Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
* 7nm fabrication
* 3 High Power Cores
* 5 Low Power Cores
* 8 Core 2nd generation Apple GPU

Predicted Geekebench scores :p
Single Core: 4,500
MultiCore: 12, 000
Compute: 45,000

All they have to do is Increase or add L3 Cache to the A11 and give it more L1 Cache. And a minor speed bump thanks to a smaller nm die will make it faster and yet more energy efficient.
 
  • Like
Reactions: gobikerider
I don't know enough to say, but is Apple leading when it comes to mobile chip development? It seems like they are continuously outdoing themselves and the competition year after year. I wonder when they will start a new series of chips dedicated to laptops or potentially a mac mini.
 
I don't know enough to say, but is Apple leading when it comes to mobile chip development? It seems like they are continuously outdoing themselves and the competition year after year. I wonder when they will start a new series of chips dedicated to laptops or potentially a mac mini.

Yes. Apple is leading the way in chip design. No, they probably won’t be moving the Mac to ARM anytime soon.
 
Why do you say that?

They’d either have to create an emulator or get all the developers to make their apps for ARM. The first likely isn’t going to happen, even if it’s more likely than the second.

And unless they can get Microsoft to sell an ARM version of Windows? Few would buy it anyway.

No Windows on ARM for sale
No apps (or emulated and runs mediocre)
No real benefit to users

(If they can fix two of the three, they’d probably do it)
 
They’d either have to create an emulator or get all the developers to make their apps for ARM. The first likely isn’t going to happen, even if it’s more likely than the second.

And unless they can get Microsoft to sell an ARM version of Windows? Few would buy it anyway.

No Windows on ARM for sale
No apps (or emulated and runs mediocre)
No real benefit to users

(If they can fix two of the three, they’d probably do it)

You might be right, but I’d imagine Apple wants a little more control over the chips in their laptops and desktops. I could even see a new OS developed.
 
You might be right, but I’d imagine Apple wants a little more control over the chips in their laptops and desktops. I could even see a new OS developed.

If we see an Apple laptop with ARM, it’ll be running an iOS variant. Because everything is nowadays that’s not a legacy OS.
 
I honestly feel like iOS 11 is a tease for iOS 12 when Apple will introduce Final Cut and Xcode for iOS

As much as I want that, I feel before I could replace my macbook, it would need something other than lightning. I need to be able to reliably plug monitors into it to get any decent amount of work done - let alone touchpad/mouse support.

11 is great, but I just can't help but feel that's it again for another 2-4 years o_O
 
They’d either have to create an emulator or get all the developers to make their apps for ARM. The first likely isn’t going to happen, even if it’s more likely than the second.

And unless they can get Microsoft to sell an ARM version of Windows? Few would buy it anyway.

No Windows on ARM for sale
No apps (or emulated and runs mediocre)
No real benefit to users

(If they can fix two of the three, they’d probably do it)

Software conversion is not as huge as it sounds, most Xcode projects are pretty much platform independent (Apple have likely done all their 1st party apps already). Many 3rd party SW houses have ARM code running on iOS which is pretty close (MS Office, Affinity Photo, Adobe-ish). Emulation would be fine for non-speed-critical apps.

Is Windows really a must on Macs these days?

Benefits are; much longer battery life, no longer at the mercy of Intel favourism/release cycles/product shakiness/ISA, lower cost, better performance, better security, always-on features.

ARM on Mac is quite compelling.
 
  • Like
Reactions: DNichter
Software conversion is not as huge as it sounds, most Xcode projects are pretty much platform independent (Apple have likely done all their 1st party apps already). Many 3rd party SW houses have ARM code running on iOS which is pretty close (MS Office, Affinity Photo, Adobe-ish). Emulation would be fine for non-speed-critical apps.

Is Windows really a must on Macs these days?

Benefits are; much longer battery life, no longer at the mercy of Intel favourism/release cycles/product shakiness/ISA, lower cost, better performance, better security, always-on features.

ARM on Mac is quite compelling.

If you actually tried to take a Mac app from today, hit compile to ARM, and run it? You’re going to have a bad time. And if you think the big apps on the Mac aren’t optimized for x86/64, you’re likely wrong.

Without Windows, yes, massively fewer people buy the Mac.

Let’s talk about that better performance. Let’s really talk about it. What makes you think the A11X will be actually more powerful than an Intel chip?
 
  • Like
Reactions: sracer
Software conversion is not as huge as it sounds, most Xcode projects are pretty much platform independent (Apple have likely done all their 1st party apps already). Many 3rd party SW houses have ARM code running on iOS which is pretty close (MS Office, Affinity Photo, Adobe-ish). Emulation would be fine for non-speed-critical apps.

Is Windows really a must on Macs these days?

Benefits are; much longer battery life, no longer at the mercy of Intel favourism/release cycles/product shakiness/ISA, lower cost, better performance, better security, always-on features.

ARM on Mac is quite compelling.
@Michael Goff nailed it and I can't really add to that.

Most of the benefits you cite are directly benefits for Apple itself not necessary Apple customers.... and seem to be based on the assumption that Apple's laptop/desktop product decisions are being adversely affected by using Intel chips rather than if they used their own. Something along the lines of Apple saying, "don't blame us for the long upgrade cycle on our laptops/desktops, we'd update the line more frequently if Intel was on the ball".

From my first hand experience I cannot agree with you that "software conversion is not as huge as it sounds". I'm not sure where you're getting your information regarding "most Xcode projects are pretty much platform independent" but I highly doubt it.

For me, ARM on Mac would be the final straw that causes me to leave the Apple world. But that's me, and I "think different". :)
 
If you actually tried to take a Mac app from today, hit compile to ARM, and run it? You’re going to have a bad time. And if you think the big apps on the Mac aren’t optimized for x86/64, you’re likely wrong.

Without Windows, yes, massively fewer people buy the Mac.

Let’s talk about that better performance. Let’s really talk about it. What makes you think the A11X will be actually more powerful than an Intel chip?

1) What proportion of implementation methods /blocks in your projects are actually optimised for x86/64? Mine have very few. Of course it depends on the nature of the App (mine only have performance-specific components) and understanding the arch-specific code in 3rd party frameworks, which is why it stick with 1st party where possible.

I’d be keen to know the proportion of the 100m OSX/MacOS install base running Windows. It was a good marketing point for switchers 10 years ago but most users I poll are rarely rebooting or VMing Windows as they realise Mac natives are usually better.

Whilst iMac Pros will do the heavy lifting, most Mac Apps won’t push an A11. I’m hearing good things about shared memory between CPU/GPU (no round-tripping) A10X/11 but have yet to compile benchmark code to compare this to AVX/SIMD.
 
@Michael Goff nailed it and I can't really add to that.

Most of the benefits you cite are directly benefits for Apple itself not necessary Apple customers.... and seem to be based on the assumption that Apple's laptop/desktop product decisions are being adversely affected by using Intel chips rather than if they used their own. Something along the lines of Apple saying, "don't blame us for the long upgrade cycle on our laptops/desktops, we'd update the line more frequently if Intel was on the ball".

From my first hand experience I cannot agree with you that "software conversion is not as huge as it sounds". I'm not sure where you're getting your information regarding "most Xcode projects are pretty much platform independent" but I highly doubt it.

For me, ARM on Mac would be the final straw that causes me to leave the Apple world. But that's me, and I "think different". :)

So what are the benefits of staying with Intel? Apple’s own silicon currently provides higher performance per watt than Intel in the 5-10W mobile space, they are untested in the 15-30W laptop space but it’s looking good. Why would you lock yourself in to runner-up technology? Where are the tangible benefits?
 
So what are the benefits of staying with Intel? Apple’s own silicon currently provides higher performance per watt than Intel in the 5-10W mobile space, they are untested in the 15-30W laptop space but it’s looking good. Why would you lock yourself in to runner-up technology? Where are the tangible benefits?

Where are you getting your information about performance per watt?
 
Where are you getting your information about performance per watt?

Tenuously in truth as I’ve not seen Ax on a rig; a Seeking Alpha reference to an Anandtech article denoting TDP at 4W for a non-‘X’ series processor (A9). Enclosure thermals dictate anything above 8-10W for iPads is unlikely so if the A11X delivers my estimated performance at <10W it’s way ahead of the Intel curve.

It’s the graphics performance of GPU family 4 that’s really impressive. In fact it’s so good it may even have forced rivals to collaborate.
 
Tenuously in truth as I’ve not seen Ax on a rig; a Seeking Alpha reference to an Anandtech article denoting TDP at 4W for a non-‘X’ series processor (A9). Enclosure thermals dictate anything above 8-10W for iPads is unlikely so if the A11X delivers my estimated performance at <10W it’s way ahead of the Intel curve.

It’s the graphics performance of GPU family 4 that’s really impressive. In fact it’s so good it may even have forced rivals to collaborate.

And how are we comparing the performance of a processor Intel vs ARM? I don’t think I ever got an answer for that. I mean, I saw Geekbench estimates but...

Geekbench is a rather poor measure of anything except the ability to run Geekbench. It also doesn’t move properly between ARM and Intel. An Android device running an 835, for example, gets roughly 2200/7000 whole Qualcomm showed off their reference for Windows getting 1202/4263 with the same chipset.
 
So what are the benefits of staying with Intel? Apple’s own silicon currently provides higher performance per watt than Intel in the 5-10W mobile space, they are untested in the 15-30W laptop space but it’s looking good. Why would you lock yourself in to runner-up technology? Where are the tangible benefits?

Running x86 code is the biggest benefit. x86 has 40 years of codebase developed for it and is backwards compatible.

ARM chips break backward compatibility every generation or two.

Emulating x86 through ARM results in a 50% penalty under GeekBench.
 
And how are we comparing the performance of a processor Intel vs ARM? I don’t think I ever got an answer for that. I mean, I saw Geekbench estimates but...

Geekbench is a rather poor measure of anything except the ability to run Geekbench. It also doesn’t move properly between ARM and Intel. An Android device running an 835, for example, gets roughly 2200/7000 whole Qualcomm showed off their reference for Windows getting 1202/4263 with the same chipset.

Geekbench is limited as any benchmark is but breaking apart its components, it’s quite broad-reaching. The only way to test real performance is to run the Apps you want to use on each hardware architecture & OS. Common products are rare but Serif’s Affinity Photo provides some good insight; Serif claim up to a 4x advantage of Ax over Quard-core i7 at a recent keynote. I ran a Macro test of the product on both platforms which bears this out (mainly thanks to shared memory across CPU/GPU). Ax performance is real.

On the Qualcomm reference platform; I wasn’t aware Geekbench for Windows was compiled for ARM which would mean its running x86 emulation and a far lower speed is to be expected. It’s still pretty good given the majority of shipping Windows PCs are not Core i5/i7 they are Pentium/Core i3.

The new rumour on iOS/MacOS backend convergence is interesting if a little dubious. This should foster more MacOS Apps but also creates not only a target-device but a target HW architecture independent IDE.

If I were Apple, I’d rollout a T3 (A11X2) in the MBP in place of the dGPU and transition the low-end MacBooks completely.
 
Running x86 code is the biggest benefit. x86 has 40 years of codebase developed for it and is backwards compatible.

ARM chips break backward compatibility every generation or two.

Emulating x86 through ARM results in a 50% penalty under GeekBench.

All of which benefits the developer but none of which benefits the user as it creates an architecture-centric development regime which literally ignores the platform the user bought.

Whilst there are an abundance of mature, stable & efficient x86 libraries the world has moved on from the monolithic codebase. Integration is the key area of development focus and in the Apple world with rich, established, performant 1st party SDKs we have far less reliance on 3rd party frameworks. This benefits our customers by giving them a real MacOS/iOS user experience.
 
They’d either have to create an emulator or get all the developers to make their apps for ARM. The first likely isn’t going to happen, even if it’s more likely than the second.

And unless they can get Microsoft to sell an ARM version of Windows? Few would buy it anyway.

No Windows on ARM for sale
No apps (or emulated and runs mediocre)
No real benefit to users

(If they can fix two of the three, they’d probably do it)

Windows ARM laptops are on sale. There are only a few but they are there. Qualcomm was the launch partner. A bunch more debuted at CES.
 
Geekbench is limited as any benchmark is but breaking apart its components, it’s quite broad-reaching. The only way to test real performance is to run the Apps you want to use on each hardware architecture & OS. Common products are rare but Serif’s Affinity Photo provides some good insight; Serif claim up to a 4x advantage of Ax over Quard-core i7 at a recent keynote. I ran a Macro test of the product on both platforms which bears this out (mainly thanks to shared memory across CPU/GPU). Ax performance is real.

On the Qualcomm reference platform; I wasn’t aware Geekbench for Windows was compiled for ARM which would mean its running x86 emulation and a far lower speed is to be expected. It’s still pretty good given the majority of shipping Windows PCs are not Core i5/i7 they are Pentium/Core i3.

The new rumour on iOS/MacOS backend convergence is interesting if a little dubious. This should foster more MacOS Apps but also creates not only a target-device but a target HW architecture independent IDE.

If I were Apple, I’d rollout a T3 (A11X2) in the MBP in place of the dGPU and transition the low-end MacBooks completely.

A10x on an iPad Pro runs affinity photo and Adobe Lightroom faster than my MacBook Pro 13” (2017). It’s performance doing heavy Raw manipulations quite astonished me.

If they can make these apps running that fast on macOS by switching to arm I’m all for it.
 
A10x on an iPad Pro runs affinity photo and Adobe Lightroom faster than my MacBook Pro 13” (2017). It’s performance doing heavy Raw manipulations quite astonished me.

If they can make these apps running that fast on macOS by switching to arm I’m all for it.

It gets a bit more complicated, since some of the things that make ARM + iOS so good in the <10W envelope mean you need to revisit things as you scale it up into laptops and desktops. There's also the pesky differences between iOS and macOS that can and do impact performance in real world scenarios. And the missing features in the integrated GPU/CPU that desktop/laptop chipsets have had, that happen to free up die space for other things or cut power requirements, but at the cost of developer complexity or limitations (making some particular bits of code good, and other bits of code bad).

One could argue that one reason that Intel chips fall so short in the low power realm is the design is optimized towards maximum peak performance when plenty of power is available. But it must give up some efficiency to reach those peaks. PowerPC had issues scaling up to keep up with the sort of crazy that Intel was reaching for back in the day, and the G5 in particular wound up stagnating, and unable to scale down in power use for laptops. Intel itself hit a dead end with the Pentium 4 design and went back to the Pentium 3 design to create the Core line of chips. There's no clear way to predict how well a particular design will scale in either direction without trying it, unfortunately. And you can hit some rather nasty walls in the process, based on history.

That said, there's a surprising amount of overlap in Apple's MacBook and cheaper end of the MBP 13" line. I wouldn't be too surprised if you could get a fast ARM Mac that was competitive with the MacBook and 13" MBP. Although it'd be nice if the laptop version of an A-series SoC addressed some of the short-comings of the GPU which have meant that certain features, like compositing filters in CoreAnimation simply haven't been supported on iOS.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.