MP 7,1 So...Who Ordered a New Mac Pro?

[deconstruct60]

I am using the R4i right now, and I like it overall. It is a little noisy, but otherwise it allows me to have a large amount of relatively fast storage within the Mac Pro itself, minimizing clutter outside.

It does fit in nicely with the Mac Pro MPX modules - I just wish it was starting that could be passed along to the new Mac Pro, and sadly it can't because of the MPX design it seems.

I outlined some more modern alternatives to the R4i in this post in another thread.

https://forums.macrumors.com/threads/2023-mac-pro-merged.2391616/page-16?post=32236584#post-32236584

For the folks with a sunk cost in the R4i , I'm sympathetic on why they would want to continue. But the headwinds upon the product are very high at this point. SSD prices in 2023 are just not what they were in 2017.

This new Sonnet product with 8 m.2 SSD slots if you throw 'somewhat dated' , PCI-e v3 4TB drives at it has same raw capacity of the R4i (32TB) at about the exact same price point.

https://www.sonnettech.com/product/m2-8x4-pcie-card/overview.html

(e.g., RAID 1+0 pair two 4's for a '4' and than stripe 4 '4''s for 16TB ). Much less noise . One slot width. One 6-pin AUX cable. Apple is probably a bigger fan of that than of the R4i at this point.

If there are 8 or more E-cores then one E-core could just run the software RAID subsystem all the time with negligible impact on the rest of the system. (probably with enough spare time for other background tasks also).

R4i was likely looked on at Apple as a transitional tool/crutch like Rosetta. Add another 3-4 years and then ask folks would give up on HDDs again ( and see how many more folks they get to take that option).

One big 20TB HDD (with NAND metadata) is still hard to beat , $/TB capacity wise, as an internal Time Machine backup target though.

Western Digital Red Pro 20TB HDD Review: OptiNAND on a Platter

WD’s own approach to high-capacity NAS HDDs

www.tomshardware.com

( Even there though ... NAND consuming some of the data is still coming.)

So J2i is still alive.

 

[deconstruct60]

I am currently 🤣 @ 6 people

Apple isn't. And since they are ones who set the R&D agenda and funds resources ... they get the last word.

A lot of the folks who are in the "you can pry my 5,1 from my cold dead fingers"... I don't think apple is loosing sleep over at all. Those folks didn't buy the 7,1 (2019 ) model either. It isn't about buy something new that Apple offers with them.

Getting folks to move off of the 7,1 after only a couple of years I highly doubt Apple had any high expectations in doing.

Mainly this is aimed at 6,1 and iMac Pro folks. Some 5,1 folks who just have to move on inside the Mac space. And folks on 7,1 who found Afterburner far , far too limiting.

It isn't a massively large group. Doesn't need to be.

 

[salamanderjuice]

I saw a YT video (I forget which) where the guy made an argument I hadn't considered:

-The 2023 Mac Pro is not a prosumer tower, there isn't one as of now
-It's a specific product for people who need the slots
-It's actually a great deal for those people, because a fully tricked-out 2023 is like 12K, compared to 60 something for the intel one

In the end, it's a slot box for media houses. Specialty product and nothing more.

Except a fully tricked out 2019 MP isn't really comparable to the 2023 one. 60k got you 1.5TB of RAM, 12k gets you 192GB. You'd need to buy 8 2023 MPs to get the same amount of RAM and then you have the problem that its 8 seperate computers.

 

[flowrider]

Ya know, the survey leaves out a very important group of folks. Those who have bought something else🧐

Lou

 

[deconstruct60]

Ya know, the survey leaves out a very important group of folks. Those who have bought something else🧐

Lou

I think most folks are treating the last option as “ I did not order a new Mac system”. But , yes it isn’t going to put a number on the folks exiting the whole ecosystem .

However, also not going to measure the new folks coming into the system either ( especially down in the deep recesses of a Mac sub forum ). Apple doesn’t need 99% of Windows users to move over , but 0.5-1% could replace a sizable fraction of those leaving . ( it is a much larger pool so 1% is relatively a different number in the Mac space. ).

 

[m1maverick]

I think most folks are treating the last option as “ I did not order a new Mac system”. But , yes it isn’t going to put a number on the folks exiting the whole ecosystem .

However, also not going to measure the new folks coming into the system either ( especially down in the deep recesses of a Mac sub forum ). Apple doesn’t need 99% of Windows users to move over , but 0.5-1% could replace a sizable fraction of those leaving . ( it is a much larger pool so 1% is relatively a different number in the Mac space. ).

I'll be interested to see how many Windows users continue to migrate to Macintosh now that it no longer natively supports x64 software. Along with improved performance x86 / x64 brought with it native support for Windows making trying out the Macintosh a safe bet.

I attend training events which require virtualization and AS Macs are specifically called out as not being suitable for class. I wonder if this will manifest itself in other areas where native Windows support is desired.

 

[basehead617]

What name could mean more than "Ultra"? Obvious: they could take the chip to the absolute maximum and hence call it the "Max"!!!...

hey, hang on a minute... 🤣

it was already teased in rumors as M2 Extreme (2 x Ultras)

 

[davidec]

Hard to justify given the 30K I outlaid on the 7,1. It's a great machine and works for me in a high end audio studio on big projects. I'm very hopeful Apple do a silent update on a 9,1 M3 in the first half of '25. If so I'll outlay another 30K. 🙏

 

[deconstruct60]

I'll be interested to see how many Windows users continue to migrate to Macintosh now that it no longer natively supports x64 software. Along with improved performance x86 / x64 brought with it native support for Windows making trying out the Macintosh a safe bet.

Vast majority if iPhone users are still on Windows. There is a deep pipeline there that has nothing to do with that initial migration wave in 2006 over from Windows at all. ( 1st generation iPhone was 2007. It was sucking up substantive xnu kernel and programmer resources in 2006 but not particularly visible on the outside yet. iPhone folks dragging in appreciable Windows users probably wasn't all that big until 2010-2013 window. iMessages came to macOS in 2012 so now had the 'blue bubble' integration going on. )

There was also an unforced error in Windows land in 2007 with Vista. And a bit less so in 2012 Windows 8. Those surges probably aren't going to happen again. Nothing Apple can really do to make those happen again. Windows 10 gets dropped in 2025 , but the folks clinging to pre 2018 Intel stuff probably won't come rushing to macOS. If primarily interest in clinging to oldest possible hardware then Apple isn't a good target.

If Qualcomm's Oryon ( Nuvia) stuff doesn't disappear down a legal rat hole and isn't priced arrogantly , that might do more to stem a fraction of the exit flow from Windows than anything that Apple does.

There are more people on iOS/Android than Windows . Windows isn't what it was in 2006.

Apple only needs about 3-4% coming out of the general Windows pool to generate lots of new Mac traffic.

Office.com , Google Docs , Zoom , etc. lots of folks don't have all their data sitting on a HDD stuck in the office/home anymore. the 'low friction' works both ways. Some folks will stomp off in a huff to Windows. And another set will come macOS way. both are going to have some churn. Folks act like if Apple looses a couple thousand mac customers it will be the end of the world for Apple. It won't.


Keeping 100% exactly the same customers all the time isn't the point.

I attend training events which require virtualization and AS Macs are specifically called out as not being suitable for class. I wonder if this will manifest itself in other areas where native Windows support is desired.

Amazon is adding far more Graviton units to their cloud than x86 ones. Same is ramping for other hyperscalers.
Training for businesses running relatively smaller internal clouds ? Yeah, that will dominantly linger on x86 a bit longer. [ Bergamo may reverse the trend, but x86 is bleeding share at a decent rate. Both AMD and Intel are frantically working on something to stop the flow. ]

But it is coming.

" ...
TSMC reportedly pledged to process an extra 10,000 CoWoS wafers for Nvidia throughout the duration of 2023. Given Nvidia gets about 60-ish A100/H100 GPUs per wafer (H100 is only slightly smaller), that would mean an additional ~600,000 top-end data center GPUs.

The projections imply an increase of about 1,000 to 2,000 wafers each month for the rest of this year. TSMC's monthly CoWoS output oscillates between 8,000 and 9,000 wafers, so supplying Nvidia with an additional 1,000 to 2,000 wafers monthly will significantly enhance the utilization rate of TSMC's high-end packaging facilities. This upsurge might lead to a supply scarcity of CoWoS services for other industry players due to the heightened demand, and that's why TSMC reportedly plans to expand its advanced packaging capacities. ..."
https://www.tomshardware.com/news/t...nced-packaging-capacity-to-meet-nvidia-demand


All of those Hoppers aren't going to be coupled to a Grace CPU, but likely a substantive fraction are. Arm is going to get pulled into that mix too.

[ If Apple was going to shift to CoWoS for a quad then they'd been fighting Nvidia tooth and nail for queue spots. ]

 

[deconstruct60]

What name could mean more than "Ultra"? Obvious: they could take the chip to the absolute maximum and hence call it the "Max"!!!...

hey, hang on a minute... 🤣

Mn Galaxtus ? (well Disney might not buy off on that one.)

Mn Ultra Plus ( decent chance something practically affordable isn't a 'double ultra' , but an incrementally bigger than the scope that Ultra-class laid down. ) . Apple can easily put an adjective on the end. ( Ultra 3 , Ultra 110 , Ultra+ , Ultra D )

The nominal rumor placeholder was 'Extreme'... which might have been indicative of the price as much as the SoC.

 

[ZombiePhysicist]

Mn Galaxtus ? (well Disney might not buy off on that one.)

Mn Ultra Plus ( decent chance something practically affordable isn't a 'double ultra' , but an incrementally bigger than the scope that Ultra-class laid down. ) . Apple can easily put an adjective on the end. ( Ultra 3 , Ultra 110 , Ultra+ , Ultra D )

The nominal rumor placeholder was 'Extreme'... which might have been indicative of the price as much as the SoC.

Seems like double ultra might be too on the nose. And it starts to sound like a beer or a condom brand.

Further down this path, we start moving towards Infinity + 1.

 

[avro707]

Ya know, the survey leaves out a very important group of folks. Those who have bought something else🧐

Or those who couldn't afford the 7,1 and now cannot also afford the 2023 machine also because it's even more expensive to start with. We don't know how much they are part of the third option.

It would be interesting to see who went to PC workstations (eg HP, Lenovo, Puget).

 

[okkibs]

And the reason they rushed a stopgap [...] so they would not have to support $$$ Intel versions of the operating system for extra years with the leftover 7,1.

Unfortunately that seems a reasonable assumption. Their entire business strategy of not offering upgrades to existing pro machines like the Mac Pro and Studio, not even offering SSD upgrade modules even though these are easily swappable by the customer (and they do swap them if faulty!), is terrible. The trashcan Pro had replaceable graphics modules and Apple just never offered any, and their solution for the Apple Silicon Mac Pro is to not allow replaceable graphics in the first place.

The bottom line probably is that they don't need the Mac Pro sales as all of the Mac sales are a minor part of the business now and the Mac Pro is a small subset of that. They just slapped something together with the parts they already had and called it a day. Just recently I posted here with a more favorable take but perhaps this is really it, they don't care to make a good Mac Pro anymore. After all, even if they made the best one yet, it won't substantially increase market share or cause a difference in the bottom line except a bad one for the increased design costs.

 

[m1maverick]

Vast majority if iPhone users are still on Windows. There is a deep pipeline there that has nothing to do with that initial migration wave in 2006 over from Windows at all. ( 1st generation iPhone was 2007. It was sucking up substantive xnu kernel and programmer resources in 2006 but not particularly visible on the outside yet. iPhone folks dragging in appreciable Windows users probably wasn't all that big until 2010-2013 window. iMessages came to macOS in 2012 so now had the 'blue bubble' integration going on. )

There was also an unforced error in Windows land in 2007 with Vista. And a bit less so in 2012 Windows 8. Those surges probably aren't going to happen again. Nothing Apple can really do to make those happen again. Windows 10 gets dropped in 2025 , but the folks clinging to pre 2018 Intel stuff probably won't come rushing to macOS. If primarily interest in clinging to oldest possible hardware then Apple isn't a good target.

If Qualcomm's Oryon ( Nuvia) stuff doesn't disappear down a legal rat hole and isn't priced arrogantly , that might do more to stem a fraction of the exit flow from Windows than anything that Apple does.

There are more people on iOS/Android than Windows . Windows isn't what it was in 2006.

Apple only needs about 3-4% coming out of the general Windows pool to generate lots of new Mac traffic.

Office.com , Google Docs , Zoom , etc. lots of folks don't have all their data sitting on a HDD stuck in the office/home anymore. the 'low friction' works both ways. Some folks will stomp off in a huff to Windows. And another set will come macOS way. both are going to have some churn. Folks act like if Apple looses a couple thousand mac customers it will be the end of the world for Apple. It won't.


Keeping 100% exactly the same customers all the time isn't the point.



Amazon is adding far more Graviton units to their cloud than x86 ones. Same is ramping for other hyperscalers.
Training for businesses running relatively smaller internal clouds ? Yeah, that will dominantly linger on x86 a bit longer. [ Bergamo may reverse the trend, but x86 is bleeding share at a decent rate. Both AMD and Intel are frantically working on something to stop the flow. ]

But it is coming.

" ...
TSMC reportedly pledged to process an extra 10,000 CoWoS wafers for Nvidia throughout the duration of 2023. Given Nvidia gets about 60-ish A100/H100 GPUs per wafer (H100 is only slightly smaller), that would mean an additional ~600,000 top-end data center GPUs.

The projections imply an increase of about 1,000 to 2,000 wafers each month for the rest of this year. TSMC's monthly CoWoS output oscillates between 8,000 and 9,000 wafers, so supplying Nvidia with an additional 1,000 to 2,000 wafers monthly will significantly enhance the utilization rate of TSMC's high-end packaging facilities. This upsurge might lead to a supply scarcity of CoWoS services for other industry players due to the heightened demand, and that's why TSMC reportedly plans to expand its advanced packaging capacities. ..."
https://www.tomshardware.com/news/t...nced-packaging-capacity-to-meet-nvidia-demand


All of those Hoppers aren't going to be coupled to a Grace CPU, but likely a substantive fraction are. Arm is going to get pulled into that mix too.

[ If Apple was going to shift to CoWoS for a quad then they'd been fighting Nvidia tooth and nail for queue spots. ]

I feel like I've just jumped back to 2013 where the same general argument was made: Few people need the capabilities of high end systems so therefore those wanting them have no right to criticize. You're relitigating it all over again.

 

[davidec]

I can tell you from first hand experience that Apple dearly values their professional music making users on many levels. They’re never going to let go of that. It’s in their DNA so I personally don’t believe the Mac Pro is an after thought. Many top music producers use them. I’d presume this was simply a must get all Macs on Apple silicon move. I’m just hopeful they update it with an M3 and if that happens they somehow allow 8,1 users to update the chipset even if it has to be done in-house. Interesting to see what happens over then next two years.

 

[deconstruct60]

Unfortunately that seems a reasonable assumption. Their entire business strategy of not offering upgrades to existing pro machines like the Mac Pro and Studio, not even offering SSD upgrade modules even though these are easily swappable by the customer (and they do swap them if faulty!), is terrible.

Apple hasn't even shipped a single MP 2023 yet. Why would there be replacement parts available for a product that nobody has yet?????

The MP 2019 does have SSD module parts .

Mac Components - Mac Accessories

Shop for a variety of components for your Mac and Mac Accessories, including Mac Pro wheels, stands, feet kits and more. Buy now with fast, free shipping.

www.apple.com

they are not marked as compatible with the MP 2023. Not particularly surprising since these modules are the INTERNAL subscomponents of an SSD. Pretty good chance there will be modules that only work in the 2023 module (and no future/past other Mac models). It is not a generic complete SSD part. From 2017-2019 to 2023 there have probably been some changes to the SSD controller. So probably would not be the same exact same of NAND packages the SSD controller is tuned to.

At this point of the initial demand bubble launch , it is rather dubious that there would be very large pile of 'extra' spare parts just lying around with no place to go. After Apple gets a feel for what the 'just in time' normalized demand rates will be on these parts , then they can create a predictable reasonably small pile of "extra parts" on the side to sell. Probably like a quarter or so amount of time to get some stabilized predictive flow. ( if they were backordered up to their eyeballs it would take longer. )


Are these modules the same ones as in the Studio. Eh, could be (it would make Apple's internal inventory simpler and same SSD controller generation. ) . There is adhesive tape over the screws keeping the Studio closed. And buried under the power supply to get to the 'module slots'. It isn't a causal user upgrade to change the modules. So shouldn't be surprising not being sold to end users that way.

The bottom line probably is that they don't need the Mac Pro sales as all of the Mac sales are a minor part of the business now

"Mac sales are a minor part of the business" is really not true. The Mac sales are not as big as iPhone sales, but if Mac sales when to $0.00 Apple's share price would drop a lot. If broke out Macs (and associated services ) into a separate business it could land in the Fortune 500 list. It is like someone who is 6.25 feet (1.9m) tall standing next to four over five 7.2 feet (2.18m) NBA centers would looks 'small' in that context.

Pointing at iPhone and iPad sales is simply just misdirection.

Mac laptop sales dominate desktop sales. That isn't an 'Apple' thing. Over in Windows same thing is true. Intel sell more laptop chips than mainstream desktop chips. When the Mini has a really bad mobile chip with a 'weak as water' GPU more folks were herded into buying a Mac Pro or an iMac. Same with the iMac when it has mobile CPU units and relatively much weaker GPUs. As those products got better internals users shifted their buys. The Mac Pro shrank in part because those folks actually didn't want to buy it in the first place. They didn't have options.

The Mac Pro core issue is that is roughly about as big as it was 12-17 years ago while the other catagories have seen lots of growth ( both rotation away and 'new' inflow into mac ecosystem). There is some 'new' inflow to the Mac ecosystem from the outside , but it is being counterbalanced by outflow into other parts of the Mac ecosystem and 'churn' out of the Mac ecosystem. It is kind of flat and 'boring' ( from a growth potential).

and the Mac Pro is a small subset of that.

They just slapped something together with the parts they already had and called it a day.

If they are substantively provisioning those 6 PCI-e v4.0 slots with bandwidth.... they didn't just slap together what they arleady had. They didn't have that before.

Did they reuse the case? Duh. What did they do from 2005-2012. The iMac 27" used the exact same screen for how many years? The Mini has been in roughtly the same dimensions for how long? That isn't a "Mac Pro is small" thing at all. Reusing components about products is a basic strategy uses across all of their product lines. That has nothing to do with 'being lazy just for the Mac Pro cause it is too small". The iPhone has been the same size in multple years also and it is no where near 'small footprint' on the balance sheet.

Just recently I posted here with a more favorable take but perhaps this is really it, they don't care to make a good Mac Pro anymore.

It think the bigger disconnect is in what the term "Mac Pro" means. For Apple I suspect that it more so means that it is a Mac Pro ( a Mac integrated/holistic system of software/firmware/hardware first and there other differentiating highbandwidth/high compute aspects added around that. ). For others it appears Mac Pro ( where 'Pro' is not being compensated for a profession, but is a code talk for 'modular'. So Mac Pro has to be first and foremost a box with as many slots. And then you happen to make macOS boot on it, but if booted Windows too that would be great. Looking for the lowest coupling of the system possible to maximize commodity parts suitability. )

The move to x86 only made that disconnect even larger. The latter saw that as even more commodity parts options and system deintegration. Whereas Apple saw it as a bigger pool from which to select a narrow subset of parts for specialized systems for profitable subsets of the PC market ( Apple holds down bill-of-material costs while layering value add system integration on top. )

 

[matdotcx]

One thing that Apple could do that would differentiate it from the Studio if they really wanted, would be to lift the (seemingly) arbitrary virtualisation cap, so that the Pro could run more than just two guest macOS VMs. Then in place of running a rack full of Studios, one could have a rack-mounted Pro with dozens (or more) macOS guest VMs for those folk that need it - just CI alone would make it worth it for a lot of folks.

 

[prefuse07]

Apple hasn't even shipped a single MP 2023 yet. Why would there be replacement parts available for a product that nobody has yet?????

Because, as I showed in THIS THREAD -- The only thing new in the M2 Mac Pro is the LoBo. Therefore, they DO already have a massive stockpile of parts from 2019.

 

[m1maverick]

One thing that Apple could do that would differentiate it from the Studio if they really wanted, would be to lift the (seemingly) arbitrary virtualisation cap, so that the Pro could run more than just two guest macOS VMs. Then in place of running a rack full of Studios, one could have a rack-mounted Pro with dozens (or more) macOS guest VMs for those folk that need it - just CI alone would make it worth it for a lot of folks.

With an upper RAM limit of 192GB the 2023 Mac Pro wouldn't make for a good virtualization server.

 

[rubberducker]

I can see if a media house uses Arri or is heavy on ProRes, that's where the Mac Pro definitely excels.

I personally use Red, so the GPUs like W6800x duo still perform better for the most part, but I can definitely see the use case for other codecs.

And having that extra utility of the PCIe ports does not hurt, either. For most professional environments $3k can be justified as the difference if the product will provide some more value I feel.

Still, Mac Studio for most places makes more sense, but I definitely can see where the Mac Pro will be useful with what it has.

My M1 Max MacBook Pro rinses my 2019 Mac Pro (with Dual w6800x’s, 192 gb ram + afterburner) playing back both .r3ds and Arri ProRes.

The M2 ultra will surely be much much better for these tasks.

 

[deconstruct60]

With an upper RAM limit of 192GB the 2023 Mac Pro wouldn't make for a good virtualization server.

A broad spectrum workload consolidation virutalization server? Yes. A good virtualization server? That is a joke. Amazon and MacStadium have run profitable Mac virtualization servers for years with less memory than that. Heterogeneous user virtual machine consolidation.... macOS isn't even licensed that way at all.

MacOS Server is a dead product. Dropped XServe even longer ago. Macs don't provide technical support for booting nothing but macOS ( not going to see VMware/HyperV/LinuxKVM down there on bare metal. ). The rack mount Mac Pros are not primarily aimed as 'data center priest' run air conditioned holy shrines. Apple isn't in general serve business.

The Mac Pro doesn't run CICS and IMS database software either. About the same issue.

It is a single user , graphical interface driven workstation. It isn't out to be 'king of everything in the datacenter".

 

[deconstruct60]

Because, as I showed in THIS THREAD -- The only thing new in the M2 Mac Pro is the LoBo. Therefore, they DO already have a massive stockpile of parts from 2019.

Whaever. The holes in the board is the same so ALL of the electronics on the board have to be exactly the same? please. In that picutre you can see , if bother to look, that what is likely the PCI-e switch is different. The SSD modules don't have to have exactly the same NAND modules on them just because the board they mounted to is exactly the same size. ( every M2 2280 board is not mandated to have exactly the same NAND modules soldered to it.).

And the other side of the board have other obvious differences. Slot 8 . Different. Slot 7 Different. The I/O boards that go into both of those . Different.

You haven't proven anything substantive at all.

 

[prefuse07]

Lmao do you even realize what you just typed? -- That you yourself just proved my point? The BOARD is the only difference, as I said.

Are you so high up on your horse that you yourself cannot even see it?

Look at what you just said (i'll even highlight for you to help you see it):

Whaever. The holes in the board is the same so ALL of the electronics on the board have to be exactly the same? please. In that picutre you can see , if bother to look, that what is likely the PCI-e switch is different. The SSD modules don't have to have exactly the same NAND modules on them just because the board they mounted to is exactly the same size. ( every M2 2280 board is not mandated to have exactly the same NAND modules soldered to it.).

And the other side of the board have other obvious differences. Slot 8 . Different. Slot 7 Different. The I/O boards that go into both of those . Different.

You haven't proven anything substantive at all.

Everything you just mentioned is on the freakin LoBo. You made zero mentions of anything within the case/hardware that are different, because there are none. They're doing as I said -- sliding in a new LoBo, but keeping everything else from the 7,1 the same.

The fans are exactly the same, the exhaust, the plastic covers, hell even the power module is the same (though I believe it has less wattage? anyone feel free to confirm this). Oh yeah, also, all of the slots on both boards line up exactly the same, which means they fit into the rear slots exactly the same, thus further providing the ability to reuse the 7,1 case.

Once again, proving my point -- the only difference between the two machines are on the LoBo.

Think about it from a business perspective -- if this is supposedly the lowest selling Mac, and all apple really has to do is design a new LoBo, but keep everything else the same, why would they pass up on that opportunity, from a business perspective?


But, continue your snark towards me, on your high horse, Mr. high and mighty. When someone finally does a hands-on teardown and proves my point, it's gonna be hilarious watching you fall off that horse.

 

[Boil]

Apple could go the board swap route...

Silicon Graphics / SGI did that "back in the day", but they expected the older components back, and there was a "core charge" until they were...

But then once Apple changed chassis / logic board layouts folks would gripe about that because they could no longer keep upgrading the guts of their nearly decade old chassis...?

;^p

 

[goMac]

Apple could go the board swap route...

Silicon Graphics / SGI did that "back in the day", but they expected the older components back, and there was a "core charge" until they were...

But then once Apple changed chassis / logic board layouts folks would gripe about that because they could no longer keep upgrading the guts of their nearly decade old chassis...?

;^p

Apple doesn't do board swaps. With the shipping and labor they'd rather just sell a new machine. Cause with the cost of the board (which is nearly the whole machine anyway) it's not going to be much cheaper. At least not so much cheaper it's going to make a tangible difference to Apple or the user.