Report: Apple could begin introducing ARM Macs as soon as 2020

Apple is no stranger to switching CPU instruction set architectures in its Macs when there's an advantage to be gained, and if a new report by Bloomberg is correct, the company could be on the verge of another such transition—this time, to ARM SoCs designed in-house. Such a move could ultimately evict Intel's CPUs and chipsets from Macs.

Bloomberg speculates that the move would begin in Apple's thin-and-light notebook PCs, followed by higher-power and higher-performance systems as Apple's chip-design efforts matured. The report suggests that Apple could begin powering Mac products with its own silicon as soon as 2020, a move that would likely need to be well under way by now given the length of silicon product-development cycles.

Although rumors of Macs with ARM SoCs inside have swirled for years, this report breaks at a time when Intel's ability to deliver next-generation CPUs and new microarchitectures appears to be hampered by nagging troubles ramping its 10-nm process. Since fall 2015, the blue team has moved its Skylake microarchitecture to refined versions of its 14-nm process and increased the number of those cores on a chip, and it's also introduced more capable versions of those cores for servers, workstations, and high-end desktop PCs in tandem with a new on-die interconnect.

Even with those moves, that work means Intel has been working on the "optimization" phase of its "process-architecture-optimization" cadence for quite some time. The company claims to have shipped 10-nm CPUs to customers last year, but those shipments have so far not translated to products that consumers can actually buy. Troubles with 10 nm have, by some accounts, already torpedoed products like the next-generation Knights Hill Xeon Phi accelerator.

Those delays might be holding back Apple's ability to design and introduce new, higher-performance Macs, and it's possible that Cupertino has had enough of waiting. Now that Apple has designed and produced its own CPUs, graphics processors, security coprocessors, and even platform controllers, it may be feeling confident enough to test the waters with a complete PC-class platform of its own.

Such a move would hardly be a stretch in a mainstream thin-and-light PC, given the performance of Apple's own iPad Pro family, though it's an open question whether Apple's silicon teams could design top-to-bottom replacements for the Xeon W CPUs and Radeon Pro Vega graphics processors that power the latest iMac Pro. That kind of performance and capability from Apple silicon likely remains many years into the future, and it's not even clear that Apple would go to the effort given the size of that market niche versus more mainstream users.

Of course, building Macs around its in-house silicon gives Apple complete control over its own platform, a philosophical point for the company that might outweigh any concerns about ISAs or performance. According to Bloomberg, the move would deprive Intel of about five percent of its annual revenue, but it's hard to say just how much it's worth to the blue team in goodwill to have its processors inside some of the most posh PCs on the planet. In any case, we have years to wait to see just how broad Apple's ambitions will be if Bloomberg's report is correct.

Comments closed
    • tipoo
    • 2 years ago

    I can’t find any hint towards the leanings of dedicated GPU models – will they make only the integrated ones, or are they going whole banana and making dedicated GPUs as well? As of the A11 they do have a fully custom GPU.

    Designing the primary API and custom GPU and whole stack of the OS could have some advantages, i.e the Scorpio brings some DX12 functions into hardware.

    • Voldenuit
    • 2 years ago

    They should call it OS X RT.

    • techguy
    • 2 years ago

    I have to laugh at all the kiddies just getting into tech and saying “no way this will happen”.

    Here’s an article on cultofmac from 7 years ago that says this rumor was old: [url<]https://www.cultofmac.com/109835/intel-apple-switching-macbooks-to-arm-is-a-very-real-and-scary-threat-for-us/[/url<] It's going to happen. Why? Because it's Apple's business model. Control the entire stack. Vertical integration. Profit.

      • blastdoor
      • 2 years ago

      I certainly agree that it could make a lot of sense.

      But if Intel were able/willing to keep offering Apple a product with the characteristics and price Apple wants, then it wouldn’t make sense for Apple to switch.

      Perhaps my failed 2017 prediction can be partially explained by Intel doing a reasonable good job of meeting Apple’s demands. It’s a little tricky, though, because things that happened in the product space in 2017 are based on design decisions and negotiations from several years ago.

      Perhaps in 2013 or so, Intel’s fabrication dominance seemed secure, and perhaps Intel made some key concessions to Apple, resulting in no ARM in Macs in 2017.

      But perhaps the strong performance of the foundries over the last several years has increased Apple’s confidence that they can stand toe to toe with Apple, and perhaps now Intel just can’t make the concessions needed to overcome that new reality. And so we get this rumor…

    • tipoo
    • 2 years ago

    Cautiously optimistic. It would let them decouple themselves from all of Intels blunders and delays, and support whatever they want when they want, i.e Intel not supporting LPDDR4 and limiting them to 16GB while Apple takes the poop for it. Or outside of Apple, only providing 12 PCI-E lanes on U series processors so Microsoft took poop for not being able to do a 4 lane GPU, 4 lane SSD, and Thunderbolt 3 at the same time in the Surface Book 2.

    They only directly compete in one space so far and that’s fanless, it would certainly be interesting to see if they could maintain that edge on say a 45 watt, actively cooled part in a 15″ rMBP. Certainly they could already make a compelling 12″ Macbook out of it, but how it scales will be the most interesting thing to see.

    Question is if Apple could avoid blunders that would put them behind PC chips on their own, but their execution on the A series has been very good year over year.

    Also wondering about the GPU side, they make a fully custom one as of the A11, would they take over that part in the dedicated GPU mac models, or only the integrated only ones?

    • srg86
    • 2 years ago

    I can see precedent for this, especially for someone that likes to lock you in to everything like Apple.

    That precedent is Commodore. In the late 70s and 80s, after buying out MOS Technology, Commodore were almost completely vertically integrated, they designed and fabbed their own chips, though they still used the same MS BASIC for years.

    This is partly why they could compete on price with the Commodore 64 so aggressively.

    Heck, from what I understand, they even had their own LCD plant!

    • psuedonymous
    • 2 years ago

    I could see an iPad Pro variant that docks into a keyboard, but ARM on Mac OS is just not going to be feasible. PPC+x86 happened because x86 was such a huge leap in performance that backward-compatibility through instruction set translation was actually viable without a massive performance penalty. That is absolutely not the case with x86 to ARM.

    Plus, Apple have hit Intel with the Guaranteed Minimum Order Volume stick so hard an Intel+AMD composite part popped out. I’d expect them to use it.

    • ronch
    • 2 years ago

    Crazy twist: Apple adopts RISC V, not ARM.

      • Redocbew
      • 2 years ago

      Hey man! Who’s side are you on anyway?

        • tipoo
        • 2 years ago

        He’s a risc taker!

          • alloyD
          • 2 years ago

          RISC is good! Hack the Gibson!

    • the
    • 2 years ago

    This should make for a very interesting WWDC this year. Any sort of platform transition starts with developers a few years earlier. The current rumor this year, independent of this one, is that Apple permit developers to release combo macOS/iOS applications. That would be a very big step in the transition. The low level features of distributing multiple binaries in one application has been part of OS X, now macOS, since it released 17 years ago. Apple has initially planned for a 32 bit to 64 bit PowerPC transition at the time but the same mechanism is what permitted a quick transition from PowerPC to x86 in a matter of a few years (initial support in OS X 10.4.6 with going x86 only in 10.6 and no more PowerPC emulation in 10.7). Including ARM binaries in macOS bundles is likely going to be a thing this year if the above rumors are true. Transitioning the Mac line up to ARM in 2020 would be trivial in that context.

    The big thing that would be lost would be booting into Windows. That was once the big feature of Apple’s line up. Back in the day the cheapest dual socket workstation you could get was from Apple: it was difficult to even white box build a dual socket workstation for less than what Apple was selling the Mac Pro for. That brought in a lot of new users and continued success in the phone and mobile market has widened their user base. No longer running Windows natively may not be that big of a deal in the end. At this stage, those that needed both Windows and Mac have likely migrated to a Windows only system due to Apple’s lack of modern professional systems. Those that remain Mac users can probably make due with emulation or a pure macOS system. If it take off, there the possibility to run Windows on ARM on Apple’s future hardware as a plan b if Apple’s transition is a bit rocky.

    • davidbowser
    • 2 years ago

    I’m really having a hard time with this. I guess I could understand if was Apple releasing a MacBook line with ARM processors, but that makes me wonder if it would be running on iOS or MacOS.

    It comes down to
    [list=a<] [*<]use case [/*<][*<]target market [/*<][*<]apps [/*<][*<]operating environment (OS and hardware) [/*<] [/list<] If the use case is mobile computing for students, then all the apps they use would need to be available and supported, which would necessitate having the OS and required hardware to run the apps. Assuming the ARM MacBook is running MacOS, then the software would be available and supported, and the OS would likely have to run the app in x86 emulation until cross-compiled binaries are released. So in that scenario, the question would be the performance of the emulation on ARM. Assuming the ARM touchscreen MacBook running iOS, then the software [i<]likely[/i<] would be available and supported, and the OS would run normally. Take those scenarios and substitute business users, and you have a tougher time with app support and the expected horsepower to run things locally. Consumers would be a mixed bag (but way better than business users) because consumers are already big users of smartphone apps, but the non-iOS apps would be a killer.

    • adisor19
    • 2 years ago

    Myess, the writing has been on the wall for quite some time.. the MacBook can be converted to A11 SoC today if Apple was willing and with the rumored macOS 10.14 and iOS 12 sharing the same APIs, things could be happening behind their lab doors.

    At this point, it’s a matter of when, not if.

    Adi

    • Takeshi7
    • 2 years ago

    I’m patiently waiting for the Power Mac G6.

      • tipoo
      • 2 years ago

      [url<]https://www.phoronix.com/scan.php?page=news_item&px=Talos-2-POWER9-Pre-Order[/url<]

    • blastdoor
    • 2 years ago

    [quote<] it's an open question whether Apple's silicon teams could design top-to-bottom replacements for the Xeon W CPUs and Radeon Pro Vega graphics processors that power the latest iMac Pro. That kind of performance and capability from Apple silicon likely remains many years into the future, and it's not even clear that Apple would go to the effort given the size of that market niche versus more mainstream users.[/quote<] It's very hard for me to guess what Apple will do here. But I think that they *should* regard the high-end as worthy of their efforts. The market for a high-end Apple-designed CPU isn't just the Mac Pro, it's also Apple's own servers. How about an "iCloud Pro" service that provides pro users with on-demand processing power from an Apple cloud?

      • derFunkenstein
      • 2 years ago

      That would go pretty poorly for the discerning video or audio pro that Apple typically caters to. Upload a file tens of GB in size to render in the cloud and download it again? No way.

        • blastdoor
        • 2 years ago

        Apple customers are routinely, seamlessly moving photos and videos across devices through Apple’s cloud. Right now, that cloud is just storing files and syncing across devices. Why not add in some processing power, too?

        Regarding a file that’s tens of GBs in size — whether it makes sense to load it up to the cloud and download it again depends on how long the processing will take and how that compares to the file transfer speed. Allegedly 1 to 10 gigabit per second will be feasible with 5g wireless.

          • chuckula
          • 2 years ago

          [quote<]Regarding a file that's tens of GBs in size -- whether it makes sense to load it up to the cloud and download it again depends on how long the processing will take and how that compares to the file transfer speed. Allegedly 1 to 10 gigabit per second will be feasible with 5g wireless.[/quote<] OK look, assuming Apple is going to magic up an ARM chip that actually outperforms a 2020 x86 part by any worthwhile amount is a clear sign of RDF. But assuming that 5G wireless is going to magically deliver you multi-gigabit links to remote cloud systems for more than about 30 seconds a month is just plain nuts.

            • K-L-Waster
            • 2 years ago

            Quiet, you. It’s Apple, so technology constraints don’t apply.

            Physics may be Laws to everyone else, but to Apple it’s just friendly recommendations…

      • davidbowser
      • 2 years ago

      Disclaimer – I work for Google, buy opinions are my own

      [url<]https://www.cnbc.com/2018/02/26/apple-confirms-it-uses-google-cloud-for-icloud.html[/url<]

        • abiprithtr
        • 2 years ago

        [i<]Disclaimer - I work for Google, buy opinions are my own[/i<] Look, not all of us are rich enough to buy your opinions. You [i<][b<]have[/b<][/i<] to let us rent them for a day or two, instead!!

      • Eversor
      • 2 years ago

      I find it very hard that Apple would build a server/workstation chip. There is a lot of hard stuff to implement that Apple would need to ship volume, for it to make economic sense.
      For laptops/iMacs they would be just using more powerful versions of the same SoC.
      For big machines they’d have to look at Qualcomm Centriq / Cavium chips for when the stack is close to transitioned from x86.

        • blastdoor
        • 2 years ago

        Could you provide examples of the kinds of things that a big machine would need that is hard to implement? I’m curious to better understand what that means.

          • Eversor
          • 2 years ago

          Sure. I mean stuff on the hardware level. Things like processor interconnects, quad channel memory controllers, multi-core fabrics, etc – stuff you have and need in server chips but not in desktop/mobile ones.
          Without a good interconnect, a quad A11 chip would be somewhat useless and a Windows/Linux based product would be much higher performance.

            • blastdoor
            • 2 years ago

            So, Apple currently makes SOCs with 6 cores that work together (2 big, 4 little). Say they switch to 8 big cores — is that a radically harder problem to solve? How about 16?

            In terms of dual socket systems, Apple has been making those for a while, both for PPC and x86.

            I dunno… I’m skeptical that this stuff is so hard that Apple can’t do it in Macs, or that it wouldn’t make economic sense to do it. Especially if they end up using it for their own servers.

            • derFunkenstein
            • 2 years ago

            no, the hard part is what constitutes “high performance”. High performance in a phone is not the same as high performance on a desktop.

            • blastdoor
            • 2 years ago

            Ok… so, what does that mean exactly?

            More HZ, more cores, more IPC?

            What?

            • derFunkenstein
            • 2 years ago

            More single-thread performance, mostly. There’s a huge gulf in, for instance, browser JavaScript performance in Safari on iOS and Mac. Gotta get there somehow.

            • blastdoor
            • 2 years ago

            [url<]https://arstechnica.com/gadgets/2017/06/review-10-5-inch-ipad-pro-is-pro-hardware-patiently-waiting-for-pro-software/[/url<] From the ArsTechnica review of the 2017 iPad Pro, it looks like the iPad Pro is ahead of the MacBook and close to a 2016 MacBook Pro in the browser benchmarks. I certainly donโ€™t see a โ€œhuge gulfโ€ in performance.

            • Redocbew
            • 2 years ago

            That’s because that review runs a single test using the macbook, and that single test is Geekbench.

            A bad benchmark doesn’t “get close”. The numbers just mean nothing.

            • tipoo
            • 2 years ago

            Everyone criticises GB but just about every A11 benchmark points the same way. Browser tests are also up there with the 12″ Macbook at least, and actually nipping at the base Macbook pro (which has a fan).

            • Redocbew
            • 2 years ago

            The thing is, everybody looks at those numbers and thinks “Wow! This is almost as fast as a desktop! Apple has a desktop chip in a phone!”. Clearly they don’t. Regardless of current performance there would obviously still be a lot of work to do there.

            There’s never been a “one size fits all” benchmark that really works. From what I can put together the single threaded performance of the A11 looks to be an incremental improvement over the A10X. That’s great, but I don’t think trying to use any of this as an indicator of what performance of a theoretical desktop chip from Apple might be is going to be anywhere near accurate.

            • blastdoor
            • 2 years ago

            Not true โ€” they also used Kraken, Jetstream, and Octane.

            • Redocbew
            • 2 years ago

            Ok, let’s say for the sake of argument that I buy these performance claims. I’m still not sure what it proves. The A11 is not a desktop chip. You can’t use it to predict performance, and the performance we see today isn’t what Apple should be targeting with whatever mystery skunkworks they may or may not have in the works. So all of this amounts to what exactly?

            At the risk of being overly Krogoth I’ll also say I’ve never really seen the point in browser benchmarks when network performance is likely to mask any difference in hardware anyway.

        • Usacomp2k3
        • 2 years ago

        You could throw 4+ A11 chips into a Mac Pro and performance would probably scale pretty well.

    • derFunkenstein
    • 2 years ago

    To really believe this is going to happen, you have to believe that everybody’s process nodes are named equally. Intel getting to 10nm is way different than TSMC or GloFo getting there. So for all of Intel’s problems (which I do agree are a real thing), they’re still leading this game. So “Intel problems” are not sufficient. If anything, this will give Apple some negotiating leverage with Intel but not much more.

    [quote<]In any case, we have years to wait to see just how broad Apple's ambitions will be if Bloomberg's report is correct.[/quote<] Especially with a new Mac Pro coming this year. We all know Apple only updates the design about once every five years. 2023 is a ways off. ๐Ÿ˜‰

      • blastdoor
      • 2 years ago

      You actually don’t have to believe that at all.

      I don’t believe that TSMC 7nm == Intel 7nm.

      But if TSMC 7nm arrives this year in iPhones and Intel is still struggling to ship any measurable quantity of anything on Intel 10nm, then I think it will be reasonable to conclude that Intel is in trouble.

        • the
        • 2 years ago

        At this point, TSMC just has to ship. Intel’s 10 nm seems to be broken as it is at least 18 months behind schedule and recently have dual core mobile chips even appeared on ARK but spotting them in the wild is like seeing a pink unicorn.

        Intel has clearly lost their lead as everyone has caught up. It is just a wait now to see who pulls ahead.

      • Redocbew
      • 2 years ago

      The last time this rumor came up it was a precursor to that ridiculous OLED touchscreen above the keyboard. It might not even be the CPU we’re talking about here at all, so count me in as part of the “wait and see” group as well.

        • derFunkenstein
        • 2 years ago

        That’s a good point that didn’t even cross my mind. Who knows what this could be, Face ID built into the webcam?

          • tipoo
          • 2 years ago

          There were existing rumors of T2 coming to three more macs in the near term, but it would be a pretty low brow reporter that confused the two issues.

            • blastdoor
            • 2 years ago

            Yeah, that would be a major reporting fail.

            • Redocbew
            • 2 years ago

            If they were able to confirm that the mystery ARM chip in the story was in fact the T2, then yes it would be. In this case we don’t know that or much of anything else.

            The only thing we really know for sure is that Apple + ARM = bonkers, again.

      • tipoo
      • 2 years ago

      Irrespective of naming, iirc there’s going to be a convergence in pitch size in the next very few years. Might be that Apple also considered that eroding lead.

      • cegras
      • 2 years ago

      They will about the same:

      [url<]https://www.semiwiki.com/forum/content/7191-iedm-2017-intel-versus-globalfoundries-leading-edge.html[/url<]

    • Kretschmer
    • 2 years ago

    I see zero evidence that Apple is interested in “design[ing] and introduc[ing] new, higher-performance Macs.” For gods sake, the Mac Pro is running an Intel Chip from *2013* on the *22nm* process.

      • chuckula
      • 2 years ago

      All the better to be used as the comparison point for Geekbench numbers in 2020!

      • blastdoor
      • 2 years ago

      A year ago I would have agreed wholeheartedly.

      But now we have evidence greater than zero that Apple absolutely is interested in designing and introducing higher performance Macs. We have the iMac Pro introduced last December and the promise of a new Mac Pro this year.

      We also have some unprecedented admissions from Apple of error and regret regarding the Mac Pro situation.

      So, it’s just not true that there’s “zero evidence” of interest in designing and introducing higher performance Macs.

        • the
        • 2 years ago

        Here is a link to the relevant article that Apple admitted to its mistakes with the 2013 Mac Pro is and is promising something in 2018:

        [url<]https://techcrunch.com/2017/04/04/apple-pushes-the-reset-button-on-the-mac-pro/[/url<]

          • derFunkenstein
          • 2 years ago

          If what’s wrong with the Mac Pro is that it was too limited, adding ARM to it seems even less helpful. If anything, it seems like a strike against ARM-based Macs.

            • the
            • 2 years ago

            You’re under the presumption that Apple would take the current trash can iteration and slap an ARM chip on it instead of a Xeon. It is clear that Apple is going to be re-adding PCIe slots and storage to the Mac Pro when it is released later this year. They admitted their fundamental mistakes with that design and are correcting.

            • derFunkenstein
            • 2 years ago

            I find it hard to believe it would simultaneously “go pro” and “go ARM” at the same time, is all.

            • the
            • 2 years ago

            They likely won’t when switching to ARM. Apple will follow the same plan as the PowerPC to X86 transition with mobile and low end desktops first, then follow up with professional machines last.

            There will be a new x86 Mac Pro before an ARM Mac Pro arrives. Unless Apple completely drops the ball again, they’ll likely have the same form factor.

            • Kretschmer
            • 2 years ago

            “There will be a new x86 Mac Pro before an ARM Mac Pro arrives.”

            If that ARM Mac Pro is released before 2023, you might not want to put money on this. ๐Ÿ˜€

            • blastdoor
            • 2 years ago

            There was a time when the same could be said of x86.

            • Kretschmer
            • 2 years ago

            Not at all. When Apple transitioned from PPC “G5s” to x86, those G5s were being thwomped by Intel and AMD chips with faster clocks. At that point, their choices were “switch” or “fall hopelessly behind.”

            • blastdoor
            • 2 years ago

            So…. maybe you’re really young?

            In 1994, x86 was viewed as totally inappropriate for a high-end workstation (which is what a Mac Pro is). So was Windows, by the way. Back then, the high-end was a Unix workstation with a RISC processor, like SPARC, POWER, or Alpha.

            The Pentium Pro was Intel’s first stab at legitimacy in the high end.

            Itanium was supposed to really be Intel’s entry into the high-end because even Intel didn’t view x86 as being “pro” level.

            It really wasn’t until AMD added 64 bit extensions that x86 moved into the real high-end.

            As the history of x86 has shown, the fact that an ISA wasn’t used in high-end systems at time T does not mean that it is impossible for it to be used in such systems at time T+1.

            • Concupiscence
            • 2 years ago

            That’s true, but x86 isn’t having any trouble dominating in performance outside of a few niche applications. That Apple’s succeeded in building capably fast dual-core chips in a tight thermal envelope doesn’t imply they’ll be able to leapfrog the PC platform’s capabilities. They may be able to offer spry speed within a small power budget, but the real appeal for Apple may be good-enough performance and attractive power consumption specs on nobody’s terms but their own.

            • blastdoor
            • 2 years ago

            A couple of points/thoughts:

            1. The A11 has 6 CPU cores and 3 GPU “cores”. All cores designed by Apple. Interconnects by Apple. Just because this is in a phone, that doesn’t mean it’s simple.

            2. Performance/watt is hugely important at all levels, from smartphone to server room. So achieving the greatest performance/watt in any given setting really is important. Apple and Intel are the best at this metric in the areas where they offer products. Based on Intel’s performance/watt with Xeons, I actually do think they could compete with Apple in performance/watt in a smartphone [b<]if they tried[/b<]. Similarly, I think Apple could compete with Xeons [b<]if they tried[/b<]. The question is just whether they'll try. Intel decided long ago not to really try to compete in the smartphone space -- terrible decision, but now it would be incredibly hard to reverse it.

            • Redocbew
            • 2 years ago

            The A11 has two “high performance” cores, and four “high efficiency” cores. Whatever that means.

            Maybe this is Apple’s way of saying “You asked. We listened. Behold! We’ve given you MOAR COREZ!

            … even though you don’t really need them in a phone.”

            The point is that the A11 being a “six core” chip doesn’t really have anything to do with the prospects of ARM on the desktop.

            • derFunkenstein
            • 2 years ago

            [quote<] Whatever that means.[/quote<] It's Apple's version of big.LITTLE where the little cores do the menial tasks and the high-performance cores do all the heavy lifting.

            • blastdoor
            • 2 years ago

            Managing heterogeneous cores is harder than managing homogeneous cores.

            • Redocbew
            • 2 years ago

            That still doesn’t have anything to do with the prospects of ARM on the desktop. Apple is frickin huge, so yeah of course they’re going to be capable of doing pretty much anything they have in mind. However, not even with full control of both the hardware and the OS does it make sense to change ISAs just because you can, and there’s nothing magical about ARM which makes it any better than x86.

            I obviously have no idea whether this is going to happen or not. I’m just not seeing many good reasons for why it would.

        • Kretschmer
        • 2 years ago

        Still, the idea that Apple will be held back by Intel is asinine when Apple has plain not cared for 5 years. Do we really expect Apple to become better at fabrication and CPU design than the world leader in the same?

          • blastdoor
          • 2 years ago

          Apple doesn’t do fabrication — the question is whether the foundries can become at least as good as Intel, and the answer based on what we know now is that they arguably are already there (TSMC 10 nm vs Intel 14nm) and that it looks like they’ll stay there as later this year we get TSMC 7nm vs Intel 10nm.

          In terms of CPU design, Apple is pretty darned good. Intel and Apple don’t design CPUs for the same classes of devices, but they both lead their respective markets.

          Finally, the question isn’t whether Apple becomes “better”. The question is whether they become “good enough.” If Apple can design a CPU core that’s approximately as good as Intel’s Core, with a transistor density approximately equal to Intel’s, then that’s enough. By adding some custom silicon for whatever software feature Apple is pushing, and getting a lower marginal cost on the SOC, it becomes worth it.

            • Kretschmer
            • 2 years ago

            I don’t buy process parity. From what I remember, each fab defines “XXnm” differently, with Intel’s definition being the most rigorous. So TSMC 7nm may not be superior to Intel 10nm. On top of that, process issues with yields, leakage, and the like make execution as important as process size.

            I can see “general purpose” Macs switching to ARM as a “good enough” solution for users who instagram, facebook, browse, and stream – those users really just need a phone with a larger screen – but it would take an unholy amount of investment and risk to unseat Intel at the top of the performance CPU heap. What’s more likely is that Apple just drops their Mac Pros and MacBook Pros altogether, becoming a purely mainstream consumer computer company. Then ARM would dovetail perfectly with their “good enough and wicked light” aspirations.

            • blastdoor
            • 2 years ago

            We won’t know for sure until we can compare products using both processes, but publicly available analyses suggest TSMC 7nm and Intel 10nm are likely to very similar — aka, “process parity.”

      • ptsant
      • 2 years ago

      Based on recent trends, Apple is mainly interested in thinner and more expensive Macs. Arm chips might fit that strategy.

      • End User
      • 2 years ago

      This is a stupefyingly inaccurate statement (SSK level bad):

      – Apple just released the Xeon based iMac Pro.
      – [url=https://daringfireball.net/2017/04/the_mac_pro_lives<]Back in 2017 Apple announced that a new Mac Pro is on the way.[/url<] - 8th gen Intel based MacBook Pros will make an appearance as well. 2018 is going to be all about high performance Pro systems from Apple.

    • chuckula
    • 2 years ago

    RYZEN IN MACBOOKS….. not confirmed.

    ๐Ÿ™

      • Growler
      • 2 years ago

      I’d be shocked to see buffalo in Cupertino.

    • Techgoudy
    • 2 years ago

    [quote<]Those delays might be holding back Apple's ability to design and introduce new, higher-performance Macs, and it's possible that Cupertino has had enough of waiting.[/quote<] No doubt in my mind that Apple has been tinkering around with this idea for quite some time really. I don't think it has much to do with Intel being stagnant either, but rather Apple doing the thing it's always done which is trying to control it's entire ecosystem.

    • DancinJack
    • 2 years ago

    [url<]https://techreport.com/forums/viewtopic.php?p=1380344#p1380344[/url<] Come yell at each other in this thread about the above topic.

      • derFunkenstein
      • 2 years ago

      In big, bold letters on the last few posts.

    • sweatshopking
    • 2 years ago

    So I said this was coming years ago, and while some agreed, many laughed at me
    #imsad

    ALSO FIRST

      • derFunkenstein
      • 2 years ago

      I’m still laughing. I’ll believe it when Timmy C says it on a stage (or possibly when ARM Macs show up in leaked benchmarks) and basically not a moment before.

        • Eversor
        • 2 years ago

        This will be much easier than the PPC to Intel transition, as they already have a lot of code running and expertise on ARMv8.
        I’d be surprised if this isn’t a 4Q 2019 release.

          • derFunkenstein
          • 2 years ago

          Then that’s when I’ll believe it.

          • blastdoor
          • 2 years ago

          I agree — it is unambiguously clear that this would be much easier than PPC to Intel, which itself was much easier than 68k to PPC.

          The thing that is confusing to me is that Apple made those earlier transitions with far fewer financial and technical resources than they have today, and so I don’t understand why they are being so cautious now. But maybe the reason is that with those earlier transitions, they really had no choice — they were jumping from sinking ships. Today, they’re on a perfectly seaworthy ship and they’re just contemplating moving to a nicer ship.

            • Eversor
            • 2 years ago

            I think it’s a technology play, like leaving PowerVR licensing behind when they felt they weren’t keeping up with their performance targets. They’re at a point where they can leverage their SoCs for better battery life without sacrificing much or any performance.
            I mean, if you roughly compare benchmarks you see iPhone chips stepping into 7th gen i7 territory – at totally different power levels.

            The iPad Pro is sort of a starting ground into that but without full desktop apps it doesn’t go that far.

      • blastdoor
      • 2 years ago

      The same thing happened to me!

      Except that I said it so long ago, that I eventually gave up and ended up laughing at myself.

      It has always made a lot of sense to me for Apple to do this. But I think they really could have done it by now, yet they haven’t. So I really don’t know what to think.

        • sweatshopking
        • 2 years ago

        Then you’re not thinking very much. Apple likes to control everything in their products. They’re making screens, socs, and moving for other components. This is coming, but just waiting on performance and macos compatibility. When that comes idk.

          • blastdoor
          • 2 years ago

          Well, maybe I’m not. But the logic of this has seemed quite compelling for a long time, and Apple has had the financial and technical resources needed to pull it off for quite some time, and yet it keeps on not happening.

          I feel like I need to understand why my prediction failed before I can confidently make another one.

            • sweatshopking
            • 2 years ago

            I don’t think it has failed. We’ve seen how hard it is to switch archs before, and apple will wait until they’re totally done.

            • blastdoor
            • 2 years ago

            Well, my prediction failed, because I specified a date — I said it would happen by the end of 2017.

            Clearly the smart prediction is to not specify a date. Then you can never be proved wrong, but always have hope of being proved right ๐Ÿ™‚

      • Eversor
      • 2 years ago

      You are not alone. Around the time the iPhone 7 was out, I was also downvoted to oblivion for suggesting x86 may be in trouble if Intel didn’t get back on track.
      It was pretty obvious then that Apple was capable of building high performance micro architectures, which also ran with much less power. Nvidia was (and still is) also doing the same by designing the micro-arch, as is Samsung.

      The A11 is a beast running web workloads on Apple’s JS engine and is neck and neck with an i7 Skylake U on what benchmarks can be compared. The quad Intel 8s have some advantage but certainly not at the same efficiency and a bigger A11 variant should be coming soon.

      If Apple does move, you’ll have all day battery life with the same performance as comparable Intel chips.

      Last week, I got sent a tweet by a friend which showed QC Centriq 2400 running a workload at the same performance as an equivalent Xeon that required 50% more power. The Meltdown debacle and Intel’s response has only added motivation to look more closely at ARM on the enterprise.

        • Rza79
        • 2 years ago

        [quote<]The Meltdown debacle and Intel's response has only added motivation to look more closely at ARM on the enterprise.[/quote<] ARM is vulnerable to meltdown and spectre too. [url<]https://www.techarp.com/guides/complete-meltdown-spectre-cpu-list/[/url<] [url<]https://www.itproportal.com/news/qualcomm-confirms-its-hardware-is-affected-by-meltdown-spectre/[/url<]

          • Eversor
          • 2 years ago

          I know, that wasn’t the point. ARM’s recent architectures that have issues can only leak data through registers and the only one doing “full Meltdown” through cache timing attacks is the upcoming A75.

          ARM’s disclosure about the issues was much better than both AMD and Intel.
          You had Linux kernel developers publicly complaining how they handled both the disclosure and the micro code updates – remember the mess that was? More than a month after, no one even knew which u-archs were affected, how, and which would receive microcode updates or not.

Pin It on Pinterest

Share This