Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
I bought a brand new MBP just a few weeks ago, but I’m really pleased about this news. It’s going to be a while until the transition is fully complete, and I’ve got no interest in being a 1st Gen beta tester. But by the time I’ll be looking to replace my intel machine there should be a well sorted ARM machine to jump to.
 
No, they won't. The Macs are not selling in large enough quantities to justify a custom chip design. It will maybe have a a few more cores enabled, but nothing dramatic.
Actually it does make sense the cost of a Apple designed chip is around $50-60 but for an i9 Apple has to pay upward of $400 per chip that's savings. Next If apple produces the same chip for all its devices and just disable few cores in mobile devices for power saving then it has to only pay once for the lithographic stencils each of which cost a lot, also a minimum unit run is required to be cost effective. All those things would be mitigated with a single design, it makes a lot of sense.
 
at this point all we can infer is because the A12z has comparable GPU in compute performance to a Xbox one S which itself has similar performance to a PC with about a Radeon HD 5850 thats the level of performance we can expect give or take for optimizations.

atleast out of the Dev Box but who knows what they will have on a future Mac Pro equivalent in 2 years or whenever the do the highend.

That level of performance times three to match the same thermal/energy envelope of a MPB with a dGPU. An iPad Pro runs 29w vs the 96w of a 15" MBP.
 
The only worry I have with this is that you'll see the Mac return to the days of PPC where you had a smaller app pool

You now have access to 2 million ios/ipados apps.

Over time, a lot of those will scale up to support the full power of mac.
[automerge]1592861769[/automerge]
They’ve only been using the instruction set from ARM for several years now (that’s my understanding). They haven’t touched a reference design from ARM since I think the A7?

Note that behaving as BIG.little isn’t the same thing as using an actual ARM design, it’s an organizational/performance scheme.

Apple has been doing completely native chip design for several years now.

Again, that’s my current understanding.
That’s absolutely correct. The last time they used an Arm reference design was A4. Everything they do is fully custom now.
 
  • Like
Reactions: Nütztjanix
I bought a brand new MBP just a few weeks ago, but I’m really pleased about this news. It’s going to be a while until the transition is fully complete, and I’ve got no interest in being a 1st Gen beta tester. But by the time I’ll be looking to replace my intel machine there should be a well sorted ARM machine to jump to.
Yeah, I think you're in the optimal position

I'm holding on to the intel. When macOS stops supporting it, i'll move to ARM
 
Metal compute scores: A12Z 10K, 5600M 50K, Vega II 100K.

3DMark Ice Storm Unlimited off screen 1280 x 720 from NotebookCheck
----

5500M = 210925

Radeon Pro 560X (2018, used in MBP) = 255217

A12Z (slight tweak on 2018 A12X) = 221182

Two year old tablet design = mid range laptop GPU in 2020. I think you are underestimating what Apple could do with a newer GPU design and a laptop sized battery.
 
  • Like
Reactions: Roode and Toby_95
Is anyone able to comment on Apple's reliance of fixed function hardware in iOS devices? From my limited understanding, this is part of the reason an iPad Pro can export 4K as fast as a MacBook Pro, because it doesn't use (or doesn't fully use) the CPU to do so, but rather a dedicated a hardware encoder.

Does this mean if a new standard comes out 6 months after purchase the Mac either won't be able to work with it or will only be able to render it dog slow on the Apple CPU? Or will this hardware not be required on a Mac since they can cram in as many cores as they effectively like and match performance that way?

I guess expansion cards are always an option on the Mac Pro but what about other Macs?
 
Yes they will. They will use the same core design, but different top level designs, different number of cores, and different SoC blocks. The whole point of SoC design methodology is to allow it to be very simple to design different chips based on the same technology.

In theory that is true, but not in practice. They need to run the wafers with these chips through the factory. It takes a few months for each wafer from start to finish. I doubt that they go through all the trouble of processing different silicon for iPhone, iPad, low-end Mac, mid-range Mac, high-end Mac and getting the numbers just right to not end up having too few or too many chips.
They can not even sell surplus silicon to other companies.

Let's see, but I doubt that there will be much variance between the chips. It would not be cost effective.
 
  • Like
Reactions: Atlantico
I think traditional Unix stuff will be ok. When the Mac transitioned from PowerPC to Intel open source projects just targeted the Mac Intel binary format.
That's not the worry here, the point is if you can install it!
 
Which defeats the entire purpose of this transition. In order to obtain the benefit of ARM one will need native ARM applications. I have no doubt many applications will make the transitions, especially the big boys. However there will be a significant amount which never will. Leaving emulation / translation as the only option.

im pretty sure the entire purpose is to get more efficient cpus. Running windows software on a Mac has never been apples primary goal, ever.

im sympathetic to what’s being “lost” (go read my posts about the then rumour of arm Macs) - I use x86 Debian VMs for about 80% of my work.

But I’m also realistic. There are likely longer term benefits for most Mac users, and the drawbacks will reduce as more developers either recompile their existing apps or fill the spots left by those who don’t recompile/port.

some things are likely gone long term, such as x86 compatibility.

I know it means I have to think about what/when I’m buying next, a lot more now, and that’s part of working in/with technology.
 
  • Like
Reactions: Nightfury326
If you observe Open source projects like say https://github.com/lifting-bits/remill What it does lift X86 or ARM ISA to LLVM Bit code and once you have that you can Jit or AOT the whole thing. Since the mentioned "during install" much of it might be happening AOT, but in the presentation they also mentioned Java VM will also run seemlessly so they might also be doing some JIT + Caching.

So you're saying if I can build from source on an Intel Mac, i'll be able to do the same on an Apple silicon Mac?
 
I have invested heavily in Thunderbolt devices. This is quite disturbing. Especially given that there is no information about that roadmap from Apple. Zero reassurance.
I think Apple will support TB for a long time. Mac Pros, MBP, XDR displays, most hardware rely heavily on it. I would be shocked if they drop it. Apple had a huge investment developing it. Also someone said the TB 4 is backward compatible. So I would think the next 10 years or so should be fine.
 
  • Like
Reactions: Roode
You’re misreading that. That’s Apache printing out the request coming from Safari, so it’s just the identification string coming from Safari on the host machine (I imagine it’s pretending to be “Mac OS Intel” so that during testing it wouldn’t accidentally reveal that it’s really ARM in server logs offsite). It‘s not 100% clear, but it appears that the Linux instance is just virtualized and not emulated (that is, it’s the ARM version of Debian).View attachment 926045

Okay, true that I misread that. It could be reasonable it's masked by intention. Or for that matter that this is a feture that is not yet functional for ARM and that's why it was demod on an Intel machine.
 
I like how during the Virtualization feature demo, they only mention Linux and not Windows. As a full stack web developer, I need all environments, so it makes me nervous. Need more info before deciding if I'm going to make the jump to their silicon or begrudgingly transition my primary machine to a PC. Here's hoping...
Windows will be fully supported under ARM, can you see a future with Apple having so advance hardware and Microsoft suffering any pain with Intel?

intel’s days on consumer are dead. Let’s see how they run on the server side the next 10 years or if they keeps on x86.

what looks odd is Apple didnt mention any percentage gains, well they didnt show any details at all further than “look hiw fluid it is!!”

im wondering if theARM movement is somethingmore related to some gains but more control over production than real world gains
 
Last edited:
That really is awesome. I’ve been fascinated with emulation, ever since I played with Basilisk II back in the 90s to emulate a 68k Mac on my PowerPC Mac.

I’ll definitely be reading up on Remill, it sounds revolutionary.
To be honest I think Apple might have a trick or two up its sleeve that we might not be aware of, since they do have previous experience in this sort of technology.
 
  • Like
Reactions: NickName99
It took Apple less than four years to obsolete the last G5s…

You cannot imagine how unenthusiastic I am after hearing this news. It’s almost like Apple is doing its best to turn us all off its products. Having moved from 68k, to PPC, and classic to MacOS X, then on to Intel… with Rosetta support being dropped in relatively short order… I am widely unenthusiastic about yet another Apple fudge-fest.

They have made obsolete whole swathes of machines, that would still be usable today but for a useable web browser and mailer (frankly who cares about the new, super-duper features that we all seemed to manage without – I'd rather see the return of WindowShade!). Sometimes you have to wonder whether this is just planned obsolescence in order to benefit their bottom line.

It’s almost a direct poke in the eye for them to be naming the next bridge “Rosetta2”. Fool me once, shame on you…
 
at this point all we can infer is because the A12z has comparable GPU in compute performance to a Xbox one S which itself has similar performance to a PC with about a Radeon HD 5850 thats the level of performance we can expect give or take for optimizations.

atleast out of the Dev Box but who knows what they will have on a future Mac Pro equivalent in 2 years or whenever the do the highend.
Yes but the A12Z GPU performance is far below the dGPU solutions Apple offers right now. On the CPU front, they are already ahead of intel, on the GPU front, we haven't seen anything that can match the Navi's and the Vega's.

But probably their GPU in SoC can scale as well.
 
  • Like
Reactions: Atlantico
I am excited but also apprehensive. We have been holding back upgrading our MacPro but with the ARM processors, they can give us excellent performance with the same upgradability. But we are concerned about the price. The recent MacPros are very, very expensive so I would assume a two ARM-processor MacPro will also be very expensive. We shall see.
 
In theory that is true, but not in practice. They need to run the wafers with these chips through the factory. It takes a few months for each wafer from start to finish. I doubt that they go through all the trouble of processing different silicon for iPhone, iPad, low-end Mac, mid-range Mac, high-end Mac and getting the numbers just right to not end up having too few or too many chips.
They can not even sell surplus silicon to other companies.

Let's see, but I doubt that there will be much variance between the chips. It would not be cost effective.
I guarantee you it is true. There are at least three different chip designs for different products already getting ready for tape-out, each with a different core configuration.

And there is no issue with supply chain management here. They know how many wafer starts to do for each product, and they are very good at managing it. They will use every die they make, for a very long time, and use excess in next year’s products for lower end devices.
 
  • Like
Reactions: MyopicPaideia
im pretty sure the entire purpose is to get more efficient cpus. Running windows software on a Mac has never been apples primary goal, ever.
The quote of my response didn't make reference to running Windows software. It was wrt native Mac OS software.
 
It will most likely be emulated Windows and I wouldn't be surprised if the performance is much worse than bootcamp on an existing Intel based Mac.

You would not want to run x86 Windows but rather ARM Windows under virtualization. If a Windows app is not available for ARM, ARM Windows would be responsible for emulating the application code, not MacOS or the Virtualizer.
This has the advantage, that not the whole Windows (like the Windows Kernel) need to be emulated but just the application code, while the Windows Kernel running native.
 
  • Like
Reactions: aid
3DMark Ice Storm Unlimited off screen 1280 x 720 from NotebookCheck
----

5500M = 210925

Radeon Pro 560X (2018, used in MBP) = 255217

A12Z (slight tweak on 2018 A12X) = 221182

Two year old tablet design = mid range laptop GPU in 2020. I think you are underestimating what Apple could do with a newer GPU design and a laptop sized battery.
Ok, but I'm comparing Apple's to Apple's.

Today Apple's dGPU offerings are 5500M, 5600M, Vega 48, Vega 56, Vega 64, Vega II, Vega II Duo and 5700XT.

Apart from 5500M, all these are 5 to 10 times as fast as 12Z.

So yes, I also expect Apple to come up with a solution. But we haven't seen anything remotely close to the speeds of the GPU's from Apple yet.
 
  • Like
Reactions: dysamoria
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.