Apple planning to ditch Intel chips in Macs for its own custom silicon in 2020

1234689

Comments

  • Reply 101 of 176
    asciiascii Posts: 5,936member
    Speaking of Intel chips, Intel just officially launched the Coffee Lake notebook CPUs. New Macbook Pros tomorrow? :smile: 

    The current Dell XPS 15 is now showing as "clearance" on their site, so it appears Dell aren't wasting any time getting these new CPUs in to people's hands. High end notebooks will have 6 cores for the first time.
    edited April 2018 cgWerks
  • Reply 102 of 176
    herbapouherbapou Posts: 2,228member
    No bootcamp no more macs for me
  • Reply 103 of 176
    Mike WuertheleMike Wuerthele Posts: 6,858administrator
    With all this BootCamp discussion, there's a key point that's getting missed -- It depends on what you need Windows for. Windows for ARM is a thing.

    edited April 2018 SolicgWerks
  • Reply 104 of 176
    macxpressmacxpress Posts: 5,801member
    wood1208 said:
    If this rumor is true and word "switch by 2020" is used than it can be end of 2019 or spring 2020 before school year starts for new Apple chip based MACs rolling out. Low cost Apple made chips can help create cheaper MACs that can be Chromebook killer in education market.
    I doubt you'll see a "cheap" Mac because of this. If anything you'll see the Mac mini with an Apple CPU in it at the same price it currently is. You'll see something like a MacBook Air (change the name I hope!) maybe starting at $799 or something. Apple is not in a race to the bottom and it's never going to compete on price. Its not like its magically free for Apple to make these chips along with every other chip Apple creates to go inside a Mac. 

    And damn it...its not MACs...its Macs. It's not an acronym. I hate when people do this. It just shows how out of touch someone is when talking about Apple quite honestly. 
    fastasleep
  • Reply 105 of 176
    macxpressmacxpress Posts: 5,801member
    karmadave said:
    Every few months a story like this pops up on AI and have the same response. Apple will NOT abandon x86 architecture for the Macintosh. It would kill their Enterprise Mac business. They will continue to develop their own specialized processors, for Mac, in addition to utilizing an x86 compatible processor from Intel (or maybe AMD).
    How exactly will this kill their enterprise business? So IBM is buying Macs because they run Intel processors. 
    docno42
  • Reply 106 of 176
    It probably makes more sense for Apple to make a netbook-style device with iOS on it. They could revive the iBook name for branding and market it to students and consumers. That way they can also continue to make the Mac & MacBook devices with Intel (or AMD) chips for several more years to support that market as the 'professional' line. I wouldn't be surprised to see Apple buy stake in AMD, but it may simply be more cost effective to partner with them.
  • Reply 107 of 176
    wizard69wizard69 Posts: 13,377member
    mjtomlin said:
    Apple has been extremely ambitious this decade with their silicon team. I'd be willing to bet there's a project to develop their own ISA that will replace both ARM and x64. There's no reason they need to stick with any kind of industry "standard" anymore, they could create a custom ISA that meets their very specific needs. There is only one kernel and core OS that runs across all their devices. Creating a highly tuned CPU tailor made for that platform would enable a level of performance and efficiency that no other company could achieve.
    I'd rather see them stick with ARM ISA and consider the extension Fijitsu and  partners are working on.   That would be advanced vector processing which would be very useful to a number of Apples customers.   The obvious reason is that going it alone doesn't always lead to grasping the larger mindset in a community.
  • Reply 108 of 176
    wozwozwozwoz Posts: 263member
    The New Apple chip will need to be not only faster, but tens of times faster for anyone to care. For most users, this will just mean disruption, confusion and yet another Apple hardware/software platform change. Even if it were faster (and don't expect Intel to stand still), one of the reasons Apple has been successful with the Mac in recent years is because of the Intel compatibility, and the ease of transition for Windows users to still run their apps, should they wish to do so. If Apple is throwing away that Intel compatibility, then I think they are also throwing away their growth path into the mass market. None of which sounds sensible to me. I don't want to run iOS apps on my Mac thanks, and I have lived through previous Apple hardware/OS transitions (remember PowerPC!?), and they just involve a lot of work for developers to make programs that already work ... keep on working. 
    avon b7cgWerks
  • Reply 109 of 176
    These kinds of leaks (“Marzipan!” “Kalamata!”) should be viewed in the light of hardware — actual products that Apple will sell. Could these developments result in a touch-screen laptop? It’s not a big step forward, and has the added advantage of cutting off what little oxygen Microsoft’s Surface has had.
    edited April 2018 watto_cobra
  • Reply 110 of 176
    frantisekfrantisek Posts: 756member
    There were writtings on wal as someone said here already. They are.

    iOS can install just reqired parts of app based on device, not sure how technology is called.
    There was probably one or two other things that will support that. I wrote about them somewhere year two back but forgot already. lol
    There is not writen Apple will ditch Intel completelly at least not initialy
    Microsoft is flirting with Windows for ARM all the time in case ARM based chips become more viable for laptops, they have to react.
    And Apple expertise on transitions is big.
    And in sending old technologies into grave as well.
  • Reply 111 of 176
    dick applebaumdick applebaum Posts: 12,527member
    ascii said:
    wizard69 said:
    I really doubt we will see emulation as the primary solution to X64 apps.    Instead I think what Apple has planed is a system where by apps are purchased and then the code segments needed for a specific machine are down loaded.   They have been building up infrastructure to do this for a couple of years.  
    This is what I think too, they even spoke about it at WWDC15 calling it Apple Bitcode.

    Basically all developers upload their apps to the Mac App Store in LLVM IR (LLVM intermediate representation - see Wikipedia) and when people download apps they get a single architecture (not FAT) binary that matches their machine. 

    What about apps that aren't distributed through the Mac App Store? Maybe they will take this opportunity to lock down the Mac to the store, just like iOS.
    Ha! Bitcode? LLVM IR?  What ever happened to BALR Using?
  • Reply 112 of 176
    dick applebaumdick applebaum Posts: 12,527member

    ascii said:
    Soli said:
    ascii said:
    What about apps that aren't distributed through the Mac App Store? Maybe they will take this opportunity to lock down the Mac to the store, just like iOS.
    I see no way to rule that out. Ultimately I think it comes to whether Apple feels the platform will be successful with only App Store apps or not. This has worked great for iOS, but can this work for a desktop OS going forward?

    My guess is that it won't be, but I also wouldn't be surprised if they did.
    It depends on the users I think. If the users put pressure on developers to distribute through the store then it could work, but if users instead decide to put pressure on Apple (to change the model) it could fail.
    Or, they could offer a Service for apps not distributed thru the Mac App Store.
  • Reply 113 of 176
    thttht Posts: 5,421member
    hmm said:
    tht said:
    My Python 3.6 plotting script execution time:

    2015 rMBP15 w/Intel Core i7-4980HQ (2.8/4.0 GHz): 91.1 sec

    2017 iPad Pro 10.5 w/Apple A10X (2.3 GHz): 91.5 sec

    This is 45 W vs a 10 W envelope or so. Use Pythonista on iPad. Terminal on macOS. Don’t know if the Core i7-4980HQ actually turbo-ed to 4 GHz. Who knows. That’s why you do a lot of testing.
    Typically python is single threaded. With cython you can use the OpenMP backend for multithreading anywhere Python api calls can be optimized out. Now Apple's version of Clang doesn't support OpenMP, so you are running on one core. Turbo varies. If you're using AVX extensions, it typically runs at native clock speeds. 

    Of course it's possible to be bound by something other than cpu. You can be bound by memory/cache bandwidth and other things. The test overall tells you a grand total of nothing without proper context.
    The “test” is my plotting script that I run tens of times per day for work. It tells me that the iPad Pro 10.5 can execute it in about the same time as my 2015 rMBP15. So, for this particular process, Apple’s ARM architecture is doing pretty well, which is basically what the benchmarks say.

    Yes. It is single core. Whether the Python runtime uses SIMD units on either architecture, I don’t know. Since iOS and macOS are basically the same for this type of program: runs on the same Mach w/BSD runtime, uses the same CLANG/LLVM toolchain, etc, it is about as fair as it gets, at least for parts of the architecture it uses.

    Moreover, the vast majority of users are going use the Python run time environments as they are. The whole purpose of the language is to not get into close to the silicon levels of programming. If scripts run faster on one architecture over the other, the ultimate conclusion is that one architecture runs python scripts faster than the other.
  • Reply 114 of 176
    jbdragonjbdragon Posts: 2,305member
    I've been able to run Macs for work because of the ability to virtualize Windows. Not sure this bodes well for the future of Macs in certain business segments unless emulation performance under these new chips will be acceptable.
    What do you think the MAC will be doing? It'll be in the transitional phase once again, this time with running Intel based software on a Arm processor. I see no reason why you couldn't continue to Virtualize Windows on the Mac. Maybe it takes a bit of a speed hit.
  • Reply 115 of 176
    wizard69wizard69 Posts: 13,377member
    Apple can 
    1) BUY AMD
    2) or INVEST HEAVILY IN AMD then PERPETUALLY AND IRREVOCABLY LICENSE ALL OF AMD's INTELLECTUAL PROPERTY

    Then Apple can build its own Intel compatible CPUs at much higher performance and customization of function.

    Why just do ARM CPU's? ARM is not proven at all to be able to handle the heavy multitasking involved with laptop and desktop processors. Intel CPUs are.

    With perpetual licenses to both the ARM and Intel architecture, Apple can simply blend, customize, and do what it sees fit to improve the customer experience. Apple can for example add AI chip processing to an Intel core along with a custom secure enclave. Apple can't wait long enough for Intel to drag its feet to do these things.  

    Intel has dropped the ball heavily. It's CPUs have not been much faster in the past decade. It is irritating for Macs to be held up in performance simply because Intel can't keep up with the technology. Intel's process is simply ages behind Samsung's process for example.

    As Steve Jobs said, skate to where the puck will be, not where the puck is. Intel simply can't do this. They have been mired in the Windows hedgemoney for decades. But now that Microsoft too acknowledges that Windows is not what it use to be, Intel is in deep trouble.
    I have to disagree with some of this.  

    Consider this Apple is proving everyday what ARM can do and they are doing that at extremely low powers.   In a desktop or laptop class processor they can balance power against performance to their hearts content.   This means caches can be sized to anything that is thermally acceptable, processor cores can be as many as needed and of course the support hardware can be expanded as needed.   To put it another way, IPad proves everyday that Apples ARM hardware is good enough for a low end laptop.

    Likewise Intel compatibility is a thing of the past.   Agian iPad proves this as they have taken considerable market share on an ARM processor running a completely different GUI from the stuff that is Windows or Mac OS.

    As for Intel I have to agree, they have dropped the ball and are far too concerned about backwards compatibility.   They should have made an effort to migrate to a trimmed down instruciton set when they went to 64 bit.
  • Reply 116 of 176
    SoliSoli Posts: 10,035member
    jbdragon said:
    I've been able to run Macs for work because of the ability to virtualize Windows. Not sure this bodes well for the future of Macs in certain business segments unless emulation performance under these new chips will be acceptable.
    What do you think the MAC will be doing? It'll be in the transitional phase once again, this time with running Intel based software on a Arm processor. I see no reason why you couldn't continue to Virtualize Windows on the Mac. Maybe it takes a bit of a speed hit.
    1) It wouldn't be virtualized, it would be emulated if you're talking about a version of Windows meant for an Intel-based machine (not their ARM variant). Windows is virtualized on today's Macs because they use Intel's architecture.

    2) I don't see any reason to say that it will be be running x86_64 code on ARM like with Rosetta. So much has changed for developers and frankly I don't see some all-or-nothing event where Apple is going to say that they're moving all Mac to ARM over the next year or so, like they did with transition to Intel. Maybe they will, but it seems more likely to me that we'll see something like an inexpensive Retina MacBook Air (mostly just a current, 12" MacBook) running on Apple silicon with the MacBook Pro and Mac Pro still going strong with Intel for many years to come. if you want or need a VM or Bootcamp for Windows than their sub-$1000 Mac notebook isn't the right choice. If the popularity is high enough we'll see a migration to more ARM-based Macs, but by the time the high-end MacBook Pro is on the chopping block I suspect the biggest complaint  will be why Apple didn't transition it sooner when their silicon was much faster while being more efficient than what Intel could offer.
    cgWerks
  • Reply 117 of 176
    dick applebaumdick applebaum Posts: 12,527member
    tht said:
    hmm said:
    tht said:
    My Python 3.6 plotting script execution time:

    2015 rMBP15 w/Intel Core i7-4980HQ (2.8/4.0 GHz): 91.1 sec

    2017 iPad Pro 10.5 w/Apple A10X (2.3 GHz): 91.5 sec

    This is 45 W vs a 10 W envelope or so. Use Pythonista on iPad. Terminal on macOS. Don’t know if the Core i7-4980HQ actually turbo-ed to 4 GHz. Who knows. That’s why you do a lot of testing.
    Typically python is single threaded. With cython you can use the OpenMP backend for multithreading anywhere Python api calls can be optimized out. Now Apple's version of Clang doesn't support OpenMP, so you are running on one core. Turbo varies. If you're using AVX extensions, it typically runs at native clock speeds. 

    Of course it's possible to be bound by something other than cpu. You can be bound by memory/cache bandwidth and other things. The test overall tells you a grand total of nothing without proper context.
    The “test” is my plotting script that I run tens of times per day for work. It tells me that the iPad Pro 10.5 can execute it in about the same time as my 2015 rMBP15. So, for this particular process, Apple’s ARM architecture is doing pretty well, which is basically what the benchmarks say.

    Yes. It is single core. Whether the Python runtime uses SIMD units on either architecture, I don’t know. Since iOS and macOS are basically the same for this type of program: runs on the same Mach w/BSD runtime, uses the same CLANG/LLVM toolchain, etc, it is about as fair as it gets, at least for parts of the architecture it uses.

    Moreover, the vast majority of users are going use the Python run time environments as they are. The whole purpose of the language is to not get into close to the silicon levels of programming. If scripts run faster on one architecture over the other, the ultimate conclusion is that one architecture runs python scripts faster than the other.
    Just out of curiosity, have you considered writing your plotting script in Swift?
  • Reply 118 of 176
    dick applebaumdick applebaum Posts: 12,527member
    With all this BootCamp discussion, there's a key point that's getting missed -- It depends on what you need Windows for. Windows for ARM is a thing.

    Thx for the link... Interesting read.  

    It appears that Windows for ARM won't be able to run x64 Windows apps (like Photoshop) for the foreseeable future.
  • Reply 119 of 176
    dick applebaumdick applebaum Posts: 12,527member
    Well, I suppose anything is possible:



    williamlondonSpamSandwich
  • Reply 120 of 176
    FolioFolio Posts: 698member

    FWIW, excerpt from BAML 02APR2018 report. I enhanced their bottom line with italics:

    Media reports suggest Apple may insource Mac chips

    Media reports today suggested that Apple might insource the production of chips for its Mac computers beginning in 2020. The project codename is supposedly “Kalamata” and is intended to make all of Apple’s devices (iPhone, iPad and Macs) more seamless and possibly run similar operating systems and Apps. While such insourcing of chips could benefit Apple by not being dependent on Intel’s processor cycles, by lowering the Mac costed bill-of-materials by ~$40-50, and by potentially streamlining and reducing R&D spend, in our opinion, if such a move is planned, Apple still has work to do in order to ensure that internally developed chips measure up to the performance characteristics required of laptop grade processors. We reiterate our Buy on multi-year rev and gross profit dollar growth and strong capital return.

    Two-fold benefit from insourcing chips

    In our opinion the benefit to Apple would be two-fold: 1) aggregating development across iOS and MacOS can help lower the overall cost of R&D by potentially combining development teams and potentially reducing time to market for new products and Apps, and, 2) internally developing the processor can help save some cost vs. purchasing the processors from Intel. Even if the cost of an internally developed chip is 2-3x the cost of processors currently used on iPad, we estimate the savings to Apple can still be $40- 50perunit. If the new chips are adopted on half of the 20mnMacs sold annually,that can reduce cost by ~$500mn or impact annual EPS by ~$0.10.

    We see phased implementation – laptops, then desktops

    While the performance of the A10X fusion chip on iPad-Pro and A11 bionic chip on iPhone X (both 64-bit architectures) are significant improvements over prior generations, and the iPad Pro is considered to rival some laptops in performance, in our opinion, Apple still has work to do to develop a processor that can replace Intel chips in high end Mac desktops and notebooks. In our opinion, Apple may choose to take a phased approach to the implementation, using the first generation of internally developed chips on lower end laptops, and subsequent generations on high end desktops. 

    cornchip
Sign In or Register to comment.