Report: Apple Silicon iMac Featuring Desktop Class 'A14T' Chip Coming First Half of 2021

[whitby]

I am really looking forward to seeing what these new Apple Silicon devices will bring to the table. I have just upgraded all my laptops and desktops to the latest Apple versions, so there is no rush to replace them and I will buy one of the new devices just to see how good they are. The question is what do we hope to get from them (apart from the inconvenience of not running virtual machines as easily ).

For me improved battery life has to be the number 1 priority, as the biggest issue I have is that the Intel powered devices barely last half a working day. My 16" MBPro is a battery hog and barely lasts 4 hours. The Air is pretty good (around 7 to 8 hours for me) and the 13" MBPro is in between at around 5 to 6 hours of usable time for me. I would like a machine which can be charged every couple of days or so and does not give me power anxiety (after a few hours I need to find my charger, we have them all over the house so one is handy). An Air with 2 days of life would be ideal. In other words pretty much what the iPad offers but with a 'proper' OS.

I have no issue with the speed of the Intel powered devices (faster will always be better, I know) and I do not play games. However I do a lot of photography work, email, software design, documents and I use VMs a lot to run test systems that only run in Windows. This last one is going to be a problem for me and I have a Microsoft Surface Laptop 3 for running systems in Windows environments just in case. I used to build powerful Windows desktop machines for many many years and I have an 8 core processor with 128GB RAM and 4 TB SSD and 15TB of rotating storage that I may have to press back into service, but it would be so much easier if using my iMac with VMs could continue, but the technical challenges are formidable. But there are options appearing but most forget that most business software runs on Intel architecture machines in either a UNIX or Windows environment. And it will not change for many many years, there are too many man-hours invested in business software systems that use the current environments and architectures to consider an easy and instantaneous change. Web based applications often have backends buried in these older systems and architectures and much as people would like to think that the world runs on web based micro architectures, they do not and are often powered by much older systems that will need to be maintained and enhanced for years.

It is all very interesting and I love what Apple are doing and will continue to experiment and adopt where appropriate. It is an interesting world we live in.

 

[deconstruct60]

Not sure how I feel about buying a $2000 imac with the same processor a snot nosed 12 year old carries around with him on his iphone 12. Yes, I realize the T means more horsepower and magic powder, its just the visual that gets me...

It is not going to be the exact same processor. The A14X isn't going to be the same as the A14. If Apple grew the SoC package even larger for an A14T then the performance gap could be even larger. Minimally Apple would need to put more pin outs on the A14T if it handled desktop class I/O. Mini , iMac , and Mac Pro have substantially more ports than an iPhone ( one port wonder). The I/O is kneecapped on the A14 to keep the size of the physical package smaller to fit inside the substantially smaller form factor. Once get into the desktop space the iPhone/iPad Pro physical logic board limitations would only "paint" Apple into a corner of rather goofy self imposed limitations. Unless Apple is out to strongly chase after the Raspberry Pi sized computer market, it is more than dubious move.

 

[ChrisA]

I'm a little surprised the Mac mini isn't going ARM early on given its lower usage in the home and industry. Sounds like a good and fairly low-impact early product for this. They have some balls to begin with laptops of all things!

What's the point of an A14 in a desktop? The A14 is for saving power and extending the life of the battery. On a computer that remains plugged into AC mains power, you are best off using a more powerful CPU

Small size notebooks that are not used for high-end stuff like 3D CAD or Final Cut Pro are the logical use for these iPhone processors. It will take time for Apple to develop higher-end ARM chips so the transition is starting at the low end of the line were the use of electric power matters the most

 

[deconstruct60]

I wonder how the release schedule for the A series chips will affect the updates of the Mac lineup. As the A series are used in the yearly refreshed iPhones and iPads they come out - surprise surprise - every year but will this mean that the Macs will be refreshed yearly or will they be more spaced out. Interesting to see and if its something Apple wants to promote

Probably not on a yearly schedule. Even the A series chips are not totally on a yearlly schedule. The 'X' variants used in the iPad Pro has more so been on a 18-24 month schedule for several years. Apple has a track record at this point of just doing a real 'X' variant upgrade when the microarchetiecture design takes a process shrink jump.

Several years ago Intel was on a 'tick-tock' schedule. They would do a microarachitecture major change on 'tick' iterations and then more so focus on a process shrink on the 'tock' cycle. The 'X' series has been doing a 'tock-tock-tock' movement when the process shrinks were available ( which is around every 18 or so months).

Where the Macs may overlap with the phones is at the lower end of the line up where there is more overlap with the iPad Pro and iPhones. The volume of the iPad Pro + the 'bottom' one or two Mac laptops may be enough additional volume for Apple to assign more resources to do a larger SoC more often than they have.

But as get higher up in the much lower volumes of the Mac Pro line up there is quite a good chance that the iteration will actually get longer than the 'X' series has been. So Mac Pro and iMac Pro zone iterating around every 3-4 years. There they are likely going to be substantively behind the iteration pace of AMD and (by 2022) Intel.

Apple sells in the range of 12M Macs a year split over multiple CPU packages ( SoC designs). The iPhone sells in a year roughly a single digit multiple of the entire Mac user base each year. The economies of scale are much, much, much different. Mac aren't going to get the same resource allocation as the iPhones do. The iPads do larger numbers than Macs do also. The iPad Pro is small enough fragment of that though that it probably does need some volume synergy with some moderately high Mac model to get more traction .

 

[DeepIn2U]

It's happening! Can't wait to see these new Macs!

Interesting to see a MacBook to return using the same CPU/internal-GPU as the 2020 iPad Pro. Now what will the excuse Apple will continue to give us that iPad is not going to run macOS 11 etc? Whom will be the first to fully move macOS to iPad to make Mr SnazzyQ super happy. Hmmm?

 

[smulji]

he 'X' variants used in the iPad Pro has more so been on a 18-24 month schedule for several years.

True. Since the A10X, they've been on a 24-month cycle.

 

[cmaier]

What's the point of an A14 in a desktop? The A14 is for saving power and extending the life of the battery. On a computer that remains plugged into AC mains power, you are best off using a more powerful CPU

Small size notebooks that are not used for high-end stuff like 3D CAD or Final Cut Pro are the logical use for these iPhone processors. It will take time for Apple to develop higher-end ARM chips so the transition is starting at the low end of the line were the use of electric power matters the most

They’ve had plenty of time to design high end processors. You are all going to be surprised.

 

[deconstruct60]

Lifuka is discrete graphics.

Not necessarily. In fact Apple going on and on about "Unified Memory" being part of their secret sauce points to that it could be an just as integrated as the rest of the A series. Just bigger. This could be a GPU chiplet attached to the same "internal" memory bus in the SoC package as the CPU ( and ML and Secure and etc. ) cores. Throw in a large GPU local cache/working space for outbound frame bufffers and Apple could get the performance substantially up without getting off their "same GPU structure as the iPhone" track.

It would consume stantailly more power and space so it simply would not fit in an iPhone or iPad Pro. But most Macs don't really have that constraint. The desktops especially so.

If it is a big "snap on" iGPU then Apple could just attach a smaller GPU chiplet to do Macs lower in the line up ( e.g., Mini or MBP 16" ) .

If the GPU is on the SoC package and uses the main system RAM for primary workspace than it is integrated graphics. A chiplet doesn't necessarily make it discrete.

Lifuka is an island in Tonga. If the code names are relationally significant "Likufa" could easily be about a sub component of a SoC ( GPU in SoC as Likufa in Tonga ). Just one that not all the SoCs will get.

The notion that Apple is going to try to jump into the mid-upper discrete GPU competition is more than somewhat dubious. There is relatively little volume ( units sold) there and it really doesn't make much sense if Apple puts at least a modest iGPU in every SoC to cover running iOS/iPadOS apps in direct virtualization mode.
Apple is way, way behind in top end GPU performance computationally wise.

 

[MikeZTM]

I think it's fair to say they will perform amazing for the apps they have been designed to run.
That's the easy bit.

If you said to Apple or even Intel or AMD, design me a chip that's runs "Software X" at blistering speed I'm sure they could do.

It's designing a general purpose chips that could be fitted into any machine and run any software well that's the trick.

Apple does not have to do this, so they have a much easier job.
They control the rest of the hardware that does with the chip and the tools to build the software.

The key is going to be just how well new macs run general purpose software that was not specifically designed from the ground up to run on them.

Usually no client side software is designed from "ground up" for any platform.
Photoshop was designed for desktop form factor instead of Intel or IBM/Moto or ARM.
Even games was designed for input mechanisms more than platforms. A controller based game can be ported from PS4 to switch when there's enough audience.

 

[kaycrystal626]

Will see how it perform on Adobe products. That's all I care.

 

[MikeZTM]

But laptops are where the low energy impact of the A series chips comes into play most.

You can think it in reverse -- low energy consumption means you can pack in more cores in a desktop workstation.
While Intel tops at 28 cores UMA and AMD tops at 64 cores UMA, you can have a 100 or more cores UMA ARM chip in the same power envelope (about 200W).

 

[MikeZTM]

What's the point of an A14 in a desktop? The A14 is for saving power and extending the life of the battery. On a computer that remains plugged into AC mains power, you are best off using a more powerful CPU

Small size notebooks that are not used for high-end stuff like 3D CAD or Final Cut Pro are the logical use for these iPhone processors. It will take time for Apple to develop higher-end ARM chips so the transition is starting at the low end of the line were the use of electric power matters the most

A14 in desktop means faster performance at same power budget.
A13 on a iPhone with limited thermal/power condition is already reaching 9900k by single core performance.

Remember Intel can never maintain that single core performance when you stressing all cores due to heat and power limit. An 8 core A13 will out-perform 9900k a lot.
A14 currently look like will have about 18% more single core performance compared to A13. And that projected to more than the performance of a 10900k.

We can forget Intel for now even on desktop and watch A14 fight AMD Ryzen 5900x on single core performance crown.

 

[MikeZTM]

Not without some haxie in there to make it work. The move to ASMacs signals a huge, possibly deadly, hit to the Hackintosh community.

You can bet Apple isn't going to do this stupidly. They will bork it to ensure iPads don't run it without major work.

Since apple claimed Apple Silicon Mac can still boot (in future) old version of macOS offline I guess they will find a way to run it on Pi.
The most tricky part becomes the GPU driver as they may not have easy way to output any graphics.

 

[nihil0]

And z from the a12z stands for what?

Zuperzpeed

 

[cmaier]

Since apple claimed Apple Silicon Mac can still boot (in future) old version of macOS offline I guess they will find a way to run it on Pi.
The most tricky part becomes the GPU driver as they may not have easy way to output any graphics.

When did apple claim apple silicon macs could boot old versions of macOS?

 

[MikeZTM]

When did apple claim apple silicon macs could boot old versions of macOS?

Explore the new system architecture of Apple silicon Macs - WWDC20 - Videos - Apple Developer

Discover how Macs with Apple silicon will deliver modern advantages using Apple's System-on-Chip (SoC) architecture. Leveraging a unified...

developer.apple.com

In this session Apple shows you can do reduced security mode to boot older version of macOS.
By old I mean "old in the future" that is still Big Sur and later.

This just officially confirmed running macOS is still not tethered to apple's signing server unlike iOS right now.

 

[boss.king]

What's all this fuss with gen1? The cpu's are already proven for years in ipads, so they are not gen1's, all the problems should come from arm macos being gen1, which will be upgradeable and gradually fixed.

There's more to hardware than the CPU. Look at what a catastrophe the 2016 MBPs were. I've have to take mine in for repairs 3 times already, making it the most unreliable computer I've ever owned. The third gen always seems to be where they really work out the kinks in my experience.

 

[Piggie]

Usually no client side software is designed from "ground up" for any platform.
Photoshop was designed for desktop form factor instead of Intel or IBM/Moto or ARM.
Even games was designed for input mechanisms more than platforms. A controller based game can be ported from PS4 to switch when there's enough audience.

Yes, that's what I'm saying.

It's going to be vastly easier to code your apps, or pay people millions to code their apps specifically for your hardware, then say, look how amazing our hardware is, look at the speed of those apps.
Whilst it's obviously true those apps do run amazing, one could, if one wanted call that cheating.

A bit like company A making a car that does pretty good on all race tracks
Then company B making a car specifically designed for one specific track, showing how fast this car is around this track and hence giving the impressing their car is better.

I do hope Apple are NOT going to play a silly game like that.
I want Apple ARM Chips to be amazing ans way better than Intel Chips.

But I also accept Intel chips are made to run anything you can throw at them, made by anyone for a million different hardware types and configurations.

I hope Apple does not just show off special apps written specifically for it, and say wow, look how much faster our chip is, when an Intel chip destroys it, with most other software not specifically coded for it.

Not long to go now, and I'm really looking forward to seeing what Apple shows off, and I hope it's amazing.
Genuinely amazing, not just great at a few carefully selected things.

 

[DeepIn2U]

Li-fuka ?

A14X confirmed™ for the MacBook! Betting that will be the first ARM Mac this year.

LMAO ... I also got the reference in this name too. Japanese gotta love the english sounds of their words ... just so cool and works!

I'm a little surprised the Mac mini isn't going ARM early on given its lower usage in the home and industry. Sounds like a good and fairly low-impact early product for this. They have some balls to begin with laptops of all things!

Not sure if they have some balls this time around, Cook and team are following the same playbook as Jobs' team back with Intel transition.

1. Calling it a transition
2. Saying it'll take 2yrs and a slow transition, "we've still got some great Intel products still in the pipeline"
3. Rosetta Stone (2) - even the name and the code is being re-used, just updated.
4. First product launched is not only a laptop but their consumer model ... MacBook. now they may rename the MacBook into something else (someone kick Tim walking by in the arse if Apple uses "Apple Book")!

 

[johngwheeler]

Can’t say I love this, as I just ordered a new iMac last night...

Buy you knew that there would be an Apple Silicon iMac in the first half of next year, surely?

You haven’t made a bad choice; a new iMac is probably 6 months away, and would you really want to be an early adopter of a completely new and unproven system architecture?

 

[ravinder08]

Have been holding off on replacing my late 2012 iMac it's till working fine, will wait for the 27" or 32" iMac to drop before replacing it. I think I will definitely notice an improvement. I can't wait!

 

[MikeZTM]

Yes, that's what I'm saying.

It's going to be vastly easier to code your apps, or pay people millions to code their apps specifically for your hardware, then say, look how amazing our hardware is, look at the speed of those apps.
Whilst it's obviously true those apps do run amazing, one could, if one wanted call that cheating.

A bit like company A making a car that does pretty good on all race tracks
Then company B making a car specifically designed for one specific track, showing how fast this car is around this track and hence giving the impressing their car is better.

I do hope Apple are NOT going to play a silly game like that.
I want Apple ARM Chips to be amazing ans way better than Intel Chips.

But I also accept Intel chips are made to run anything you can throw at them, made by anyone for a million different hardware types and configurations.

I hope Apple does not just show off special apps written specifically for it, and say wow, look how much faster our chip is, when an Intel chip destroys it, with most other software not specifically coded for it.

Not long to go now, and I'm really looking forward to seeing what Apple shows off, and I hope it's amazing.
Genuinely amazing, not just great at a few carefully selected things.

By SPECperf I can safely claim Apple Silicon is at least better than Intel's lakes by a huge margin on general propose computing.

BTW SPECPerf is exactly what Intel used to show it's generational difference.

 

[Fomalhaut]

I'm a little surprised the Mac mini isn't going ARM early on given its lower usage in the home and industry. Sounds like a good and fairly low-impact early product for this. They have some balls to begin with laptops of all things!

I think Apple realises that they need to enter "with a bang", and mid-spec laptops are their biggest sellers.

 

[thejoshuabradshaw]

Cannot wait to see the implementation. My 2015 is still a beast for my needs but it's exciting to see the new evolution of the iMac lineup. Face ID, T2, and Apple silica. Don't mind the size as it never leaves my desk so the blocky size will do just fine as long as it's upgradable.

 

[Fomalhaut]

Not to mention the fact that they've already made them for developers. It seems like minis should be one of the first Arm releases for Apple, and I'd be very surprised if they weren't.

Very unlikely I think. The Mini-based DTK is a Franken-Mac used to host a rapidly-produced motherboard with an iPad SoC. The Mini is the natural choice for this kind of low volume temporary solution because it is (a) quite cheap to produce, (b) has large enough internal volume to contain a custom board without extensive design (c) is small and cheap to ship to developers

Bear in mind that the DTK is missing some features - no Thunderbolt / USB-C - and is using an existing iPad chip, not one customised for laptop/desktop use, which Apple Silicon will be.

It's about as close to a production Apple Silicon Mac as a Rasberry Pi....so not very.