Agreed, though the issue is not just that Mac Pros come out on a sporadic schedule, but they have major changes in form factor. The last few versions have gone from 4-slot tower > 0-slot SFF > (arguably*) 0-slot iMac Pro > 8-slot tower > (likely) mini tower with 0 GPU slots.
* I still believe Apple intended to replace the Mac Pro with the iMac Pro, before getting cold feet; they were clearly caught on the hop.
That belief is dubious when Apple explicitly said otherwise.
" .... Notebooks are by far and away our most popular systems used by pros.
Second on the list is iMacs — used by pros, again by the people who use professional software day in, day out, not just casually.
Third on the list is Mac Pro. Now, Mac Pro is actually a small percentage of our CPUs — ...
....
....
Next up: we have plans on iMac, to move that line ahead, and do great things on iMac. It’s core to our Mac business and our customers, including making configurations of iMac specifically with the pro customer in mind and acknowledging that our most popular desktop with pros is an iMac.
...."
You have already read the news. But we thought we would also use this opportunity to share a transcript of the interview with Phil Schiller, Senior Vice
techcrunch.com
Pretty explicitly saying here that they are going to do an "iMac Pro" to address the iMac market segment without saying 'iMac Pro' is going to be the exact brand name. This is a growing segment that they will pursue very shortly (i.e., has higher priority).
from 2008-2017 as Apple shifted the iMacs from using mobile processors to desktop processors the sales went up and more customers transitioned from Mac Pros boxes to iMacs. Previously, Apple herded folks who needed a desktop CPU level of performance into Mac Pro because no other desktop in the line up had one. Once got a desktop processor match up in a more affordable desktop... more folks bought that. A very similar thing happened with Mini 2018 . And with the high end MBP as their mobile processors started to top the performance level of desktop processors 2-4 years previous. (i.e., why MBP was top selling pro systems by 2017 ).
The iMac Pro reduced the urgency for a Mac Pro 2012 replacement, but Apple was clearly positioning it as an expansion of iMac line up. The classic iMac chassis was 'painted into a corner' with it one and only one Fan and outlet hidden behind the pedestal arm. The iMac Pro make a trade-off for a bigger exit fan (tossing easy user accessible ram) for two Fans. That allowed a "bigger single GPU" and largely decoupled the CPU thermals from GPU thermals (a problem the MP 2013 had. mandatory two GPUs and coupled GPU/CPU thermals on a single fan).
The iMac Pro was a substitute for a large fraction of folks very happy with a MP 2013 , but many in the legacy Mac Pro legacy user base didn't view the MP 2013 as a 'Mac Pro' either. For some there is a MP 2013 ---> iMac Pro --> Mac Studio path. But that actually starts with the MP 2013 as an origin; not in the 2008-2010 era. In 2017, it likely was pretty clear in their 5 year plan that the Mac Studio was coming (or something close to that).
What shifted in 2018-2020 for the iMac Pro is that Apple started to think about getting back into selling discrete display and that helped block a direct iMac Pro replacement. Supply chain issues with next gen screen likely also played a major role. When those better stabilize, it is decently likely that the 'large screen' , high performance iMac will come back. ( unless the 'thinnest design' politboro kills it. It is clearly doable with the classic iMac Pro chassis (dual fans , big vent and not an iPad on a stick. ) ). If the Studio combo ( Mac + Display) combos sell far better than expected then it likely won't ( will get large screen iMac on a stick capped at Mx Pro ... or less. ).
With the non-kneecapped Mini and Mac Studio the whole collective iMac category isn't going to be as dominantly large a Mac desktop player as it once was. That is what is having a most direct impact on a speedy transition replacement for the iMac Pro; not that the MP 2019 was a replacement. Apple is allowing for alot more fratricide in the desktop line up these todays. M1 strung out over the MBA, MBP 13, Mini , and iMac 24. The iMac 24" doesn't 'buy' you a significant performance gap. Studio M1 Max and MBP 14/16" Max ... no huge gap there either.
It is not just solely about the Mac Pro. Apple is dragging more Mac users into the lower 'half' of the Mac line up than every before. That is where they are 'skating to where the puck is going'. The iMac Pro was the same thing as skating toward where users were generally going.
The MP 2019 was not a direct replacement for iMac Pro. It shot the old MP 2013 entry price up by 100% (and higher still relative to 2012 starting point). Even relatively to the iMac Pro it was at least a 20% price increase. If even out the starting SSD capacity size and GPU ( iMac Pro 32GB RAM , 1TB SSD , Vega 56 8GB HBM vs modern MP 2019 32GB RAM , 1TB , W6600 ( which is a step down from a Vega56 by match on VRAM capacity. ) it is a 30% price increase. That is pretty steep increase to claim trying to cover the same user budget base as the previous product. There is definitely an intersection of the user base budgets, but it isn't the same user base.
Apple also knew in 2016 that they'd need eGPUs to stop gap some of the performance zones for the iMac Pro (only having one GPU limitation when some even MP 2013 users found two GPUs useful). Side effect of that would also likely open door of stop-gapping the Mac Pro 2012 into running a bit long on unofficial support window ( GPUs going into eGPU service would happen to work on 2012 also ). So again, a contributor to why not a highest priority for essentially a hobby product. They could slide a Mac Pro update into 2019 , or so , without major increase in problems .
The lack of regular MP releases is also a major issue for those looking for an affordable, expandable desktop Mac. If the MP had received bi-annual releases throughout the 2010s, the lack of a desktop (i7/i9) Mac wouldn't be an issue -
people could just buy e.g. a used 2018 Mac Pro.
Buying used Mac Pro does relatively little to grow Apple's revenue numbers year over year. That notion that the used market is some top 3 priority for Apple R&D dollars is disconnected from reality. Turning out a commodity container every two years is not something Apple is going to put tons of effort into. They aren't a commodity motherboard and case vendor.
Apple's move to the mainstream Wintel foundation for major components was not to enter the commodity Windows market. It was just to source major component R&D costs over more systems vendors at an effective cheaper rate for Apple ( Apple gets better parts for less money. ). They were trying to be the best Windows box on the market.
There has been over 10 years of "Apple has gotten build the xMac" assertions. And Apple only increased the overall Mac ecosystem success over time without one. Yes, for a long while the workaround strategy has been to press older generations of Mac Pros into filling that xMac role. However, the MP 2019 was a huge departure for that by Apple.
Cranking up the entry price by 100% impacts that two ways. First, substantially fewer Mac Pro as going to get sold as new. That means substantially fewer of them will be sold as used. Second, the generally available used price will take even longer to sag all the way back to affordable 'xMac' levels. [ There will be scattered small exceptions where some will be firesale pricing that has to be dumped quickly or was completely written down for tax purposes. Sinking the number of eligible units possible for sale will help offset that. ]
MP 2019 is more of transition prototyping system than some direct replacement of the 2012 Mac Pro.
Apple is transitioning to dumping UEFI. T2 chip is a transition to an Apple proprietary boot set up. Apple is unhappy enough with UEFI security that T2 validates the UEFI firmwrae entirely before handing the Intel chip a copy (not the original or access to it). That is not signaling a future toward commodity boot hardware from generic retail, race-to-the-bottom pricing, sources.
Apple security data stored on Apple SSD being critical to full system function. Can offload daily work from Apple SSD , but if it goes completely toast and there are no support components replacements available then it is not a generic PC using entirely generic components.
Sure there is PCI-e slots but
Apple Afterburner ... more so a field prototype before weaving it into SoC die.
Duo GPU cards ... again throughly exploring gaps to making a multiple die single GPU.
MPX connector ... elegant work around until can put Thunderbolt controller provisioning into the SoC die. And get rid of dangling power cables don't like anyway.
Future door for non boot subsystem critical PCI-e cards ... yes. Future door for critical boot subsystem critical cards ... no.
As it is, the 6-year rein of the trashcan means there's nothing between the ancient 2012 machines and the still very expensive / current 2019 machines.
Apple put the 2013 on obsolete list (not getting macOS 13 ). That gap didn't help the MP 2013 as life as an 'xMac' either. End of 2022 probably starts a countdown clock on macOS on Intel. Lack of a 2021 foundational upgrade for MP 2019 means macOS on Intel isn't going to get a life extension even if Apple continues to coast on the MP 2019 chassis for another year.
The GPU trickle down upgrades that folks used to move old Mac Pro's forward as xMac systems was entirely dependent upon new Intel Macs evolving and using newer GPU families from AMD. There is no AMD driver support on macOS on Apple Silicon. There is not necessarily anything coming for MP 2019 either if Apple is going to coast on already spent R&D and focused on cost containment for macOS on Intel expenditures ( doing updates for several years , but on a very fixed budget that is shrinking over time.)
The whole 'old MP as xMac' scheme was largely just a "happens to work" scheme there was firmly grounded on Apple evolving along the commodity PC component parts path. CPU is very clearly out. GPU are minimally in zombie status if not totally out. UEFI boot is clearly out. That is a far more impactful factor that the width of the MP 2013 gap. It is not a new or pressing problem for Apple.
just assuming the worst to avoid disappointment. When it comes to the Mac Pro, the best policy is to take the worst estimate and add a year to it. With the last one, Apple announced an upcoming, all-new MP in April 2017; it arrived at the very end of 2019, with most shipping in 2020 (in time for the Apple Silicon transition).
Huge assumptions don't often lead to better expectations management. Apple gave themselves a deadline of "about 2022" to finish the transition. Pretty good chance that they had a plan framework by end of 2019 that supported that timeline. The pandemic probably caused them to put a stronger caveat about when in 2022 they were targeting, but likely still trying for 2022.
All the way to the end of 2023 would only happen if they structured that original plan on something pretty goofy. Like Thunderbolt 5 ( kind of how MP 2013 slid to end of that year waiting on TBv2). Some 'tail wags the dog' premise for a deadline. Likely there were strongly hoping for 2022 and hiccups slid it into 2023. Unless it is a fundamentally major screw up all the way to the end of 2023 is just pushing back date so far as to just going for a far future date in and of itself. Could just as easily say something 'before late 2025'.
The MP 2019 foundation isn't going to age like the MP 2013 did. There is far more competition between Intel/AMD now in the server/high-end workstation space. The W-6300 series from Intel really didn't get wide spread adopted by major vendors because AMD had better stuff. It only gets worse in 2023 as even Intel deploys better and AMD increments in volume. Even more so on GPU front if Apple goes Rip van Winkle on GPU driver updates on macOS on Intel. That was clear looking at roadmaps 3 years ago so not sure why Apple would have a plan that coasting on MP 2019 into late 2023 would be a 'good idea'.
Whether Intel get their **** together or not doesn't really matter in the PC space. If AMD's CPUs are better, you just buy one of those. Workstations are mainly Intel-based, though Lenovo makes good AMD machines.
No, it really does. TSMC can't handle Intel's foundary output being shifted over. It is very clear as to why the AMD sat on the Threadripper 5000 until Apple announced the M1 Ultra. AMD didn't have the capacity (wafter throughput from TSMC) to expand the Threadripper 5000 into the market and keep up with Epyc 7003 demand at the same time.
If Samsung had there fab processes running more efficiently then perhaps AMD could spread the load over TSMC and Samsung to cover Intel's output, but either one of those by themselves isn't really plausible( given their other customer workload demands).
If the overall PC SoC volume goes down then the average price of PCs will go up. That would have an impact.
Workstations are a major problem source for Intel. They are on the verge of losing tons of share there. There was practically no adoption of the Xeon W-6300 by the top 3 workstation vendors ( Dell/HP/Lenovo). All the models updates from mid-2021 from those vendors at the higher end are all AMD. Dell's 'new' workstation is AMD.
Intel stumbled on 'Fishhawk Falls'/'Sapphire Rapids' and it is sliding significantly into 2023. There is nothing new coming to challenge these AMD workstation SoCs and AMD is on track to have enough wafer starts to they won't have to squat on Threadripper 7000 for as long. ( Best Intel can hope for is that Eypc 7004 series going past sales expectations impedes it, not that Intel has slows it down directly competitively).
lower-mid-range Intel has some hopes with mainstream Gen 13 (Raptor Lake) , but those have moved during 2021.
AMD sockets do tend to last several generations, though obviously must change sometimes, and now is one of those times. In any case, upgrading piecemeal is practical with PC hardware - and there's a wide supply of used CPUs / GPUs to do so cheaply (and if you buy from reputable sellers, is perfectly reliable).
The Ryzen desktop 7000 series picked up a realtively very small iGPU. Intel has had a decent sized iGPU for several generations. The general PC hardware market is laptops. The long term trend away from dGPU is coming to the PC desktop solutions. Slower than the speed of Apple's adoption but that is the general direct that the 'puck' (market) is going. Same basic fundamentals technology drivers as the laptop becoming dominate over desktops trend over last two decades.
iGPUs are generally going to 'eat' the bottom of the dGPU line ups. As chiplet packaging processing costs fall, that will only gain more momentum.
