Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Technically MacOS on x86 isnt allowed to run on non-Apple hardware either, that’s the Hack part of Hackintosh

Yes, but once Intel CPUs are no longer supported, I could see Apple putting in validation requirements that those versions of macOS only run on an Apple Silicon SoC. And if those validations are handled by, say, the Secure Enclave, it might be too difficult to create a workaround.
 
I suspect that by then the Hackintosh community will move to commercial ARM machines for their Hackintoshes, they dont have to stick with x86 either as more things move to ARM
...probably won't work.

Intel Macs were, pretty much, PC clones with nicer trackpads, some of the legacy stuff omitted, hence Windows could run on Macs, and building a Hackintosh mainly meant choosing hardware with the same 3rd party chips for graphics, storage controllers, networking, sound etc. that Macs used (not to dismiss for one moment all the fine work the Hackintosh community did in making this as close as possible to a click-and-drool experience).

M1 Macs differ substantially from both PCs and other ARM-based systems. M1 may use the ARM instruction set, but Apple have added their own features so it's doubtful that MacOS will run at all on other ARM processors. In any case, MacOS on ARM will depend on Apple-written drivers for M1 graphics and the M1 SSD controller - let alone all the security stuff that depends on the M1's security features.

(I'm not sure that there is some all-encompassing hardware standard for ARM-based systems to compare to the 'x86 PC compatible' de-facto standard)

Even with Intel, reality is that the writing was on the wall for Hackintosh as soon as Apple started using the T1/T2 chip - once pre-T1 machines pass out of support, Apple will be free to remove all the x86 drivers for pre-T1, third-party controllers from future versions of MacOS and, potentially, lock down MacOS to require the security features in T1/T2. The "good" news is that Apple are still selling pre-T1 Intel Mac Minis so they won't be dropping pre-T1 x86 support just yet - but those old x86 drivers are probably never going to see ARM versions.
 
Technically MacOS on x86 isnt allowed to run on non-Apple hardware either, that’s the Hack part of Hackintosh
Practically, MacOS for Apple Silicon most likely won't run on non-Apple ARM chips, even if Apple never lift a finger to deliberately prevent it.
 
  • Like
Reactions: freedomlinux
Gonna be insane amounts of power per watt, however, how does Apple address the GPU given it’s all integrated? On a laptop, makes sense, but in a desktop?
This is the question that's burning for me. I use discreet GPUs for 3D professional work – the on chip graphics are children's toys compared to what I need.
 
I'm honestly not sure how much energy Apple will put in to stopping "workarounds", etc

They aren't losing money on the OS (since it's free), and it's not at all clear to me that folks still doing Hacks in this day and age are likely to suddenly go buy much more expensive and restrictive (the latter for sure) Apple hardware if their ability to Hack gets cut off.

I actually think the remaining Hack holdouts over the years ahead are just as likely to not even be macOS users in a few years if they head down the full iOS style lockdown path (or close).

I'm not even really sure where I myself stand on all that.
 
  • Like
Reactions: Snahbah
That's why I'm inclined to believe this neglected sentence from the article:
According to Gurman, the larger Mac Pro will continue to be equipped with Intel chips while the smaller model will adopt Apple silicon.
My guess is that the Apple Silicon Mac Pro will be Apple's second run at the Trashcan concept - that doesn't mean another triangular-cored cylinder, but still a compact desktop processor that relies on external Thunderbolt expansion, primarily designed as an "appliance" FCPX/Logic Pro and other software that can be customised for its unusual architecture. The big difference this time is that they've still got an up-to-date (and relatively easy to update), full PCIe tower system on the books for the people for whom the Trashcan model doesn't work.
Good points, I also really liked your points from a enterprise perspective of the risk of changing over processor platforms altogether with the expensive Mac Pros.
 
I'm honestly not sure how much energy Apple will put in to stopping "workarounds", etc
If you mean ARM-based Hackintosh they won't have to put any energy into stopping workarounds, but the Hackintosh community would likely have to put an enormous amount of effort to get MacOS to run on a non-Apple Silicon ARM chip, if it were possible at all.
 
I'm ammuning they smaller Mac Pro will have not PCI-e slots, which is a good way to slim it down. It will probably be a square version of the trash can design. Hopefully it isn't throttled down for the sake of heat issues like the trash can was.

One of several significant problems with the "not slots" position is that it runs 100% counter to Apple's move and focus on SSD drives as being the primary target storage media. Having one, and only one, internal drive doesn't really met the requirements of a lot of "high bandwidth" workloads. Apple even admitted as such in their April 2017 "dog ate my homework" session on why the MP 2013 had stumbled.

So admitting that and they hypocritically releasing yet another Mac Pro with exactly the same problem will get tons of blowback. Maybe not necessarily. AUX powered. , triple wide slots. But enough to stuff 2-4 M.2 SSDs inside on a x8 PCI-e v4 socket would be a basic minimal not to repeat the same mistake they clearly said they saw before. It doesn't have to provisioning for fire-breathing GPU cores, but there needs to be something.

And some 2.5" SATA 3 SSD really isn't a modern , high bandwidth load storage drive anymore. That's a secondary, maybe even back up drive.

AFPS is highly skewed to SSD properties. They've put macOS on a SSD path ... just one SSD for an iPhone/iPad is fine but in "Mac Pro" space that's lame. If Apple wants to do that call it something else. "Mac Max" or something.


And frankly "half size" means there is room for a card. The large hang up is that the M1 Max die doesn't provision for substantive PCI-e lanes. ( capped at relatively paltry for this class of system : x4 PCIe v4) They need another die. ( it could be about the same size die and swaps 3-4 Thunderbolt controllers for just a 1-2 x8 PCI-e v4 controllers. ). They can't just pour Thunderbolt ports on the system like ketchup. Past 4 ports probably has relatively low utility versus actually provisioning some internal PCI-e v4 (or better). Again Apple admitted they "leaned too hard on Thunderbolt" back in 2013. To do it again on the same product is a bozo move.

Apple made a big deal about being able to host 4-5 Pro Tools HDI cards in a Mac Pro 2019

pcie-card-diagram.jpg



power-plus-more-power-852x532-v2.jpg




They don't have to do 4-5 HDX cards, but they should be able to do at least one. It is not a high wattage or ultra bandwidth card but it is a more than half length one. It largely needs a full length, full height slot. It fills the slot bay in the picture above.

If can't do a "Mac Pro thing like HDX cards" then don't have a Mac Pro. Apple themselves set that baseline metric in their 2019 dog and pony show.

If Apple chucks that down the toilet then pushing groups of folks away. Probably on a "3rd strike".

At some point Apple has to make a die that is not just super optimized for laptops. If the MBP 16" is going to be their almost sole focus then they aren't going to hold onto what is left of the core Mac Pro user base.

Dumping video cards will not please the AMD and Nvidia fanboys (and Apple probably can live with that) , but all the cards is a bridge too far. Literally dozens of high utility cards that do not have 'crazy thermal drama' problems and tossing down them down the drain. ( also substantively helps the ecosystem for the TB expansion boxes for laptops and Minis also. )
 
@deconstruct60

Such a great post

Essentially, as you've so eloquently laid out there, there is no "Mac Pro" if we are just going to neuter it down to be another system on a chip with little to no expansion.

Being this different and oddly useful is the whole point of this particular product line (MP)
 
Not easily make with the same motherboard. This rack model is just rotated 90 and a new container built for essentially all the same internal components.

The hard part would be getting another motherboard "green lit" as a project with relatively small volume. The XServe disappeared because Jobs said "Nobody was buying". That is probably not literal, but on the weekly dashboard of Mac sales the Apple execs do if the pie chart slice so small you can't see it ... that amounts to "nobody". Apple isn't run like Dell/HP/Lenovo.

The Mac Pro 2009-2012 was "rack hostile" with the handles getting in the way of rotating and inserting into a standard width rack. This 2019 "rack option" is mostly just an option of a better optimized container for the horizontal position. Other than that it isn't trying to be a different product. There is a use-case where folks put Mac Pros into equipment bays , utility carts , etc. The racking here is just to enable that. (same target group that wants several cards , but mounted horizontal. Mostly this is not super hard core data center temple folks. More ruggedized Macs than trying to fit in next to the Mainframe. Similar for the sky high priced wheels for the tower version. Folks rolling it from place to place. )

Similarly, the Mac Pro 2009-2012 had a CPU tray not so much because Apple worships at the alter of ultra-modularity , but so that they could do a single socket and dual socket versions with one shared logic board and chassis. Apple didn't want to build multiples. Apple isn't out to build everything for everybody with dozens of Mac products to choose from.


macOS isn't trying to be a scale out pizza box operating system. So not sure who these 1U models would really be gear too. Likely some of the first complaints will be that it doesn't have dual , hot swappable power supplies. Can't native boot VMware VSphere/ESXi ... etc. etc.


As far as microserver density ....

image

https://www.macstadium.com/virtualization

23 Mini per shelf. 10 shelves ... 230 Macs per rack side. Double sides that 460. 1U pizza boxes isn't going to beat that. Granted a bit abnormal tall.
[ standard 42U rack would if had double Max count SoC , that would be 84 and two side-by-side => 168. Still several hundred of that 460 mark. Even at 3 x 42 = 126 1Us. ]

If put a M1 Max in each of those , then that 460 Max's in a rack. That is probably enough for a sizable number of folks and workloads.


If the "half sized" Mac Pro shaved off half the box height, then they could fix two 5U on custom slider mount that took two side by side. In a 42U rack, they could get up to get in 16 systems instead of 8. If had need for a set of systems with dual socket 40GbE and a large inerternal SSD those would probably be "good enough" for a decent number of folks.

As a 1U that is out to displace generic Supermicro , Inspur , Dell , HP 1U boxes running some non-MacOS hypervisor on bare-metal , I highly doubt Apple is interested in that.





the size of the "Mac Pro" box is probably more going to be driven by which and how many PCI-e cards they want to throw out the "supported" window. If it is 'all of them' then the that 3-stack-mini would be about the right size.
If throwing out true full length and aux powered cards then can shrink each dimension of the current MP 2019 case.

If they are trying to hold onto the full height , full length cards ( e.g. a large multiple M.2 SSD holder RAID card. ) that will be at least as large ( or bigger ) constraint that the cooling issue. Although 4 x 85W => 340W would not be a "laptop like" thermal problem.

Is there a gap between a Mac Pro and Mini for servers? Yes. Is it big enough for Apple to do yet another product? Probably not. Apple is not going to be poor house if they skip that product.

If Mac users are going to call out Intel for using Macs to create TV commercials, or Microsoft for using Macs to produce ads for Windows products, then the same standard should apply when Apple uses anything other than Macs running MacOS in any part of their business, including server rooms and datacenters. I doubt that Apple, with their security requirements and size of the company, just lets all their employees bring in their personal computers and connect to the internal company network. For companies that are interested in how to run their IT the way Apple does, Apple could be publishing whitepapers detailing how they manage user computers, user accounts, what server hardware and OS they use, network monitoring tools, etc. But instead, the best that Apple offers is a web page saying “buy Macs because they’re great”.
 
  • Love
Reactions: turbineseaplane
Well ML relying on CUDA.

An "M1 Max Quadro" will have (at least) 64 Neural Engines so if your ML workloads use Core ML or Create ML it should do quite well.
But CUDA is the "de facto" standard for machine learning.
I'm honestly not sure how much energy Apple will put in to stopping "workarounds", etc

They aren't losing money on the OS (since it's free), and it's not at all clear to me that folks still doing Hacks in this day and age are likely to suddenly go buy much more expensive and restrictive (the latter for sure) Apple hardware if their ability to Hack gets cut off.

I actually think the remaining Hack holdouts over the years ahead are just as likely to not even be macOS users in a few years if they head down the full iOS style lockdown path (or close).

I'm not even really sure where I myself stand on all that.
Guess you kind of underestimate all of this. Running on standard Intel CPUs with standard AMD GPUs is easy.

On a Hackintosh you do not have the integrated Apple GPU. A standard ARM CPU doesn’t have Apples secret instruction set. Apple CPU as well as Apple CPU is proprietary and not documented. People trying to run Linux on a M1 Apple facing similar prpblems and try to reengineer the GPU to write a driver.
And there will be no Metal driver from AMD, since Apple doesn‘t use AMD graphics anymore.

Apples ARM chips have special commands to improve FFT and other special operations.

just to name a few problems …

Face it, Hackintosh is dead as soon as x86 is no longer supported.
 
Last edited:
I pine for the days when the Mac Pro wasn't the sole domain of the money-is-no-object people.

Most people can't use more than 4 or 6 CPU cores, but want a configurably-beefy GPU. However, Apple wants to tie CPU and GPU performance together and has a skewed notion of the relative need for single threaded vs. multi-threaded vs. GPU performance.

The M1 Max is eye-wateringly expensive as it is, but only has the GPU performance of a low-end discrete card. 128 graphics cores would probably put you closer to the performance of a 3070 or 3080, but it looks like you'll have to shell out for 40 CPU cores in order to get it. Who the heck needs 40 CPU cores?? It's great for those five or ten people, but what about everyone else? A 40/128 core Apple Silicon computer would probably cost at least 4x the top-spec M1 Max MBP. That's just crazy if you just want a GPU that won't have all the $2000 PC users laughing at you.

Apple, do everyone a favor and make a version of the chip with low CPU cores and as many GPU cores as you can throw at it. 4/128 or 6/128 would probably be perfect for a lot of people. Even better, put the GPU on an actual card and then **shocker!** update it regularly so people can keep up with GPU advancements. Maybe spin off a focused Apple GPU division.
 
But CUDA is the "de facto" standard for machine learning.

Fair enough, but if your ML workflow is based on CUDA, you have not been using a Mac / macOS for a decade or longer so what a Mac Pro has for ML support is irrelevant to you. :)

Who it will be relevant to is those using Apple's ML language and tools (Core ML / Create ML).
 
Last edited:
Face it, Hackintosh is dead as soon as x86 is no longer supported.

I don’t dispute that entirely…

That said, there are many more years of runway still..

OpenCore is already enabling Hacks on hardware that wasn’t officially supported by Apple in the form of a released Mac.

Nothing is forever and I personally don’t worry about 3-4 years out on tech.
I’ll evaluate the landscape at that time
 


Since 2020, Apple has been working to eliminate Intel chips by transitioning the entire Mac lineup to Apple silicon chips, and 2022 is expected to be the year when the transition is completed. One of the major Mac lines still using Intel chips is the Mac Pro, but a refreshed model is in the works for 2022.

mac-pro-mini-feature.jpg

This guide covers everything that we know about Apple's plans for the updated Mac Pro.

Design

Apple is rumored to be working on two updates for the Mac Pro. The first machine is a direct successor to the 2019 Mac Pro with its modern, lattice design and modular casing, while the second is a new addition to the lineup that's smaller in size.

macproback.jpg

The larger Mac Pro is expected to continue to use the same stainless steel frame and aluminum housing with dual-sided logic board and easy access to the interior for adding and removing components. There's been no word on whether the thermal architecture will change, and it will still feature the same three-dimensional interlocking hemispheres for heat dissipation purposes.

As for the smaller Mac Pro, it's been described as looking similar to the existing Mac Pro, but with a more compact enclosure that's half the size. It will have a mostly aluminum exterior, and Bloomberg's Mark Gurman has said that it could "invoke nostalgia" for the Power Mac G4 Cube.

power-mac-g4-cube.jpg


Hit-and-miss leaker Jon Prosser has claimed that Apple's smaller Mac Pro could look like "three to four Mac minis stacked on top of one another" with a "compute unit on the bottom" and a "big heat sink on top."

jon-prosser-mac-pro-mini.jpg


Apple Silicon Chips

According to Gurman, the larger Mac Pro will continue to be equipped with Intel chips while the smaller model will adopt Apple silicon.

Apple is developing some super high powered Apple silicon chips for the Mac Pro, but the earliest versions may not be able to compete with Xeon processors for heavy duty workloads, and there may also be concerns about software compatibility. For that reason, we could see an Intel Mac Pro and an Apple silicon Mac Pro released side by side to meet all of the needs of professional users.

Apple-Silicon-Teal-Feature.jpg

The half-sized Mac Pro is expected to come with the "equivalent of either two or four M1 Max chips," which will make it much more powerful than the 2021 MacBook Pro models that include the M1 Max chips. The first Apple silicon chip is expected to feature 20 CPU cores and 64 graphics cores, while the second more powerful chip could feature 40 CPU cores and 128 graphics cores.

Back in November, The Information said that Apple would adopt a version of the M1 Max chip that has at least two dies to support a higher number of cores than are available with the standard M1 Max in the MacBook Pro, and that future versions would have up to four dies.

For the Intel-based Mac Pro, Apple could use the Intel Lake SP chips, which are Intel's third-generation Xeon Scalable processors. Signs of these chips were spotted in an Xcode 13 beta prior to the launch of macOS Monterey.

Release Date

The Mac Pro is expected at some point in 2022, and it could be introduced as early as June, perhaps at the Worldwide Developers Conference. Apple could preview the machine at the event and then release it later this fall, giving developers some time to prepare pro software for the more powerful Apple silicon chips.

Article Link: Apple's 2022 Mac Pro Refresh: What We Know So Far
Adding more cores only does so much depending on what is being run on the computer, how the OS is managing the processes, and how the chip is managing the cores. That's why in the current M1's hitting a point of diminishing returns for the average user. For pro-consumer work in audio and video they are close to hitting point of diminishing return. For pro-level work in video then more cores could benefit, but all depends on the management at both OS and chip levels. They are like the old mainframe computers with lots of processors, memory, fast buses, but those mainframe were running many app's at a time. Like an mainframe would be a waste for typical user so will be a huge SoC. So who does Apple see its future customer being and for what types of applications. All we know is AI is their big focus, what AI needs that much computing power? Interesting to think about.
 
They will find a way to **** it over, like no nvidia drivers or external video cards.....
soldered in cpu /ram/ ssd...

Just want something to replace the 2010, with intel/AMD so i can continue to run game servers & development...
Vmware/ k8
minimum 64gb ram, since 4 game servers is aready that....
 
  • Love
Reactions: turbineseaplane
Adding more cores only does so much depending on what is being run on the computer, how the OS is managing the processes, and how the chip is managing the cores. That's why in the current M1's hitting a point of diminishing returns for the average user. For pro-consumer work in audio and video they are close to hitting point of diminishing return. For pro-level work in video then more cores could benefit, but all depends on the management at both OS and chip levels. They are like the old mainframe computers with lots of processors, memory, fast buses, but those mainframe were running many app's at a time. Like an mainframe would be a waste for typical user so will be a huge SoC. So who does Apple see its future customer being and for what types of applications. All we know is AI is their big focus, what AI needs that much computing power? Interesting to think about.
Yep.. only you look at it as an investment.... but I think they are just too dumb...

I had exactly the same argument with Synology... fine you want to screw over the platform with hte 7.0 update
and remove things like GITLAB ..cos like most of your users don't use it.......in favor of some music *****, and totally destroy the A.D integration, so >300 users no longer sync correctly.
So F*** you no more muti-site & branch replacement orders for you

They fail to understand that the people that place the BIG corporate orders, in excess of a million, every 3 years
also have their requirements....
Said the same to Apple "corporate sales team" when they whined about why we no longer place orders for >100 macs
Simple... where is my replacement mac pro at a reasonable price, that I can swap pci cards & ram..
Oh and i still aint forgiven you for ****ing over Nvidia drivers after we spent tens of thousands on new cards, becasue you had a hissyfit over your own internal Q.A failure.
have these dumb asses ever tried to get an video processing card in a uni-body?

I don't want some all in 1 ***** for the corporate environment, where the CEO thinks they have the same ***** as a secretary cos the uni-body screen is the same.... nor is the CEO gonna sign off on a 7,000 USD single computer
 
  • Like
Reactions: turbineseaplane
Personally I dont see the definition of a Mac Pro to be one that can be accessed and things added in / taken away. I see that more as a hobby computer or a legacy device.

However, I have zero issues with a full / half sized pro that does provides PCIE 5 slots and allows dual full GPU's into it for example.

The thing is, I just dont see it happening. The only reason we have the current Mac Pro is because they couldn't build the device they wanted.

They can make a cube [or trashcan] now, and they will.
Not saying I entirely want this, but it is what I expect, and all evidence points towards it.
 
I pine for the days when the Mac Pro wasn't the sole domain of the money-is-no-object people.

Most people can't use more than 4 or 6 CPU cores, but want a configurably-beefy GPU. However, Apple wants to tie CPU and GPU performance together and has a skewed notion of the relative need for single threaded vs. multi-threaded vs. GPU performance.

The M1 Max is eye-wateringly expensive as it is, but only has the GPU performance of a low-end discrete card. 128 graphics cores would probably put you closer to the performance of a 3070 or 3080, but it looks like you'll have to shell out for 40 CPU cores in order to get it. Who the heck needs 40 CPU cores?? It's great for those five or ten people, but what about everyone else? A 40/128 core Apple Silicon computer would probably cost at least 4x the top-spec M1 Max MBP. That's just crazy if you just want a GPU that won't have all the $2000 PC users laughing at you.

Apple, do everyone a favor and make a version of the chip with low CPU cores and as many GPU cores as you can throw at it. 4/128 or 6/128 would probably be perfect for a lot of people. Even better, put the GPU on an actual card and then **shocker!** update it regularly so people can keep up with GPU advancements. Maybe spin off a focused Apple GPU division.
I have been thinking this for a while. They had the CEO of OTOY (Octane) featured on the MacBook Pro launch film. His product thrives on discrete graphics. The AS MBPs without eGPU support don’t make any sense for Octane. Octane starts to work well with two very powerful cards - better with four. So I couldn’t square his appearance with the M1 Max capabilities apart from thinking there must be some sort of Apple GPU solution in the pipeline.
 
Last edited:
One of several significant problems with the "not slots" position is that it runs 100% counter to Apple's move and focus on SSD drives as being the primary target storage media. Having one, and only one, internal drive doesn't really met the requirements of a lot of "high bandwidth" workloads. Apple even admitted as such in their April 2017 "dog ate my homework" session on why the MP 2013 had stumbled.

So admitting that and they hypocritically releasing yet another Mac Pro with exactly the same problem will get tons of blowback. Maybe not necessarily. AUX powered. , triple wide slots. But enough to stuff 2-4 M.2 SSDs inside on a x8 PCI-e v4 socket would be a basic minimal not to repeat the same mistake they clearly said they saw before. It doesn't have to provisioning for fire-breathing GPU cores, but there needs to be something.

And some 2.5" SATA 3 SSD really isn't a modern , high bandwidth load storage drive anymore. That's a secondary, maybe even back up drive.

AFPS is highly skewed to SSD properties. They've put macOS on a SSD path ... just one SSD for an iPhone/iPad is fine but in "Mac Pro" space that's lame. If Apple wants to do that call it something else. "Mac Max" or something.


And frankly "half size" means there is room for a card. The large hang up is that the M1 Max die doesn't provision for substantive PCI-e lanes. ( capped at relatively paltry for this class of system : x4 PCIe v4) They need another die. ( it could be about the same size die and swaps 3-4 Thunderbolt controllers for just a 1-2 x8 PCI-e v4 controllers. ). They can't just pour Thunderbolt ports on the system like ketchup. Past 4 ports probably has relatively low utility versus actually provisioning some internal PCI-e v4 (or better). Again Apple admitted they "leaned too hard on Thunderbolt" back in 2013. To do it again on the same product is a bozo move.

Apple made a big deal about being able to host 4-5 Pro Tools HDI cards in a Mac Pro 2019

pcie-card-diagram.jpg



power-plus-more-power-852x532-v2.jpg




They don't have to do 4-5 HDX cards, but they should be able to do at least one. It is not a high wattage or ultra bandwidth card but it is a more than half length one. It largely needs a full length, full height slot. It fills the slot bay in the picture above.

If can't do a "Mac Pro thing like HDX cards" then don't have a Mac Pro. Apple themselves set that baseline metric in their 2019 dog and pony show.

If Apple chucks that down the toilet then pushing groups of folks away. Probably on a "3rd strike".

At some point Apple has to make a die that is not just super optimized for laptops. If the MBP 16" is going to be their almost sole focus then they aren't going to hold onto what is left of the core Mac Pro user base.

Dumping video cards will not please the AMD and Nvidia fanboys (and Apple probably can live with that) , but all the cards is a bridge too far. Literally dozens of high utility cards that do not have 'crazy thermal drama' problems and tossing down them down the drain. ( also substantively helps the ecosystem for the TB expansion boxes for laptops and Minis also. )
Loved the post. However, since I’m no expert, I have a question on my mind regarding SoC: Why have “dedicated” computing power/cores in an unified architecture? Conceptually, wouldn’t it be more effective to have computing power assigned on task basis, which would adapt to workload, rather than have “pre-assign” cores? I see this “compartment computing” as a “legacy” from 3rd party modular components architecture, which unified architecture could avoid. Off course probably there may be tech constraints to prevent this that I’m uneducated about. Also wouldn’t the modular architecture be “highly “ bottlenecked by PCI-e v5 x16 (hopefully ?) bandwidth? 64GB/s vs 200-400GB/s?
 
Personally I dont see the definition of a Mac Pro to be one that can be accessed and things added in / taken away. I see that more as a hobby computer or a legacy device.

However, I have zero issues with a full / half sized pro that does provides PCIE 5 slots and allows dual full GPU's into it for example.

The thing is, I just dont see it happening. The only reason we have the current Mac Pro is because they couldn't build the device they wanted.

They can make a cube [or trashcan] now, and they will.
Not saying I entirely want this, but it is what I expect, and all evidence points towards it.
Hopefully they’ll do both: “mini” Mac Pro aka “the cube”; “regular” Mac Pro. Maybe modularity will be offered in a new concept, more “in sync” with SoC unified architecture…
 
Adding more cores only does so much depending on what is being run on the computer, how the OS is managing the processes, and how the chip is managing the cores. That's why in the current M1's hitting a point of diminishing returns for the average user. For pro-consumer work in audio and video they are close to hitting point of diminishing return. For pro-level work in video then more cores could benefit, but all depends on the management at both OS and chip levels. They are like the old mainframe computers with lots of processors, memory, fast buses, but those mainframe were running many app's at a time. Like an mainframe would be a waste for typical user so will be a huge SoC. So who does Apple see its future customer being and for what types of applications. All we know is AI is their big focus, what AI needs that much computing power? Interesting to think about.
I feel unified architecture presumes a new approach to computing power management. A way to this would be to let AI neural engine assign computing power according to momentum workloads.
 
A few people here still talking about M-series power per watt being a consideration. Sad truth is away from mobile systems like laptops and iPads/iPhones literally no one cares about that. What matters is overall power and flexibility. What we know so far is M-series devices have a very walled-off architecture with virtually no opportunity to upgrade literally anything, and even if Apple put a gate in the wall, currently nothing out there is compatible. All that is going to have to change.
 
  • Like
Reactions: Mr.PT
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.