ARM: Why!?

[sbarton]

What it means in the short/mid term is insignificant to 10/15 years when the reality of Apple owning own all aspects of system design and integration become apparent. It sounds like a good thing, and certainly in some aspects it is, however i remain dubious having lived through the powerpc era of Apple.

Technology moves on and in a decade or two wonderful new io, graphics, bus, chipset and system technology yada yada will evolve in the mainstream PC ecosystem. Apple will face mounting pressure to integrate this technology or offer similar features to their users and i suspect the enormous burden of owning that solution end-to-end will begin to show cracks just as it did before. We suffered so many oddball graphics api's and limited GPU options during the PPC era and were always paying 2 and 3 times as much for system upgrades which ended up with sub-par performance. I remember paying $800 for a 3dfx card with buggy Apple Glide drivers on my PowerMac 9600 so i could play the handful of games available for OS9 at the time. Meanwhile PC users had isles of great graphics cards to choose from and of course much better games, could easily add expansion IO like USB, memory, storage, etc. Mac users today take for granted the immense compatibility of upgrades and peripherals and i fear fail to realize that most of that is due to the commonality of platforms.

When SJ came back Apple was on it's death bed having squandered their fortune chasing the PC industry and failing miserably. IMO (and i know this is controversial so again - In My Opinion) SJ brought Apple the OS it couldn't design on it's own and pivoted the company over to the Intel ecosystem to provide the foundational system engineering it couldn't manage on it's own so that the company could focus it's limited resources on making amazing desktop experience we all wanted - and it worked brilliantly and the rest is history. I understand that the computing landscape has changed significantly and that Apple does a stellar job at doing the core engineering for mobile devices today, and one could easily point to them as the defacto leader in this space. If the needs of personal/business/desktop computing continue to merge and coalesce then perhaps this is a viable way forward for the company. I don't know.

I could be wrong. Hope i am, but really seems like I've seen this movie before. I do know that as a user who's use-case for a desktop/laptop includes about 30/40% gaming that Macs haven't changed much if at all (maybe actually worse) in the past 30 years. I've always managed to make them work in some oddball way or another mainly because I'm a huge fan of design of the HW/OS like most of you.

Removing bootcamp really bothers me though and with the cost of Mac these days being what they are, truly concerned about my future relationship with them. Time will tell i guess.

 

[Zdigital2015]

A lot of people seem to (incorrectly) believe that it means they’ll be able to just magically run iOS/iPadOS apps.

as for making x86 chips, they’d need a licence from each of AMD and Intel I believe, to make 64bit x86 processors, and I doubt it’s worth the effort.

Users literally will be able to magically run iOS and iPadOS apps as long as the developer allows it. Did you miss that part of the keynote? Not being snarky, but they went over this explicitly in the demo of Bug Sur using three distinct iOS apps.

 

[Stephen.R]

Did you miss that part of the keynote?

Did you miss the date of that post? Yes, it turns out, Apple are doing the software legwork to make that work. The context of the comment was people suggesting/implying/assuming that adopting Arm CPUs was all that's needed to run iOS apps on a Mac.

 

[Zdigital2015]

Did you miss the date of that post? Yes, it turns out, Apple are doing the software legwork to make that work. The context of the comment was people suggesting/implying/assuming that adopting Arm CPUs was all that's needed to run iOS apps on a Mac.

S***! I did miss the date. I just got an alert that this was a new MR forum, so I assumed that any post made were from today onward, not from March! Sorry about that. My bad.

 

[Stephen.R]

S***! I did miss the date. I just got an alert that this was a new MR forum, so I assumed that any post made were from today onward, not from March! Sorry about that. My bad.

No worries.

I guess they moved a heap of existing threads after creating the new sub forum.

 

[MikhailT]

People are fixating on the "ARM" and CPU parts, that's not really the whole picture here.

Remove the "ARM" part, this is not a standard ARM CPU, you cannot run any ARM OS on this at all.

This is a custom, complete, system on a chip (SoC). Not only the CPU is fully customized by Apple, so is the GPU, neural engine, ISP, disk controller, caches, memory model, etc.

Apple can start to tackle the slowest parts of their ecosystem and build hardware accelerators into their SoC.

Exporting 4k videos too slow? Build hardware encoder/decoder into their SoC and now exports are some time faster than what either Intel CPU or AMD GPU can do on their own.

Previewing 5 4k videos is too slow? Build the hardware to do just that.

Have 8k@144hz monitor but no port that can support it because DP/HDMI will take a while to catch up? With a complete vertical integration, Apple can build it on their own.

Look at the success of the secure enclave on their A chips, 4k video recording on their iOS devices, 5k display in iMac (which has a custom timing controller built by Apple), etc.

 

[leman]

People are fixating on the "ARM" and CPU parts, that's not really the whole picture here.

Spot on! That’s why they say Apple Silicon, not just ARM. It’s the entire stack as you say.

 

[perplx]

I don't think anyone has talk about the cloud yet here, right now you can rent arm based server from AWS that are up to 40% better price performance than x86. They haven't taken over because developers don't have arm workstations and cross compiling is a lot of extra work when you can just rent an x86 server. Now apple is going to ship the first real arm workstations and every company with a huge cloud bill is going to buy arm macs for development because the saving are just too big.

 

[Novus John]

Apple has long since not been willing to play ball with other OEMs and technology companies. They dropped opengl and opencl, blocked nvidia and amd from offering vulkan and other drivers on the mac, haven't offered real support for external graphic cards despite tb3 being capable of that, pushed metal, slashed 32bit libs thus disabling a lot of games and legacy software, banned kexts, introduced a bunch of mostly hidden lockdown features in the name of "security", etc.

Their goal isn't to keep the mac a commodity hardware, but to make it a very expensive, proprietary and completely incompatible computing product just like the ones we had back in the 80s and 90s. This has nothing to do with Intel or AMD, but rather Apple's desire to make the mac an "exceptional" product and not in a good way.

 

[windowsblowsass]

I would just like to clarify that when I started this thread, ARM was still just a rumor. I feel like it makes my tone sound a bit different now that it is a fact.

I’m still not sure how I feel about this, but I’m definitely open to at least looking at benchmarks when the new models are actually released.

Go look at a13 benchmarks right now and see what it matches/beats intel laptop chips in a damn cell phone. Extrapolate from there

 

[Novus John]

Go look at a13 benchmarks right now and see what it matches/beats intel laptop chips in a damn cell phone. Extrapolate from there

This is a comment worthy of a mighty facepalm. If there were any rewards here I'd have given you "The Facepalm of The Year" award.

 

[windowsblowsass]

This is a comment worthy of a mighty facepalm. If there were any rewards here I'd have given you "The Facepalm of The Year" award.

Luv u 2 bb

 

[burgerrecords]

People are fixating on the "ARM" and CPU parts, that's not really the whole picture here.

Remove the "ARM" part, this is not a standard ARM CPU, you cannot run any ARM OS on this at all.

This is a custom, complete, system on a chip (SoC). Not only the CPU is fully customized by Apple, so is the GPU, neural engine, ISP, disk controller, caches, memory model, etc.

Apple can start to tackle the slowest parts of their ecosystem and build hardware accelerators into their SoC.

Exporting 4k videos too slow? Build hardware encoder/decoder into their SoC and now exports are some time faster than what either Intel CPU or AMD GPU can do on their own.

Previewing 5 4k videos is too slow? Build the hardware to do just that.

Have 8k@144hz monitor but no port that can support it because DP/HDMI will take a while? With a complete vertical integration, Apple can build it on their own.

Look at the success of the secure enclave on their A chips, 4k video recording on their iOS devices, 5k display in iMac (which has a custom timing controller built by Apple), etc.

This is definitely Apple being Apple - which is to innovate for a dedicated, loyal and not terribly cost conscious user base which then pushes other technology players to copy from apple - and then we find in the not too distant future it is both commoditized but still different {enough.?}

The challenge is that Apple needs to continue to keep Mac essential/relevant enough for certain key professions to keep real Mac alive - there are more good alternatives than ever. I hope this transition goes well.

 

[endlessike]

What it means in the short/mid term is insignificant to 10/15 years when the reality of Apple owning own all aspects of system design and integration become apparent. It sounds like a good thing, and certainly in some aspects it is, however i remain dubious having lived through the powerpc era of Apple.

Technology moves on and in a decade or two wonderful new io, graphics, bus, chipset and system technology yada yada will evolve in the mainstream PC ecosystem. Apple will face mounting pressure to integrate this technology or offer similar features to their users and i suspect the enormous burden of owning that solution end-to-end will begin to show cracks just as it did before. We suffered so many oddball graphics api's and limited GPU options during the PPC era and were always paying 2 and 3 times as much for system upgrades which ended up with sub-par performance. I remember paying $800 for a 3dfx card with buggy Apple Glide drivers on my PowerMac 9600 so i could play the handful of games available for OS9 at the time. Meanwhile PC users had isles of great graphics cards to choose from and of course much better games, could easily add expansion IO like USB, memory, storage, etc. Mac users today take for granted the immense compatibility of upgrades and peripherals and i fear fail to realize that most of that is due to the commonality of platforms.

When SJ came back Apple was on it's death bed having squandered their fortune chasing the PC industry and failing miserably. IMO (and i know this is controversial so again - In My Opinion) SJ brought Apple the OS it couldn't design on it's own and pivoted the company over to the Intel ecosystem to provide the foundational system engineering it couldn't manage on it's own so that the company could focus it's limited resources on making amazing desktop experience we all wanted - and it worked brilliantly and the rest is history. I understand that the computing landscape has changed significantly and that Apple does a stellar job at doing the core engineering for mobile devices today, and one could easily point to them as the defacto leader in this space. If the needs of personal/business/desktop computing continue to merge and coalesce then perhaps this is a viable way forward for the company. I don't know.

I could be wrong. Hope i am, but really seems like I've seen this movie before. I do know that as a user who's use-case for a desktop/laptop includes about 30/40% gaming that Macs haven't changed much if at all (maybe actually worse) in the past 30 years. I've always managed to make them work in some oddball way or another mainly because I'm a huge fan of design of the HW/OS like most of you.

Removing bootcamp really bothers me though and with the cost of Mac these days being what they are, truly concerned about my future relationship with them. Time will tell i guess.

I was very impressed with Shadow for streaming PC video games when I played around with it. If your internet connection is good enough this may be a solution for you.

 

[harvester32]

It seems like a lot of people are excited at the rumors of a possible upcoming ARM-based Mac. I have to say, I'm finding it hard to understand why this is a good thing. Won't it just mean that we will lose the ability to run Windows, and we will have to migrate our entire software library a fourth time? Plus a lot of existing hardware could needlessly be made obsolete.

The previous migrations made sense because Apple was leaving behind technologies that had reached their end of life (680X0, Mac OS Classic, and PowerPC). In my opinion, Intel processors still have excellent performance and a bright future. They also seem to be very popular with developers. I just don't understand why we would abandon them now.

Even if Apple were insistent on making its own CPUs, couldn't Apple just start making x86 chips?

Let's not forget the sheer power of the graphics in the A series chips. One of the fundamental flaws in the Mac hardware has been the introduction of myriad Macs with graphics that are 1-5 generations behind other solutions. The A series graphics have been benchmarked to be quite powerful and scale with the processor. That said, it would be nice if they'd build a Mac that had a socket so you could upgrade your processor, and graphics chip at the same time by going from an A15 to an A18 or something similar. But either way, we don't get that now with the integrated chips in the intel processors and let's face it, those are no speed demons!

 

[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

 

[Darth Tulhu]

Turns out Apple did not go Ryzen for a reason.

I can't wait to see what they do with their own chips in Macs.

This is the beginning of full OS convergence between Apple devices.

I know it's a time for great fear among several of you pros, but this looks different than the PowerPC era to me.

Apple makes it's money on non-Macs. Thus, I believe the pro Macs that they'll build in the future will be to primarily facilitate development for that market. Will these new Macs be even more niche like the Mac Pro is?

Interesting times.

 

[Stephen.R]

Apple makes it's money on non-Macs.

Revenue from Macs was about $25B for FY2019, (around 10%) roughly the same as the iPad.

yes, iPhones ($143B, 57%) and services ($46B, 17%) both generate more revenue. But Apple services in particular, and in some markets the iOS/iPadOS lines too, make the most sense/value (for the customer) if you buy into the “Apple ecosystem”.

People have been claiming the Mac is doomed because iPhones are popular since the iPhone became popular.

 

[leman]

Apple has long since not been willing to play ball with other OEMs and technology companies. They dropped opengl and opencl, blocked nvidia and amd from offering vulkan and other drivers on the mac, haven't offered real support for external graphic cards despite tb3 being capable of that, pushed metal, slashed 32bit libs thus disabling a lot of games and legacy software, banned kexts, introduced a bunch of mostly hidden lockdown features in the name of "security", etc.

This is rather one-sided perspective. Yes, in the end Mac is developing to be a separate platform with its own API. But then again... OpenGL is a based on a 20+ year old design and has been morally obsolete since at least a decade... OpenCL (API developed by Apple itself) has been mismanaged by Khronos and sabotaged by Nvidia who is pushing their own vendor-locked API. Vulkan was a great thing that Apple initially supported... which ended up becoming an extremely complicated, unergonomic framework. Apples support of eGPU is second to none (if you are using a supported GPU of course) and the depreciation of 32 is a long time coming. I mean, if you want to blame someone, blame Microsoft for continuing to sell an OS for an obsolete architecture. I don‘t see any problem with signed-only kexts, only makes sense for software that has access to my kernel, and as to other security features, I haven’t been bothered by them in practice. These arguments usually boil down to “it‘s my computer so it’s my right to put stuff in /bin”, which I consider to be incredibly silly.

So yeah, while I’d be happy about more cross-platform compatibility, I can understand why Apple is doing what they are doing and it also makes a lot of sense to me. I was initially very upset about them pushing Metal and dropping Vulkan (before I actually saw what mess Vulkan is), but in the meantime... Metal is probably the most ergonomic GPU API that I saw and it allows you to do some rather crazy things at a fraction of complexity of DX12 and Vulkan. Not to mention the full access to Apple GPUs internal caches etc. I can code an advanced render path using complex data structures that fully leverages async nature of the GPU in Metal and be done before I write the basic device initialization boilerplate with Vulkan (yes, I hyperbolische, but you get the point).

 

[maflynn]

Their goal isn't to keep the mac a commodity hardware, but to make it a very expensive, proprietary and completely incompatible computing product just like the ones we had back in the 80s and 90s. This has nothing to do with Intel or AMD, but rather Apple's desire to make the mac an "exceptional" product and not in a good way.

Yep and now you'll not longer be able to install an [arm] operating system of your choice - only macOS

Apple's ARM switch will be the end of Boot Camp

When, on John Gruber's WWDC Talk Show, Craig Federighi confirmed that Apple would not support Boot Camp on ARM Macs:

"We're not direct booting an alternate operating system. Purely virtualization is the route. These hypervisors can be very efficient, so the need to direct boot shouldn't really be the concern."

 

[DWomack]

More ARM...

Amazon Launches a Killer ARM Server Chip With the Graviton2 - ExtremeTech

For years, we've been hearing about the potential of ARM products in the conventional server market. Amazon finally built a chip that might live up to expectations.

www.extremetech.com

 

[Darth Tulhu]

Revenue from Macs was about $25B for FY2019, (around 10%) roughly the same as the iPad.

yes, iPhones ($143B, 57%) and services ($46B, 17%) both generate more revenue. But Apple services in particular, and in some markets the iOS/iPadOS lines too, make the most sense/value (for the customer) if you buy into the “Apple ecosystem”.

People have been claiming the Mac is doomed because iPhones are popular since the iPhone became popular.

I concede, I oversimplified a bit (OK, maybe a lot).

But I think Apple is (and has been since "back to the Mac") coming at the Mac from the iOS side, not the other way around. The Mac will (or already has) become the satellite around iOS. Thus it makes sense for Apple to position the Mac software to be as iOS-like as possible, because Apple is obviously positioning iOS as the future.

According to Steve in the iPhone introduction, iOS IS OS X. So no, the Mac is not, and never will be doomed.

From a certain point of view, of course, because for many after the announcement of the move to "ARM" and Big Sur, it actually is.

I think it is no coincidence Apple marketing has increasingly been asking "what's a computer?"

The Mac will run Apple apps and the big third-party ones effectively, I'm sure.

But those that need Windows, they need to get a PC, which they should already have one. I don't think Apple is interested in those customers (that need a PC) anymore.

 

[leman]

But those that need Windows, they need to get a PC, which they should already have one. I don't think Apple is interested in those customers (that need a PC) anymore.

Or they will run Windows in a VM. It all depends what kind of solutions will be out there and how they will perform.

 

[jerwin]

Intel insider claims it finally lost Apple because Skylake QA 'was abnormally bad'

"The quality assurance of Skylake was more than a problem," says Piednoël during a casual Xplane chat and stream session. "It was abnormally bad. We were getting way too much citing for little things inside Skylake. Basically our buddies at Apple became the number one filer of problems in the architecture. And that went really, really bad.

 

[stevenaaus]

The idea that Intel's CPU failings are the reason for the Arm switch is a little laughable.
For starters, what are these failings ? The article has no particulars , and reads like a gossip magazine. And of course they could switch to AMD's (Apple's GPU partner) awesome Ryzens with no effort.

The switch is obviously another deprecation of the Macbook line for monetary reasons. AMD64 is the performance king with perfectly acceptable power draws. But Apple already maintain Arm code for the cash cows of iPad and phones, so easier for them to spin some publicity BS and count the dollars with the CPU switch.