Editorial: Intel CPU constraints are sign on the road to ARM chips in the Mac

13

Comments

  • Reply 41 of 79
    Intel is not the culprit here. The real issue with that 5% hit is the problems with Mac. Now MacBook/Pro comes with unreliable keyboard, crappy webcam, display issues, T2 crash issues and many more and Apple is doing almost nothing to address these. Why someone with right mind will buy a machine with so many issues. Even if he/she gets Apple Care, after 3 years, the repair bill would be massive due to multiple failures. Lots of people are holding back for these reasons.

    It's time for Apple to address these and bring "reliable" machine like previous years so that people want to buy again. Otherwise they will lose customers to PC.
  • Reply 42 of 79
    DuhSesameDuhSesame Posts: 1,278member
    Currently I owned a 2016 MacBook, and I will say that it could be a perfect machine except the processor (and the keyboard if it’s trouble-free).  Even A12 is sufficient as long as most applications can run in good speeds.

    Yeah I know some people said faster is always better, but that’s just one aspect of a product, and fast enough is fast enough.  If you can do things fairly quick then people won’t complain.
    edited May 2019
  • Reply 43 of 79
    rotateleftbyterotateleftbyte Posts: 1,630member
    It is all fine and dandy for Apple to move to ARM based CPU's but they do need to take every other company who supplies software for MacOS with them apart from naturally Adobe Flash.
    I think that Apple will move in 2020 with an announcement in a few months to give time for developers to prepare with even some seed units before the end of the year.


  • Reply 44 of 79
    In my opinion, Apple will be taking advantage of TMSC's processes for phones, ipads, and some CPUs.  If AMD delivers some great 7nm products this fall anything is possible.  AMD is gaining speed to catch Intel in several markets.  An ARM based CPU could have its benefits as iPad Pros are becoming more powerful than some computers.
  • Reply 45 of 79
    About 5-10 years ago, Intel bragged they could produce 10nm and maybe 7nm and 5nm chips without a large investment in new foundry equipment or even larger amounts in new foundries. Looks like TSMC is leaps and bounds ahead. Read that Intel was outsourcing some 7nm work (may have been for cell phones) to TSMC. TSMC is working on 5nm chip designs. Could Intel go down the path of AMD, ultimately becoming a fabless chip designer that farms out most chip manufacturing? GlobalFoundries some years ago bought a large portion of the chip foundries formerly owned by AMD, IBM and several other firms. GlobalFoundries then spent billions upgrading them and receives outsourcing work from a number of companies. Years ago (before Apple was born), most computer manufacturers made their own processors, disk drives and memory (very few, if any, still do). Years ago, Ford took almost exclusively raw materials into a plant, and spit out a car (Model T?) at the endpoint. Times change. Eventually Intel's 38+ year old family will be just a quaint memory. Amazon's Graviton server chips (ARM based) are produced by TSMC. AMZN can offer web services on Graviton at 45% lower prices than Xeon. Google is also working on a third generation of chips. https://www.wired.com/story/new-amazon-chips-cloud-computing/
    edited May 2019
  • Reply 46 of 79
    wizard69wizard69 Posts: 13,377member
    sirozha said:
    LordZedd said:
    I've been through the 6502-68K, 68K-PPC and PPC-Intel switches. If there is a switch to ARM, I'm out. No bootcamp, no VM, no games at all, no professional apps, no drivers for anything.
    The only way for Apple to ensure that there are plenty of apps for the ARM-based macOS is
    to make a hybrid iOS/macOS device. With the Marzipan framework, the same app can run in iOS and macOS with minimal effort. That’s the only way to leverage the plethora of existing iOS apps to run in the desktop (aka ARM-based macOS) mode. Otherwise, few developers will want to develop for the ARM-based macOS. 

    By the way, the ARM-based Macs will only happen with lower-end Macs, such as MacBook and perhaps MacBook Air. The MacBook Pros will continue to be intel-based for a foreseeable future. 
    I really don’t think people understand the differences between now and the last transition.  Developers that have followed Apples guidelines will have ARM versions of their apps out very quickly.  This will not be like previous transitions where even simple apps hard machine specific code that had to be rooted out.  Beyond that ARM will be perfectly capable of emulating x86 for the few apps that need it.   I’m completely convinced that the advantages of ARM far out weigh any disadvantages.  

    As for Mac Book Pros Apple could very well make the transition there first.   The reason would be the huge win in performance they can leverage and not just with the ARM cores.    Modern software is all about Machine Learning and other AI techniques.   In this regard Apples AI acceleration hardware is extremely important moving forward.    Done right Apple can get a five year lead on the Personal computing industry.   Again we need to be looking towards the future here ARM based Macs are not about past software solutions.   
  • Reply 47 of 79
    wizard69wizard69 Posts: 13,377member
    crimguy said:
    I'd go AMD in a heartbeat.  They don't run that cool though.  I have an 1800x desktop and it is a fine performer for an incredible price.  

    Can arm hit those kind of performance numbers?  I'm asking honestly - I've seen incredible ios benchmarks but that's apples/oranges.
    AMD has made huge strides of late.   They now match Intel performance wise and are sometimes better.     Power usage is vastly improved too.   When the TSMC manufactures chips come out on the 7nm process they may well be the best choice for i86.  

    ARM has new architectures  out that vastly improve high core count machines.    Info on those was just released a few weeks ago.   While focused on server and appliance duty the tech would improve any high core count machine.   However let’s not forget that this is Apple.  

    So what is different about Apple?    Well the obvious thing is that there SOC designs are now uniquely Apples.     Effectively they can build a solution to any performance level that current technology allows. They could also join with the ARM/Fujitsu alliance to put advanced vector tech into their chips.    Let’s not forget Neural Engine and other AI related tech that Apple can implement.    Frankly Neural Engine is one of the more important reasons to get Mac on ARM real soon.   Intel isn’t holding Apple back because of poor performance they are holding Apple back due to an inability to innovate.  Neural Engine is like a 6502 at the moment, just a drop in the bucket compared to today’s 64 bit hardware.   Imagine Neural Engine with 10 or even 100 times today’s performance and features.  Apple could do 10X easy tomorrow on an ARM based desktop chip.  

    What  I’m trying to get at is that looking at the past means nothing here!!!!    The whole concept of what a processor is and how it should function is being blown out of the water.  In the past Apple transitioned to slightly better hardware that did basically the same thing.  If they go ARM don’t see that being the case.    Rather ARM will be the start of a transition to much more complex and capable chipsets.   These will support progressively more advanced AI features and other specialized hardware.  
    glynh
  • Reply 48 of 79
    wizard69wizard69 Posts: 13,377member
    bsimpsen said:

    davgreg said:
    The possibility also exists for Apple to develop its own x86 chips. They certainly have the people capable of doing the work.
    While that's technically possible, it's unlikely Intel would license x86 to Apple, and so that's legally a no-go. It also makes no financial sense to pour significant resources into another CPU architecture when the A-family is within striking distance of x86 already. There is just too much intellectual economy of scale enticing Apple to put all its wood behind the ARM arrow. Macs will go ARM or go away.
    If Apple really thought that emulation was important (it isn’t ) they could do a few things within the ARM cores to speed things up.   ARM/Apple came up with instructions to speed up JavaScript so it is very possible they could do the same thing for X86 emulation.  A couple of instructions, maybe a dedicated register array and a few other tweaks can make a huge difference.  

    The problem here is that this garbage isn’t needed.  Apple will have all of their key software native at launch.  They will also have all open source software running native.   People forget that Linux has been native on ARM for a long time, so plenty of software is already native.  

    I really think people are getting too worked up over nothing.  If Apple goes ARM it will be for things far more important than a silly instruction set.  
  • Reply 49 of 79
    mr lizardmr lizard Posts: 354member
    I wonder which Mac models were out of stock due to Intel’s supply constraints. 
  • Reply 50 of 79
    tmaytmay Posts: 6,244member
    wizard69 said:
    bsimpsen said:

    davgreg said:
    The possibility also exists for Apple to develop its own x86 chips. They certainly have the people capable of doing the work.
    While that's technically possible, it's unlikely Intel would license x86 to Apple, and so that's legally a no-go. It also makes no financial sense to pour significant resources into another CPU architecture when the A-family is within striking distance of x86 already. There is just too much intellectual economy of scale enticing Apple to put all its wood behind the ARM arrow. Macs will go ARM or go away.
    If Apple really thought that emulation was important (it isn’t ) they could do a few things within the ARM cores to speed things up.   ARM/Apple came up with instructions to speed up JavaScript so it is very possible they could do the same thing for X86 emulation.  A couple of instructions, maybe a dedicated register array and a few other tweaks can make a huge difference.  

    The problem here is that this garbage isn’t needed.  Apple will have all of their key software native at launch.  They will also have all open source software running native.   People forget that Linux has been native on ARM for a long time, so plenty of software is already native.  

    I really think people are getting too worked up over nothing.  If Apple goes ARM it will be for things far more important than a silly instruction set.  
    I'd argue that some behind the scenes discussions with MS, who also wants to move faster to ARM, would be a worthwhile path for Apple.
    Would Apple sweeten the deal with a first look, and exclusive access to Apple's Mac SOC for a Windows 10 Product? Note that Apple is very keen on the services business today, and Marzipan on a Windows 10 machine would be a "thing" as well.
  • Reply 51 of 79
    mr minskymr minsky Posts: 11member
    Ok maybe not relevant on the laptop side, but ARM grew out of work started by Acorn Computers (ARM = Acorn RISC Machine). The Acorn RiscPC computer that Acorn released powered by early ARM chips also allowed a riser board with a 2nd, or co, processor. Acorn released a 486 co-processor board that could be harnessed via the main OS (riscOS) as an independent, but of course usable, window running (for example) MS Windows. If this could be done 20 something years ago at the birth of ARM then I‘m pretty certain that Apple could do this today if they wanted... at least on the desktop side. Everyone would be a winner in this instance, and it would allow for a beast of a (modular) pro-machine.
  • Reply 52 of 79
    wizard69wizard69 Posts: 13,377member
    tmay said:
    wizard69 said:
    bsimpsen said:

    davgreg said:
    The possibility also exists for Apple to develop its own x86 chips. They certainly have the people capable of doing the work.
    While that's technically possible, it's unlikely Intel would license x86 to Apple, and so that's legally a no-go. It also makes no financial sense to pour significant resources into another CPU architecture when the A-family is within striking distance of x86 already. There is just too much intellectual economy of scale enticing Apple to put all its wood behind the ARM arrow. Macs will go ARM or go away.
    If Apple really thought that emulation was important (it isn’t ) they could do a few things within the ARM cores to speed things up.   ARM/Apple came up with instructions to speed up JavaScript so it is very possible they could do the same thing for X86 emulation.  A couple of instructions, maybe a dedicated register array and a few other tweaks can make a huge difference.  

    The problem here is that this garbage isn’t needed.  Apple will have all of their key software native at launch.  They will also have all open source software running native.   People forget that Linux has been native on ARM for a long time, so plenty of software is already native.  

    I really think people are getting too worked up over nothing.  If Apple goes ARM it will be for things far more important than a silly instruction set.  
    I'd argue that some behind the scenes discussions with MS, who also wants to move faster to ARM, would be a worthwhile path for Apple.
    Would Apple sweeten the deal with a first look, and exclusive access to Apple's Mac SOC for a Windows 10 Product? Note that Apple is very keen on the services business today, and Marzipan on a Windows 10 machine would be a "thing" as well.
    I’m not sure we will ever see Apples OS and key libraries running on other systems.   I suspect that prior experience there would destroy any attempt to do so.  

    As for MS early access that wouldn’t be handled any differently than in the past.  However I still don’t think people grasp that a transition for most developers will be far easier than with previous transitions. Apple has been warning and imploring developers for years to stay away from hardware specific code.   If they have done so porting will be a minor issue for most developers.      It simply isn’t common to write whole apps in assembler anymore.  More so most Mac development has revolved around Objective C, C++ and now Swift, these are high level languages that will ease the transition. 

    Now im not saying that poorly written code doesn’t exist or that assembly doesn’t exist, all I’m saying is that life will be far easier for developers than in the past.  So all this nonsense about emulation, and poor performance simply isn’t justified.  
  • Reply 53 of 79
    wizard69wizard69 Posts: 13,377member
    mr minsky said:
    Ok maybe not relevant on the laptop side, but ARM grew out of work started by Acorn Computers (ARM = Acorn RISC Machine). The Acorn RiscPC computer that Acorn released powered by early ARM chips also allowed a riser board with a 2nd, or co, processor. Acorn released a 486 co-processor board that could be harnessed via the main OS (riscOS) as an independent, but of course usable, window running (for example) MS Windows. If this could be done 20 something years ago at the birth of ARM then I‘m pretty certain that Apple could do this today if they wanted... at least on the desktop side. Everyone would be a winner in this instance, and it would allow for a beast of a (modular) pro-machine.
    There are all sorts of “they could do this” solutions but my point is they will not have too.  The vast majority of important software will be quickly running native.  

    Look at it this way developers on a daily basis build and run iPad software on their Macs.   It isn’t a big deal to cross compile these days.  The reverse, that is building on a Mac and cross compiling for an ARM based Mac is equally easy to do.  Of course we need for an ARM based Mac to exist but that isn’t an impossibility, it just doesn’t exist at the moment.  
  • Reply 54 of 79
    frantisekfrantisek Posts: 755member
    There is one point. There is huge group of teenagers living on iPhone/iPad becoming adult with own money. Apple need HW to feed their future purchases. Something closer to iPad experience can be appealing to them. iOS and macOS are merging slowly but steadily step by step. And Apple is working to make apps universal. So for target audience ARM mac means more apps, more iOS integration and better speed, probably much better graphic speed so much better gaming experience ....
    edited May 2019
  • Reply 55 of 79
    DuhSesameDuhSesame Posts: 1,278member
    JWSC said:
    sumergo said:
    wood1208 said:
    If Intel executes 10nm chips schedule properly than it will be last Intel chips to go in Macbook/PRO. Intel in conference call said it will make 10nm ICE processors based on Lakefield packaging for one specific customer and my guess that could be Apple.
    We're coming up on two years since the first estimated delivery date has passed on this one, so I'm not holding my breath.
    But DEMINSD does have a point.  Apple's current wish for thin-at-all-costs and damn the functionality/usability/laws-of-physics causes problems for their suppliers - and thence the customer.

    "Thin & quiet" is a Jony Ive industrial design / hardware view - it neglects the software/usability of devices - how they actually work - in favour of their "sexy looks".

    That is such a bogus narrative that it hardly deserves a response.  But I’ll bite.

    The overwhelming majority of users do not care about and do not need bleeding edge performance.  They want mobility, portability, and a processor that doesn’t suck up battery life.

    You and a few others keep going at Jony Ive like he’s some crazy nutter who understands nothing about those who use Apple products.  And that’s just flat out wrong.  Jony and his team are usability experts.  They know shit you haven’t even dreamed of.  What do you know about usability?

    Intel missed the mobility boat, which is one reason they are in such trouble today.  Wishing Apple down that same path is bonkers!

    This is all BS.  People who criticize it with “thickness obsession” often have no idea what they’re talking about, they just want to find something that looks “obvious” to give them a reason to blame.  I think the article states well, it is designed for Intel’s failed roadmap, and they probably didn’t realized that when they released the new MacBook Pro.
    Did it mean I’m okay with the current dimensions?  No, at least not 2018, but it’s not like the XPS15/ThinkPad X1s/etc. did a better job, if they did score little higher, so unless you want twice the size and tons of heat pipe, you aren’t going to cool that i9.  But of course people always ignored this then keep saying what the heck they want (One can always argue why they didn’t redesigned the thing two years after, but even that is more clever than “obsessed with thin” crap).

    Like everything else, performance is only one of the factors that goes into the final product.  Yes, some “pros” claimed they want infinite processing powers, but in the end you have to set the line.  Simply put, if it’s fast enough, then it’s fast enough.

    Now I have no interest to talk about Intel at this point, it’s all past stories.  I do interested in how Apple going to set those lines.  Personally, I would say 70~80 Celsius when stress testing is good, possibly way better than majority of the x86 machines, as they all throttled with the same test.  Considering the architecture is way more efficient, it will probably cools well with current dimensions, if not thinner.  Then again, if it needs to be a few mm thicker, I’m sure they will do it.
    I will judge myself when the new Pro comes out.
    edited May 2019
  • Reply 56 of 79
    scatzscatz Posts: 30member
    Someone may have already mentioned this, but the critical issue is software, all else is secondary. Apple has the ability and agility I think to create ARM based macs, it's just all of those developers out there with their intel based software. So, if they do this, it's going to be slow and measured. Tip-toes in the water style. Maybe create an iBook like they used to years ago with colours and ruggedness which can be used by families at home, students and children. Over time, developers will skill up and investigate how to get around creating their software on ARM macs. Once a tipping point is reached, then Apple already has the ability and I'm sure the raw CPU power to create powerful laptops. Creating a fun basic ARM based iBook as well will really grab peoples interest. Should not cost the earth to develop either, and will create a new buzz around something very different. That's my take.
  • Reply 57 of 79
    sirozhasirozha Posts: 801member
    sirozha said:
    LordZedd said:
    I've been through the 6502-68K, 68K-PPC and PPC-Intel switches. If there is a switch to ARM, I'm out. No bootcamp, no VM, no games at all, no professional apps, no drivers for anything.
    The only way for Apple to ensure that there are plenty of apps for the ARM-based macOS is
    to make a hybrid iOS/macOS device. With the Marzipan framework, the same app can run in iOS and macOS with minimal effort. That’s the only way to leverage the plethora of existing iOS apps to run in the desktop (aka ARM-based macOS) mode. Otherwise, few developers will want to develop for the ARM-based macOS. 

    By the way, the ARM-based Macs will only happen with lower-end Macs, such as MacBook and perhaps MacBook Air. The MacBook Pros will continue to be intel-based for a foreseeable future. 
    "By the way, the ARM-based Macs will only happen with lower-end Macs, such as MacBook and perhaps MacBook Air. The MacBook Pros will continue to be intel-based for a foreseeable future. "

    That's an assumption.  When Apple transitioned from PPC to Intel, they transitioned the entire Mac line in 1.5 years.  I don't expect anything different if (or when) they transition the Mac line to ARM
    This whole article is an assumption. When I said that only lower-end Macs would initially be transitioned to the ARM architecture, I based it on the fact that higher-end MacBook Pros use the Intel CPUs that cannot yet be touched by the Apple’s ARM-based CPUs. Back in the early 2000, when Apple transitioned from PPC to Intel, the reason was that Intel’s CPUs were more powerful and power-efficient than PPC. Today, the Apple ARM-based CPUs are competitive only with lower-end Intel CPUs. 
    edited May 2019 ElCapitan
  • Reply 58 of 79
    ElCapitanElCapitan Posts: 372member
    LordZedd said:
    I've been through the 6502-68K, 68K-PPC and PPC-Intel switches. If there is a switch to ARM, I'm out. No bootcamp, no VM, no games at all, no professional apps, no drivers for anything.
    Not only that but the 32-bit "appocalypse" is coinciding with the no OpenGL "appocalypse" which could have a significant impact on availability of a whole range of applications. (Given Apple has deprecated OpenGL, it is unlikely they will have any support for it in their own chips). 
  • Reply 59 of 79
    DuhSesameDuhSesame Posts: 1,278member
    ElCapitan said:
    LordZedd said:
    I've been through the 6502-68K, 68K-PPC and PPC-Intel switches. If there is a switch to ARM, I'm out. No bootcamp, no VM, no games at all, no professional apps, no drivers for anything.

    LordZedd said:
    I've been through the 6502-68K, 68K-PPC and PPC-Intel switches. If there is a switch to ARM, I'm out. No bootcamp, no VM, no games at all, no professional apps, no drivers for anything.
    Not only that but the 32-bit "appocalypse" is coinciding with the no OpenGL "appocalypse" which could have a significant impact on availability of a whole range of applications. (Given Apple has deprecated OpenGL, it is unlikely they will have any support for it in their own chips). 
    I don’t think staying with a company that failed in every way would work, but software compatibility is something you can work on.  That’s short-sighted.

    Plus, how do you know it’s definitely “No boot camp, games, VMs?”
    edited May 2019
  • Reply 60 of 79
    michelb76michelb76 Posts: 588member
    Software compatibility isn't really being worked on, or we would have CUDA support for ML and other scientific purposes. Apple is going to do their own thing as they always have and people and industries will adjust accordingly, abandon or support. ARM based macs could mean we could see more games on mac as iOS ports are then commercially more interesting. As for the scientific and certain pro communities, we will have to see. AMD gpu's aren't really populair there, and an Apple gpu will have a tough time gaining proper support unless it blows everything else out of the water.

    That said, you can bet we will have VM's on ARM based macs.
Sign In or Register to comment.