Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Apple Expected to Announce Arm-Based Mac Plans at WWDC, Transition Away From Intel to Begin in 2021
- Thread starter MacRumors
- Start date
- Sort by reaction score
Intel brought in developers. Some of us will have to leave. Can't run docker images and code emulated when it is being deployed to Xeon's. I know you can do lots of general tasks on any processor, but I am not going to be able force all of our applications over to ARM just so that I can have my Mac.
Dramatic much?
If they are good enough for servers now I think Apple can find a way to get good performance in a desktop.
Having a unified processor ecosystem has huge benefits over the possible risk they might see a performance hit in the first few releases of hardware.
I'm still an avid Apple fan, but now dual boot into Linux, and may do so full time in the future.
When Apple switched to Intel, the competitive landscape was much different than it is now. I was much less suspicious of an Apple walled garden during that Intel switch than I am now. Not to mention forced use of a poorly designed touchbar and $200 headphones that lasted a year. Let's just say I'm still an Apple fan but have a much greater healthy skepticism about remaining so.
You can also argue that they wouldn't be going through this trouble if there weren't significant demand for it. They could easily deprecate Boot Camp while still on x86.
Quite. But that was not my point.
The cmaier person was claiming that ARM "products", as he put it, were outselling AMD64 CPUs (aka x86-64) at Apple by an order of magnitude, therefore nobody has to worry about ARM on Macs flopping. Strange logic, but I didn't come up with it.
But my point was that virtually all of those ARM products, excepting the iPad pro, are designed for consumption and facilitation of consumption, not creation. That you can make an ARM device to create, is neither here nor there. Of course that's possible and has been done many times, including the very first ARM based computer.
He was simply comparing apples to oranges. Consumption and lifestyle devices outsell more expensive professional devices, and if you include the iPad pro in that statement, this still holds true.
[automerge]1591743724[/automerge]
Dramatic much?
If they are good enough for servers now I think Apple can find a way to get good performance in a desktop.
Having a unified processor ecosystem has huge benefits over the possible risk they might see a performance hit in the first few releases of hardware.
I quoted a poster who wrote "ARM products", he did not write ARM computers.
This was specifically referring to Apple products, not anyone else's ARM products.
If you're trying to point out that ARM based computers exist, you're barking up the wrong tree. I'm well aware of ARM and it's history, having used an Acorn Archemedes A3000 back in the day. ARM was originally a desktop PC CPU.
Last edited:
I don't think the amount of people running bootcamp is that small. There are tons of people who use macOS for their work machine, and then run Windows for being able to play old games or the likes. Then on top of that you have all of the people running windows in parallels to be able to test web browsers while creating web sites. Also, maybe this is an unpopular opinion, but I could not care less about running iOS apps on a Mac. Most iOS apps that I have installed on my iPad are simply due to the website being gimped on iPads. I cannot think of a single iOS app that I would want to install on my Mac as opposed to using the corresponding existing Mac app or website.
[automerge]1591744132[/automerge]
There will always be something new. At some point there will be 128-bit processors, and that alone will be a new architecture jump. The only reason arm is so popular now is due to them eclipsing intel in efficiency. Someone will invent a more efficient architecture, and it will eclipse arm the exact same way without a doubt. That is the historical trend for nearly 80 years, and until proven wrong, it will be inevitable.
Last edited:
I don’t think ANYONE should be taking Intel’s launch date targets seriously for awhile.
Because Apple can modify the processor to add Apple specific private instructions (their license allows it as long as it adheres to the entirety of the ARM instruction set otherwise), the processor can literally become ANYTHING Apple wants it to be. So, yeah, for Apple, ARM could be the logical endpoint.
For example, if they determine that they need a specific set of instructions at the CPU level for better performance, they can add those and have it align to some future OS release. OR, they can determine that they will deprecate a specific CPU feature in 5 years and replace it with a better implementation and have the hardware, the compiler, and Xcode all aligned to make the switch seamlessly on their schedule.
Amazon Graviton 2 is a server CPU that have 64 ARM cores on it.
You are the fanboyism yourself when talking about Intel. Nobody talk like that for Intel 20 years ago when DEC and Power was still in the PC market.
There's nothing magic about x86 and ARM can do everything x86 can -- since ARM could emulate x86. If emulated performance is times faster than native then who care about native or not?
ARM is not bind to low power device. Graviton 2 is already faster than the fastest Xeon in server market.
Intel is slowing down a lot these years and for past 10 years they only increased IPC by 15%.
Today a A13 have 80% more IPC than a Skylake, that's a shame for Intel.
AMD is doing well for the past 3 years, but not as good as Apple's ARM was for the same time.
And since Apple definitely knows how Intel and AMD's next few years secret products, they are making this decision with much more knowledge than us.
It means you might be an idiot for buying a Macbook between now and WWDC. I'd wait to see what the transition path will look like.
I never said they didn't.
I was stating that when you start adding 64 lanes of PCIe (which you need to do in a desktop processor) things heat up very quickly.
So my statement wasn't about what they had.
[automerge]1591745192[/automerge]
Dramatic much?
If they are good enough for servers now I think Apple can find a way to get good performance in a desktop.
Having a unified processor ecosystem has huge benefits over the possible risk they might see a performance hit in the first few releases of hardware.
Why? You seem to miss that MacOS is processor agnostic.
It currently runs on ARM and Intel. The underpinnings are the same.
Mac never supported docker.
You was running a centos VM for docker already.
And since server will shift to ARM sooner than your laptop, I guess that's a plus for ARM laptop.
BTW, we are already running different CPUs from the server. Xeon-EP support AVX512 which does not exist on 2019 MacBook.
You already have to deal with software difference for that. If that kind of performance is not your consideration then docker file support multi-arch and you can test in a ARM docker and deploy on Intel servers if all your dependency support that.
They can modify implementation.I don’t think ANYONE should be taking Intel’s launch date targets seriously for awhile.
Because Apple can modify the processor to add Apple specific private instructions (their license allows it as long as it adheres to the entirety of the ARM instruction set otherwise), the processor can literally become ANYTHING Apple wants it to be. So, yeah, for Apple, ARM could be the logical endpoint.
For example, if they determine that they need a specific set of instructions at the CPU level for better performance, they can add those and have it align to some future OS release. OR, they can determine that they will deprecate a specific CPU feature in 5 years and replace it with a better implementation and have the hardware, the compiler, and Xcode all aligned to make the switch seamlessly on their schedule.
They cannot add instructions, only Cortex-M embedded allows custom instructions.
You say something that is contradictory.
"Because Apple can modify the processor to add Apple specific private instructions (their license allows it as long as it adheres to the entirety of the ARM instruction set otherwise)"
If it is a custom instruction it doesn't comply to the ARM instruction set.
They can chose to implement their hardware any way they want; number of pipelines, speculative execution, number of cores in a cluster, number and size of caches, but they cannot ADD Apple only instructions to the instruction set.
Currently 64 lanes is only provided by AMD on EPYC/HEDT platforms.
Desktop CPU only have 20-24 lanes.
I do not think a 20-24 lanes implementation will heat up that much -- especially when power saving feature will keep most device running at 1x until needed.
[automerge]1591745743[/automerge]
At least we will not have a MacBook Pro 2016 with 16GB RAM maxed out.
Apple already support LPDDR4 ram in iPhone 6s back in 2015.
That was the "PowerBook G5" moment for Intel.
I may be wrong, but I think Apple could add instructions since they design the core from the ground up because they do not simply pick an off-the-shelf core (Cortex A or M or R, etc.) Also, it would probably depend on their license terms, but they may be able to add instructions provided that it stays compliant with the rest of the licensed instructions.
Thanks for the clarification -- I was indeed referring to software emulation and apologize for any confusion. That said, whether the performance hit for emulation is an issue will depend on the user's particular purpose. It's possible that Apple's processors will be faster than what we have now, which would make emulation a more compelling option for some who need to run alternative operating systems.
Interesting. All Apple computer designs with CPUs from any company but Intel were commercial failures. We'll see if they can change this pattern.
Apple wellmise say bye to a portion of their Mac profits and say bye to Macs being truly Macs. With ARM they are less Macs more glorified iPads.
I was there with my G5 in the PowerPC days and the only software was from OS X developers and it was rather thin. While Wintel enjoyed amazing apps, Mac OS X didn't have much unless it came from Apple, Microsoft or dedicated Mac developers. The transition to Intel brought so much awesome software. I really hate that they're not continuing with intel. I loved the ability to run pretty much any software and it was easy to codevelop for Windows and Mac, which brought some very obscure software to the Mac platform for the first time.
I'm also worried about how this transition will go now vs in 2005. My brand new iPhone 11 has messed up coverart due to a sync issue they haven't fixed since the rollout of IOS 13 and the IOS team has the most developers on it compared to other teams at Apple. Now they're developing Mac OS X for two codebases and they're spreading the already thin OS X team thinner still. Catalina was buggy and eliminated so much good software due to 64 bit limitation. OS X hasn't had the same stability it had in Snow Leopard. I still have to use a third party app or command line to not let the computer shut down or sleep when I'm copying lots of files, which is outrageous as disk activity and network activity can be easily sensed and the computer should prevent sleep (this all broke with Mountain Lion and hasn't been fixed despite multiple bug reports sent to Apple).
I've not been happy with the direction Apple has been going of late, but I'm so deep in the Apple ecosystem, I'm not sure where to go next. I haven't touched a linux distro in years and Windows 10 still isn't the home run it could be. This really puts users like me at a pivot point looking forward, which absolutely sucks and is painful.
Actually Xeons have 40-48 lanes.
Skylake desktop are 28/44.
You don't run at 1x until demand dictates.
PCIe negotiates link width and speed as the link comes up.
You need to do emphasis, pre-emphasis and equalization.
To do what you say, the link would need to go down and renegotiate speed and width.
That's not what devices do. I worked on and designed PCIe Gen3 switches.
Moden CPUs burn a significant amount of power driving I/O.
That is one reason why LPDDR is popular. Standard DDR consumes more power.
[automerge]1591748096[/automerge]
Once again.
If you add an instruction; the processor will no longer be compliant.
Arm only allows custom instruction in the Cortex-M embedded processor.
Just because they design the processor from the ground up means nothing about instruction set.
It allows them to chose implementation.
If you want to call a custom processor ARM (A12); it must pass the compatibility suite for the class of processor A7x, etc.
You can argue if you like, but I design chips for a living and some of them in the ARM ecosystem.
Not really what I’m saying. When I buy a Mac Pro which I did, personally I want to still be using it in a few years. But if arm comes I won’t want it anymore. And that’s a bummer!