Hacker News Reader

Comment by nakovet | Hacker News Reader

This is industry revolution happening before your eyes.

Excluding smartphones/tablets, X86 has been the defacto computer architecture of personal computers. Microsoft tried an ARM processor but the tradeoffs were evident and it was considered half baked, leading to poor reviews.

For quite some time in our industry the main limitation has been clock speeds, nowadays the main limitation is temperature control.

Apple launched an ARM chipset that remains as cool as 50C when intel reaches 95C, this will forever change the industry of personal computing.

That's why people are raving about it.

View on HN

Naive but genuine question: What are the major advantages of switching to an ARM processor? From Apple’s keynote, I gathered that there’s the benefit of instant-on, and allowing iPhone/iPad apps to just work in MacOS, but are there some other fundamental advantages to be had with ARM for Mac? What difference does it make for both power users and real consumers at the end of the day?

Longer battery life for individual consumers, cheaper electricity bills at scale

For Apple, deeper control of their vertically integrated stack. Which depending on who you ask, is a good or bad thing.

Other commenters here covered other benefits as well.

The benefits don't really have that much to do with ARM specifically, more that Apple can tune their silicon designs for their very niche applications, while ARM Holdings designs and other silicon IP designers have to design much more generally as they're selling to a broader customer base integrating the cores into everything from smartphones to edge switches.

I'm not sure to what extent they've done so, but with their own silicon Apple has the freedom to simply not implement the unnecessary optional bits of ARM, like 32 bit support, and the optional extensions to the ISA not applicable to desktop class general purpose compute.

The saved space in transistor budget could allow them to save precious time and energy in performance critical spaces like the instruction decoder/fetcher.

No idea what they're actually doing (they're not exactly very open about their designs) but with their own processors they're free to optimize for a specific use case and a specific kernel. A specific example of this is how they've reduced the time spent in garbage collection by a factor of 3-5 IIRC, which has pretty dramatic ramifications for both performance and memory usage (as you can do GC quicker and more often)

Just a few clarifications:

- M1 is indeed 64 bit only

- Apple's software doesn't use GC - it uses reference counting which has indeed been speeded up dramatically.

The M1's (two) CPU designs are Apple's own and there is a lot of information out there on them e.g.

https://www.anandtech.com/show/16226/apple-silicon-m1-a14-de...

Thanks for the info :)

I counted ARC as a sort of GC, though I've heard the opinion it is and isn't a type of GC, and see both sides' points

I love that article and the research behind it, but it's worth pointing out the details are sort of "reverse engineered" by probing XNU and running some diagnostic programs, not from apple supplied docs. I doubt apple will ever directly document it like IBM or Intel document their CPUs, though I hope I'm wrong!

It's a great piece and as you say it's a shame that we won't see more detailed info from Apple but I've seen a couple of interesting comments (leaks?) from Apple employees with more details (e.g. the tweets embedded in the article below). I suspect that we'll know an awful lot about the M1 by the time we've finished!

https://blog.metaobject.com/2020/11/m1-memory-and-performanc...

Another big one: not relying on 1 or 2 large companies for CPUs. Intel and especially AMD can’t afford to make (what they perceive to be) niche features on their CPUs. They have to design for the mass market, or if offering something more unusual or specialized they have to sell them at very high prices.

With ARM, any company is able to get an architecture license and go hog wild. The architecture (sticking to AArch64 since nobody cares about 32 bit ARM anymore for general purpose computing) is much cleaner and simpler with less cruft than x86. This makes it easier for a variety of companies to offer different solutions at different price, performance, and power points. Plus with the weight of Apple behind it there will surely be much more development in both the ARM software and hardware space as it applies to general purpose desktop computing.

Yeah, but Apple's stuff won't necessarily translate to generic ARM, they have their own extensions. Plus their own OSes.

I don't really see how this benefits Linux or Windows.

They have their own extensions, but having a mainstream desktop platform that runs on ARM will certainly be a push for compilers and software to have better support for ARM in general. For example, right now Docker ARM images are basically an afterthought and if something doesn't work in the ARM version a maintainer might just shrug their shoulders and view ARM as a niche use case. With a critical mass of ARM desktops that will no longer be the case.

It was surely not a given that any already compiled non-trivial code must run as is on Apple's M1. Here the whole Linux kernel and the CLI parts of the distribution are apparently running virtual just using a small number of the API calls.

It's, for my understanding, more than amazing, knowing what's already known about M1 and Apple's newest OS and surely newsworthy.

This. And now begins the decline of x86 and Intel.

AMD is making a bigger dent in Intel than Apple ever will. Apple doesn't even have 10% of laptop marketshare whereas AMD are currently at 20% for laptops and on a steady climb. Not to mention both the desktop and server markets where AMD is doing even better. So unless you think the whole PC world is going to switch to macOS or there exists another company capable of competing with AMD using ARM, we're not going to see a whole lot of change.

Broadly agree with a few caveats:

- If this gives Apple a major competitive advantage at the higher end (where presumably most of the margins are) then we may see some moves to try to compete with Arm based machines.

- A lot will depend on where MS takes ARM Windows - could MS do a competitive Rosetta style layer to give backwards compatibility?

- Much less lock-in in servers - eg AWS Gravitron and Ampere.

A tangential correction: it’s AWS Graviton (one “r”), not Gravitron (two “r”s). I used to misread it as Gravitron too initially.

Thanks - I think I've built muscle memory for Gravitron now which is very unhelpful!

> If this gives Apple a major competitive advantage at the higher end

Whatever about the highest end (really have to wait for the M1X or whatever to evaluate that), it clearly gives Apple a _huge_ advantage for the $1000-ish ultrabook market; battery life. There's simply no thin-and-light to rival the M1 MacBook Air's battery life without very severe compromises. I think the market will have to respond to that, or just cede most of the ultrabook market (which is significant).

I doubt the industry really cares all that much about the 45W 15" laptop market; no-one buys them. Mid-range ultrabooks, though, are a good combination of margin and volume.

I don't think the implication was that Apple itself is killing Intel, but rather that ARM is - this is just the first step in what many see as a trend that will have Intel scrambling.

> or there exists another company capable of competing with AMD using ARM

There's a very long history in consumer computing where the industry as a whole tries something, executes badly, says "ah well, we tried" and kind of gives up on it, then Apple does it well, then the industry notices "oh, it's possible" and does it well.

Obvious examples: Smartphones: WinMob/Symbian, then years later iPhone, then almost immediately after Android (with WinMob and Symbian quietly dying)

Personal media players: Creative Nomad et al, then iPod, then lots of stuff.

Ultrabooks: A variety of unusable hideously expensive compact PC laptops going back to the 90s and mostly abandoned by the late noughties, then MacBook Air, then every PC manufacturer makes a MacBook Air clone.

ARM computers: Surface RT (ridiculously slow, no software support), modern ARM Surface (expensive, slow, poor software support), M1 (good), ???

Not to say it's inevitable, but once Apple shows it can be done, history suggests that it will be done. Possibly complicated by the chicken and egg problem here; Microsoft needs Qualcomm to make good laptop chips to invest much in ARM Windows, while Qualcomm needs Microsoft to make good ARM windows to invest much in good laptop chips.

This is a really interesting point. I wonder if there is another route though. If we assume that Intel (for process reasons) are out of it, then could AMD produce an 'x86 version' of the M1 on TSMC 5nm using Zen3 and successors which would help other firms compete with Apple.

I guess the extent to which M1 is great because its ARM isn't clear - there are some advantages in instruction decoding, having bigLITTLE etc - but the impact of all these together hasn't been quantified. Plus there is the deeper integration with Apple's software.

> then could AMD produce an 'x86 version' of the M1 on TSMC 5nm using Zen3 and successors

I mean, they could, but the A12Z in the devkits, produced on TSMC's 7nm process, also showed very impressive results. It's not clear how much gain the M1 is getting from 5nm, but I'd be _extremely_ skeptical that it's the whole story, especially given limited improvements between the directly comparable A13 (7nm) and A14 (5nm, some microarch changes).

> Plus there is the deeper integration with Apple's software.

This is relevant to some OS niceties (eg the disconcertingly instant wakeup and resolution shifts) but should have no bearing on, say, SPECInt2017.

Agree that they couldn't match - but if they got within say 15% on performance and battery life would that be good enough?

I think the software integration goes a bit further than wake-up etc e.g. they seem to have speeded up handling of reference counting through architectural changes (not sure I could explain how though and even Apple's software engineers seem a bit in the dark!)

> but if they got within say 15% on performance and battery life would that be good enough?

I mean, depends what you mean by 'good enough'. It would, obviously, be good for AMD; I don't see that it would make it competitive in perf per watt terms, though.

> e.g. they seem to have speeded up handling of reference counting through architectural changes

They have, but that wouldn't be relevant to (most) synthetic benchmarks, I wouldn't have thought. SpecInt2017 won't be doing much if any reference counting, for instance :) The M1's advantage in ObjC and Swift should be expected to be even greater than its advantage in 'normal' (ie C) code, but beyond refcounting microbenchmarks I don't think there are many tools to demonstrate that (and the vast bulk of performance-sensitive code running on MacOS is C/C++ anyway).

>Ultrabooks: A variety of unusable hideously expensive compact PC laptops going back to the 90s and mostly abandoned by the late noughties, then MacBook Air

PowerBook 100 before the Air. 12" PowerBook G4, too.

We need an AMD+ARM combo. Intel fatality.

I ain't keeping up with hardware but from I can remember Surface Pro X that came out last year was received positively, packed enough power, worked fine but wasn't trending for some reason. Thus the real question is, though there are ARM laptops since mid-2010s why is Apple taking once again credit for making a "revolution"?

Surface Pro X is a great hardware device but it is definitely not a great software experience. Same problems with the old Surface RT with the ARM chip, backward compat with existing software made the whole experience difficult for almost all folks. As for SPX with Windows 10 on ARM, you can't use any 64bit app and existing x86 32-bit apps were incredibly slow and not reliable (look at photoshop for an example). 64-bit emulation support is coming next year[^1]. Many of their development tools are not native either, some folks are still waiting for MS to ship Visual Studio to ARM [^2], .NET 5 was just shipped a week ago with some basic ARM support but no WPF for ARM until next year.

The difference here is Apple has a very fast x86 translation layer called Rosetta 2 that can run almost all Intel apps at 80% or so of the native M1 performance. Most folks don't even notice the difference. Windows 10 on ARM (SPX) had to emulate it at all time, with massive performance regressions.

What Apple has done is a complete transition that includes both hardware/software to ARM at the same time. All of their apps including development tools (Xcode) were native, huge number of macOS devs already shipped native support on day 1 or within the first week and Rosetta 2 made waiting for native support okay.

And that's just the software. The hardware? Super fast, long battery life, fanless in MBA, very quiet fan in MB and more. For a lot of people, it's a complete experience that worked out for them.

Note, not all developers are going to be ready for this, there's no docker support, some tooling will take awhile to update but they're going at much faster pace than MS at this point.

[1]: https://www.theverge.com/2020/9/30/21495510/microsoft-window... [2]: https://docs.microsoft.com/en-us/visualstudio/install/visual....

I take that it is a superior experience to what was available. My issue was with calling it a "revolution". From what you're saying the most important improvements have been the emulation and that it is better supported. That said, are there any benchmarks on performance hit during emulation for those and other devices? I searched but couldn't find something.

Here's one from Surface Pro X with Microsoft's SQ2 chip that they customized with Qualcomm: https://www.windowscentral.com/surface-pro-x-sq2-review#benc...

Switch to Geekbench 4 for x86 vs. ARM for emulation on the same device.

One of the definitions of revolution is:

> a dramatic and wide-reaching change in the way something works or is organized or in people's ideas about it.

Apple didn't change from a x86 CPU to a ARM CPU, they went from a third party Intel CPU to a complete Apple-designed/customized/built system on a chip (SoC) that includes the following: an Apple CPU (4 high-pref/4 high-effiency cores), Apple GPU that's using Apple's TBDR implementation, Neural engine (ML accelerators), ISP, media en/decoder, disk controller, Secure Enclave, Unified Memory Architecture with memory dies on the same package that allows all of the cores to talk to the same memory space without copying back and forth, and so much more. They did all of this in a familiar package that runs faster, quieter and last much longer without major impact on the customer's experience.

A complete package that was executed well on the first try and for some, this is considered to be a game-changer. Especially as it is changing many people's understanding of how a laptop should behave, specifically not having to worry about heat, battery life and having too low of memory as Apple's UMA actually makes swapping to disk a non-issue, there's no difference in the overall experience.

SPX isn't a complete package either, it uses Microsoft customizations on top of Qualcomm's Snapdragon 8cx CPU but solely Qualcomm's GPU, Adreno. I don't think there's anything else that MS has customized for SPX.

> From what you're saying the most important improvements have been the emulation and that it is better supported.

That's just one part. The overall experience matters. You can invent a technology that changes the world but if no one can use it well, it wouldn't matter much until someone take that technology and make it useful. Sadly, it's the latter that matters the most.

It's the same thing with iPhone being called one of the most important technological revolutions. iPhone was definitely not the first nor the second touch screen mobile device but it was a complete package that sold it, just like these Macs.

Just to clarify, I don't see it as a revolution at the moment but I do think my M1 MBA is a game-changer for me. The biggest change is that I haven't had to worry about battery and heat at all. My old rMBP wouldn't last me half a day (it heats up quickly within the hour) but this one can and have gone for two days before I had to charge it and even then, I only did it during work hours with the dock. I never had to use my charging cable outside of my desk with the dock. Plus, no fan either, god my old rMBP would kick up the fan all the time. (yes, I cleared out the fans every 6 months and had it cleaned by Apple as well).

Thanks for the well-written replies. From the benchmark mentioned SPX has score ratio x86/ARM (emulated/native) ~60%, thus ~80% claimed by Apple is indeed a big step.

Because Apple have made the first one that has real world performance gains over mid-range x86 processors on native apps, built an x86 emulation layer that delivers close-to-native performance AND delivered a huge uplift in battery life over even lower powered x86 devices.

I think you're trying as hard as possible to convince yourself Apple haven't done anything special here, but that just isn't true.

>real world performance gains over mid-range x86 processors on native apps

Does it? As I said, I ain't keeping up with hardware news.

>I think you're trying as hard as possible to convince yourself Apple haven't done anything special here, but that just isn't true.

I didn't say that. I specifically referred to calling it a revolution. If what you mentioned above is true then it will be one.

In some benchmarks the M1 is faster than any other laptop CPU you can buy (and even when it loses it’s usually close and runs at significantly lower wattage). In many single-threaded benchmarks (web/Javascript stuff for example) it’s faster than any x86 CPU you can buy.

I haven't been keeping up either. How does the M1 compare price wise to all the other laptop CPUs you can buy? Aren't the newest processors always the fastest?

M1 isn't for sale except as part of an SOC in Apple hardware, so there's no way to compare prices directly.

And it isn't just that the Apple hardware is some of the fastest available, it's that it does so with far lower power requirements. That's the really significant piece to all this.

The newest passively cooled super thin notebook ARM chips are not usually faster than Intels flagship chips, no.

But this is.

Three reasons:

- Quite slow relative to similarly priced Intel laptops.

- Quite expensive; started at $1500, so 1.5x the cost of the MacBook Air. And that $1500 didn't even get you a keyboard; if you wanted one of those it was an extra $150 or something.

- Very poor support for x86 software by Microsoft.

There was little convincing reason to buy one over an Intel equivalent. Realistic starting price was $1650 for something that was slower than Intel laptops for ARM software, was essentially unusable for x86 software, didn't have significant battery life advantages, and wasn't in a normal laptop form factor.

> Thus the real question is, though there are ARM laptops since mid-2010s why is Apple taking once again credit for making a "revolution"?

The usual. Apple took a thing that people had previously done, badly, and did it well. That is more or less Apple's thing; they rarely invent truly new markets, but they do take existing but irrelevant markets and make them relevant.

Reply on news.ycombinator.com