Apple may shun Intel for custom A-series chips in new Macs within 1-2 years

1356710

Comments

  • Reply 41 of 183
    chadbagchadbag Posts: 1,076member
    Quote:

    Originally Posted by bappo View Post



    Hardware wise, current chip have a very different thermal envelope than a desktop chip; there is probably quite a lot computing

    power that can be squeezed out of the Ax architecture in the next few years. But do not count in a high number of cores, like 32 or 64; this works very well in a server

    environment, it doesn't in a desktop application, unless the application has been rewritten to work on massive parallel machine; and it is not necessarily

    easy. But a 4/8 core machine could be the sweet spot.

     

    A 32 or 64 core ARM would work very well in a desktop.   Power hungry apps are already parallelized, but even ones that aren't can run on their own core so you don't have to share your CPU with other apps, or with as many other apps.   On my desktop I have 12 apps currently marked as "running".  If they were each in their own core, plus the OS processes used a few cores, then your average desktop app itself would not have to be optimized for many cores in order to get a boost from having lots of cores.  Spreading the normal work of many apps across many cores is a benefit.   None of my running apps, except maybe Xcode and a mysql access app need multiple cores themselves.

  • Reply 42 of 183
    blastdoorblastdoor Posts: 1,893member

    1. Porting software is not an obstacle. OS X is now totally free to classic MacOS baggage. Carbon is gone. Apple has total control over the developer tools. 

     

    2. Windows compatibility isn't a big deal for most users. It matters a great deal for a vocal minority, but for most users it's a non-issue. I can see Apple deciding that losing that niche is an acceptable loss. Also, emulation can be a solution for some of those users. 

     

    3. Comparing Cyclone to the Pentium-3 is absurd. Cyclone is at worst a Core 2 Duo. If you want to ignore Geekbench, fine. Then look at sun spider and tell me how Cyclone is no better than a Pentium 3. 

     

    4. The A10X that goes into a Mac will probably be about twice as fast (in terms of CPU) as the A8X. I'd guess 2/3 of that improvement will come from core improvements, 1/3 from clock speed increases made possible by die shrinks and the higher power envelope in a Mac. I'll guess the GPU will be four times as powerful as the A8X (GPU is easy to scale -- just add more transistors). That's enough for a MacBook Air. 

     

    5. I dream of a Mac Pro with a lot of A-cores.... I'm thinking two 16 core chips. It might not ever happen, but I don't think it's as crazy as it sounds. The Mac Pro by itself is a low volume product that cannot justify the investment by Apple into a custom CPU just for it. However, Apple has a lot of data centers and will continue to add more. And those Xeons are darned expensive. It might make sense to design a high-end A-chip that could be used both in Apple's data centers and in the Mac Pro. Of course, if it's in the Mac Pro, then why not the iMac? 

  • Reply 43 of 183
    So the question is: What would it take for this to become a reality and does this reality make any sense?

    Based on current PC/Mac shipments and the overall market at the current point in time, this does not make sense. The bulk of the PCs shipping today are still WinTel based, seems like Macs to fit into the environment would still need to run windows software.

    If its only windows software that needs to be replaced, then they need to come out with the ability for developers to port windows code to OS X easily. Much easier than it is today. Maybe iOS is the stepping stone to this shift. Maybe work with Microsoft to keep building the ARM version of windows. Did I just type that????

    Next step make iOS a desktop class OS. What does this mean? True multitasking, advanced UI,... what? Not sure this makes sense either. We use keyboards for typing and we have not been using voice. Would it end up at the same point that the surface is at, with a keyboard and touch screen?

    OK so maybe all of this does not matter.... We still think that it matters what compute platform is used and that market share is key to success. Apple has proven in the mobile market that market share does not mean profit share. But I think the key to this is that Apple can not move back to being a separate system that does not fit in. i.e. When we had PPC based Macs, getting a Mac was difficult to get when you worked in an all WinTel environment. Usually only marketing types used them. It was only after they adopted the ability to run windows that it became easier to 'fit' in. I believe this is still the case with the old guard IT systems in place for the majority of the companies and I don't see this changing in 1-2 years.

    Every time I see one of these articles I still have to ask, how does this make sense and why would Apple want to spend their time trying to build more hardware of this caliber? There are so many other mobile opportunities that it seems a waste to take on Intel in the big hardware arena. Keep on building great mobile and IoT devices and eventually they will take over Intel on the desktop, but it won't be a direct frontal attack as this article assumes.
  • Reply 44 of 183
    maecvs wrote: »
    That would be an incredibly stupid move. It would be like going back to the days of the Power PC chip. Windows compatibility would be gone, the critical reason the mac has made the in-roads into the PC market that it has today. The Mac would go back to being a niche product.

    You're wasting your time writing posts on this forum. You just showed you have the creative chops to write sci-fi disaster stories. Here's a story tip: San Francisco is hit with a R9 earthquake while a skyscraper is on fire... Title it "Shake and Bake,"

    Don't bother thanking me.
  • Reply 45 of 183
    What newbie wrote this? Apple already made such a switch twice, not once - from 68000 to PPC, then from PPC to Intel. In both cases they ran legacy apps under emulation while developers slowly tranisitioned to fat binaries before being able to concentrate only on the newer architecture. There's no reason to believe a similar switch wouldn't go the same way - Apple don't like customers having to worry about what software will or won't run on their machine, which is also why unlike windows we never had 32 bit and 64 bit versions of apps. A switch from Intel to ARM has one massive problem here though: in both the 68000 to PPC and the PPC to Intel transitions the newer architecture had enough power to run old software under emulation to various degrees of acceptability. ARM remains slower than the Core i3, a chip Apple has never deigned to use; what chance has an ARM architecture got of emulating Intel code in a usable way in the foreseeable future? You'd need an ARM chip that is considerably faster than a competing Intel to carry the overhead, and that looks unlikely. And no, chucking multiple cores at it won't help for the many use cases where tasks are single threaded.
  • Reply 46 of 183
    mjtomlinmjtomlin Posts: 1,819member

    Eh, of course it's entirely possible for this to happen - never mind all the people complaining about the need for Windows compatibility - but I seriously doubt Apple will release an ARM based Mac on the market before giving developers time to update their apps to target the platform. Apple could very well announce such a system is coming and then tell developers they have X amount time to update their apps in the Mac App Store to support it.

     

    One thing though, all current computers need to remain Intel-based and any new ARM-based systems will need to be a completely different line of computers, not called MacBook, iMac, etc. There needs to be a clear way for consumers (and developers) to identify which is which.

  • Reply 47 of 183
    patrickrs wrote: »
    Quote:
    nhughes wrote: »
     


    I fail to see how this is "link bait." It's a news story, and it's fairly covered as such, with zero sensationalism. I welcome criticism of our content and editorial policies, but a generic "link bait" complaint isn't contributing to the conversation here.

    The issue for me is that this is "news" about the analyst -- they just sent out a new research note -- but simply "speculation" about Apple. If your article was intended to be about the analyst, then "Breaking" (assumed short for "Breaking News") would be appropriate. If the article is about Apple, which to me it clearly is, then it should be labelled as "Speculation", not "Breaking News".

    Give it up. You're wasting oxygen the rest of us can use...
  • Reply 48 of 183
    I'll go one better. I think an ARM chip is going to power the new MacBook Air redesign.
  • Reply 49 of 183
    Great news that Apple might be considering a change from Intel. However, it will be essential to keep x86_64 code compatibility otherwise no more running Windows from the Mac and more importantly, having to recompile/rewrite applications for ARM. It might also kill a lot of Hackintosh efforts which would be a possible problem (not for Apple obviously).

    Ha ha, iOS compatibility has become more important then Windows in the mobile world. There is now more iOS apps in the app store then available on Android, Windows mobile, and Blackberry COMBINED.

    Sometime in the next 24 months iOS sales will be out-pacing PC Windows sales. Meanwhile MS would DIE for a killer app on their Surface or Phone.
  • Reply 50 of 183
    bdkennedy1 wrote: »
    I'll go one better. I think an ARM chip is going to power the new MacBook Air redesign.

    Not the expected MBA because it's been delayed waiting for the intel chip... but sometimes this year it could happen. Let's see what comes out at the developer conference.
  • Reply 51 of 183
    i could see apple positioning their macbook airs with arm chips, running their custom OS X, would be something like a google chrome killer, while the macbook pro remained on intel, positioned for professionals and those who need to run windows...but it's a stretch.

    probably most of the benefit goes to apple in terms of cost and control and integration, but i don't see the customer benefiting all that much.
  • Reply 52 of 183
    mjtomlin wrote: »
    One thing though, all current computers need to remain Intel-based and any new ARM-based systems will need to be a completely different line of computers, not called MacBook, iMac, etc. There needs to be a clear way for consumers (and developers) to identify which is which.

    That was one avenue I suggested when I made the speculation about the potential for an ARM-based Mac-like computer years ago after the A4* was introduced, but I think it can go either way: not calling it Mac does help prevent confusion, HOWEVER, since I see this placed in the low-end of the Mac market, not necessary a budget machine I could see why Apple would want to call it Mac.

    On top of that, when they talk about their traditional "PC" sales it's easier and better for them to say x-many Macs sold, that doesn't show that unit sales of Macs (i.e.: Intel-based Apple "PCs") declining, which would certainly be the case even if their Intel-based prosumer to professional models were still growing in unit sales, revenues and profits. (This assumes the MBA would be axed and would be some called something else with an Apple A-series chip.)

    With the advent of the Mac App Store (MAS) the conversion from x86_64 to a new Universal standard (x84_64/AArch64) might be tough. Apple has had success with this in the past, especially with Rosetta, although the jump in performance over PPC was so much that the emulation from Rosetta was mostly obscured from the user's perspective. I don't think they'd get that this time, but with so much vertical integration I wouldn't be surprised if Apple did have some tricks up their sleeve in this regard.


    * I was not popular for that comment, but I think people are slowly coming around to the idea.

    bdkennedy1 wrote: »
    I'll go one better. I think an ARM chip is going to power the new MacBook Air redesign.

    I'm not as certain but I certainly wouldn't be surprised.
  • Reply 53 of 183
    hattighattig Posts: 830member

    I think it is likely that at some point, Apple will migrate to using their ARM designs in Macs. They will start off with a single, low-end, consumer device, such as a 12" MacBook Air - maybe not the 2015 edition, the 2016 or 2017 editions are definite candidates.

     

    Seamless compatibility with iOS applications will be a feature - but will require the device to have a touchscreen. This will give access to some apps that might not be ARM-ised in the Mac App Store for a while, such as Microsoft Office.  I agree with the poster above that this compatibility could be more important than Windows compatibility for many consumers.

     

    Expect the version of Mac OS X that is for ARM to be themed closer/exactly like iOS. A large step towards this has already occurred with Yosemite.

     

    But what about Intel Mac OS X applications? Two options: (1) Apple requires them to recompile and update via the App Store to provide an ARM version. (2) Apple provides a AOT recompiler/emulator for the Intel apps that runs on the spare cores in the device when needed. Most desktop apps spend more time in system libraries than their own code, so performance drops will not be noticed.

     

    As regards performance - low-power Intel chips aren't very powerful, and I'm including Broadwell-U chips in this category. These are dual-core, quad-thread (but these extra threads don't really translate into a lot more performance) CPUs. We haven't seen an A8X that have been given a 15W TDP headroom, that could raise the performance of the chip. Of course, I expect the device would use it's own SoC, I'm speculating quad or octo-core CPU with 16 GPU cores (instead of 8 in the A8X). This would be around 200mm^2 on TSMC's 20nm process. A lot smaller on a 14nm process such as Samsung's.

     

    And the main reason for doing this? Reducing costs. Intel's chips are expensive. Apple gets them cheaper certainly, but making your own SoC is cheaper still. This translates to either higher profits for Apple, or cheaper devices for consumers.

     

    It may be that Apple's professional offerings stay on Intel for many years to come, and only the consumer devices transition.

  • Reply 54 of 183
    Quote:

    Originally Posted by nhughes View Post

     



    The research note was sent out this morning, we immediately began working on a story on it, and it's a big story, hence the "breaking" tag.


    So, if anyone of us put out a research note,  you'd put a story out on it?

     

    At least 10 of use have stated the 2016-2018 timeframe for ASeries intersecting with the Intel performance curve.

  • Reply 55 of 183
    Quote:

    Originally Posted by Macky the Macky View Post





    Let me spin you a more successful sales pitch: You walk into an Apple store and ask to see a laptop. The genius asks you what are you using currently, and you say an iPhone and iPad. He then asks you if you'd like to use all your current programs on your new laptop, or do you want Intel compatibility at the cost of needing new software for your new purchase. When you learn that the A-series laptop also costs less and runs longer on a single charge, what's going to stop you from making a quick decision?



    Meanwhile Apple just stopped any thoughts by you of comparing an Apple Intel-based laptop to whatever else is out there because you want the A-series Laptop. At checkout, you are asked if you'd like to buy an Apple Watch with all the money you just saved on HW and SW.



    * The A-series hardware will become faster, yes... but don't forget Apple is speeding up iOS with Metal and Swift in such a way as to get much more performance without upgrading the processor or taking a battery hit... the combo is where the future lies when Apple controls the the chips and the iOS and the programming that makes it possible for the user-designed software to run like low-level-code speeds...

     

    Ack - can't imagine this would be a comfortable conversation to have before dropping that much money. (A laptop or desktop isn't a quick purchase.) "Let's see, so what you're saying is that I can save a few bucks and buy this new MBA with Apple's chip and use it for 14/16/18 hours or stick with the Intel and use all my existing software? Hmmm, I think I'm going to have to consider this a little longer. OK, well, since I'm here maybe I'll pick up a couple of iTunes cards for my nieces."  :\

     

    I get the perspective: cooler, faster, longer, graphics, development, etc. I'm astonished at the development of the A-series. It's really come so far, so fast, and no one else is doing their own hardware and software in such a powerful, beautiful way. A couple of immediate problems though: people don't want to think about the CPU and there will be people who feel cheated when their software doesn't work (even printers or other peripherals might not work). What about those to consider themselves power users? When will the A-series be enough? And will it be a question about brawn (Intel) vs. smarts (A-series and lots of others!) Customizing gets a lot harder with a chip choice.

     

    For these reasons, I think having them coexist for OS X is a non-starter. Making a switch might make sense but you're looking at a large product line. I wonder if they would transition slowly from lower-end devices all the way to the Mac Pro. Nah, that seems fraught with road bumps, too. Interesting speculation but until there's other development I'm still skeptical.

  • Reply 56 of 183
    chadbag wrote: »
    macman1984 wrote: »
    The last line of this article is the killer for me: there is NO WAY an Ax (10,11, 12 whatever) could ever run a virtualized Windows. I need this for my work. Granted, most people don't.

    I am not sure why it couldn't.  PPC Macs ran virtualized Windows.   You probably wouldn't be playing games but you could virtualize the CPU and have 9 or so cores running it for decent performance I would guess.

    I suspect that the first manifestation of this for desktop Macs will be something like an iMac with:
    • standard n A9X APUs
    • BTO option for additional A9X APUs
    • BTO option for an additional Intel CPU
    • BTO option for GPU for use with Intel CPU

    So, you could configure to your needs and the price you are willing to pay.

    Some additional things to consider ...

    Today, most of the major 3rd-party apps are available for Mac and Web as well as for Windows.

    MS Office already runs on Apple ARM APUs.

    AFAICT, all current iOS and OS X apps run on a virtual machine ... the LLVM runtime ...

    LLVM compilers are available for Windows, iOS and OS X.

    Languages with compilers that use LLVM include Common Lisp, ActionScript, Ada, D, Fortran, Ocaml, OpenGL Shading Language, Go, Haskell, Java bytecode, Julia, Objective-C, Swift, Python, Ruby, Rust, Scala,[3] C#[4][5] and Lua.

    The Swift language is purported to execute faster than other compiled languages. it appears that Swift and Apple's LLVM have a symbiotic relationship -- where, because Swift code is safer, the compiler can generate more concise, faster-executing code (less error-checking overhead).

    It is significant that much of LLVM and Swift development was done by the same person -- Chris Lattner.

    https://en.wikipedia.org/wiki/LLVM


    Finally, the world is different, today, than it was in 2006 when Apple switched to Intel:
    • The A9X CPU/GPU will be more powerful than the Intel chips first used in the 2006 Macs
    • Most "must have" 3rd-party apps, have OS X source code -- which could be easily compiled to run on A9X ARM.
    • The action has moved from the desktop to mobile.


    I suspect, that the Apple/IBM partnership using Swift and ARM-based mobile Macs and iDevices (including iPad Pro) will penetrate the enterprise and never look back ... Much as the IBM/PC did in the 1980s.
  • Reply 57 of 183
    thttht Posts: 3,100member

    It's at least a good strategy to put pressure on Intel, for both design and economic reasons.

     

    Apple wants to have super thin, fanless desktops and laptops with the highest performance possible. I think their vision for the iMac is to have it be a monolithic thin slab of display with whatever design accompaniments like the current chin, etc. Then, they still want the laptops to be thinner. So, the current lineup of 15 W TDP chips to 65 W TDP chips all need to be driven down to 5 W to 10 W. Apple is catching up with their custom ARM SoCs while Intel is just trying to hold steady with performance while driving TDP down. It's possible Apple can develop higher performance chips at sub 10 Watt TDP levels than Intel can.

     

    Then, Intel's chips are $200 to $300 per chip or thereabouts. Apple's A8X chip probably costs them $30. That's a huge order of magnitude cost difference. Meanwhile, AMD hasn't come close to really putting any kind of price pressure on Intel. If Apple ships a Core M chip in an MBA for $999 or $899, that Core M would represent ~30% of the retail cost of the laptop. That's craziness. So, a risk mitigation of developing competitive ARM chips that can be used in a Mac would be worth it just to put pressure on Intel.

  • Reply 58 of 183
    appexappex Posts: 687member
    Disaster ahead. It seems that Apple did not learn past lessons. We need x86 for full true and real compatibility with the world (read Windows).
  • Reply 59 of 183
    Quote:
    Originally Posted by Maecvs View Post



    That would be an incredibly stupid move. It would be like going back to the days of the Power PC chip. Windows compatibility would be gone, the critical reason the mac has made the in-roads into the PC market that it has today. The Mac would go back to being a niche product.



    The Mac never really stopped being a niche product if you only look at market share numbers-- but that metric is silly (Porsche is DOOMED!), and this assertion that the only reason people bought Macs was because they could run Windows, is false.  Sure, back before iOS existed and Apple was a smaller player, being able to run Windows on a Mac was a selling point-- but that's it, it was one point.  People bought Macs because they were trying to get away from the crashy, virusy, awfulness that is Windows-- but occasionally they had to run a program that had no analogue on the Mac side.  Now that people are finally starting to understand that there are alternatives to MS junk, and developers develop like crazy for iOS and a lot more for MacOS than they used to, loosing the ability to run Windows natively would hardly hinder most people.  iPhones, iPads, Android phones, Chrome notebooks-- none of these run Windows, and the people that buy them do not care.  

  • Reply 60 of 183
    melgrossmelgross Posts: 31,471member
    chadbag wrote: »
    A 32 or 64 core ARM would work very well in a desktop.   Power hungry apps are already parallelized, but even ones that aren't can run on their own core so you don't have to share your CPU with other apps, or with as many other apps.   On my desktop I have 12 apps currently marked as "running".  If they were each in their own core, plus the OS processes used a few cores, then your average desktop app itself would not have to be optimized for many cores in order to get a boost from having lots of cores.  Spreading the normal work of many apps across many cores is a benefit.   None of my running apps, except maybe Xcode and a mysql access app need multiple cores themselves.

    This is where we get a problem. Few apps are optimized for more than two cores. That will remain true for some time. Remember that there are still a lot of two core CPU's out there. Perhaps half of the current pc's have two core chips. Developers are finding it hard to parallel most apps meaningfully past two cores.

    There are apps such as rendering apps, video editing apps, and a few others that do use numerous cores efficiently. But most don't.

    A problem in assuming that 32 or more cores would be a great thing is in assuming that so many cores would be good for multitasking, but even there, it's questionable, for most users. I have a Mac Pro with two Xeon chips, each with four cores, and eight virtual cores. Since it's easy to monitor how many cores are working, and by how much, I do that fairly often.

    Even during the most arduous tasks, I rarely find most cores are busy, even to the slightest extent. And as for all sixteen cores, well, that almost never happens, and when it does, most of them are running at very low levels.

    So having eight cores would be more than enough for most people.
Sign In or Register to comment.