. Me - woodworking...

    I have a woodworking hobby and native Mac apps like SketchUp or Mac/Java apps met my needs... Then I bought a woodworking CNC machine!  

    My long-term goal was to eventually run both CAD and CAM on an iPad Pro* -- It seemed like a natural to use the Apple Pencil for the CAD (Computer-Aided Design) and the iPad certainly had the horsepower to run the CAM  (Computer-Aided Manufacturing) -- which is reading a text file (gcode) program and issue simple X,Y,Z axis commands to control the CNC machine (similar to Turtle Graphics).   There are gcode interpreters for the Raspberry Pi.

    To promote my long-term goal, I attempted to run entirely on the Mac.  The CAM was easy, as the major software was a Java app available for Mac and Windows.   The best Mac CAD app is Autodesk Fusion 360 -- but it has a very, very steep learning curve.  There are a few others and some on-line apps, but they are limited.

    Long story, shorter:   The CAD scene is dominated by Vectric apps: VCarve and Aspire.  These are Windows only!

    So, I bit the bullet and bought:

    1. VMWare Fusuion Emulator
    2. VCarve Desktop CAD app
    3. TotalAV Mac Virus Scanner

    Yes, sadly, after a few days my Mac had been infected by viruses...

    So, there it is:  If you want to do CAM, you need to do Windows (and all that comes with it)...

    Disruption anyone?
    edited June 2019 iqatedo
  • Reply 29 of 154
    racerhomie3racerhomie3 Posts: 1,264member
    I totally agree with William. A $600-900 ARM MacBook would definitely establish the Mac market to a whole new audience.
  • Reply 30 of 154
    sirbryansirbryan Posts: 33member
    Do you see companies like Adobe, Avid, Autodesk, Microsoft, Unity etc. embracing this? If someone can’t use Pro Tools or AutoCad on the Mac because it’s not ARM compatible I can’t see it being successful. So maybe the pro Mac line stays Intel.
    The point of the Intel vs. ARM discussion isn't about maintaining Intel compatibility for the sake of third-party apps that run on macOS. Many of those companies had Mac apps running on 68000 and PowerPC. It is really about the declining need for people to run Windows on their Macs (natively or virtualized). Apps designed to run natively on macOS that are compiled against the OS libraries (i.e. no hardware-specific calls) will run on whatever hardware Apple wants to support. For many, it would require not much more than a recompile and re-upload.
  • Reply 31 of 154
    dick applebaumdick applebaum Posts: 12,527member
    sflocal said:
    It’s inevitable that Apple will throw Intel under the bus.  Intel just doesn’t have the chips to keep up in the industry anymore.  They’re like IBM.

    that being said, there must be a big enough market for Windows running on Macs to keep companies like VMWare and Parallels busy developing their virtual machine packages.

    I went to Macs because they could run Windows.  I don’t use it much but for certain development tools and software I use it’s imperative.  I would have to buy a Wintel PC if Apple goes this route.

    I’m hoping Apple’s next chips for desktops are not ARM chips but their own home-grown x86-compatible version with Intel’s blessing.
    That's certainly a possibility...  An earlier post mentioned that a barrier was the need co convert x86 CISC to RISC -- but Intel does just that at execution time thru a proprietary process.   As, I read it Apple, with Intels blessing (Apple $), could do either ARM or another Apple Chip running [native] x86.
  • Reply 32 of 154
    A bit more on A12 performance, from earlier comments I wrote in https://forums.appleinsider.com/discussion/comment/3154207/#Comment_3154207 :

    "Apple's ARM design is *already* able to compete with Intel's best consumer chip (the 8-core i9-9900k - I'm not talking about the ridiculous X-series Xeons-in-disguise). I wrote about this once before, but in short, the iPad Pro shows that Apple's uncore and interconnect capabilities are superior to Intel's, at least for up to 8 cores. So they could easily put eight of their high-performance cores into an "A12XX" and have something competitive with the i9. This wouldn't even be a major job. Now, there are still questions to be asked: How much headroom do they have for clocks at higher power? Pipeline length may be too short to enable 4GHz, or even 3.5, and there are other less obvious issues as well. Can the memory interface and caches grow organically without a major redesign? Could they scale to even more cores (probably not)? Etc. So, we don't know if Apple can beat the current i9 with their existing design, but we do know it's in the right ballpark.
    [...]
    There's actually an interesting historical parallel here. In the old days, "server" CPUs (actually mini/mainframe, servers were just big PCs and not relevant) defined the state of the art and provided the most performance (all the once-great RISC families: Sparc, MIPS, Alpha, etc., and various mainframe/supercomputers). But as volume in the PC business grew, Intel kept pushing its CPUs forward, and then applying those lessons learned to its server chips. Even though server ship development lagged consumer processors by 1-2 years, those advances eventually pulled Intel Xeons ahead of all the other architectures (excepting POWER, because IBM is stubborn and has lots of talent).

    And now, we're seeing the same thing: Mobile chip development has the advantage in volume, and so the lessons being learned there will have repercussions on the desktop processor market. Only now the volume winner (at least in high-performance mobile) is Apple (and to some extent TSMC), and the loser is likely to be Intel."

    In case the first bit in that quote isn't clear: we can learn something very interesting by contrasting multicore performance of the A12 and A12X. You can see nearly linear performance improvement with the two additional big cores. This is important and underappreciated. It shows that Apple's capable of building an interconnect that scales to at least 8 cores, and what's more, that power use for the interconnect is scaling well too. That's a very big deal as the interconnect becomes a dominant factor in power use with the largest CPUs.

  • Reply 33 of 154
    SoliSoli Posts: 10,035member
    jeremy c said:
    @damk, agreed. There are linux tools that a lot of us use in macOS that would require porting over to ARM from x86.
    What "linux tools" are you referring to that are not part of any Linux distro that is ARM compatible?
    docno42
  • Reply 34 of 154
    SoliSoli Posts: 10,035member
    melgross said:
    Soli said:
    melgross said:
    firstly, no, the ‘a serir]es still has a long way to go before it can compete
    1) It's an odd assumption to think that the A-series chip is the most powerful chip Apple can design seeing as how those are designed for handheld devices with no options for fans, very limited batteries, and incredibly small spaces, but I find it even more odd that you assume that it would be an A-series chip and not another chip designation, like all the others Apple has come up with already.

    2) We know that Apple's chip designs can compete because we already have evidence that their chips are outpacing what they're putting into many of their Macs, but you're incorrectly assuming that they only way Apple could come out with an ARM-based Macs on the lower-end (where compatibility for Bootcamp and virtualization won't have to be supported) is for Apple to also best Intel on their upcoming Mac Pro performance.

    if Apple does begin this process, it will be a very difficult one.
    Of course Apple has begun this process. Do you really not see all these changes to their codebase at WWDC as not leading to this eventual transition to offer some ARM-baed Macs?
    I’ve never said it was the most powerful chip they can design. But it’s not as easy to design a complex chip. As performances goes up, so, will cost, power draw and heat. I’m not saying they can’t do it. But even if they do, developers are still going to have to go native. Yes, no? Who knows how that would go?
    It sure reads that way when you say the A-series chips have a long way to go to compete with Macs (which is clearly false and has been for years) by not once mentioning that Apple's A-series chips are designed for small, handheld devices with very limited battery life and a small thermal envelope.

    Why do you think Apple is incapable of designing a P-series chip that shoots ahead of the A12X in all the ways that a traditional, low-end notebook would need, like considerably more RAM? I see absolutely no hurdle other than Apple needing to line up all the SW properly to make this transition into a multi-architecture Mac platform (or Mac-like category) feasible.
    JWSC
  • Reply 35 of 154
    SoliSoli Posts: 10,035member
    sflocal said:
    It’s inevitable that Apple will throw Intel under the bus.  Intel just doesn’t have the chips to keep up in the industry anymore.  They’re like IBM.

    that being said, there must be a big enough market for Windows running on Macs to keep companies like VMWare and Parallels busy developing their virtual machine packages.

    I went to Macs because they could run Windows.  I don’t use it much but for certain development tools and software I use it’s imperative.  I would have to buy a Wintel PC if Apple goes this route.

    I’m hoping Apple’s next chips for desktops are not ARM chips but their own home-grown x86-compatible version with Intel’s blessing.
    That's certainly a possibility...  An earlier post mentioned that a barrier was the need co convert x86 CISC to RISC -- but Intel does just that at execution time thru a proprietary process.   As, I read it Apple, with Intels blessing (Apple $), could do either ARM or another Apple Chip running [native] x86.
    They may be negotiating that and that may also be a real hold up (especially with the Qualcomm's claim's that Apple isn't pay their proper licensing fees), but I have to wonder why a low-end notebook needs to even have x86 support going forward? How many users of the 12" MacBook are using BootCamp, Parallels, or VMWare? Can these users not simply go to a MacBook Pro to get their dual-boot or virtualization needs met? I'm sure Apple knows. And why are ignoring Windows has been ARM-compatible for many years now which means that running Windows on a dual-boot or virtualized ARM-based Mac are not impossible.
    docno42
  • Reply 36 of 154
    dick applebaumdick applebaum Posts: 12,527member
    Soli said:
    sflocal said:
    It’s inevitable that Apple will throw Intel under the bus.  Intel just doesn’t have the chips to keep up in the industry anymore.  They’re like IBM.

    that being said, there must be a big enough market for Windows running on Macs to keep companies like VMWare and Parallels busy developing their virtual machine packages.

    I went to Macs because they could run Windows.  I don’t use it much but for certain development tools and software I use it’s imperative.  I would have to buy a Wintel PC if Apple goes this route.

    I’m hoping Apple’s next chips for desktops are not ARM chips but their own home-grown x86-compatible version with Intel’s blessing.
    That's certainly a possibility...  An earlier post mentioned that a barrier was the need co convert x86 CISC to RISC -- but Intel does just that at execution time thru a proprietary process.   As, I read it Apple, with Intels blessing (Apple $), could do either ARM or another Apple Chip running [native] x86.
    They may be negotiating that and that may also be a real hold up (especially with the Qualcomm's claim's that Apple isn't pay their proper licensing fees), but I have to wonder why a low-end notebook needs to even have x86 support going forward? How many users of the 12" MacBook are using BootCamp, Parallels, or VMWare? Can these users not simply go to a MacBook Pro to get their dual-boot or virtualization needs met? I'm sure Apple knows. And why are ignoring Windows has been ARM-compatible for many years now which means that running Windows on a dual-boot or virtualized ARM-based Mac are not impossible.

    As someone who has recently revisited running Windows on a Mac -- I think it is more than [on an ARM Mac] "can you do it?" -- rather "do you really want do it?".

    Windows comes with a lot of downsides in addition to Viruses.   To sustain my woodworking hobby CNC, I need to flip among 3 different UIs:  Mac, Windows, Java.  They are all similar, but just enough different to frustrate finger memory... And I have no idea [or desire to learn] what to do when something goes wrong on Windows or Java -- just reboot.

    I think there is more promise for companies (Microsoft, Adobe, AutoDesk, etc.) to port their apps to iOS and ipadOS to take advantage of the very large install base.
    edited June 2019 raoulduke42
  • Reply 37 of 154
    DAalsethDAalseth Posts: 2,783member
    larrya said:
    The question I have is, if you're kissing off 35% of your customers, where do the replacements come from?  I doubt you can grow the base back by adding a bunch of ported iOS apps.  I will probably be one of them because, although I stopped running Parallels a long time ago, I need the option for work stuff.

    ...clip...

    Long story, shorter:   The CAD scene is dominated by Vectric apps: VCarve and Aspire.  These are Windows only!

    So, I bit the bullet and bought:

    1. VMWare Fusion Emulator
    2. VCarve Desktop CAD app
    3. TotalAV Mac Virus Scanner

    Yes, sadly, after a few days my Mac had been infected by viruses...

    ...clip...
    I know that it's a bit late, but I would have approached the problem slightly differently. Rather than emulation, I would have used a virtual environment. Run Windows in this virtual environment. That would have two advantages. First it is a sealed box, malware can't get out. Secondly if your virtual machine gets infected you can restore to the last known good backup from before the infection. Now getting ports in and out of a virtual machine can sometimes be a problem. I've fought with that several times when I was supporting servers. But for me the added security is worth it.

    That said, yes there are some times when Windows software is the only option. It sucks, but that's the way it is.
  • Reply 38 of 154
    hmlongcohmlongco Posts: 533member
    I'd love to see what happens if you were to take today's A12 chip from a iPad Pro and put it into an environment with fewer battery/power constraints and much better thermal characteristics (i.e. larger heat sink and fans).

    One thing I've not seen mentioned is that Apple's newest Bionic chips have also have 8-core Neural Engines currently capable of doing 5-trillion operations per second. This makes them extremely powerful at supporting many on-device AI driven tasks.

    What if Apple made a notebook/desktop-class chip with 16 neural engine cores? 32? 20-trillions ops per second, anyone?
    iqatedo
  • Reply 39 of 154
    wallymwallym Posts: 33member
    As a developer, I need both mac and windows support.  To openly campaign to remove Windows compat is to be irresponsible to the marketplace.  If users don't need Windows, that's fine.  Don't penalize me for your lack of needs.
    macplusplus
  • Reply 40 of 154
    SoliSoli Posts: 10,035member
    wallym said:
    As a developer, I need both mac and windows support.  To openly campaign to remove Windows compat is to be irresponsible to the marketplace.  If users don't need Windows, that's fine.  Don't penalize me for your lack of needs.
    Wanting ARM-based Macs on the power end doesn’t mean that you’re losing Windows support unless you’re cheap.
    docno42
Sign In or Register to comment.