A few notes:
ECC is a big one, that should have been on my original list, thank for pointing it out. However, capacity is a big considering of some tasks. I saw many people stating at the time that they couldn't upgrade until Apple offered more memory for computational tasks.
Yeah, I should have elaborated that it was later added, but at the time Apple was likely working on this last Intel Mac Pro, it was probably a consideration. So, timing is key here.
In 2020-2021 it was a bigger job to get folks off the MP 2009-2012 models than to get them onto Apple silicon. MP 2019 was either going to be effective there or not. The vast majority of folks who had put money into a 2019 model were likely not going anywhere for years. There was initial a shortage on 2019 Mac Pros so they really were more so pragmatically a 2020 product.
The Mac Studio with the Ultra did not appear until March 2022. Just as a couple of months before the 2 year anniversary of "about two years " declaration for the transition. There was not as much memory , but there was also really no deliberately more direct replacement for the Mac Pro either. The Studio was more so of a large screen iMac replacement.
iMac Pro 2017 --> 256GB ECC
support.apple.com
iMac 2020 --> 128GB (no ECC )
support.apple.com
[ NOTE: generally Intel doesn't even let processors without EEC even get over 128GB in capacity. The notion that 200-300-400GB of RAM doesn't need ECC is more a GPU frame of reference than a CPU one. Different approach to data integrity versus single focus on maximizing bandwidth. ]
Mac Studio M1 Ultra. --> 128GB max memory
support.apple.com
The M1 Ultra even falls a bit short of the iMac Pro let alone the MP 2019. However, for the bulk of the entry level Mac Pro configurations, it would work. The M2 Ultra was relatively close 192GB.
The "Extreme" really going to solve the > 1TB problem. Even at 256GB for an utlra that would still land you at 512GB.
the > 1TB MP 2019 was at least about making Apple more money than it was about some 'large' user base. Intel had two versions of the the 28 core chip. One had a limitation under 1TB and another had a >1 TB tax applied to it. Apple just laid their tax on top of the Intel tax. And did not at all offer the non taxed CPU model as a configuration offfering. That > 1TB is largely an artifact of a feature design for multiple user servers, not workstations. Vast majority of workstation did NOT need it which is EXACTLY why Intel offered the non-tax model in the line up.
I would not ever consider Rosetta a long term options, but many people may have wanted one last Mac Pro as a hold over (little did they know that the Mac Pro as a product was doomed, anyway).
As the macOS on Intel support windows gets smaller the used Mac Pro 2019 will get cheaper. Folks would want to sit in a time bubble avoiding chance will have ample opportunities to jump on board. Those are just not systems Apple has any significant interest in selling. ( and it really won't be very many folks. Perhaps vocal folks on these folks, but relative to aggregate Mac market not that many. )
More so the issue with mentioning the hypervisor is that the Apple Silicon Hypervisor is made for virtualizing AArch64 only, not x86/x64. This is why I have a 2013 Mac Pro siting on the other end of my desk from my Mac Studio.
You seem to be working was a substantially different definition for the word 'hypervisor' than what it technically means.
" ...A
hypervisor, also known as a
virtual machine monitor (
VMM), is a type of computer
software,
firmware or
hardware that creates and runs
virtual machines. A computer on which a hypervisor runs one or more virtual machines is called a
host machine or
virtualization server, and each virtual machine is called a
guest machine. The hypervisor presents the guest operating systems with a
virtual operating platform and manages the execution of the guest operating systems. Unlike an
emulator, the guest executes most instructions on the native hardware. ..."
en.wikipedia.org
A Hypervisor is nominally a very thin/shallow layer for allocating the privileged level of resources of a system for virtual machines. Its primary job is to hand out slices of the actually machine running on to the virtual machine images running above. Virtual machines are emulated any more than virtual memory is emulated.
It is just a minor indirection from the direct physical resource. The type 2 versions (like Apple's) are not as 'thin' as the type 1 but view from above (the VMs layered on top) there is not much of an substantive difference. Mainly a difference in how much overhead the hypervisor hogs up on the core resources. Type 2 is a bigger 'hog'.
It is not the hypervisors job anywhere I am aware of to "translate the opcodes' of anything that isn't present in the machine. Their might be a emulating virtual image running on top of the hypervisor , but it is now of the hypervisors business what the virtual machines are doing on top. The job is to stop those VM from abusing/usurping shared resources.
On some systems like an iBM mainframe there is no option but to boot into a hypervisor and then initialize VM allocations from there. LPARs and VMs are all physical allocations. Perhaps there is a "emulation' privilege mode of the processor that might need to be managed , but that is a property of the processor; not some direct emulation activity.
"Abandoned apps and drivers are not really in Apple's "skate to where the puck is going" approach to most things.
If it wasn't viable for these developers to spend money on it , why is it going to viable for Apple to do it. ( other than take money from a large group of folks who do pay and hand to do groups of folks who don't pay.)"
I'm not 100% sure what you mean by this, so I'm going to reply based on what I infer: that is a valid statement from a business philosophical perspective, but in reality high end professionals aren't just buying computers to upgrade to the new thing, or because it's feeling old, like consumers do, they're replacing faulty ones or expanding their infrastructure, and many times certain critical software or drivers for expansion cards and peripherals may not be available, or available yet (since we're talking a transitional machine here), so for many, moving to Apple Silicon wasn't an option,
The MP 2019 was/is a transitional system. There was more than ample warning that Apple was going to stop selling them. Apple openly stated a "transition in about two years." so by early 2023 a person would have to have their head in the sand not to believe that there was a need to buy a spare system soon if really needed one for substantive business requirements for some legacy drivers that are comatose.
It was also more than obvious at that point (having not surprising skipping Ice Lake that there was no new system coming. By 2023 no new MPX units, no new AMD GPUs used in other macs. And even Intel mini on last legs. )
I wasn't talking about 'change for change sake". I'm more so referring to the somewhat common practice of taking "if it isn't broken don't fix it" from an expression to being dogma. And then to the place where it is more so just an excuse.
By 2022 the transition kit software had been available for two years. If a business critical software vendor doesn't even have a plan for the transition then have a problem that needs fixing. By 2023 and new Mac Pro if still no software then a slacker for a vendor. By 2023-2024 if no stated plan or action at all then there is a "plan" that is extremely likely to be "no plan" ; just being somewhat deceitful in not communicating it. That is most definitely a problem. During this time window, the MP 2019 is fully supported with OS with update, so have a transition target. (in additional to a variety other other used Intel Macs. )
In 2021-2023 Apple is shipping about 20M apple silicon systems per year. What kind of software business is this 'critical' software vendor running that is ignoring 10's of millions of potential customers? The product. will die without new customer and they are not trying to get any. If it is already 'dead' that is not a surprising course of action. This is where "ain't broke' often drifts into being an excuse. Because no one has clearly reported the blazing forest fire it doesn't exist. And start looking for excuses like "well Apple has money ... they should solve this problem of my dead software." Or just pretending that the software isn't already dead/comatose.
Professional IT managers who are actively evaluating 'weak' vendors for removal/replacement are likely leaving over time. Or at least coming up with a plan of how to put the system into some kind of 'time bubble' where there is no transition and just hope nothing needs any active support in the future.
due to a significantly larger investments prior to the announcements to Apple Silicon, so a transitional machine would have been very important. Again, we saw this with the move to the G5 and the removal of native OS boot support.
folks who had 'sunk costs' problems before Apple's 2020 transition induction probably were going to have 'sunk costs' problem for 2021, 2022, 2023, and 2024. This is all well past anything Intel and far along the path of macOS on Intel phase out.
The G5 example seems to indicative the opposite of what you hope. The G5 dumped the legacy stuff. As the highest in PowerPC model wouldn't the "Extreme" highest end Apple Silicon model do the same thing? Apple did not have to introduce another Intel model to cover the 1TB RAM space as the MP 2019 already existed. Don't have to 'backtrack' on the sequence if did the MP 2019 before the 2020 launch in first place. [ Similar who why there was one more last gasp iMac 5K in 7/2020 because was going to be a long wait until Mac Studio in 2022. ] To refresh the MP 2019 all Apple had to do is drop some more MPX cards and vast majority of folks were happy.
the G4 was. also a cheaper model. It addressed the OS9 stuff, but also the 'even cheaper' market. It can at the same time as G5 so a cheaper was also a selling point. A "Ice Lake" Intel Mac Pro probably would have been more expensive; not cheaper.
8 core W-3223 was $823 range.
12 core W-3235 was about $1,500 range.
16 core W-3245 was about $2,200 range. 205W
whereas the in Ice lake
12 core W-3323. was $1,043 range (. about $200
more expensive than the 3223 ).
16 core W-3335 was $1,400 range. 250W
Apple very unlikely would have 'ate' that price increase and kept the base system price the same. They would have passed that along. Relative to the older 12 it was a price decrease , but I suspect Apple would have substitute more expensive tRAM and SDD to keep the price levels for overall system just as high. AMD competitiveness was putting some downward pressure on paper. The reality was that these were more 'paper tigers' and Intel held the limited die supply for the server systems with better margins and more critical losses to AMD to cover.
A 2021 Mac Pro Intel would have helped far more with reputation repair than 'transition' issues. the timeline for Intel support really would not have moved at all in the long term. Getting onto a more regular schedule of new Mac Pro every 2 years might have helped. After going Rip van Winkle for 6 years, lots of folks were not coming because presumed Apple didn't care. After 6 years did a 4 year. If 4 years was going to be the reoccurring target a substantive more folks were just going to leave.
