Poll: When will there be an arm mac that exceeds a 2019 mac pro in performance?

[JMacHack]

Your entire belief that it will happen is that Apple wants to completely rearchitect cache, on-chip, and potentially external interconnects for a niche product that sells less units a year than iPhones in 1-2 days.

Particularly when multiple vendors sell the exact product you need.

Everyone arguing completely misestimates the cost of CPU engineering and how niche the Mac Pro is.

Rearchitecting is what they're gonna have to do anyways for the Mac line of chips. In fact, they said the Mac line of chips will be a new line in the keynote. Somehow I doubt, highly doubt, that after this the Mac Silicon will be beefed-up iPhone/iPad chips. They explicitly called it a "scalable architecture" in the keynote.

It isn't just the Mac Pro that's getting these new chips, it's the entire lineup, likely sharing many features as such. The cost is going to be spread out among the entire lineup.

Also, this is the same argument we heard before (and even after) the 7,1 launched. "Oh it's too niche Apple's gonna kill it off anyway!" The Mac Pro is a lynchpin in the lineup, no other mac has any PCI slots. After the whole 2017 "mea culpa" and reversal of the trashcan design you think they're gonna just drop it?

 

[benshive]

It's replacing Intel's chip so the real question is when an ARM chip from Apple will be more powerful than a 28-core Intel Xeon W. I have to imagine it'll be quite awhile. Especially considering the new Mac Pro just came out, they're probably going to focus on lower end devices like MacBooks first and work their way up to whenever the Mac Pro is redesigned again.

 

[konqerror]

It isn't just the Mac Pro that's getting these new chips, it's the entire lineup, likely sharing many features as such. The cost is going to be spread out among the entire lineup.

No it's not. Again, look at Intel Core vs Xeon today. Everything from the L2 cache and beyond was rearchitected. Everything is the same from your Core Y up to the desktop i9s. At that point, the architecture stops scaling.

Meanwhile, the Xeons burn about 100x more power idling than the Core Y. Not only that, memory access latency is significantly higher.

Computer architecture 101: making trade-offs.

 

[leman]

Your entire belief that it will happen is that Apple wants to completely rearchitect cache, on-chip, and potentially external interconnects for a niche product that sells less units a year than iPhones in 1-2 days.

Exactly. I firmly believe they would. Why? Because that would be the most impactful thing for the brand. Forget about money or difficulty. Think about psychology.

And the secondary reason: because there is no alternative. Your idea to license a HPC chip does not make any sense. You take about tradeoffs and yet you fail to show any understanding for the fact that a supercomputer and a hybrid workstation such as Mac Pro exist in different domains.

No it's not. Again, look at Intel Core vs Xeon today. Everything from the L2 cache and beyond was rearchitected. Everything is the same from your Core Y up to the desktop i9s. At that point, the architecture stops scaling.

Meanwhile, the Xeons burn about 100x more power idling than the Core Y. Not only that, memory access latency is significantly higher.

Computer architecture 101: making trade-offs.

Apple has something that Intel does not: experience with advanced power management. This is something you should not underestimate. There is no other company in the world that know more about this topic. The amount of data Apple has about running high-performance circuitry on low-power platforms is staggering. There is a good reason to assume that this knowledge is Apple's real secret weapon: they might just know how to scale stuff without it starting to fall apart.

And of course, according to the tech sites analysis, current iPhone's memory level parallelism is state of the art. So they probably have the good stuff there as well.

Somebody just denied the existence of AVX-512 on Intel processors.

Please leave your fantasies of what I might have denied or not to yourself (I assume you mean me since we were talking about vector units). This discussion is already more heated than it should be. Let's at least stay constructive.

 

[SegNerd]

Apple wouldn't introduce a new machine that is slower than the previous generation. It would defeat the purpose.

That’s really the question, isn’t it. I sure hope Apple wouldn’t do that, but since it has a vested interest in ARM CPUs, I don’t feel like it’s a sure thing.

 

[Waragainstsleep]

Everyone arguing completely misestimates the cost of CPU engineering and how niche the Mac Pro is.

Actually we've told you repeatedly that Apple can afford it and actively likes to invest in research and engineering. The have piles of cash they can't easily repatriate without a hug tax hit, they can likely spend it on talent or invest it in engineering work, parts, prototypes, components and materials without taking the same hit. Their accountants are as talented as their engineers.

Apple could try to license the tech from existing interconnects, design their own from scratch, reverse engineer one or more others or all of the above at once to see which option works out. They developed and test OS X on Intel for 5 years in case it became prudent to switch. They've been building their own CPUs for a decade now. You really think they haven't addressed this issue yet? If they aren't already testing early prototypes they are not far off by now. They wouldn't make that two year promise otherwise.

 

[psingh01]

They said the transition of their lineup will take two years. So in two years you’ll definitely have a Mac Pro with Apple Silicon. Maybe sooner. They wouldn’t have set that timeframe if they didn’t believe they could meet it and they aren’t going to replace the Mac Pro with something slower. They’ll want to showcase what they can do. Now they might not update the Mac Pro until then, so they’ll just have to beat a two year old system.

 

[dburkhanaev]

I still believe it’s possible that doesn’t mean a Mac Pro replacement in 2 years; it’s weird they released this halo piece for such a small window. Or maybe Mac Pro just stops being sold in two years and isn’t replaced for a period of time (it’s low enough volume and it’s not like Apple’s semantics are not run through PR filters)

Im as excited as anyone to see how thorough up and down the pure technical dominance is going to be (including their gpu) Usually there are trade offs somewhere? I’m glad Apple exist though.

During the last transition, the announced timeframe was the time Apple expected to ship their last PPC Mac. They did it in less than two years. I think they mean two years will be the end of introducing or shipping any Intel Macs. You’ll be on soft vintage mode at that point.
[automerge]1595908388[/automerge]

First, it's a flaw thinking that Apple doesn't outsource. The majority of Apple's chips today are licensed IP. Things like the video codecs, secure element, AOP, Lightning controller, even the Neural Engine is licensed from third parties.

Fundamentally the ARM architecture is licensed IP. Apple has to pay good money to use that.

Second, it costs a huge amount of money to engineer things. I think you don't understand how much. Half of the cost of each Intel chip is paying for engineering costs.

I simply cannot see Apple having the budget to engineer a server-class chip for a insignificant sliver of their sales, particularly when they can buy it.

Buy it from whom? TSMC will be building the Apple Silicon chips (whatever they‘ll be called). Apple is designing the chip. They’ve been building their own silicon for years. The iPad Pro processor is nearly Desktop class in some performance areas. And Apple licenses the ARM instruction set for the base architecture, but the chip is entirely Apple design. Why do you think there still isn’t a smartphone with a processor that can outclass an iPhone SoC? Because most ARM chips are the licensed ARM chip itself, from which others start with.

It’s why Apple moved completely to 64 bit. It’s why Snapdragon, good as some of them are, can’t hold a match to Apple A series chips. The fact is that Apple has already invested the engineering for years, only to be held up by Intel. I have my own reservations about the chip transition, but Apple is no slouch in the custom chip space And you better believe they will release Pro devices with pro power.

 

[vigilant]

This is two edged sword, true they can maximize design goals offering a paired OS to hardware, but at the same time, if Apple lacks competition that keeps pushing them forward you just end up what they want to sell as a computer.

One has to wonder if you will get what you really want in the end because perhaps Apple wants to take this decision away from you, referring to your early comment about being satisfied in a meaniful way. Apple has done this in the past, computer in a box with everything they think you need. Look at their design history.

I like it when Apple is having motivation for its users, not motivate for its shareholders. Their past keynotes are famous for everything sounds awesome, but is it?

So let me separate this into two very different topics.

Apple’s move to Intel was at the time was based off of the need of the users. Have they had issues with that under Intel and IBMs watch? Inarguably.

Looking at what they thought they would get from Intel with the release of the MacBook I think speaks largely to this. From that moment moving forward there was an expectation that chips that could be passively cooled for a entry level product was definitely a thing. Fast forward to today and we see rumors of Intel using TSMC for chip fabs, and the long timeline of things not going well with what they promised for years. To me it seems very clear that Apple’s bet on Intel was largely based on the ability to get scalable chips that would get more cores over a very predictable period of time. That didn’t happen. My MacBook Pro 16 is 10% faster than the Mac Mini in benchmarks. How does that happen. Seriously. Apple’s goal isn’t to make “the fastest thing for a given benchmark”. Their goal is to meet the needs of the people buying their products. I think the current Mac Pro is a concession that they can’t get there fast enough on Apple Silicon. It seems apparent to me that while the silicon won’t be the same the philosophy will be for the Mac Pro long term.

In terms of shareholders, I think Apple cares, but not enough to change their direction. Apple doing Apple things will result in people buying products. Period. Full stop. They tried doing that with Intel, and thermal constraints were an issue. Then they made the absolutely beastly Mac Pro.

The Apple Silicon thing is something that has the potential to give them a set of components that they can mix and match for the given use case. Will it be perfect? No, do you remember the Mac Pro Trashcan? But it gives them more control to adjust things in a way they can control.

A given Mac today with Intel has a long life. Longer than PowerPC had best I can wager. My expectation is that a similar “value system” not in terms of cost, but in terms of expectation will be in Apple Silicon Macs.

How long does an iPhone with Apple Silicon last in useful updates and usability compared to Android? I’d bet that there are actually more iPhones at least in the US that are old compared to Android. The ecosystem takes that into account.

 

[MyopicPaideia]

Just want to echo @dburkhanaev here in reiterating that there are absolutely ZERO off the shelf, third party outsourced design elements in the A series Apple Silicon at this point. Too lazy to go look it up at this point, but IIRC the A6 was the last chip that used off the shelf ARM cores or any other third party IP. Since the A7, (which took the entire industry by surprise by going 64 bit, btw) Apple‘s A series chips have been 100% in-house designs of the entire SoC. They even stopped outsourcing GPU architecture at that point.

I know this has been written and said way too many times already - and at least a lot of people are finally letting it sink in, but the ARM license is simply for the ISA (Instruction Set Architecture) - i.e. the lowest level, most fundamental programming language that tells the SoC what to do almost at the binary transistor level. It has absolutely nothing to do with the hardware design architecture of the SoC itself, any of the cores or any other elements of the SoC.

As far as this costing money, Apple bounces back and forth as the largest company in the world (use Forbes Global 2000 list from this past May as a reality check benchmark) that is not an financial institution or oil oligarchy. They are even in the top 10 if you do include those two categories! If anyone can afford it, they can. They also have been doing this for a decade now, and have invested heavily in this area already ($10.7B in 2019 - 7th highest in the world), and will be taking advantage of all that they have learned and accomplished. From my personal perspective, looking at it through the eyes of a finance executive, they have already been reaping a pretty darn acceptable return on investment I would have to say...why do you think iPad Pro’s and iPhones using the latest chips cost what they do? Exactly. AND - they are going to be able to leverage all the extra R&D they put towards the new Mac series of SoC’s to develop their A series chips to make them even better as well. Take a look at their profit margins - they have been not only high, but remarkably consistent over many years - it is a CFO’s wet dream!

Remember a couple of years back when Apple Execs were being all coy about having a product pipeline in the works that they were super excited about? I have a feeling this is what they were talking about. My hunch is that by the end of 2021 there is going to be some game changing stuff coming out of Apple‘s product line.

 

[vigilant]

Your information is outdated. ARM officially allowed it's users to implement custom ISA extensions last year. Apple uses it's own NPU IP since A12. Apple A12 CPU cores have also introduced specialized DSP hardware and SIMD instructions called AMX

Also, Apple implements ARMv8.4-A on their CPUs, Fujistu A64FX for example is ARMv8.2-A



Precisely. And given Apple's expertise in cache and interconnect designs, plus the fact that they sit on a large pile of cash, I see no reason why they can't pull it off.



Exactly. Same core. Different performance profile configuration. Intel server CPUs are configured to trade peak single-core performance for sustained parallel performance. The W-class CPUs Apple uses offer both high single-core peaks and good sustained performance. The ARM server CPUs you like to mention are optimized for parallel throughput. Different use case, different profile.



You are tunnel visioning this. I will repeat again: the Mac Pro itself does not have to be profitable. Apple can afford to subsidize the development of workstation-class chip from it's other profits. You are thinking in terms of pure momentary profits and ignoring the psychological or ideological factors. Apple will not announce a move to it's int-house designed chips and then buy someone else's hardware, especially one that does not support Apple's latest features and does not integrate well with their design vision.

This is one of the better posts I’ve seen, and was ready to say largely the same thing.

I’ll add my thoughts now.

Apple doesn’t need to change their attempts at Apple Silicon CPUs or GPUs for that matter to make Apple Silicon work.

The largest pieces that I do see they’ll need to work on is feeding high volume of cores in a way that makes it worth it, which doesn’t need to happen at the silicon raw cost level.

Apple could create a high core count SOC, and for low yields disable the cores that don’t make the mark and set firmware for the rest to hum along as expected.

Most ISA agreements I’ve read in the past year take into account custom instruction sets, I see this as a non-issue.

The investment Apple needs to make isn’t in overly scrutinizing small things on the per SOC level. It’s scrutinizing things on the per core level, whether that be image processing, neural engine, CPU or GPU. Scheduling and cache is a per product line. Everything else comes down to implementation and fab.

A 128 core SOC intended for a Mac Pro that isn’t high yield could be binned to be a killer 40 core iMac Pro.

Intel has been doing it for years.

Core CPU, GPU, and Neural Engine can be treated that way.

We know this, it’s been on the market for years.

 

[vigilant]

You're just spewing pseudo-technical BS at this point. I'm done.

You can rage quit, but Intel does reuse aggressively core design across their chip lines.

It’s a known fact

Binning process could arguably be different, but the actual core design is right.

Of course you can tell us after you’re done rage quitting about nuances, and cache design, but it doesn’t change anything.

Have fun, rage quit.

 

[jerryk]

Very soon after the 1st Apple Silicon MBP ships. Might even be the first Macbook Pro to ship with the new processor. The 15"/16" has often been the first one to get hardware updates like Retina displays, CPUs, etc.

 

[DanMan619]

Just going solely on them saying the transition will be complete in 2 years. I'd have to assume some ARM Mac Pro is coming at the tail end of 2022. I don't think they'd have verbally committed to that if what they have in the lab wasn't demonstrated to potentially be ready to go by the end of that time frame. I also am of the belief that others have mentioned here, that they wouldn't pull the trigger on this transition if what they had wasn't going to be at the very least equal, if not better then the current Mac Pro. Assuming they stick to the schedule, i honestly don't see it being an impossibility that they release that Mac Pro late 2022. I don't think they'd need to artificially wait to release it in terms of the 2019 Mac Pro still being new either.

Contrary to the potentially popular opinion. Releasing it that soon after the 2019 Mac Pro wouldn't really affect anything in terms of making it "obsolete" or enraging 2019 Mac Pro buyers. By the time the ARM Mac Pro would theoretically be out in late 2022, the 2019 Mac Pro will be 3 years old. Which should be enough time for people to not feel upset like people do when Apple release a spec bumped Macbook Pro 6 months after the prior one. And Apple has also already made it clear there will be overlap in this transition with another few Intel Macs still coming co-existing alongside the ARM versions. I think having that couple year overlap with the 2019 Intel Mac Pro and the possible 2022 ARM Mac Pro makes a lot of sense. There's no real reason for them to wait till say 2026 to release an ARM Mac Pro.

 

[jinnyman]

Well Apple has tons of cash alright.
But we are yet to see their dedication in MP grade niche market as clearly seen how long Apple took to introduce 7,1 since 6,1.

I'm certain Apple will eventually surpass current 10 cores intel chips or more, but I seriously doubt Apple is planning to compete in EPYC and Xeon category. Based on Apple's Business model, it's too expensive and too niche.

I think Apple may come up with something that compete in 10core category with specialized units so that their chip actually excels at specific tasks such as FCX and call it a day.

 

[theorist9]

I suspect that pro users who've bought the Intel Mac Pro have not been made unhappy with their decision by Apple's transition to AS. The machine meets their performance needs, and runs the software they need to run. In particular, I'm sure they are happy to have a machine they know works with their software during the transition/shakeout period. By the time they need to upgrade its processors/GPUs (say, four or five years after purchase), pro application developers will have had time to reoptimize their software for AS and work out the major kinks.

The one thing that would make Mac Pro customers unhappy is if, at that time, Apple doesn't allow these highly modular machines to be upgraded to AS, by not offering AS logic boards and AS-compatible I/O modules for the Mac Pro. That would be a bad look for Apple, since it would betray the two key messages they've linked to this machine:

1) It's highly modular and upgradeable (indeed, a lot of what you are paying for is the box), so you will be able to keep this machine current for years to come.

2) We really care about pro users; we're not going to ignore your needs like we did in the past.

It would also not be very green for Apple to turn such a large hunk of metal obsolete so quickly.

 

[Boil]

I suspect that pro users who've bought the Intel Mac Pro have not been made unhappy with their decision by Apple's transition to AS. The machine meets their performance needs, and runs the software they need to run. In particular, I'm sure they are happy to have a machine they know works with their software during the transition/shakeout period. By the time they need to upgrade its processors/GPUs (say, four or five years after purchase), pro application developers will have had time to reoptimize their software for AS and work out the major kinks.

The one thing that would make Mac Pro customers unhappy is if, at that time, Apple doesn't allow these highly modular machines to be upgraded to AS, by not offering AS motherboards and AS-compatible I/O and RAM modules for the Mac Pro. That would be a bad look for Apple, since it would betray the two key messages they've linked to this machine:

1) It's highly modular and upgradeable (indeed, a lot of what you are paying for is the box), so you will be able to keep this machine current for years to come.

2) We really care about pro users; we're not going to ignore your needs like we did in the past.

It would also not be very green for Apple to turn such a large hunk of metal obsolete so quickly.

Apple is not going to offer ASi upgrades for the 2019 Mac Pro.

Apple has never offered logic board upgrade paths (nor any other manufacturer that I can recall). If you want to place new mobos in existing cases, you are in Windows DIY land...

Apple wants to sell you a new ASi Mac Pro (Cube).

The modular concept of the 2019 Mac Pro is more about being able to add RAM, upgrade your SSDs (at the Apple Store at Apple prices), add additional spinning platters of rust, & to add capabilities via the assorted PCIe / MPX slots.

As for turning "such a large hunk of metal obsolete", a move to ASi does not stop macOS from running on that Cheesegrater 2.0, nor does it stop one from running Windows via Boot Camp until the machine dies...

I fear your theory is wrong... ;^p

 

[theorist9]

Apple is not going to offer ASi upgrades for the 2019 Mac Pro.

Apple has never offered logic board upgrade paths (nor any other manufacturer that I can recall). If you want to place new mobos in existing cases, you are in Windows DIY land...

Apple wants to sell you a new ASi Mac Pro (Cube).

The modular concept of the 2019 Mac Pro is more about being able to add RAM, upgrade your SSDs (at the Apple Store at Apple prices), add additional spinning platters of rust, & to add capabilities via the assorted PCIe / MPX slots.

As for turning "such a large hunk of metal obsolete", a move to ASi does not stop macOS from running on that Cheesegrater 2.0, nor does it stop one from running Windows via Boot Camp until the machine dies...

I fear your theory is wrong... ;^p

Of course we're all speculating here, and you may be right about what Apple will do.

But: I would disagree with your characterization of the Mac Pro's modular concept, which makes it sound like it's about being able to add more stuff rather than upgrading the capabilities of core components. Indeed, Apple explained that what motivated their new Mac Pro design was the central error in their old design: The GPUs weren't upgradeable because the old MacPro's thermal solution couldn't handle the new high-TDP models. Thus a key design consideration in the Mac Pro was to allow upgradability of core components, not merely expandability.

Also, why not allow an upgradable logic board? I could see it not being possible if the AS Mac Pro uses iGPUs rather than dGPUs, since in that case the Mac Pro's configuration wouldn't work. But if Apple decides to keep its AS GPUs physically separate, such that they could slot into the current spaces, what's the technical barrier with offering a replacement logic board?

And if there is no technical barrier, why not do it? The only argument then would be economic -- that Apple will want to force people to buy an entirely new machine. But Apple doesn't make much money on the Mac Pro -- it's more of a halo product that is important to keep its core pro users happy. And if that's its purpose, then allowing upgrades (assuming they are technically feasible) makes more sense than not.

As far as no one else doing it, or Apple never having done it before, that's not by itself a reason for Apple not to do it now. They are, after all, supposed to be innovators.

 

[Boil]

Of course we're all speculating here, and you may be right about what Apple will do.

But: I would disagree with your characterization of the Mac Pro's modular concept, which makes it sound like it's about being able to add more stuff rather than upgrading the capabilities of core components. Indeed, Apple explained that what motivated their new Mac Pro design was the central error in their old design: The GPUs weren't upgradeable because the old MacPro's thermal solution couldn't handle the new high-TDP models. Thus a key design consideration in the Mac Pro was to allow upgradability of core components, not merely expandability.

Also, why not allow an upgradable logic board? I could see it not being possible if the AS Mac Pro uses iGPUs rather than dGPUs, since in that case the Mac Pro's configuration wouldn't work. But if Apple decides to keep its AS GPUs physically separate, such that they could slot into the current spaces, what's the technical barrier with offering a replacement logic board?

And if there is no technical barrier, why not do it? The only argument then would be economic -- that Apple will want to force people to buy an entirely new machine. But Apple doesn't make much money on the Mac Pro -- it's more of a halo product that is important to keep its core pro users happy. And if that's its purpose, then allowing upgrades (assuming they are technically feasible) makes more sense than not.

As far as no one else doing it, or Apple never having done it before, that's not by itself a reason for Apple not to do it now. They are, after all, supposed to be innovators.

Because it is just bad business. Apple is in the business of making money by selling computers & the entire ecosystem / lifestyle that is associated with Apple. And who is to say that an ASi Mac Pro NEEDS to be in such a huge chassis?

To add:

I really feel that the transition to ASi Macs needs all new chassis / designs to accentuate the transition. Unfortunately that means the death of the Cheesegrater 2.0. But who knows...?!?

 

[theorist9]

Because it is just bad business. Apple is in the business of making money by selling computers & the entire ecosystem / lifestyle that is associated with Apple.

That's not actually an argument -- it doesn't address any of the economic and marketing points I made. If they could make it upgradeable, I think that would be better business, for the reasons I gave above.

And who is to say that an ASi Mac Pro NEEDS to be in such a huge chassis?

That's a very good point -- if the new machine has far less TDP than the old one then, in that instance, Apple certainly would want to switch to an entirely different case design.

But who knows...?!?

Indeed

 

[Waragainstsleep]

Also, why not allow an upgradable logic board? I could see it not being possible if the AS Mac Pro uses iGPUs rather than dGPUs, since in that case the Mac Pro's configuration wouldn't work. But if Apple decides to keep its AS GPUs physically separate, such that they could slot into the current spaces, what's the technical barrier with offering a replacement logic board?

And if there is no technical barrier, why not do it? The only argument then would be economic -- that Apple will want to force people to buy an entirely new machine. But Apple doesn't make much money on the Mac Pro -- it's more of a halo product that is important to keep its core pro users happy. And if that's its purpose, then allowing upgrades (assuming they are technically feasible) makes more sense than not.

If they could make it upgradeable, I think that would be better business, for the reasons I gave above.

Its obviously technically possible. Apple doesn't really do upgrades and never really has. They used to sell Apple branded memory a long time ago but that was always it. Swapping GPUs in a Mac Pro is barely more advanced than swapping USB peripherals.

Apple has never allowed for upgrades to hard drives, optical drives, CPUs, PSUs, wireless cards or displays even when the Mac was being serviced anyway. They certainly never allowed logic board upgrades.
When an Apple tech replaces a logic board, they always had to match a code on the old board when ordering the new one. Same for many other parts actually.
Apple has always been unbelievably strict about things like this. The reason they would give you (if they gave you any) is about compatibility. Board XYZ has been tested with HDD ABC and ODD PRQ and LCD EFG but not with HDD 123 or ODD 789 etc etc. The pinouts are the same, the specifications should be the same (sometimes LCDs needed the right display cable actually) and as far as most people know stuff like hard drives are universal (ask someone who looks after RAID units and you might get a different answer) and the truth is that if you swap in a different one it will almost certainly work but if Apple hasn't tested it to their stringent standards, they won't allow it.
Sometimes parts with different codes were supplied by Apple due to stock levels but you can be sure they had tested the part with that code with all the other parts with other codes to make sure the unit was going to run the way they intended afterwards.

In a way its good, diligent engineering practice. In another way its big company bureaucracy gone completely mad. Most consumers who bother to think about it go with the latter, I won't speak for the engineers but I think most of them know there is an element of truth to the former, even if they prefer not to admit it.

Apple testified in a US court not so long ago that they don't make any money from aftermarket repairs. This is an interesting claim. Their prices are typically cheaper than their AASPs but still not cheap for a logic board or display assembly. It could be the way they do their accounts and the paid repairs are lumped in with the free warranty ones in which case, they would for sure be a big loss.
If repairs in general were a loss for Apple then it would indeed make sense to make them easier to work on wouldn't it? More modular for one thing, fewer custom screen types, fewer different screws. Just 3 different screws across the ranges: small medium large. Theres a dozen different screws in any iPhone, there's no need for that. Less glue would be nice. Reduces Apple's labour time/costs, reduces parts costs if you swap out RAM or SSD instead of a whole logic board which has to go back to be analysed and re-engineered and re-tested in China.

It has to be about reducing the lifespan of the devices to increase turnover of new ones. Its a habit I'd love to see Apple break. Given the portion of revenue the Macs account for, a holistic philosophy where they build them to last and be easier to keep going would be really nice. Its environmentally sound and would incentivise them to make bigger improvements to tempt people to upgrade. They could at least offer better trade in prices on older Macs (with a receipt) to encourage upgrades and recycling. Its not like they don't have the margins to do that.

Would it really make better business sense (by which we mean better profits) though? I'm not sure it would sadly.

 

[Waragainstsleep]

I got rather sidetracked.
I wonder how many families of Mac chips Apple plans to use.
I can see a MacBook Air equivalent not really needing its own but just using whatever the iPad Pro has with some extra RAM thrown in. Like the DTK.
The MacBook Pro range could be one family with the model differentiator being GPU size (perhaps moreso than clock speeds)
The iMacs could use the same as the MBPs, which I expect they will if they get much thinner, but if they keep a significantly higher cooling capacity might get their own family of chips. Or just allow an extra option at the top end.
The Mac Pro is clearly going to need its own. Loads of cores, monster GPU plus ability to play nicely with other GPUS or PCI-E extras.

I wonder if the AS iMacs will have any RAM slots. I'm guessing not.
The Mac Pros surely will though, I can't see Apple soldering 3TB of RAM to anything.

I'm guessing the iMac Pro was always going to be a one-off had Apple stuck to Intel. Now I'm wondering if their product matrix might blur a lot of lines.
A MacBook Air with an MBP spec GPU for better gaming?
An iMac with an iPad level chip for basic office use?
A Mac (Super) Mini with an iPad chip but an iMac GPU for use as a console?

 

[JMacHack]

I got rather sidetracked.
I wonder how many families of Mac chips Apple plans to use.
I can see a MacBook Air equivalent not really needing its own but just using whatever the iPad Pro has with some extra RAM thrown in. Like the DTK.
The MacBook Pro range could be one family with the model differentiator being GPU size (perhaps moreso than clock speeds)
The iMacs could use the same as the MBPs, which I expect they will if they get much thinner, but if they keep a significantly higher cooling capacity might get their own family of chips. Or just allow an extra option at the top end.
The Mac Pro is clearly going to need its own. Loads of cores, monster GPU plus ability to play nicely with other GPUS or PCI-E extras.

I wonder if the AS iMacs will have any RAM slots. I'm guessing not.
The Mac Pros surely will though, I can't see Apple soldering 3TB of RAM to anything.

I'm guessing the iMac Pro was always going to be a one-off had Apple stuck to Intel. Now I'm wondering if their product matrix might blur a lot of lines.
A MacBook Air with an MBP spec GPU for better gaming?
An iMac with an iPad level chip for basic office use?
A Mac (Super) Mini with an iPad chip but an iMac GPU for use as a console?

If I had to take a crack at it, I'd guess we're going to see no less than two, but no more than four chip lines.
Line 1: "Consumer"
iGPU, fewer cores, smaller amounts of cache and gpu cores. Less die space allocated to other ASICs like the Neural Engine. Probably show up in the MBA, 14" MBP, 24" iMac, and Mac Mini.

Line 2: "Pro"
No GPU or beefed up GPU (depending on the dedicated GPU situation I'm leaning toward dedicated GPU though), more cores, larger amounts of everything. For the 16" MBP, 27" iMac, and Mac Pro (I don't think the iMac Pro will live to see Apple Silicon)

Potential lines:
Line 3: "Consumer+"
Similar to Line 1 but with a little more "oomph" probably higher-binned "Consumer" chips.

Line4: "Pro+"
Mac Pro/top iMac only. Ridiculous amounts of cores, cache, ASICs for the Neural Engine (I'm thinking the Neural Engine is Apple's big "pro line" strategy)

Frankly, I don't see Apple discontinuing the Mac Pro. But if they don't they're going to need CPUs capable of high core counts (at least 28/56) and able to address at least 1.5TB of RAM to keep parity with the current top of the line Mac Pro. And I doubt that they're gonna put chips like this into say, the Macbook Air or even the lower end "Pro" machines.

I do see Apple trying to use a few different lines of chips as possible, but I don't see a feasible way to do this without two, maybe three different lines of CPUs. Potentially more, considering how many different lines of Intel chips they use now.

Well Apple has tons of cash alright.
But we are yet to see their dedication in MP grade niche market as clearly seen how long Apple took to introduce 7,1 since 6,1.

I'm certain Apple will eventually surpass current 10 cores intel chips or more, but I seriously doubt Apple is planning to compete in EPYC and Xeon category. Based on Apple's Business model, it's too expensive and too niche.

I think Apple may come up with something that compete in 10core category with specialized units so that their chip actually excels at specific tasks such as FCX and call it a day.

Fair point, letting the 6,1 languish was an awful mark on Apple's reputation and the 7,1 is only a year old. We don't have any indication of their long-term dedication to the "Pro Market" I'll give you that.

However, just peacing out of the Mac Pro segment after doing a giant "Mea Culpa" and creating a team to assess "pro" needs? I just can't see that as likely. The "too expensive and too niche" argument was made for the Intel 7,1, and that actually happened, so why not for the ASi 8,1?

Honestly I think the Xeon/EPYC workstation line fits better for ASi than the consumer i7/i9 series. Lots of cores, lower clock speeds (2.0-3.5GHz range that is), more cache seems to be Apple's forte given the current iPhone and iPad chips. Remove the video encoders and GPU, allow more space for other things and you've got yourself an excellent workstation CPU. If Apple's IPC is as far ahead of Intel's then they could theoretically clock to something like 1.8GHz and match the lower-end Xeons. That's well within the range of even the passively-cooled ASi.

 

[dburkhanaev]

Just going solely on them saying the transition will be complete in 2 years. I'd have to assume some ARM Mac Pro is coming at the tail end of 2022. I don't think they'd have verbally committed to that if what they have in the lab wasn't demonstrated to potentially be ready to go by the end of that time frame. I also am of the belief that others have mentioned here, that they wouldn't pull the trigger on this transition if what they had wasn't going to be at the very least equal, if not better then the current Mac Pro. Assuming they stick to the schedule, i honestly don't see it being an impossibility that they release that Mac Pro late 2022. I don't think they'd need to artificially wait to release it in terms of the 2019 Mac Pro still being new either.

Contrary to the potentially popular opinion. Releasing it that soon after the 2019 Mac Pro wouldn't really affect anything in terms of making it "obsolete" or enraging 2019 Mac Pro buyers. By the time the ARM Mac Pro would theoretically be out in late 2022, the 2019 Mac Pro will be 3 years old. Which should be enough time for people to not feel upset like people do when Apple release a spec bumped Macbook Pro 6 months after the prior one. And Apple has also already made it clear there will be overlap in this transition with another few Intel Macs still coming co-existing alongside the ARM versions. I think having that couple year overlap with the 2019 Intel Mac Pro and the possible 2022 ARM Mac Pro makes a lot of sense. There's no real reason for them to wait till say 2026 to release an ARM Mac Pro.

I think you’re probably right, if the premise is that Apple won’t have an analog to every product on the Line-up that exists now. I think that might even be likely. The only thing that I have to add about the two year transition comments made by Apple is that this is their timetable to be out of Intel products, and Apple has a history of making it ahead of transition timelines.

while there may not be a Mac Pro on AS in 2021, we won’t see them ship any more Intel Macs after the two year window. What could be seen is an improved Intel Mac Pro just at the tail end of the transition. That is what happened with the PPC-Intel transition by the way. In August, at the tail end of the PPC transition, Apple released an updated quad-core PPC PowerMac. Then the Intel Macs shipped exclusively thereafter.

I think that they want a cut-off date, because there is likely a time frame for how long they want to release MacOS to surviving Intel Macs and-a cut-off date for when security patches will cease to release for the last binary version of MacOS. Plus they will likely support hardware and repair for about 7 years after that cutoff. Apple tends to like for these transitions to be clean-cut.

 

[Waragainstsleep]

I find it weird they plan to ship new Intel Macs and the first AS Macs this year. To me it would have made sense to ship the last Intel Macs first, then ship the first AS Macs with redesigns where they plan to redesign them. I can see them losing a few sales on the upcoming Intel iMac if they release a crazy good MacBook/Air/Pro soon afterwards.

I notice Kuo is sticking to his 13" MBP will be first guns but assures us there will be a 14" model. this to me makes little sense.
As far as I can tell, he gets his clues from other suppliers, LCD sizes are a good clue from Samsung or LG, there was a rumour about a MacBook Air based on a battery order recently, I'm sure there are others. I don't think (m)any of these rumours are coming out of Foxconn and this means that there is some guesswork. It may be that this new 13" laptop rumour is based on the screen order but it could be another Intel model or it could be a MacBook Air. I'm not sure what he could have heard that would tell them the difference.
A 13" Air and then a 14" AS MBP would make more sense to me. The Air can be crazy thin and sexy and thus get away without being the absolute spec monster we are hoping for, its also most likely to be ready to go since it doesn't need tone massively different inside to an iPad Pro. The the 14" and 16" MBPs can be next with much more impressive CPUs.