Are ARM Macs Good?

[thekev]

Yes and no.

If for example the application uses frameworks (like uh... every macOS and iOS app) it can call native versions of the frameworks included in the OS.

So... sure... if you are writing AVX code directly you might see an issue. But if you are using say, Metal or OpenCL, or any one of the other provided macOS frameworks for doing the heavy lifting, as recommended, then your application simply links to the provided framework and uses whatever the best hardware the end device has to do the job, the best way it can.

Didn't Apple deprecate OpenCL? I can't find the mention in their documentation at the moment. You're right that it shouldn't be an issue if they're linking to macOS frameworks for the costly portions. It still doesn't make emulation fast in any sense, as you are still effectively running this portion via an interpreter.

I didn't say anything about writing in assembly, apart from an earlier comment that referenced a low level library.

I'm going to rephrase slightly.

If you're trying to achieve ABI compatibility with some older application code, which requires more than trivial changes to be recompiled for ARM, the parts that were compiled for an older architecture are likely to be much slower, both due to having an interpreter layer in between and due to other details such as scheduling differences between the two architectures.

Things that would require more than a trivial recompile are stuff like removal of deprecated components and other various api changes.

As you mention though, parts that link to .dylib files aren't really impacted, because those are natively compiled from day one.

 

[ssgbryan]

Posted this before in another thread but I might just do it again here.

With what we know about the first MacBook ARM chip (based on A14 architecture, 8 performance cores + 4 efficiency cores) this mythical MacBook can have the CPU performance of a 7K+ 12 core Mac Pro.

Without any optimization a Surface Pro X with Qualcomm SQ1 runs x86 version GeekBench4 at roughly 60% speed of ARM GB4 (2200 vs. 3500 single core, and 6750 vs. 11500 multi core). After all it’s a heavily multithreaded and SIMD dependent benchmark running on Win32 emulation layer…

And the single core performance of this alleged “phone chip” times 60%, is pretty much the base line 8 core Mac Pro…
(That is, without considering sustained load)

Without porting AVX or SSE extension codepaths from x64 one can just *brute force* everything, aka literally emulating a x64 chip much faster than a real i5
[automerge]1588555056[/automerge]


In case you forgot a thing or two about Newton, Apple deliberately abandoned the use of ACORN in favor of Arm after some unsuccessful trials.

Sustained loads are why people like me owned Mac Pros (and Power Macs before). We won't be moving to ARM, we are/will be moving to Windows or Linux.

An ARM based Macbook will be version 1.0 hardware running version 1.0 software - at best. Good luck with that.

Early adopters will either be getting by on either iOS apps (Go Candy Crush!) or some sort of emulation for a minimum of 18 months. Assuming of course, that the software house decides to actually port the software to ARM. That isn't a given, especially when one considers the small size of the Mac marketshare and the fact that a Mac Book isn't actually designed to do heavy lifting.

 

[throAU]

Didn't Apple deprecate OpenCL?

I think it's still present? Deprecated doesn't necessarily mean removed. Just means not recommended to use.

 

[thekev]

I think it's still present? Deprecated doesn't necessarily mean removed. Just means not recommended to use.

I know that. It's usually reserved though for things that will eventually be removed though. The articles mentioning this seem to be from around 2018. If a hypothetical ARM mac debuts next year, I could see Apple encouraging people to fully transition to Metal, seeing as they appear to be sticking with it. It would be nice to see good simd interfaces at a higher level.

 

[konqerror]

Early adopters will either be getting by on either iOS apps (Go Candy Crush!) or some sort of emulation for a minimum of 18 months. Assuming of course, that the software house decides to actually port the software to ARM.

Nope. You're ignoring webapps, which is 90% of what consumers and education runs. See also quasi-webapps, namely Electron-based apps. Port Microsoft Office and you've covered probably 3/4 of the business market.

 

[jinnyman]

Well I wouldnt want to waste my time speculating on a non existing platform. That said, I do hope Whatever direction it’s going would be beneficial to customers.

based on this year’s apple development, I do have doubt on Apple’s willingness and ability to significantly improve Performance of chips just like we are seeing in x86 market.

 

[Roman78]

Well.... Microsoft did the step and released an ARM based Surface. Installed is Windows 10 Pro 1909. Problem at this point is -laughter- it only runs 32-Bit x86 applications, no 64-Bit. Problem is the ARM only emulates 32-Bit x86 and this much slower as native Intel x86 CPU's.

Pro is the much lower power consumption of the CPU. And maybe, when there is ARM-based software on the market, the ARM could be a good alternative. But at this point of time... nope...

 

[DrRadon]

Well, lets be real here.... if this goes back to PowerPC standards of Mac not being able to run Windows they fn suck.
I hope the new iMac will just be a specs bump with the new intel line of chips, new GPU... and all this experimental "god knows what"BS is not going to be a hassle for me till my 2015 MacBook Pro needs replacement. (wich hopefully will not happen before dongle mania has been dealt with)

 

[throAU]

Well.... Microsoft did the step and released an ARM based Surface. Installed is Windows 10 Pro 1909. Problem at this point is -laughter- it only runs 32-Bit x86 applications, no 64-Bit. Problem is the ARM only emulates 32-Bit x86 and this much slower as native Intel x86 CPU's.

Not all ARM is the same.

Apple's ARM processors are heavily customised and outperform everything else in the market.

And that's not just apple fanboyism - its fact. They've spent a HUGE amount of effort on them in the past 10 years. Microsoft has not, and are using ARM as a bargain basement cheap part, not as their primary platform.

Fortunately, apple's software architecture has been through multiple CPU changes before and has various features inherent to the platform to make this easier for apple than it would be for Microsoft.

 

[Moonjumper]

Not all ARM is the same.

Apple's ARM processors are heavily customised and outperform everything else in the market.

And that's not just apple fanboyism - its fact. They've spent a HUGE amount of effort on them in the past 10 years. Microsoft has not, and are using ARM as a bargain basement cheap part, not as their primary platform.

Fortunately, apple's software architecture has been through multiple CPU changes before and has various features inherent to the platform to make this easier for apple than it would be for Microsoft.

I agree. Microsoft failing on something is no indication that Apple will fail in the same area. Look at the iPad as an example. There were many expecting it to fail because Microsoft tablets had failed. Apple is very different in structure and approach.

 

[Maximara]

This is all going to come down to will MS Office apps run on this new setup? Mac was a toy computer and not taken seriously until MS decided to make nice with Apple. Keep this is mind. We have endured chip changes twice... WE don't need another CF in this regard.

The problem with that mindset is that was back then MS Office didn't have anything resembling a reasonable competitor but today they do from the Open source market in the form of LibreOffice. For 90% of what people do Libreoffice fits the bill and it opens a reasonable number of formats and you can't beat free.

 

[Moonjumper]

The problem with that mindset is that was back then MS Office didn't have anything resembling a reasonable competitor but today they do from the Open source market in the form of LibreOffice. For 90% of what people do Libreoffice fits the bill and it opens a reasonable number of formats and you can't beat free.

I run into some formatting issues when opening MS Office files in LibreOffice, or vice versa. LibreOffice is a great piece of software, but many will stay with the standard when operating in the business space. For now at least.

The default software choice is bring eroded in multiple areas. For example, Blender encroaching on Autodesk in the 3D modelling space, and Affinity software taking customers from Adobe (as good as the Affinity suite is, people are making the change because of the expensive Adobe subscriptions). What this means is that a new chip architecture will have to make sure a wider range of software takes hold. Although most have experience of Arm from their iOS apps. 3D modelling being the notable exception here.

 

[ApfelKuchen]

The problem with that mindset is that was back then MS Office didn't have anything resembling a reasonable competitor but today they do from the Open source market in the form of LibreOffice. For 90% of what people do Libreoffice fits the bill and it opens a reasonable number of formats and you can't beat free.

Regardless of whether LibreOffice is as good or better than MS Office, you will not see wholesale abandonment of MS Office. Most of the world is using MS Windows, despite the fact that Linux is free (or can be - corporate tends to buy paid distributions anyway - paid distributions come with tech support - not just for the end-users, but also for the IT departments). Big business sticks with the tried-and-true, even if it does cost money. Change also costs money - training an entire workforce to use a new platform/app, modifying business processes to match that new platform/app, etc.

If Apple introduces ARM-based Macs, MS is not going to abandon 10% of its users. Apple will be providing all the tools necessary for Office to run on ARM, just as they did when moving from Motorola to Power PC and from Power PC to Intel.

 

[Krevnik]

Apple will be providing all the tools necessary for Office to run on ARM, just as they did when moving from Motorola to Power PC and from Power PC to Intel.

This. Folks shouldn’t confuse ARM with mobile-focused UI and frameworks.

The Raspberry Pi is a smartphone-class CPU that runs desktop or server Linux. The iPad has a CPU that exchanges blows with ultraportable laptops despite having no fan, but runs a touch-first UI. The CPU architecture itself isn’t the issue.

As long as macOS on ARM uses AppKit, porting most stuff should be pretty straightforward.

 

[Homy]

Don't underestimate ARM because it's a big family of CPUs and not only about mobile chips in iPhone.

Good news for future Macs: The fastest supercomputer uses Arm chips

The same processor family that powers phones and likely the next generation of Apple computers is at the heart of the world's fastest machine.

www.cnet.com

 

[scubafish]

Looking through The various threads it’s quite clear that people think that Apple don’t have the more the powerful silicon ready just yet, but a base system running A12X is within reach more easily. I suspect Apple may simply run multiples of these processors in a localised GRID to achieve massive scaling of processing power (or another similar scaling idea). This way they simply need to optimize the software to recognise and use the multiple cores on offer and then we have a very capable system based around existing ARM chips. Who says each machine needs a dedicated massive processor?

 

[robbieduncan]

Looking through The various threads it’s quite clear that people think that Apple don’t have the more the powerful silicon ready just yet, but a base system running A12X is within reach more easily. I suspect Apple may simply run multiples of these processors in a localised GRID to achieve massive scaling of processing power (or another similar scaling idea). This way they simply need to optimize the software to recognise and use the multiple cores on offer and then we have a very capable system based around existing ARM chips. Who says each machine needs a dedicated massive processor?

Even assuming they do have the Mac targeting variants of the A14 ready why show their hand now? The A14 will debut in the iPhone later this year as expected. Then the A14*** variant will be unveiled for the Mac. Probably with more cores and higher clock speeds. I'd bet there will never be a shipping, production Mac with an A12 in it.

 

[Pacific1972]

I want to have the Apple Silicon Machine

 

[verticalines]

Looking through The various threads it’s quite clear that people think that Apple don’t have the more the powerful silicon ready just yet, but a base system running A12X is within reach more easily. I suspect Apple may simply run multiples of these processors in a localised GRID to achieve massive scaling of processing power (or another similar scaling idea). This way they simply need to optimize the software to recognise and use the multiple cores on offer and then we have a very capable system based around existing ARM chips. Who says each machine needs a dedicated massive processor?

We've never really seen Apple's own chips pushed with dedicated cooling either so hopefully we learn more what the Mac Mini setup can do in the weeks ahead. Designed for 65W.

It's all pretty much been passive. The iPhone has a 5W passive chip--if you open the cooling ceiling to even say 20W, what can it do alone? Do the core speed scale with cooling or does it just allow Apple to add a bunch more coprocessors and accelerators without much issue? The thing with ever smaller chips is cooling them efficiently due to the concentration of heat. And having 30 cores doesn't matter in serial operations when the fastest machine is waiting for the human user to think and click. Not to mention few programs really leverage that many cores and those that do can arguably be using accelerators too. Many ways to go about this especially for a company looking to vertically integrate it all from top to bottom.

I don't know why people aren't excited overall. Whatever happens, it's something different in the computing world for better or worse. Maybe I should just ignore the first-dibs reaction pundits.

 

[weckart]

I'd bet there will never be a shipping, production Mac with an A12 in it.

That's a given. Remember the Intel Developer kit with that consumer grade chip and board in a G5 tower? If developers can make a usable version of their software to run on the A12, it will fly on the A14x or whatever Apple actually produces to sell (assuming complete backward compatibility) much as the MacPro 1,1 showed the Developer Kit a clean pair of heels.

I also bet a lot of devs will do what they can to hang onto these A12 minis to sell as a rare retro piece in years to come.

 

[RegularGuy09]

iPhones need a lot less RAM compared to Android phones, because of how well the hardware and software is optimised together. With the move to ARM Macs, is there any possibility of seeing that for the Mac as well?

I'm not very informed on the topic, so any help would be appreciated.

 

[weckart]

iPhones need a lot less RAM compared to Android phones, because of how well the hardware and software is optimised together. With the move to ARM Macs, is there any possibility of seeing that for the Mac as well?

Not so much sure of need as much as of shipped with. Apple has always had a much more aggressive approach to battery life than Android. Early iPhones wouldn't multitask - if you minimised the window, you quit the application. Even after multitasking came in, Apple hibernated background apps to disk so that you often had to reload Safari tabs etc. Not true multitasking in my book. I still see a lot of reloading under iOS13, so it's Apple's parsimony. Smaller batteries and less RAM = $$$$$.

 

[throAU]

In case you forgot a thing or two about Newton, Apple deliberately abandoned the use of ACORN in favor of Arm after some unsuccessful trials.

the ARM processors are the "Acord Risc Machine" originally.

Modern Arm is descended from the acorn RISC processors in the archimedes.

 

[RegularGuy09]

Not so much sure of need as much as of shipped with. Apple has always had a much more aggressive approach to battery life than Android. Early iPhones wouldn't multitask - if you minimised the window, you quit the application. Even after multitasking came in, Apple hibernated background apps to disk so that you often had to reload Safari tabs etc. Not true multitasking in my book. I still see a lot of reloading under iOS13, so it's Apple's parsimony. Smaller batteries and less RAM = $$$$$.

I understand what you're saying but in general iPhones are much faster than many android phones with much beefier specs. While Android phones have 8-12GB RAM, for eg : iPhone SE has just 3GB RAM, and is zippy as hell. Wondering if the ARM Macs will be the same i.e not much impressive specs, but really good performance and optimisation.

 

[weckart]

Apple has been promising big so for once I am going to take its word on it. Previous attempts at putting full fat OSes on ARM have been less than stellar, including Windows RT and 10. They were certainly usable if nowhere near the speed demons of their beefier Intel brethren. Maybe Apple's inhouse designs have worked wonders rather than the oracle. If so, then they have got to here a lot sooner than I would have thought possible.

As for leaks, just watch Intel's share price. Apple isn't a big enough customer alone to make that much of a difference to Intel's future but if Apple can demonstrate that ARM can scale better than Intel, then there will be ramifications.