What Today's 27" Intel iMac Refresh Tells Us About the Apple Silicon Mac Release Order

[Realityck]

What about GPUs? Does anyone think, due to the increased power savings, that they will be able to include dedicated graphics in the Arm MBP 13"?

See Apple's Homegrown Chips Could Be the End for AMD Graphics in Macs

TBDR captures the entire scene before it starts to render it, splitting it up into multiple small regions, or tiles, that get processed separately, so it processes information pretty fast and doesn’t require a lot of memory bandwidth. From there, the architecture won’t actually render the scene until it rejects any and all occluded pixels.

On the other hand, IMR does things the opposite way, rendering the entire scene before it decides what pixels need to be thrown out. As you probably guessed, this method is inefficient, yet it’s how modern discrete GPUs operate, and they need a lot of bandwidth to do so.

For Apple Silicon ARM architecture, TBDR is a much better match because its focus is on speed and lower power consumption – not to mention the GPU is on the same chip as the CPU, hence the term SoC. This is probably why Apple wrote, “Don’t assume a discrete GPU means better performance,” in its developer support document. It’s all that dang bandwidth it doesn’t need.

Its a question of real comparisons to both discrete GPU's as it is now, and going the TBDR route.

 

[Falhófnir]

What about GPUs? Does anyone think, due to the increased power savings, that they will be able to include dedicated graphics in the Arm MBP 13"?

It seems the GPUs will be Apple custom integrated ones, but likely to be competitive with the best available dGPUs currently on the high end. Not quite to the level of the 16" Pro, but I would still expect a substantial step up on the Air!

 

[awesomedeluxe]

It seems the GPUs will be Apple custom integrated ones, but likely to be competitive with the best available dGPUs currently on the high end. Not quite to the level of the 16" Pro, but I would still expect a substantial step up on the Air!

Yep. The larger machine size, the more guesswork involved. It's reasonable to assume the MBP13 might start with an A14X with an 8-10 core GPU similar to iPad Pros. This will already stomp Tiger Lake graphics.

It's not-unreasonable to assume the MBP13 will have a step-up model using the Bloomberg APU, which probably has 12-20 GPU cores and is almost dGPU class.

What will hold its graphics back from being totally on par with dGPUs is the lack of dedicated high bandwidth memory. I'd say 10:1 the MBP13 exclusively uses LPDDR5.

 

[MisterMe]

You do know that NextStep what OS X is based on was running on both Motorola and Intel processors before Steve Jobs rejoined Apple? In fact they ran it on multiple processors. This was a normal process with gauging performace on other processor platforms in case you needed to change the processors you use.
“NeXTStep ran on PA-RISC, Intel x86, Motorola 68K and SPARC architectures since well before 2000.”
Apple was having a terrible time trying to develop a next generation OS, and if you look at the history of Apple, before they choose NextStep over Gassée BeOs you can see it’s wasn’t long term planning that they were good at.

IIRC, NeXT planned to switch from the Motorola 680x0 to the Motorola 88000 before it decided to drop its hardware and go with Intel white boxes instead. NeXTstep had run on numerous processors before NeXT stopped selling its own hardware and continued to do so after it stopped selling its own hardware.

After Apple merged with NeXT, the new Apple continued this strategy. Apple didn't tell the general public what it was doing until it had announced two processor transitions--MC 680x0 to PowerPC and then PowerPC to Intel x86 more than a decade later. Apple was very clear why it ran MacOS X on multiple processors. However, the fanboys ignored what they were told then and they continue to ignore what they were told now.

Apple said that it ran its OS on multiple processors to ensure that processor dependencies did not creep into the OS. Among the benefits of this strategy is that MacOS X could run on any processor with a compile and no need to optimize for the specific processor. However, the Intel fanboys thought that Intel was the very bestest ever processor and that Apple would never ever make another processor transition.

However, the clear implication of what Apple said during the PPC to Intel transition is that Apple continued to port MacOS X to processors that will never see the inside of a commercial product. This should have been clear when Apple introduced the first iPhone. Now that Apple has announced the switch to its own ARM-based SoCs that our favorite fruit company continues to run macOS on multiple processors in its labs. Intel engineers have told us that Apple was the No. 1 poster of errors for its processors. The implication here is that Apple does not and has never just pulled processors out of the shrink-wrap and slapped them into the motherboard. Apple engineers know at least as much about each processor that it uses as the vendor engineers.

So if you have a question about why Apple chose this model processor or graphics card or whatever over than model of the competition, it is because Apple engineers have thoroughly examined those two and every other viable candidate and made their choice based on detailed information.

One other thing:

There are those here who seem to think that Intel Xeon is the ultimate in processor performance. When you talk about performance, nothing beats supercomputing clusters. As of June 2020, the Top 5 clusters from the Top500 are as follows:

1. Supercomputer Fugaku-Supercomputer Fugaku - A64FX 48C 2.2GHz, Tofu interconnect D, Fujitsu RIKEN Center for Computational Science, Japan (ARM-based)
2. Summit - IBM Power System AC922, IBM POWER9 22C 3.07GHz, NVIDIA Volta GV100, Dual-rail Mellanox EDR Infiniband, IBM DOE/SC/Oak Ridge National Laboratory, United States
3. Sierra - IBM Power System AC922, IBM POWER9 22C 3.1GHz, NVIDIA Volta GV100, Dual-rail Mellanox EDR Infiniband, IBM / NVIDIA / Mellanox DOE/NNSA/LLNL, United States
4. Sunway TaihuLight - Sunway MPP, Sunway SW26010 260C 1.45GHz, Sunway, NRCPC National Supercomputing Center in Wuxi, China
5. Tianhe-2A - TH-IVB-FEP Cluster, Intel Xeon E5-2692v2 12C 2.2GHz, TH Express-2, Matrix-2000, NUDT National Super Computer Center in Guangzhou, China

The most powerful superconducting cluster on Earth is ARM-based. This is not the same as Apple Silicon, but it is closely related. The next two slots go to IBM POWER9-based clusters. You don't get to Xeon until No. 5.

 

[leman]

There are those here who seem to think that Intel Xeon is the ultimate in processor performance. When you talk about performance, nothing beats supercomputing clusters. As of June 2020, the Top 5 clusters from the Top500 are as follows:

It all depends on how you define performance. Supercomputing has very different needs than, say, general purpose office usage or software development. All these clusters are parallel number crunchers that assume predictable data access and computation-heavy algorithms.

Besides, the ranking you show is affected by the cluster size.
The number one cluster on your list, Fugaku, is a total beast with almost 8 million ARM cores.

 

[funkahdafi]

It seems the GPUs will be Apple custom integrated ones, but likely to be competitive with the best available dGPUs currently on the high end.

Man, I really want some of that stuff y'all smokin.

Seriously, you guys need to take off your rose-tinted glasses. The above statement is beyond ridiculous.

 

[leman]

Man, I really want some of that stuff y'all smokin.

Seriously, you guys need to take off your rose-tinted glasses. The above statement is beyond ridiculous.

Best available GPUs? I agree, that’s a bit far fetched. But something like 2070GTX? Shouldn’t be a big challenge.

 

[funkahdafi]

The most powerful superconducting cluster on Earth is ARM-based. This is not the same as Apple Silicon, but it is closely related.

No, it's not.

Absolutely not.

Most ARM designs are highly specific designs to do one particular thing really well. Those chips in those super computers would do nothing in a Mac Pro. Your photoshop would probably underperform.

 

[Realityck]

So if you have a question about why Apple chose this model processor or graphics card or whatever over than model of the competition, it is because Apple engineers have gone of those two and every other via candidate and made their choice based on detailed information.

I think this part of your response is more interesting given the choice of SoC with embedded GPU, versus X86 Intel with external discrete GPU.

What we have is the here and now with consumer and business computers/workstations made by Apple and what software they will be able to run with this transition to achieve todays goals with photography, video, computerized modeling, beside the usual mundane business software?

 

[JMacHack]

Most ARM designs are highly specific designs to do one particular thing really well. Those chips in those super computers would do nothing in a Mac Pro. Your photoshop would probably underperform.

That's kinda the point. Apple's been working on specific designs for their own processors to do some things particularly well, that's their leverage over x86.

 

[funkahdafi]

That's kinda the point. Apple's been working on specific designs for their own processors to do some things particularly well, that's their leverage over x86.

Of course, but please check the context. People here are referring to specialized, highly expensive CPUs and think because those exist Apple must already have something up their sleeves. That's just not how it works.

 

[awesomedeluxe]

Man, I really want some of that stuff y'all smokin.

Seriously, you guys need to take off your rose-tinted glasses. The above statement is beyond ridiculous.

I think what he means is that Macs will use APUs competitive with the best dGPUs currently available in their respective x86 Mac model. His next statement adds the right context, which is that the AS MBP13 will have graphics substantially above what's currently available in Apple's 13" devices but not at the level offered by the dGPUs in the MBP16. I don't really see anything wrong with that statement. If you still disagree:

1. Look at benchmarks for the A12X and A12Z

2. Consider that the A14 GPU cores will have two generations of architectural development and a full process node on those

3. Consider that larger Macs can accommodate way more graphics cores than the iPad can, as well as a stack of HBM2E cache or even HBM2E as all-system-memory

and compare the estimated performance of an appropriately-sized Mac APU to the Intel or AMD graphics currently available. Many Macs use aging AMD parts. Seriously, do you think Apple can't make an APU with graphics more powerful than the Vega 20 in the 21" iMac?

 

[funkahdafi]

I think what he means is that Macs will use APUs competitive with the best dGPUs currently available in their respective x86 Mac model. His next statement adds the right context, which is that the AS MBP13 will have graphics substantially above what's currently available in Apple's 13" devices but not at the level offered by the dGPUs in the MBP16. I don't really see anything wrong with that statement. If you still disagree:

1. Look at benchmarks for the A12X and A12Z

2. Consider that the A14 GPU cores will have two generations of architectural development and a full process node on those

3. Consider that larger Macs can accommodate way more graphics cores than the iPad can, as well as a stack of HBM2E cache or even HBM2E as all-system-memory

and compare the estimated performance of an appropriately-sized Mac APU to the Intel or AMD graphics currently available. Many Macs use aging AMD parts. Seriously, do you think Apple can't make an APU with graphics more powerful than the Vega 20 in the 21" iMac?

That makes sense and if that's what he meant, then I misread and apologize. What I understood was the Apple's GPUs will rival the best dGPUs on the market.

 

[jdb8167]

You do know that NextStep what OS X is based on was running on both Motorola and Intel processors before Steve Jobs rejoined Apple? In fact they ran it on multiple processors. This was a normal process with gauging performace on other processor platforms in case you needed to change the processors you use.
“NeXTStep ran on PA-RISC, Intel x86, Motorola 68K and SPARC architectures since well before 2000.”
Apple was having a terrible time trying to develop a next generation OS, and if you look at the history of Apple, before they choose NextStep over Gassée BeOs you can see it’s wasn’t long term planning that they were good at.

A little history here. At the time of the rumors about switching to Intel everyone who was a developer knew that Apple could have easily kept the Cocoa APIs running on Intel since the NeXT purchase. There were roadmaps early on for OS X that showed Cocoa being used across a number of different architectures. But what threw a monkey wrench into the works was that Apple had spent major development effort on Carbon. Carbon was the final port of the original MacOS APIs to OS X.

There were rumors that Apple had ported Carbon but no one outside of Apple had seen it. I thought it was going to be a problem to move Carbon to x86 until just about a month before the x86 transition announcement when someone pointed out that Quicktime and iTunes for Windows were pretty much a port of Carbon for x86. After that it all clicked and I was not surprised when Steve Jobs took the stage.

Getting the NeXT APIs (Cocoa) ported to x86 was never the problem, it was all the additional work that Apple had put in over several years into backwards compatibility with Carbon for the big names like Microsoft and Adobe. For a hint on what kind of effort it must have been, Apple never did get 64-bit Carbon working well enough to ship after spending a couple of years on it. Apple just bit the bullet and deprecated it and took the complaints from the developer community (again mostly Microsoft and Adobe).

 

[nicfle]

”...for years to come” I was wondering, what if Apple is not really abandoning Intel completely.

Would it be imposible to make a fusion computer with arm and x86 prosessors? There is the T2 chip for instance. The T2 carries out many of the tasks in a mac. Apple has been playing with this x86+arm fusion for over 5 years now. What if the T2 chip was not only for the known tasks, but also for developing a future Pro computer - the basic operating system would run from Apple silicon, but the pro user power would come from the Intel x86 prosessor. Think a hybrid car! This might be a way to offer pro users that extra speed they need until Apple has a equivalent arm chip to offer. Forgive me if this sounds imposible, I`m no computer engineer. I would like to know whatwould be the biggest problem with a design like that.

 

[KPOM]

If as Tim Cook said the Apple Silicon transition will happen in the next two years surely Apple must have been working on these new Macs for at least two or three years now. I believe this latest iMac refresh is the end of the intel Macs. Would be great to see a 24 Apple Silicon iMac by the end of this year.

I wouldn’t be surprised if the 16” MacBook Pro gets updated to Comet Lake before making the jump to Apple Silicon.

 

[KPOM]

”...for years to come” I was wondering, what if Apple is not really abandoning Intel completely.

Would it be imposible to make a fusion computer with arm and x86 prosessors? There is the T2 chip for instance. The T2 carries out many of the tasks in a mac. Apple has been playing with this x86+arm fusion for over 5 years now. What is the T2 chip was not only for the known tasks, but also for developing a future Pro computer - the basic operating system would run from Apple silicon, but the pro user power would come from the Intel x86 prosessor. Think a hybrid car! This might be a way to offer pro users that extra speed they need until Apple has a equivalent arm chip to offer. Forgive me if this sounds imposible, I`m no computer engineer. I would like to know whatwould be the biggest problem with a design like that.

Having a second processor would draw additional wattage, and keeping Intel around for much longer would discourage developers from recompiling their existing software. Apple likely was trying to reassure people that the Intel-based Macs they buy today will still be supported, meaning they shouldn’t hesitate to buy what they are offering today if they need to upgrade now.

 

[nicfle]

Having a second processor would draw additional wattage, and keeping Intel around for much longer would discourage developers from recompiling their existing software. Apple likely was trying to reassure people that the Intel-based Macs they buy today will still be supported, meaning they shouldn’t hesitate to buy what they are offering today if they need to upgrade now.

Yes, you are right. What I was thinking was more of a pro option for those marginal users that need the power from a desk computer and use heavy load software or x86 support. I don`think that would discourage anyone. Most of the software developers would follow their user base to arm. The professionals that need that power would stay with the Apple ecosystem and the developers that see the profit from this group would continue to produce pro software. In a best scenario that kind of fusion prosessor would support existing software without full emulation.

Is the wattage a deal breaker?

 

[leman]

I wouldn’t be surprised if the 16” MacBook Pro gets updated to Comet Lake before making the jump to Apple Silicon.

I would. Comet Lake is literally pointless as an upgrade from Coffee Lake R. All the tests I’ve seen show them performing within the margin of error.

 

[ChromeCloud]

What I was thinking was more of a pro option for those marginal users that need the power from a desk computer and use heavy load software or x86 support.

It's not going to be necessary. Apple Silicon Macs will run x86 apps in emulation under Rosetta 2 faster than what would be possible by running x86 apps natively on Intel chips.

 

[nicfle]

It's not going to be necessary. Apple Silicon Macs will run x86 apps in emulation under Rosetta 2 faster than what would be possible by running x86 apps natively on Intel chips.

I hope you are right. I don`t personally do any heavy load 3d or video editing, but I would be sad if Apple does it`s Final Cut Pro X thing to their pro macs. Cripple and alienate their pro users for a basic user product. Many of my friends left the Apple wagon after the Final Cut Pro X release.

 

[Waragainstsleep]

You do know that NextStep what OS X is based on was running on both Motorola and Intel processors before Steve Jobs rejoined Apple? In fact they ran it on multiple processors. This was a normal process with gauging performace on other processor platforms in case you needed to change the processors you use.
“NeXTStep ran on PA-RISC, Intel x86, Motorola 68K and SPARC architectures since well before 2000.”
Apple was having a terrible time trying to develop a next generation OS, and if you look at the history of Apple, before they choose NextStep over Gassée BeOs you can see it’s wasn’t long term planning that they were good at.

I remember it was a big deal when everyone found out they'd been running it in secret.

 

[Shivetya]

all this performance hype will be meaningless if the apps we want are not compatible and Rosetta emulation is no guarantee because we don't know its full limitations.

 

[leman]

all this performance hype will be meaningless if the apps we want are not compatible and Rosetta emulation is no guarantee because we don't know its full limitations.

What are the apps you are worried about? As to Rosetta, we know quite a lot about it works. If your Intel app uses supported machine instructions, it will run.

 

[Yebubbleman]

I’m curious what Apple are going to do about graphics processing for the Pro machines. Apple’s current Pro lineup is already hobbled by the lack of NVidia cards. Advanced 3D shaders and render engines Like Redshift need an eGPU for Mac. 3D tools are no longer niche. It’s just part of the pipeline, much like Photoshop. Until they have a solution for this, Apple silicon is going to be a novelty for creatives.

If developers develop for Metal, there won't be a problem. It's a big if. And frankly, they will alienate the PC/Console gaming developers (or rather Aspyr, Feral Interactive, Valve, and pretty much any big name developer that isn't Blizzard) much in the same way that the console-graphics-equipped Apple TV has yet to make a serious inroad with gaming the way that other living room platforms have.

But if developers optimize for Apple GPUs (which really are very different from the AMD Radeon 5000 series GPUs we see in the Intel 16" MacBook Pros and 27" iMacs of today), then apps are going to have no difficulty running well on them.

Again, think about what Apple said: Intel Macs are here to stay for *many* years to come.

You're misquoting Apple and Tim Cook here. Tim Cook said that Intel Macs would continue to be supported for many years to come. NOT that Intel Macs would continue to be sold for many years to come. Tim Cook said it would be a two-year transition. If we're judging Apple's definition of completeness (as they did back in 2006 with the PowerPC to Intel transition) as to when every model has an Apple Silicon option rather than when the Intel models disappear from sale, then two years makes sense. I do think that the Intel Mac Pro and the Intel 16" MacBook Pro will linger in sale for a bit longer after the transition completes for those that need them. But they will not be kept separately up to date with Intel and AMD's latest. That would defeat the point of it being a TRANSITION.

(NB: Ubuntu Server for ARM has been available for a while. I don't know if that allows you to install the desktop packages - as you can with Intel server - but if you can't it is probably because there is a shortage of desktop ARM systems to run it on... which ARM-based Macs will provide. Most of the desktop packages are already up and running on Debian/Raspbian and Apple actually showed Debian being virtualized on Apple Silicon so no rocket science is involved...)

I think Apple's gamble is that, by 2023, anybody in the "must have Parallels/Bootcamp" boat will be some sort of IT professional/developer who won't balk at having two computers in the house... There will probably be that one Intel Mac that hangs around in Apple's range for a few years (...like the "classic" 13" MacBook with spinning rust). Or, if you are a "serious" gamer, you should know by now that Apple hates you.

In terms of "consumer" need for running Windows - I think that's what has changed since the Intel transition in 2006: then, being able to run Windows was a real selling point for all those work apps, websites that only worked in Internet Explorer, banking, personal finance apps etc. Today, outside of the corporate world, "requires MS Windows" is already being supplanted by "Available for iOS and Android" so losing x86 support for Windows is more than outweighed by gaining support for iOS apps (and, potentially, native virtualization of Android... although that's partly CPU agnostic anyway) - and give it another couple of years that will be even more marked.

Even with "I need Windows for work" - give it a couple of years and - even if you have an x86 PC - your only option (because "data protection") will be to use virtual desktop to connect to your machine at work (or, most likely, your virtual PC in the cloud which your employer has outsourced to MS Azure). Yea, verily, even if the current policy is "instant dismissal if your data touches the cloud" that will have been turned inside out as soon as the PTB realise how much cheaper it is to have Microsoft or Amazon tick all of your compliance boxes for you (not that 'no cloud but you can walk around with the data on your PC' is rational to start with).

Anyway, on to new Apple range prediction and - if it hasn't been said already:

(NB: - "AppleBook, iApple" etc. are just names I'm using to distinguish AS from Intel - I doubt they'll drop the "Mac" but they might change the "i" and "Book" bits)

Apple range by mid 2021:

13" AppleBook (AS - ultra thin)
14" AppleBook (AS +touchbar)
16" MacBook Pro (Intel - but only with the higher-end CPU/GPU options)
21" (or thereabouts) iApple (AS)
27" iMac (Intel - as announced)
Apple Mini (AS)
iMac Pro (Intel - no updates)
Mac Pro (Intel - maybe with minor bumps)
Mac Mini (buried in a link off the main Apple Mini page for developers needing to support x86)

...with each of the AS machines available in good/better/best configurations but all running basically the same SoC with the lower-end options being underclocked and maybe having some CPU/GPU cores disabled (maybe justified by lower yields of fully functional chips, maybe not).

Apple range mid 2022 -:

Standard range - all with the same AS SoC:
13" AppleBook
14" AppleBook (+touchbar)
21" (or thereabouts) iApple
Apple Mini (AS)

Pro range - with new "workstation class" AS SoC:
16" AppleBook Pro (Apple Silicon Pro)
27" (or 30") iApple Pro (Apple Silicon Pro)
Apple Mini Pro (Apple Silicon Pro)

Legacy Range:
Mac Pro (Intel - entry level now 12 core but otherwise unchanged - Apple Mini Pro is faster)
Mac Mini (while sticks last - buried in a link off the main Apple Mini page for developers needing to support x86)

Note: no 12" MacBook because you don't want that, you want an iPad. Yes, you do. Look into my eyes (not around my eyes, into my eyes) Tim says you want an iPad... but maybe the 2022 iPad Pro will support modern MacOS apps...

I'm in line with you that the 12" MacBook is silly and that the current Air is a perfectly suitable replacement for it. Also that Apple hates gamers. I do see that there is a Server variant of Ubuntu for ARM64, but currently, that's all they have. Get me an ARM64 variant of the non-server edition and then I'll be satisfied (as not every Ubuntu box I want to run should be a server and have server packages installed).

They could build one, sure, but there are numerous problems with that 128 core CPU that is described in the article you linked to. To name just two of them:

1.) Cost. That CPU is more than just high-end. If you look at the prices of 48 core Xeons, you will know what I mean. This is not built for consumer nor prosumer devices. Pricing of the final product would be even more ridiculous than the current Mac Pro. By a big margin.

2.) Amdahl's Law. Google it. It describes the concept of CPU core scaling. Once you reach a certain amount of cores, it becomes exponentially more difficult to scale software. In other words, a 128 core CPU isn't necessarily faster than a 64 core CPU. Most certainly not in the kind of software you and I use.

This stuff is incredibly complicated and it takes more than just slapping 100 cores onto a die.

I am not saying Apple won't get there, eventually. But 2 years seem overly optimistic. If not to say unrealistic.

There are many kinds of Xeons. Apple doesn't employ server variants. They use Workstation caliber Xeons. And yes, if the A12X from two years ago can best every Mac that isn't a 16" MacBook Pro, a 27" iMac, or an iMac Pro or Mac Pro, then a CPU that can best all of those Macs can certainly be ready by two years from now. Especially if you consider the rate at which their silicon is advancing. If you compare the A8X from the iPad Air 2 and the A12X from the 11" iPad Pro and third generation 12.9" iPad Pro, you'll see a huge jump. At that SAME rate, they can easily outperform the rest of the current crop of Intel Macs.

I agree that Apple is going to simplify their lineup to what this person said. I think Apple drops the MacBook Air for the MacBook and keeps the 13” MacBook Pro (Intel) and the latest 27” iMac (Intel) as options for legacy users.

I really hope that 24” iMac drops this fall. I am extremely tempted to grab the new 27” from this week, but I think I’m going to hold off and impatiently wait (let’s face it, who patiently waits for anything)!

Apple "dropping the MacBook Air for the MacBook" is a name change, not a simplification or a merger. Also, they learned their lesson in that the larger laptop is more important to be kept around for legacy purposes. It didn't make sense for them to do it with the 13" in 2012, so they did it with the 15" in 2016 and will likely do so again this time around.

I can't wait to see the Apple Silicon iMac, I have an iMac 27 "Late 2012 and I could also decide to give up a few inches of monitor from 27" to 24".

I wanted to focus on the GPU.

At WWDC 2020 we saw "Shadow of the Tomb Raider" on Rosetta 2, quite fluid and indeed even emulated.

These are the requirements posted on Steam:



The demonstration tells us that in emulation we have seen something as powerful as at least a 2015 Q3 GPU, I repeat at least. In your opinion, were they performances from AMD Radeon R9 M290 or from Intel Iris 540 (Q4 integrated GPU)?

I don't think that comparison means anything; they're not emulating specific video cards. They're taking Metal instructions in the apps and the Apple GPU is acting accordingly. That's why the performance looked so good in that demo. You do this with any non-metal game and it won't look as good.

You are making assumptions. Based on Apple's marketing.

How do you define "Xeon replacement ARM chips"? Are you comparing to low end, low energy, low-core Xeons? Are you comparing to HEDT Xeons like they are being used in the current Mac Pro? Are you comparing to AMD's Zen 4 Epycs that are going to hit the market next year (which you should)?

I am sure they are working on this in their R&D labs, but I am also 100% certain that Apple won't have an ARM chip that will outperform 2022 HEDT CPUs like Zen 4 - which is what they are going to have to compete against. Let's not forget that there are incredibly fast x86 CPUs out there, and they will only get better in the next two years, with major developments like AMD's Zen 4 architecture just around the corner.

I can only repeat myself here - Listen closely to what Apple has to say about the transition: "Intel Macs are here to stay for *many* years to come". With that in mind, I think people should start to separate wishful thinking from actual reality.

Again, you get it wrong. Rewatch the keynote. Intel Macs aren't here to stay for many years to come. Intel Macs are going to be supported for many years to come. This doesn't mean produced for many years to come (though I'm pretty sure they'll keep one laptop, again, likely a 16" MacBook Pro model, and one desktop, again, likely the Mac Pro, around for what could be considered many years; but they won't be updating them with newer Intel processors or AMD graphics; it'll be for assisting with the move to Apple Silicon). This means that Apple is committing to producing new versions of macOS past Big Sur that will also run on Intel Macs. That's it.

Any thought or worry that with the move to ARM and new form factor that memory will not be user upgradeable across the board - including the 27" iMac replacement?

Well, right now, the only Macs that are user-upgradable are the Mac mini, 27" iMac and the Mac Pro. You can remove/replace the RAM in the iMac Pro and 21.5" iMac, but that's not considered user replaceable, but rather technician replaceable given the process to do such a repair. So, really, we're talking about three models of Mac at best. I wouldn't be surprised to see removable RAM go away on newer iMacs. It might stay on the Mac mini and the Mac Pro. But I could also see it disappearing there too. There hasn't been anything announced in this regard as far as the architecture not supporting it. Removable/replacable storage is likely done and over with though; with the Mac Pro being an exception to the rule. That much certainly isn't changing on the laptops, sadly.

Very thoughtful read. Looking closely at the lowend iMac, I agree with you that Apple has set a low bar for themselves to clear with Apple Silicon. The Vega 20 is old. But I would also be surprised to see Apple launching two Mac APUs this year, so I'm mulling over whether or not the Bloomberg APU can accommodate both the MBP13(14) and the iMac.

I think it can.

The CPU is no problem. The MBP13 would reserve around 10W for its 8 core CPU, scaling back the amount of power it gives to cores in three stages (5W / 2; 2.5W / 4; 1.25W / 8) and hitting clocks of roughly 2 @ 3GHz, 4 @ 2.87GHz, and 8 @ 2.7GHz. The iMac would reserve around 40W and scale in two stages (10W / 4; 5W / 8) hitting clocks around 4 @ 3.1GHz and 8 @ 3GHz.

The GPU side is trickier. I do think the MBP13 can accommodate twice as many GPU cores as the iPad Pro, especially since it will be a full node ahead of that device. 16 GPU cores is a good start, and the iMac can encourage them to run at higher clocks than the MBP13. But the clencher would be to slap down a 4-8GB stack of HBM2 as cache right beside the iMac's APU.

So there you go - one APU die, two devices, configurable to outperform the high end iMac 21.

I thank you. And yeah, I think it will be easy to outperform the lower-end Macs (let alone the three 8th Gen Intel based Macs that Apple still sells. They will not have to work hard to best those Macs. The Air's 10th gen chip is already pretty slow, even for 10th gen, so that'll be easy as well.

Let's point out that Intel are discontinuing the Skylake Xeon CPUs in the current iMac Pro with orders being stopped in January 2021 before the final products are shipped in July 2021. Vega56/64 is also on the chopping block and might be unavailable sooner - AMD must be doing Apple a favour if they still keep making them until deep into next year.

And if you look at the 2013 Mac Pro as a template, they bumped the base spec model to match the upper SKU before discontinuing it shortly before the 2019 Mac Pro came out. I reckon the iMac Pro will go away when the Apple Silicon Mac Pro comes out.

I didn't realize that the CPUs for the iMac Pro were that close to discontinuation; but it makes perfect sense. So, the iMac Pro's days are numbered. Though I disagree that it will be kept around THAT long (the Mac Pro will take the longest to move over). Unless Apple is going to place a HUGE order in January to last it that much more time. But I don't see that happening either. I do think that the Apple Silicon replacement to the 27" iMac will be out by that deadline. I'm wondering if part of the idea behind bumping the 21.5" iMac to 24" is so that Apple can position the 27" iMac replacement as a more reasonably priced iMac Pro replacement rather than simply a higher-end consumer iMac. Though I guess that becomes a semantics argument more than anything else.

Good breakdown.
The one thing I would question is the stipulation that the new ARM chip has to be significantly faster than the previous Intel chip. I don't believe this to be true for laptops. I think if Apple can get an ARM chip out in a laptop that is equal to or just a little faster, but moderate to significantly cooler than Intel, that is all that matters.

IN saying that, I wouldn't be surprised if the 16 Arm isn't release until July-Oct next year. I hope not, but I'm prepping for that.

Apple needs to justify the architecture switch. Merely saying "lower wattage on notebooks" and "the ability to run iOS and iPadOS apps natively on macOS" isn't enough to justify moving away from Intel to tech journalists, if not enthusiasts, if not general consumers. They need to prove that this is the correct move to be making and performance improvements are the most universal way of doing so.

Given there are already 80 and 128-core ARM CPUs in production from Ampere, Marvell, AWS which roundly beat the best AMD Epyc & Xeon CPUs in the data center on compute/watt and are close or better in absolute performance, I don't think it is at all inconceivable that an ARM-based desktop CPU can beat the best workstation Xeon or even AMD Threadripper within a couple of years.

ARM is still not best in all workloads, and needs more development, but it is already competing and in some cases winning against AMD & Intel.

Go and do some research.

The important distinction to make here is that this isn't just ARM. This is Apple's implementation of ARM, which is already markedly superior in performance metrics to most other implementations of ARM.

What about GPUs? Does anyone think, due to the increased power savings, that they will be able to include dedicated graphics in the Arm MBP 13"?

You're thinking about this from the standpoint of it still being an Intel-based Mac. The Apple GPU doesn't even work in the same way that your typical AMD, NVIDIA, or even Intel integrated GPU does. The Apple GPUs work more efficiently, so while their on-paper specs won't be impressive compared to, say, the AMD Radeon Pro 5000 series that's present in the 16" MacBook Pro or the 27" iMac, it'll still match or exceed performance because it is running many times more efficiently by design. The caveat to that is that developers need to build around Metal and actually optimize for that kind of a GPU (also that this kind of way of designing GPUs and engineering GPU performance is not terribly common and that developers may find difficulty/annoyance in doing so that wouldn't otherwise be present with AMD GPUs because that mode of operation has been around much longer).

Apple has already stated that all GPUs in Apple Silicon Macs will be integrated.

Man, I really want some of that stuff y'all smokin.

Seriously, you guys need to take off your rose-tinted glasses. The above statement is beyond ridiculous.

Watch WWDC videos and do your research on what Apple is producing and how it benchmarks in the wild compared to other PCs. Also, you have to figure that Apple is not about to attempt to complete a transition (that includes its highest end Xeon-based Macs) if it can't actually deliver the goods. The last time they promised us a powerful Mac and couldn't deliver was when Steve Jobs promised a 3GHz Power Mac G5 (which ultimately prompted the move to Intel to begin with).

”...for years to come” I was wondering, what if Apple is not really abandoning Intel completely.

Would it be imposible to make a fusion computer with arm and x86 prosessors? There is the T2 chip for instance. The T2 carries out many of the tasks in a mac. Apple has been playing with this x86+arm fusion for over 5 years now. What if the T2 chip was not only for the known tasks, but also for developing a future Pro computer - the basic operating system would run from Apple silicon, but the pro user power would come from the Intel x86 prosessor. Think a hybrid car! This might be a way to offer pro users that extra speed they need until Apple has a equivalent arm chip to offer. Forgive me if this sounds imposible, I`m no computer engineer. I would like to know whatwould be the biggest problem with a design like that.

I had that thought prior to the announcement of the switch. That the Intel processor would stick around in higher-end models while the T2 would become a T3 all across the board (with the T3 replacing the Intel processor outright on anything that wasn't a 27" iMac, 16" MacBook Pro, or Mac Pro). But it seems they're ditching Intel altogether.

As for the "for years to come" bit, that really just refers to how long they'll keep allowing Intel Macs to run future releases of macOS. Nothing more. Eventually Apple will discontinue the last Intel Mac and every Mac in the lineup will be Apple Silicon based.

I wouldn’t be surprised if the 16” MacBook Pro gets updated to Comet Lake before making the jump to Apple Silicon.

I'm in agreement here and I meant to include that in my original post but didn't. The performance gains between it and the 9th Gen chips isn't substantial. But as a stop-gap to give them time (the way they clearly did with the 27" iMac this week) would make sense. I don't see them going past that though; I can't fathom that an Apple Silicon replacement to the 16" MacBook Pro will take more than a year from now to get ready and release.

I remember it was a big deal when everyone found out they'd been running it in secret.

Such a wonderful plot twist that was...you don't ordinarily expect such things as plot twists in Apple keynotes like that.