Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
No doubt Apple will try and forge an all-ARM future but I wouldn't be surprised if they are prototyping an x86 coprocessor (which could still be considered Apple Silicon if they do it as an AMD semi-custom) just in case.

How would an Apple x86 be any “safer” than an Apple Arm? What expertise does Apple have designing x86’s? Where do they get the logic verification traces? Where do they get a license to actually make x86 chips? And why would Intel and AMD license them to do so?
 
I don't think many people read the article correctly. It said things like VMWare won't run on Rosetta. It's didn't say that VMWare won't be able to rewrite/recompile to natively run on ARM.
[automerge]1593042030[/automerge]

I laughed at "cheaper Macs"
No it will not. USB 4.0 does not have direct PCI-E bus support which is critical with TB3 and low latency hardware.
USB 4 has a direct PCI-E bus support, but it’s optional in USB 4. It’s mandatory in TB 4.
 
  • Like
Reactions: Altherekho
i think we will obviously see ARM versions of parallels and fusion, for virtualisation of OSes compiled for arm. So Linux arm and possibly windows on arm.

As above, I would be surprised if an arm-based intel co-processor hadn't at least been considered even if rejected (indeed, I think apple has mixed instruction sets in iphones previously, no? arm iOS and x86 baseband?). Or maybe more likely some sort of hardware translation acceleration, maybe an FPGA of sorts.

While writing an emulator for Intel shouldn't be too difficult, it could be complicated by x86_64 and also for a high-quality product it will take some work to optimise it for the Apple Silicon and whatever happens with the GPUs.

Also, although the A12Z was in the development kits, presumably the A silicon in the actual units will be even faster..... at least octa-core 3GHz+, or maybe 12/16 slower cores, in which case that should be pretty fast no?
 
Last edited:
  • Like
Reactions: Detnator
Software support on arm macs will be the same. Hate to break it to you. I can barely justify buying a Mac now. In the future there’s simply no way. Apple is aiming at the folks who think iPad apps are amazing or something. Think fisher price type apps. We saw this reimagining already with Big Sur. Dumbed down for the typical ignorant mainstream user.

This is so entirely wrong. The Apple Silicon Macs will have access to all of the applications you have today – likely with 95% of these running as native ARM on the day that the first new Macs can be purchased.
[automerge]1593075646[/automerge]
The demo shows Parallels? Are you sure? I thought it was the virtualization app built into macOS.

Yes, it was Parallels, but it looks like Parallels was leveraging Apple's Hypervisor framework (built into macOS) – I presume to remove the requirement for 3rd party kernel extensions.
 
One thing to keep in mind. Macs are now a small fraction of the Apple's bottom line. They don't care if they lose 20% of Mac customers because of this transition.
You will find Apple's goal is to gain a lot more than 20% new Mac customers.
[automerge]1593076025[/automerge]
The Corporate world is Wintel apps. Macs have made inroads into that world over the resistance of many IT departments by having the ability to run Wintel apps. Apple's ARM decision has basically discounted those inroads.
The corporate world is running Wintel apps using Citrix. Which runs better on a Mac than on a PC. Which will run even better on an ARM Mac.
[automerge]1593076296[/automerge]
Also, although the A12Z was in the development kits, presumably the A silicon in the actual units will be even faster..... at least octa-core 3GHz+, or maybe 12/16 slower cores, in which case that should be pretty fast no?
I think just like today, there will be different chips. If you ignore MacPro and iMac Pro for the moment, you can buy Macs today with 2, 4, 6 and 8 cores. I'd expect three fast cores like the iPad to replace all dual core and some quad core Macs, six fast cores replacing everything up to 8 cores, and eight fast cores to exceed the fastest non-pro specs today. (Plus four tiny cores on every chip; having more than four doesn't make much sense).

That tweet also says that the VMs will be ARM only. I have no doubt that they will make an ARM version of VMWare Fusion, but it won't be able to run x86 VMs.
Imagine I built a VM for Linux ARM. And then I ran MacOS ARM in that VM. That MacOS ARM would have Rosetta built in, and therefore would be able to run all MacOS Intel applications. On a non-Apple ARM computer running Linux. Now instead of this build a Linux ARM computer with Rosetta built in. That could just run Linux Intel code. And you can take this and put it into an ARM VM running on MacOS ARM. And it can run Linux Intel apps on an ARM Mac.
 
Last edited:
  • Like
Reactions: firewood
Beginning of end for Apple in the desktop/laptop space. No one will want to develop for a dead platform and be subject to App Store rules like on iOS. Although Apple is getting attacked and rightly so on their antitrust behavior.

Good thing they’re still doing intel macs because Apple silicon won’t go anywhere.

Once again the lack of any must have services or software doesn’t bode well for Apple. This is why being able to accommodate windows and other third party services should be vital to them. The macOS / hardware integration is great but ends there because off the bat Apple has very little of their own to offer that most users find compelling or essential in macOS.

Sorry mate but that is just about the biggest pile of nonsense I have heard in a while. Macs will go from strength to strength when not constrained by Intel. The vast (95%) majority won't care if the chips aren't intel ..... why should they?
 
  • Like
Reactions: Detnator
i think we will obviously see ARM versions of parallels and fusion, for virtualisation of OSes compiled for arm. So Linux arm and possibly windows on arm.

As above, I would be surprised if an arm-based intel co-processor hadn't at least been considered even if rejected (indeed, I think apple has mixed instruction sets in iphones previously, no? arm iOS and x86 baseband?). Or maybe more likely some sort of hardware translation acceleration, maybe an FPGA of sorts.

While writing an emulator for Intel shouldn't be too difficult, it could be complicated by x86_64 and also for a high-quality product it will take some work to optimise it for the Apple Silicon and whatever happens with the GPUs.

Also, although the A12Z was in the development kits, presumably the A silicon in the actual units will be even faster..... at least octa-core 3GHz+, or maybe 12/16 slower cores, in which case that should be pretty fast no?

The x86 core in the baseband was provided by intel, so not exactly helpful experience for apple. I’m sure they never even considered any sort of hardware x86 in these things.
 
So , am i understanding well or not , can you help me ?

Separately, Federighi acknowledged that users were concerned about the future of running Windows on Macs that are based on Apple Silicon. "We're not direct-booting an alternate operating system," he said, confirming that Boot Camp will be run on the new Macs. However, he implied that Windows will be able to run under virtualization. "The fact that we mentioned virtualization in the keynote was partly a nod to people's interest in the topic."

 
So , am i understanding well or not , can you help me ?

Separately, Federighi acknowledged that users were concerned about the future of running Windows on Macs that are based on Apple Silicon. "We're not direct-booting an alternate operating system," he said, confirming that Boot Camp will be run on the new Macs. However, he implied that Windows will be able to run under virtualization. "The fact that we mentioned virtualization in the keynote was partly a nod to people's interest in the topic."


Statement doesn’t mean much. You can run the arm version of windows or Linux or whatnot under virtualization. That’s a “nod” toward users who want to run windows, but nothing more than a nod.

Unless parallels or VMware or someone builds an x86 emulator, that won’t do those people any good.
 
  • Like
Reactions: arkitect
Ironically, my biggest use for VMWare on Mac is to virtualize older versions of MacOS to run 32-bit apps that have been abandoned by the developer... I literally run my pre-Catalina Macbook Pro install in VMWare Fusion... Occasionally I've virtualized Windows 2000 to interface with "legacy devices" like a Commodore 64 disk drive to do data restores :)

If anyone's interested, the Mac app is Neat Desktop - they moved to some crooked rent seeking cloud model that I refuse to buy into... Sure, you can hack VMWare Player to run MacOS VMs on Windows machines, but that just seems like cheating...

I also worry about losing eGPU support for accelerating media work and computational work ... Strange that they didn't mention eGPU at all, given that Apple themselves sell those BlackMagic eGPUs and have been pushing Thunderbolt so hard for years now...
 
So , am i understanding well or not , can you help me ?

Separately, Federighi acknowledged that users were concerned about the future of running Windows on Macs that are based on Apple Silicon. "We're not direct-booting an alternate operating system," he said, confirming that Boot Camp will be run on the new Macs. However, he implied that Windows will be able to run under virtualization. "The fact that we mentioned virtualization in the keynote was partly a nod to people's interest in the topic."

Right, but ARM Windows like this [edit not an ARM]:
 
Last edited:
What if a VM can still run, with the only restriction that it will be 'emulated'? If Apple Silicon is as fast as Apple claims, then the end user may not even notice a big difference. I'd wait to see how this whole migration turns out. It might not be as bad as some of us now expect it to become.

Oh, I'm fully open minded. I do think this is the right approach for Apple and I'm pretty excited about what it might mean. Whether it is possible to run windows x86 emulated or not is unknown outside of apple at this point. I guess that Parallels / VM Ware will be working on this as a priority. IF they do get it working, it seems there will inevitably be a performance hit, but if the apple silicon is sufficiently powerful then perhaps it doesn't matter.

The app we need windows for is not very taxing performance wise. What will be more critical is whether it is stable and reliable.
 
  • Like
Reactions: Detnator
Imagine I built a VM for Linux ARM. And then I ran MacOS ARM in that VM. That MacOS ARM would have Rosetta built in, and therefore would be able to run all MacOS Intel applications. On a non-Apple ARM computer running Linux. Now instead of this build a Linux ARM computer with Rosetta built in. That could just run Linux Intel code. And you can take this and put it into an ARM VM running on MacOS ARM. And it can run Linux Intel apps on an ARM Mac.
That is a very long-winded way of saying that it is possible to build an x86 emulation layer for ARM. ;) Sure it's possible. But the performance will likely suck, and it will not support all x86 instructions (e.g. Rosetta does not support AVX vector instructions), and it will very likely have a number of incompatibilities. Rosetta will also likely not just emulate the CPU, but also translate certain system calls to the native ARM MacOS to get at least reasonable performance, which would have to be redone for Linux. Now put yourself into the shoes of a developer targeting Intel platforms, for whom time is money. Would they really want to deal with these issues and add additional testing to make sure the code runs on the actual target platform instead of just switching to an Intel-based machine for development? Or go through the effort of building multi-architecture containers for the miniscule number of Mac users that use containers?

Realistically this move will help nobody but Apple. There may or may not be some minor performance improvements, but first and foremost it's a way for Apple to save money (both in terms or hardware and of software development costs) and consolidate even more of the value chain in the company.
 
Last edited:
Realistically this move will help nobody but Apple. There may or may not be some minor performance improvements, but first and foremost it's a way for Apple to save money (both in terms or hardware and of software development costs) and consolidate even more of the value chain in the company.
Strongly agree - what I really don't undersand is why people seem to think that a stagnant Intel CPU roadmap only affects Apple. After all, it's not like anyone else is coughing up faster Intel laptops... At least the products are mature and relatively stable - Apple CPUs will likely continue thier annual release cycle, resulting in people just having to buy new laptops every 2~3 years. Which I guess was the point. :(
 
  • Like
Reactions: Altherekho
Now put yourself into the shoes of a developer targeting Intel platforms, for whom time is money. Would they really want to deal with these issues and add additional testing to make sure the code runs on the actual target platform instead of just switching to an Intel-based machine for development?

Your premise is just wrong. Developer do not target a CPU they target a platform like Windows, MacOS, Linux, iOS, Android etc. So i wonder what these mysterious "Intel" developer really target, do you mean Windows?
If they target Android or iOS, they better be on an ARM development platform. If they target MacOS, they better be on an ARM platform in the future as well. Only if you target Windows or Linux, then an x86 based developent platform might be ok - but ARM would not hurt either.
Besides by far the most developer target iOS and Android, so they clearly benefit.
 
  • Like
Reactions: dhershberger
Your premise is just wrong. Developer do not target a CPU they target a platform like Windows, MacOS, Linux, iOS, Android etc. So i wonder what these mysterious "Intel" developer really target, do you mean Windows?
Is it? Personally I develop NFV software for a Linux-based cloud environment that runs on x86 CPUs (like 95%+ of cloud servers out there). Data plane performance optimization using things like AVX is a big part of it. Macs today are very popular among developers in my field in part because it's easy to run x86 VMs and use tools like Vagrant.
 
Last edited:
  • Like
Reactions: Nugget
I switched from parellels constantly milking me for upgrades and use teamviewer to connect to a desktop. Also save space on my MacBook Air!
 
Realistically this move will help nobody but Apple. There may or may not be some minor performance improvements, but first and foremost it's a way for Apple to save money

Performance improvements will not be minor, but major (single core performance of Intel chips saw really little improvements in the last 3 years, I expect a 100% faster single core performance with arm chips). Of course a lot of us for various reasons are still going to need the possibility to run an X86 environment, and we'll need a solution.
 
  • Like
Reactions: Detnator
Performance improvements will not be minor, but major (single core performance of Intel chips saw really little improvements in the last 3 years, I expect a 100% faster single core performance with arm chips). Of course a lot of us for various reasons are still going to need the possibility to run an X86 environment, and we'll need a solution.

I don't think we can draw clear inferences on performance differences until we have hardware in front of us to run real world tests on. For instance, Intel has added quite a lot of optimized instruction sets over the years that enhance performance quite a lot. Will those be just a performance via Rosetta 2? Running an optimized workload on Intel and on ARM which will perform better? How well does ARM handle the desktop application structure and flow (meaning there is far more down on fewer cores than on many cores).

So many unknowns, but I'm expecting modest performance gains, if any, except for very specific workloads.
 
Performance improvements will not be minor, but major (single core performance of Intel chips saw really little improvements in the last 3 years, I expect a 100% faster single core performance with arm chips).
I doubt that very much. With its current generation ARM CPU, Apple has closed the gap, but is still slower (sometimes dramatically) than Intel in most single-core tests (see e.g. here). They will be able to benefit from very tight software/hardware customization for specific tasks, and maybe keep an advantage in the performance/watt ratio thanks to TSMC's better process, but that is less important for computers than it is for phones and tablets, and may be short-lived anyway as Intel's "Tiger Lake" with their new process comes out later this year (which supposedly also has a much improved iGPU).

In any case, leaving artificial benchmarks aside, I very much doubt that the average user will perceive a performance difference due to the CPU in practice. CPU performance has simply not been a problem for most types of consumer applications for several years now (except games and a handful of niche things).
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.