Tech Apple switches to ARM for new MacBooks: why?

Who I am
Judit Llordés
@juditllordés

The new MacBooks drop Intel after 15 years.
Is this the first step towards a unified platform?

It is now official, Apple has decided to change the hardware platform for the second time in its history. The new MacBooks will speak the language of ARM. Same spoken by their mobile counterparts on iOS, and that's no accident. What lies ahead for the future of the lineup is therefore a transition period, where on the one hand support will be guaranteed to the Intel MacBooks on sale to date, while on the other hand we look forward with the new platform.



Apple's new MacBooks switch to ARM: what does it mean?

  • Why are the new MacBooks switching to ARM?
  • Advantages and disadvantages
  • Machine Learning
  • In practice, what does this mean?

Why are the new MacBooks switching to ARM?


To learn more:
Those things you can only discover by switching from Windows to Mac

Essentially, there are two reasons why Apple decided to switch to ARM for its new MacBooks. The first is purely economic: thus making the Cupertino house it no longer depends on Intel for its hardware. In fact, Apple already independently produces System on a Chip ARM, those that are then used on the iPhone and iPad. This is good - clear, from Apple's point of view, Intel would hardly agree - because it means Apple owns its hardware 100%. can he decide the roadmaps, model it on the needs of its software and ultimately produce better MacBooks.



Not only that: as mentioned, ARM is the same architecture that is used on iOS. iPhones and iPads already speak this language. More specifically, those who develop apps for iPhone and iPad speak it. The second big advantage is just that: it is a decisive step forward towards a unique ecosystem. Towards desktop applications that mimic mobile ones, imitating their stylistic features and above all putting their advantages into practice. Lighter, more intuitive, best from the point of view ofUser Experience (which is macOS's historic battle horse). Making these two worlds converge can only make macOS a more attractive product for users, who are in the situation of being able to access the amount of apps already present under iOS. And it's an assist also for developers, who see their potential catchment area expand.

Advantages and disadvantages


To learn more:
Better MacBook Pro or MacBook Air? A matter of practice

There is no practical reason, at the moment, why an app designed for the iPad Pro cannot also be used on a MacBook, if not the hardware platform that will fail with the transition to the new ARM terminals. The biggest advantage is undoubtedly this. For both ecosystems, because the idea of ​​being able to see applications that until now have remained confined to macOS also arrive on iOS - especially on the iPad - goes to also enrich the offer on mobile. Not in a massive way, not with billboard features, but it is yet another step towards a 'fluid experience in the Apple ecosystem. Something that has been pursued since the days of the iPod, born to be a multimedia “module” to be used away from your Mac.



The other advantage is that at this point a truly entry level solution for the ecosystem cannot be ruled out. Something that goes into to understand us competition with Chromebooks, a line that - as of March 2018 - corresponds to 60% of computers purchased by schools in the United States. Apple at the moment it does not have an entry-level on the same range, if not the basic version of iPad. A product that implies several compromises, especially from the point of view of connectivity - trivially, there are no USB ports natively. The new MacBook ARM, to be placed side by side with Air and Pro, would not be so senseless at this point. There is no longer a need to "pay" Intel e costs can be contained.


To learn more:
Fortnite against Apple, but the fault lies with TikTok

It is clearly not a choice without its drawbacks. First, the compatibility between the two ecosystems. Current applications, written for the Intel architecture, they just don't run on ARM. It will be necessary to compile ad-hoc versions, or to resort to some emulator - along the lines of what we saw in the transition from PowerPC to Intel. It is difficult to imagine that the market does not follow Apple and that the company itself does not personally care about compatibility with major applications, such as the Adobe suite. But the market is full of small minor apps living in their niches, and thenightmare of having to relive what happened when Apple has eliminated all applications 32 bit can not be done to exclude.


The biggest drawback, however, concerns gaming. This maneuver definitively cuts the bridges with the gaming population. True, macOS has never been a gaming platform, but something has somehow arrived anyway. Imagine one Steam for ARM, with even some games converted ad-hoc for the new platform, is science fiction. Also because it is highly probable that Apple's maneuver was designed precisely (as mentioned) to bring the two ecosystems closer and pass the bulk of applications from the App Store, intercepting their revenue. And considering what happened with Epic this summer ...


What changes for those who do Machine Learning?


To learn more:
What are teraFLOPS and why they are not talked about on PC

For those who carry out tasks a little more "demanding" at the computational level (Machine Learning but also general AI), these new MacBook ARMs promise great things. First, the RAM is shared between the CPU and GPU eliminate latency which previously existed between the two different types of RAM. This is then combined with the "neural engine", already present in the A-series chips of the iPad and iPhone, which promises to reach 11 TOPS (Tera Operations Per Second), a metric that is quite similar to the FLOPS used in gaming ..

Unfortunately, the multi-core results are very uninteresting for those who are involved in these things, or would like to start. Many libraries that pre-process data they do not use more than one core, making the presence of more computing units unnecessary and not allowing parallel operations properly so-called. In the case of the Deep Learning, the presence of the neural engine will speed up the inference phase, that is when you have to make predictions starting from a model that has already been trained.

The real catch with the new ARM MacBook models is that they stop at 16GB RAM. Many professionals already own models that have 32GB of RAM, and this downgrade it makes itself felt. For those who have this need, the advice can only be to wait for the next models, which are very likely to enrich the line with terminals equipped with 32GB of RAM. Or alternatively, to marry for the moment still in Intel: the transition will not be immediate, as already mentioned.

Furthermore, the neural engine only helps in the inference phase, not in training your model. Quite important difference, since they are two different phases.

The software on the new ARM MacBooks

We understand that ARM M1 has potential over Intel. However, professionals don't care about quibbles, but the results. Python should run smoothly, while VSCode will have a fully compatible build, it even has an experimental build at the moment. Little is known about Jupyter Notebook which may have to run with the help of Docker (hopefully not), while the evergreen Excel will have confirmed support.

What to say, here too time helps. We have to wait for the feedback from the first users to understand if all this difference is noticed compared to the Intel models. There shouldn't be any serious compatibility problems, at least on paper, but computer science has always been a very practical science.

In practice

In practice we would have to wait a little longer. Maybe even a full two-year period. Intel MacBooks on the market will still need to be supported, albeit gradually replaced by new ARM-based MacBooks. And it will be necessary to understand what will happen outside the MacBook line. Easy to expect the same transition when it comes to iMacs, but the Mac Pros instead?

There is another big consequence of giving up Intel. Windows is also given up. Bootcamp on new MacBooks with ARM is absolutely out of the question, as the OS is not compatible with the architecture. Sure, there's an ad-hoc build of Windows 10 for ARM, but it's kind of like reliving the Vietnam vet flashbacks of Windows RT. It's something Microsoft has recently launched and we don't know to what extent it will want to support. However, we are talking about the house that has not had great scruples in abandoning Windows Phone 7, and in general the Windows Phone project, despite having bought Nokia.

Speaking of things that will have a long life on the new ARM MacBooks, well, we can't help but talk about the battery

Apple itself said: This transition will bring MacBooks with the longest battery life ever. We could spoil you with the technical details, talk about how it was possible to combine the usual performance of a laptop with reduced consumption. But the really interesting thing is the practical implications. 18 hours of video playback on a single charge, compared to 12 for current models. 15 hours of internet browsing (versus 11). In addition, a practically instant “wake up” time is guaranteed from sleep mode. Just like on iPad.

What is certain is that macOS is preparing to change its face, perhaps forever. Overlapping more and more to iOS, of which, however, already with the update Big Sur is recalling the aesthetics.

The idea is to succeed where Microsoft has failed.
Forge the One Ring, a single platform to rule them all.

add a comment of Tech Apple switches to ARM for new MacBooks: why?
Comment sent successfully! We will review it in the next few hours.