Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
I’m pretty sure Apple wouldn’t mind Microsoft eventually releasing a Windows version that supports the M1. They are already supportive of Linux being installed on their M1 systems (see Asahi Linux). I think PCs will eventually head over to ARM or RISCV. It just seems the tide is inevitably turning that direction. May be a decade though.
I would imagine Apple’s happy not having to support creating Windows drivers for their hardware anymore. :) If Microsoft wants to do this, they’ll be taking it the whole way themselves, without Apple’s help. And sure, Apple wouldn’t mind but, if Microsoft were to go to that effort to get their OS to run on Apple’s hardware, there might be some upset OEMs, the ones that have gone out of their way to create systems designed to run Windows ARM. :)
 
ASi Mac Pro (8.1):
  • M2 Extreme (quad M2 Max SoCs/3nm)
  • 48-core CPU (32P/8E)
  • 160-core GPU
  • 64-core Neural Engine
  • 384GB RAM
  • 8TB SSD
  • (3 or 4) PCIe (??x/Gen?) expansion slots
 
  • Wow
Reactions: George Dawes
All I know is Apple has a desktop customer for life after my experience with the 16-core 7,1 - got it in January 2020 and it’s been absolutely flawless, silent, and blazing fast all day everyday since then..
 
As long as Apple is still selling Intel/T2 Macs they're pretty unlikely to launch a new MacOS without Intel/T2 support. But once the last Intel Mac is removed from sale then it's feasible that the very next MacOS will drop all Intel support.

last T2 and last Intel Mac are two different things. Last Intel removed from sales and very next macOS drop all Intel is relatively infeasible to rational look at Apple support policies. The countdown clock doesn't even start until it is removed from sale. So the notion that within 12 months Apple is going to utterly abandon something they relatively just sold is in no way shape or form in compliance with what their policies outline. It is going to be dropped eventually. But 1 year? Probably not.

Providing 'software' support long after have dropped hardware support doesn't make much business sense. And vice versa. For apple to commit to 4 years of hardware coverage and drop macOS after only 1-2 is silly if view Macs as a system ( a combination of the two. Which is exactly what Apple licensing and language does ).


Pretty sure the Last non T2 Mac was the antiquated , non-Retina , "educational" iMac 21.5" dating from 2017 . Dropped in October 2021.


That system is likely going to get cut off in a similar fashion to how the Mac Pro 2013 was cut off. A product that was stretched way, way past normal updates cycles for Apple ( a fork of a MacBook Pro 13" two port foundation that Apple iterated past on a regular fashion for the laptop variant). It was over 4 years old when dropped (while other iMacs and laptops had gotten more regular updates. ). That system might macOS update coverage to go one year past when most of the other 2017 models get chopped, but it isn't going to get an extension out to 2026 . Or drag the other 2017 models out to 2026.

Pretty good chance that is the window where kernel extension will die off on the macOS Intel side (if not sooner, they were put on the deprecated status even before the transition really started. )
The bulk of the 21.5" iMacs (Retina) got dropped with the iMac 24" in April 2021 . Those are already over a year on the countdown clock. There is likely less than 4 years there. Pretty good chance that is the window where kernel extension will die off on the macOS Intel side (if not sooner, they were put on the deprecated status even before the transition really started. )

The bigger block of non T2 Macs that ran "longer" than should have been expected was the 2020 iMac 27" which got replaced by the Mac Studio in 2022. So about 2027..

If Apple does the Intel Mini in 2022 also that's probably it for macOS Intel extended time. [ Yes, the Mac Pro looks to be sliding into 2023 , but probably not enough to grab an 'extra' macOS upgrade cycle. If Apple does side-by-side sales then probably a small amount of extended time. ]



Until then, your particular Mac may still get dropped but that's not inevitable, esp. if its a T2 (I'm sure non-T1/T2 machines will be the next to go) - but when the Intel code is gone from the OS, it's gone...

Macs are going to get dropped because that is the general way the mechanism worked even before the transition to Apple Silicon. Primarily all that Apple Silicon is doing is pruning off the 'possible extra grace period' in the range that Apple offered ( 5-7 years is now pragmatically closer to a 'hard' 5 . Apple isn't chopping down the whole 'tree' all at once but they are pruning aggressively. )

Even the Apple Silicon Macs are inevitable. The M1 MacBook Pro 13" owners are now "on the countdown clock". Not as short as the intel ones, but inevitable none the less. This Fall Apple will likely toss the M1 Minis onto that clock also. iMac 24" probably shortly into 2023 if not in 2022 also.


AFAIK the "5 years after last sold" and "vintage/obsolete" is only about parts and service & is required by consumer law in some jurisdictions, which doesn't entitle anybody to new software features not promised when they bought their machine.

But if the Mac is really a holistic system does it really make any business sense to do one half of the system while dropping the other half. Neither half is really generating substantive revenue so it is all a net loss on costs. Apple treats them basically the same. macOS is licensed directly to the hardware upon which it is sold. It is largely folks outside of Apple that try to throw up huge borders between the software and the hardware.


They have more wiggle room on the macOS update side if they choose to take it. ( because goofy corner cases do show up like the Mac Pro 2013 that go far too long without going onto "last sold" status. ) Mac hardware is released over several parts of the year and macOS only in Sept-Oct so there are mismatches in arrival/end-of-sale windows which leads to not exact year marks, but there is nothing about macOS support that goes way past the boundaries set by the hardware policies.


bounces around 7 years. And can eek out some extra time with willing to sit on every smaller in scope security upgrades. Probably going to see that aggregated number shrink as the Intel systems get something much closer to a 'hard 5' cutoffs ( and Rip van Winkle products like the MP 2013 disappear 'early' because their product update cycles got screwed up before Apple Silicon arrived. . ) .
 
I have A LOT of apprehension with the rumours of a reduced size [volume] MacPro. The whole point of the professional end of the market is to use specialist PCI-e cards like sound cards, capture cards etc.

You can’t fit those in a smaller unit. So it would effectively be a Mac Studio in a tower format.

And yet the Mac Pro 2009-2012 has half the slots as the Mac Pro 2019 and did OK in the market. There is a slot survey thread in the Mac Pro forum that had a decent number of entries where there was still an open slot even with four. Even more where one slot was consumed because the motherboard support for USB was too old (needed USB3.0 or SATA 6Gbps. where modern logicboard would more than fit those needs. ). Another common occurance in the slot surver was one 'boot' GPU and one newer/faster GPU. If you take all the dGPUs out of the slot survey there are empty slots in every single one of those 4 slot systems.



Going to a "half sized" Mac Pro If Apple took off the extended feet and handle and one MPX bay (and associated fan), they still have 4 slots left and a 300+W SoC. If shave the height in half than volume would drop in half but still would be able to take standards complaint full length and full height cards. A significant issue if counting just the central box of the system or the whole tower height including the extensions.

If they added 2 M.2 slots on the backside of the back into addition to their proprietary SSD modules, then that would probably alleviate some of the pressure causing one of the x8 slots being filled with a M.2 carrier card for SSD storage ( need more than just one internal SSD drive). If they provisioned two Apple SSDs on the logicboard ( only using Duo/Quad SoC with mulitple SSD controllers), then that two would cut down on need for slots consumed by storage (but somewhat dismal $/GB costs for end users. )


That MPX storage module that soaked up 4 slot widths really didn't work all that well (space efficiency wise). There are no macOS M-series GPU card driver support so 2-3.5+ wide GPU cards likely won't be soaking up volume/space either. If they remove the legacy spec busting cards then the internal volume requirements are not as high.


There was a rumor about a one-slot-wonder that also now also appears to be in the "never gonna ship" status.
If there is only one slot then there is a significant backslide issue. ( to get to half the volume but only one slot Apple would have had to bring back drive bays... which some folks would stand up and clap for. I doubt that is coming though. More like more than half as small with just one slot. )


There are some advocates for the "half size" meant chopping height , width and depth. Chopping away full length cards ( maybe going do to 3/4 or bit longer than 1/2 length cards). Some I/O cards would fit but some also would get pruned.



The MacPro 2019 is a really an industrial design masterclass. The collaboration and engineering that went into it were unprecedented for Apple. To turn around and provide one that can’t be used with all the professional hardware would seem incredibly foolish.

If not enough folks buy it then it would incredibly foolish for Apple to make it.

Apple increased the entry price by 100% , that is highly unlikely not going to increase sales.
The number of folks who need 4 HDX sound cards in a single box is far smaller than the number than need just 1 or 2.

The Mac Pro 2019 has a "low volume tax" built into the price. If Apple has at this point made back all the engineering costs of making the 2019 model then it has been in now way shape or form a foolish effort. They have made a profit. End of foolishness. With MPX W6900 modules priced higher than MPX W6800 Duos (and RAM and SSD pricing, and ">1 TB RAM" tax on 24/28 core models ), what is a bit foolish is to think they have not made their money back yet at this point. Getting paid in full for work that you do is not incredibly foolish.

While there are fair number of PCI-e cards that have gotten updated drivers for macOS on Apple Silicon there are sizable number of other ones that have not. Before the Apple Silicon was even introduced Apple announced at WWDD 2019 that kernel extensions (kexts) are deprecated on Intel ( and implicitly on Apple Silicon). Some cards have macOS on Apple Silicon have driver support via kexts. So those drivers were in deprecated mode on day 0 on Apple Silicon. Maybe those card vendors are kicking the can down the road. Some of them probably are not . If Mac Pro 2019 is where the log jam will form where some folks stop writing drivers that work... then an abnormally large number of card will dead end there. So that slot number expansion has some "legacy card collector" dimensions to it rather than forward looking professional utility. I don't think the Apple thinks the number of PCI-e cards supported are going to shrink to zero. But I do think that they expect some number of cards to be "garbage collected" as defacto abandoned in terms of drivers under active development. They are likely not laboring under a notion that Mac Pro of future will have every possible "Pro" card out in the market covered by macOS. Apple isn't building everything for everybody.



It wouldn't be very surprising if Apple only included more than 4 slots only in a $1-1.5K more expensive rack version that was provisioned via an internal slot expansion add in board that "snaps" to the baseline board. All the audio card that don't really stress PCI-e v3 and are relatively low power , higher volume consumers would still fit. (but at higher prices, but still a relatively low bandwidth lane provision out of the main SoC. ). Lots of A/V focused stuff is racked anyway.
 
What goes in the slots?

If look at the Sonnet Tech PCI-e card compatibility list ( a PDF file linked below )

https://www.sonnettech.com/product/echo-1-desktop/techspecs.html#techspecs

There is a column on the listing for "M1 Mac Compatible" that has "yes" and "no" and "-" markings. Everything marked "yes" would likely work day one on new Mac Pro with an Apple Silicon SoC. (Thunderbolt support is primarily PCI-e card support with optional "hot plug" standards actually covered by the implementation. Not skipping standards (PCI-e and macOS) is what drive compatibility. )

It isn't as long as the macOS Intel one , but it is no where close to zero either. Sonnets June released McFive card works.



Cards that are highly tied to the UEFI boot services environment likely won't work well. There is no UEFI on new Macs viable the nominal boot path. Cards with abandoned software support aren't going to work either.
 
Cards that are highly tied to the UEFI boot services environment likely won't work well. There is no UEFI on new Macs viable the nominal boot path. Cards with abandoned software support aren't going to work either.
Once you get outside GPU’s and AfterBurner and the PCI Interface cards, the only possibilities left are network, audio interfaces and storage. Due to the faster external interfaces these days, they can be handled adequately externally, and, since most old cards wouldn’t work anymore anyway, there may not be a lot of interest from the vendors updating those hardware solutions. Slots don’t feel to be as important on a MacPro device as they’d were before. If any, I’d guess no more than two. I’m curious about the future PCI situation than I am about anything else. :)
 
Pretty sure the Last non T2 Mac was the antiquated , non-Retina , "educational" iMac 21.5" dating from 2017 . Dropped in October 2021.
The 5k iMac only got T2 in 2020 (https://support.apple.com/en-gb/HT208862). You're quite right that "T2" and "Intel" should be separate things, but time-wise they're only a year or two apart on the road to obsolescence.

For apple to commit to 4 years of hardware coverage and drop macOS after only 1-2 is silly if view Macs as a system

Dropping support for no-longer-sold Macs from the next version of MacOS is not "utterly abandoning" them. The previous OS will still be supported with at least security updates for (typically) the next 2 years - and it's actually up to Apple how long to extend that - they could easily announce "long term support" for the final Intel-supporting Mac OS (heck, lots of people would be delighted with a LTS version of MacOS, even on Apple Silicon).

In terms of Apple's effort, they'd only have to worry about implementing critical security patches and maintaining the old clients for any Apple services they change. Versus developing and testing everything in each new major MacOS release for Intel as well as Apple Silicon.

As for "Macs as a system" that also means MacOS evolving and becoming better optimised for Apple Silicon. Maintaing Intel support could put a big damper on that and leave MacOS saddled with a lot of Intel-related clutter.

Bear in mind that x86 to Apple Silicon - especially at the system level that MacOS has to worry about - is a far more profound change than your usual OS update and developing to "production" level for both platforms will be a significant extra effort.

The last such change was PPC to Intel - for that it depends whether you count years or releases. Time wise it was 3-3.5 years but there were only two major dual-platform PPC/x86 MacOS releases (and that's counting Tiger 10.4.4 - the first x86-compatible version - as a "major release"). We're already at three x86/ARM MacOS releases (including Ventura). However, the PPC to x86 hardware transition was done and dusted in a year, this one is dragging on badly (probably due to "unexpected factors") so it's hard to compare... I suspect anybody buying an Intel Mini or Mac Pro since March 2022, however, is doing so because of legacy issues and might not be first in the queue to load up the next version of MacOS...
 
If the new Pro uses the same approach as all the other Apple Silicon chips to date then the RAM can't be upgradeable. LPDDR5 RAM doesn't come in plug-in form - it is designed to be surface mounted as close to the CPU as possible (on the CPU package in the case of Apple Silicon) to keep the connections short, fast & power efficient. Certainly the M2 Max-based configurations suggested in the "rumour" article would presumably be LPDDR5-based.

OTOH, if Apple wants to offer 1TB+ RAM configurations to replace the Intel Mac Pro they'll have to add regular DDR5 slots somehow - but that would need a more conventional SoC design.

I think people forget that traditionally Macs had some soldered RAM and some upgradable RAM. It was an improvement that for a decade or so it was all upgradable but this situation merits going back to this way
 
Once you get outside GPU’s and AfterBurner and the PCI Interface cards, the only possibilities left are network, audio interfaces and storage. Due to the faster external interfaces these days, they can be handled adequately externally,

The old legacy PCI cards are the ones that can be easily be shifted outside. The "faster external interfaces" are actually getting rather stale these days. TB4 is just as higher fraction of the same old x4 PCI-e v3 provision feed that have been plateaued at for a while.

A small, substantive subset of the cards missing on that "M1 -- Yes" but have macOS(MacPro) coverage on the previously linked in list are those that pragmatically require a minimum of x8 PCI-e v3 to get lots of value add with their usage. An external PCI-e closure that is defacto x16 PCI-e v3/4 to x16 PCI-e v3/4 would 'solve' that issue but Thunderbolt won't. And TB likely is not going anywhere any time soon since it is now more tightly coupled to USB-IF and a much larger committee that has to almost unanimously approve improvements.

TBv4 is being outstripped on bandwidth by x4 PCI-e v4 SSDs. About a year away from that getting to being an even deeper 'hole'. At least not in drives with higher pricing budgets. ( for sub $200-300 storage drives and/or HDDs ... yes it has largely plateaued . )

https://www.tomshardware.com/news/gigabyte-announces-aorus-gen5-10000-ssd



Apple seems to be on a path to ignoring CXL (and likely will 'slow roll' PCI-e v5 since there is relatively little added value in the laptop space for those. ) , but workload shifting products are appearing.

[ and EDSFF modules as a category to range past the limitations of M.2 ]

https://www.tomshardware.com/news/samsung-memory-semantic-cxl-ssd-brings-20x-performance-uplift

Intel has killed Optane , but something similar with a CXL 2.0-3.0 is likely to arise from those ashes later. ( either from Intel or someone else. )



Having one and only one internal drive for the Mac Pro is problematical in 2022 and going forward. (Apple basically said that in 2017 so it would be strange for them to abandoned that insight at this point ). Apple's SSD controller work is no where near the bleeding edge. Nor does it really have to be for vast majority of Apple's products. And if have a sub 512GB RAM capacity cap then issues are only multiplied (as some folks are going to need something like a DRAM based SSD to hide latencies being forced out through TBv2 would cause.). Apple is getting some large Pref/Watt wins but they are not 'free'; they are making some trade-offs that probably need at least some compensation in systems at the Mac Pro level. ( Doesn't mean throw Perf/Watt out the window but allocating some power for some relatively decent amount of I/O will not completely reverse things. And only need just so many TB controllers. After six controllers are seriously in greatly diminishing returns zone. There is a substantive chuck of area and transistors they could devote to do some more general, relatively local (so limited power overhead ) I/O. )


Apple just opened up the Metal 3.0 API to where the GPU can pull data straight off the drive. Having a faster drive will help their iGPU cover more ground. Even their SoC has some need here over the longer term future. And TBv4 isn't going to help there.


The other factor on external, on some of those use cases is that it is just that more expensive. Apple can duck being in the container building business, but that is just a ballon squeeze for the end user. Apple also is going to need some product segmentation from the Mac Studio. It doesn't mean they have to try to contain everything, but trying o contain nothing is just another extreme that is about just as much off target. Mac Pro is already very expensive. To drive even higher costs is likely to push more users to the tipping point of looking elsewhere for affordability ( and Perf/$ ) reasons.




and, since most old cards wouldn’t work anymore anyway, there may not be a lot of interest from the vendors updating those hardware solutions.

There is a chicken and egg issue. If choke off bandwidth to x8 (or more ) PCI-e v3(or higher) cards then there is no place for those vendors to sell those cards in the Mac ecosystem. Has to be somewhere for their card to land that will get traction in order to commit to doing drivers. MacOS isn't gong to win all of those. Graviton/Ampere Computing/other Arm server Neoverse family implements (and descendant forks ) are going to be a more attractive market ( nobody 'afraid' of CXL 2.0 there. )


Focusing too much on old cards is flawed. Brand new cards which don't have any driver are more easily targeted to DriverKit foundation because they don't have any kernel extension (kext) overhead. Missing out on future I/O is not just bad for the Mac Pro. It is bad for the whole Mac ecosystem that has TB expansion boxes also. A wider base of folks writing Driver Kit general PCIe drivers means more TB qualified drivers also.



Slots don’t feel to be as important on a MacPro device as they’d were before. If any, I’d guess no more than two. I’m curious about the future PCI situation than I am about anything else. :)

They are not as important ( as 10-15 years ago), but the Mac Pro isn't sitting as big of a market share either. Lots of folks who didn't need them have shifted to MBP 16" and Mac Studio like systems. But once 'boil off' that more flexible demographic, Apple is left with a more skewed demographic of remaining users. If Apple tosses discrete GPU then they will have boiled down to an even smaller demographic. That is going to skew the remainder demographic even more; not less .
 
TB4 is just as higher fraction of the same old x4 PCI-e v3 provision feed that have been plateaued at for a while.
Yes, but go beyond GPU’s, Accelerators, and interface cards, you end up in an area where TB4 is sufficient, if not more so, networking and audio. It could be that the only MUST have use case is for high speed storage.
Apple just opened up the Metal 3.0 API to where the GPU can pull data straight off the drive. Having a faster drive will help their iGPU cover more ground.
Does it pull straight off “any” drive? From the WWDC video:
With Metal 3's fast resource loading, your games and apps can load assets with low latency and high throughput by taking advantage of the Apple silicon unified memory architecture and fast SSD storage included with Apple platforms.
It doesn’t specifically call out non-Apple storage, but I haven’t watched the entire video yet.
 
Yes, but go beyond GPU’s, Accelerators, and interface cards, you end up in an area where TB4 is sufficient, if not more so, networking and audio. It could be that the only MUST have use case is for high speed storage.

Chuckle. So 'storage' suddenly drops off your list because I enumerated several examples where storage didn't fit.
Guess what? That wasn't the full enumeration. That was more to indicative that you haven't done as much homework as you think. Everything on the SonnetTech compatibility list that is marked with footnote #8 has TBv4 bandwidth issues.

Go look at HDR 12-bit 8K video RAW capture from multiple cameras. ( https://www.aja.com/products/kona-5 )
40GbE networking. ( connections to higher end SAN/NAS networks. I don't think there are any macOS single 40GbE socket options. There are some OS driver level issues but also beause don't really have targets. )

Yes there are lots of older cards from 6+ years ago that can be pumped through TBv4 to get some utility from them. Audio is more a latency than modern large bandwidth issue.




Does it pull straight off “any” drive? From the WWDC video:

Hard to tell because there is no Apple Silicon system that is not restricted to just one internal drive.
Other than laziness, there really isn't a good reason why this wouldn't work for the W6000 GPUs that Apple has support for. To use it all need to do is create a Metal MTLIOFileHandle (Metal I/O File Handle) and use that.

MTLIOFileHandle works with Mac Catalyst so it should work with Macs that run Catalyst ( which isn't only Apple Silicon macs).

Might be limited on compression formats, but Nvidia has a nvcomp library that using LZ4. An article from 2020:



Finally,

Metal 3 GPUs include the newer AMD GPUs. and Fast Resource Loading as a check box for the Metal3 column and no footnote indicating exceptions to blocks the non Apple GPUs.


Most GPUs have some texture specific hardware but if there is decent GPGPU capability decompressing on the GPU shouldn't be a huge hurdle. Unified Memory is not inartistically what makes this work at all. The Apple Accelerate library having support for these formats and Apple doing to work to extend the library support to their GPU is what matter. Pretty likely chance though that this is more optimized written software than buried , built-in hardware de/compressor hardware.

Apple's fast loading API isn't all that unique. Nvida, Microsoft , etc all have them. This is more 'catch up' by Apple than something that only Apple hardware can do.





It doesn’t specifically call out non-Apple storage, but I haven’t watched the entire video yet.

No magic voodoo really necessary on the NVME SSD end either for the Apple SSD. The SSD is on the PCI-e bus. Send the data to return address X.
 
And yet the Mac Pro 2009-2012 has half the slots as the Mac Pro 2019 and did OK in the market. There is a slot survey thread in the Mac Pro forum that had a decent number of entries where there was still an open slot even with four. Even more where one slot was consumed because the motherboard support for USB was too old (needed USB3.0 or SATA 6Gbps. where modern logicboard would more than fit those needs. ). Another common occurance in the slot surver was one 'boot' GPU and one newer/faster GPU. If you take all the dGPUs out of the slot survey there are empty slots in every single one of those 4 slot systems.



Going to a "half sized" Mac Pro If Apple took off the extended feet and handle and one MPX bay (and associated fan), they still have 4 slots left and a 300+W SoC. If shave the height in half than volume would drop in half but still would be able to take standards complaint full length and full height cards. A significant issue if counting just the central box of the system or the whole tower height including the extensions.

If they added 2 M.2 slots on the backside of the back into addition to their proprietary SSD modules, then that would probably alleviate some of the pressure causing one of the x8 slots being filled with a M.2 carrier card for SSD storage ( need more than just one internal SSD drive). If they provisioned two Apple SSDs on the logicboard ( only using Duo/Quad SoC with mulitple SSD controllers), then that two would cut down on need for slots consumed by storage (but somewhat dismal $/GB costs for end users. )


That MPX storage module that soaked up 4 slot widths really didn't work all that well (space efficiency wise). There are no macOS M-series GPU card driver support so 2-3.5+ wide GPU cards likely won't be soaking up volume/space either. If they remove the legacy spec busting cards then the internal volume requirements are not as high.


There was a rumor about a one-slot-wonder that also now also appears to be in the "never gonna ship" status.
If there is only one slot then there is a significant backslide issue. ( to get to half the volume but only one slot Apple would have had to bring back drive bays... which some folks would stand up and clap for. I doubt that is coming though. More like more than half as small with just one slot. )


There are some advocates for the "half size" meant chopping height , width and depth. Chopping away full length cards ( maybe going do to 3/4 or bit longer than 1/2 length cards). Some I/O cards would fit but some also would get pruned.





If not enough folks buy it then it would incredibly foolish for Apple to make it.

Apple increased the entry price by 100% , that is highly unlikely not going to increase sales.
The number of folks who need 4 HDX sound cards in a single box is far smaller than the number than need just 1 or 2.

The Mac Pro 2019 has a "low volume tax" built into the price. If Apple has at this point made back all the engineering costs of making the 2019 model then it has been in now way shape or form a foolish effort. They have made a profit. End of foolishness. With MPX W6900 modules priced higher than MPX W6800 Duos (and RAM and SSD pricing, and ">1 TB RAM" tax on 24/28 core models ), what is a bit foolish is to think they have not made their money back yet at this point. Getting paid in full for work that you do is not incredibly foolish.

While there are fair number of PCI-e cards that have gotten updated drivers for macOS on Apple Silicon there are sizable number of other ones that have not. Before the Apple Silicon was even introduced Apple announced at WWDD 2019 that kernel extensions (kexts) are deprecated on Intel ( and implicitly on Apple Silicon). Some cards have macOS on Apple Silicon have driver support via kexts. So those drivers were in deprecated mode on day 0 on Apple Silicon. Maybe those card vendors are kicking the can down the road. Some of them probably are not . If Mac Pro 2019 is where the log jam will form where some folks stop writing drivers that work... then an abnormally large number of card will dead end there. So that slot number expansion has some "legacy card collector" dimensions to it rather than forward looking professional utility. I don't think the Apple thinks the number of PCI-e cards supported are going to shrink to zero. But I do think that they expect some number of cards to be "garbage collected" as defacto abandoned in terms of drivers under active development. They are likely not laboring under a notion that Mac Pro of future will have every possible "Pro" card out in the market covered by macOS. Apple isn't building everything for everybody.



It wouldn't be very surprising if Apple only included more than 4 slots only in a $1-1.5K more expensive rack version that was provisioned via an internal slot expansion add in board that "snaps" to the baseline board. All the audio card that don't really stress PCI-e v3 and are relatively low power , higher volume consumers would still fit. (but at higher prices, but still a relatively low bandwidth lane provision out of the main SoC. ). Lots of A/V focused stuff is racked anyway.
Well if you consider the need for a sound card, an SDI I/O card, potentially a M.2 raid card, a dedicated GPU (depending on support) we’re all out of slots should the availability be reduced.

I happen to think that the cost of the 2019 MacPro is heart breaking because I personally can’t afford one and believe it is mainly driven by the enormous cost of the Xeon processor. But the industrial design throughout is near perfect.

I would just hate to see what has been such a design and engineering success for professionals after the flop and admittedly personal purchase of the MacPro 2009 for them to make a repeat on account of the the Apple Silicon allowing them to reduce the size.

I’d buy a MacPro the size of a wardrobe if it meant I could do all the things in a Mac environment.
 
I really fail to see a market spot for another tiny Mac Pro – now that Apple already has the Mac Studio out. In my opinion the Studio is the tiny Mac Pro we heard the rumors about.
If not for expandability i.e. with PCI-e cards, what's the goal of releasing another Mac model for Pro users with a different design?
The Studio was already designed with a huge cooling system, I'd be really surprised if Apple designed the Studio so poorly that it won't be able to house an M2 or M3 ultra in the future… I can see Apple adding something extra to the Studio, like more ports and then call it Mac Studio Pro or Mac Studio Max – this would fit their current naming terminology…

But if they keep bringing us a real Mac Pro with AS – which they already explicitly teased us with – I fail to see a niche in the market, if not for expanability. Either through industry standards like PCI-e, or something proprietary… like the MPX modules. This would ensure expandability to at least some degree while keeping cashflow close to Apple…

Something I could imagine for a new AS Mac Pro is a new connector similar to UltraFusion, which they use right now to "connect" 2 M1 Max to make an M1 Ultra. If Apple can connect 2 of their M1 chips together at great performance, why shouldn't this also work with other components. Something like that could possibly work for adding extra RAM later, but not regular RAM as we know it from the Mac Pro today, but some special SOC RAM components that we can get only from Apple – and of course with a premium price tag…
I can mostly agree. But as an editor / VFX student / technical assistant you can never have too much compute power.

If the MacPro now has to surpass the Mac Studio M1 Ultra then it’s going to be going toe to toe with Threadripper. It will need to exceed 512GB or RAM for simulations and caching (sims in Houdini / large renders / machine learning data sets) it bodes well for the capabilities of Mac Pro in the future. But that falls short if I can’t put a PCI-E SDI card in to interface with my reference monitor and loop out to the review room. Or even add in a M.2 raid card for obscene write speeds.
 
Well if you consider the need for a sound card, an SDI I/O card, potentially a M.2 raid card, a dedicated GPU (depending on support) we’re all out of slots should the availability be reduced.

Three out of four ain't bad...?

I am thinking the first ASi Mac Pro might have three or four PCIe slots...

As for discrete GPUs, if there are any at all, they would most likely be an ASi GPGPU...?

I happen to think that the cost of the 2019 MacPro is heart breaking because I personally can’t afford one and believe it is mainly driven by the enormous cost of the Xeon processor. But the industrial design throughout is near perfect.

Three other things contributing to the cost of the 2019 Intel Mac Pro; the custom motherboard, the custom PSU, and the highly machined chassis...

I would just hate to see what has been such a design and engineering success for professionals after the flop and admittedly personal purchase of the MacPro 2009 for them to make a repeat on account of the the Apple Silicon allowing them to reduce the size.

I feel the ASi Mac Pro will be reduced in size, but only one fans worth of height removed; going from eight to four PCIe slots...

I’d buy a MacPro the size of a wardrobe if it meant I could do all the things in a Mac environment.

LOL, like an old Silicon Graphics Onyx "fridge", needing 3-phase power and dedicated cooling...! ;^p

If the MacPro now has to surpass the Mac Studio M1 Ultra then it’s going to be going toe to toe with Threadripper. It will need to exceed 512GB or RAM for simulations and caching (sims in Houdini / large renders / machine learning data sets) it bodes well for the capabilities of Mac Pro in the future. But that falls short if I can’t put a PCI-E SDI card in to interface with my reference monitor and loop out to the review room. Or even add in a M.2 raid card for obscene write speeds.

From the RAM bump in the M2 SoC, one could interpret that the ASi Mac Pro might go up to 384GB of RAM for a M2 Extreme (quad M2 Max SoCs) configuration...

Now, if Apple has the ability to go with high-density LPDDR5X SDRAM, then the maximum RAM could go up to 1TB, but it will cost all your extremities...! ;^p
 
Well if you consider the need for a sound card, an SDI I/O card, potentially a M.2 raid card, a dedicated GPU (depending on support) we’re all out of slots should the availability be reduced.

There are zero macOS on Apple Silicon drivers for a discrete GPU card so not sure why counting that slot as used here. That is not on the likely possible path. So even in this example there is a likely empty slot. Part of the trade off in making a "system on a chip" is that some stuff goes into the SoC. The upside to that is that it frees up PCI-e slots. Some folks will point to it as a negative, but it has two sides.

In the Mac Pro (2008-2012) slot survey ( which is limited demographic sampling, but something concrete to looking at. ) there were there were about at least as many occurrences of two GPU set ups than one GPU ones. One of those GPUs was typically tagged as the "boot" GPU because it was a officially supported card by Apple and presented a hack-free access to a boot screen. If the "boot GPU' is in the SoC that too has the upside of freeing a slot.

Afterburner card soaking up a slot ... gone with M2 era SoC.

The computational deficient of a 2022+ era M2 generation SoC to a couple of 8-10 year old DSP processor is going to be what for the average small , limited budget home studio?

Apple has been skimming use cases that would have strictly required a Mac Pro in 2008-2009 into other Mac products over the last decade. That is only accelerating with the move to the M-series SoCs. Those SoC don't have to cover "everything" to peel off more than decent sized percentages.


I happen to think that the cost of the 2019 MacPro is heart breaking because I personally can’t afford one and believe it is mainly driven by the enormous cost of the Xeon processor. But the industrial design throughout is near perfect.

It isn't driven by the Xeon processor. The entry Mac Pro 2019 has a 8 core W-3223



The tray price for that is about $750 and Apple has around a 30% mark-up on that so $975. $390/$5999 == 16.3% . It is less than 20% of the system cost. Even if throw in the DIMM slot costs that are likely not being used it likely still under 20%. The chipset that is required with the processor.

https://www.intel.com/content/www/u...l-c621-chipset/specifications.html?wapkw=c621

$61 . Plus mark up is $80 . So processor+"have to buy chipset" is 18% of total system price. There is still 80% of the cost that is just not the immediate processor costs.

If the Xeon W-3223 processor+chipset cost was zero a $4,999 Mac Pro still would have been a 66% increase over the $2,999 starting point for the MP 2013 (with a Xeon E5v2 processor in it).

Pointing that the hyper costs for the 24-28 core models is not where the main driver for the 100% entry price comes from. That is just misdirection. Additionally, hand waving at ECC RAM isn't doing it either (apple's mark up on RAM is even more ).

That double sided , relatively very large, complex circuit layout motherboard probably costs almost as much as the entry processor does. Can try to blame that on the processor for that, but that is a stretch.


Apple just has a "low volume" tax on the Mac Pro. The volume of systems sold in and of itself probably isn't interesting financially to Apple. Back in the 2009-2012 era there was a CPU daughter card that allowed the same chassis to service folks who just needed a limited number of DIMM slots and one CPU package and those who needed lots of DIMMs slots and two CPU packages with just one product across a relative (for a Mac product) wide range of prices. The 2019 takes those "max DIMM slot , double CPU package" entry prices and puts a single CPU package solution to for. The iMac Pro - 2020 iMac more so the pricing point for the old school single CPU folks. Extremely likely Apple knew that was going to lead to less "Mac Pro" sold and hence the "low volume" tax.



I would just hate to see what has been such a design and engineering success for professionals after the flop and admittedly personal purchase of the MacPro 2009 for them to make a repeat on account of the the Apple Silicon allowing them to reduce the size.

The whole entire history of the general PC industry has been amount doing more with computers that were smaller size. Started off with don't need mini or mainframe computer to do work. Has shifted to laptops can do work previously classified as "desktop only required" . Now folks have personal computers in their hand. Saying the size can't shrink any lower than 'xyz' is largely a form over function viewpoint.


It is 100% not a repeat if go back to the same number of slots the 2009 had. Zero slots of MP 2013 and 1-4 of a new Mac Pro is a change. If it is only one , then it id not much of a change, but it would not be retreated back to what they did in 2013. If 3-4 slots then pragmatically it is the same as what was there, that was a viable product.

Also just because the "tower" Mac Pro gets smaller doesn't necessarily mean the "rack" Mac Pro would get smaller. Reducing the width of a "rack" model make it harder to rack since the rack rail distance is standard. If exactly half the width Apple could build a special rail chassis that took two of these. The would get give slightly less "bad" rack density (still way off Mini's density for most rack use cases) . Or Apple could take a half size 1-4 board and add on a 2-4 expansion board to flush out the missing rack width.

If the folks with abnormally high number of cards are highly correlated with rack models then there is about zero design and engineering success 'loss' there. It won't be cheap. Rack is $500 increase now. Probably could go up $800-1000 increase ( not shifting costs to users who never use that many slots. Have to self pay) . But Apple isn't aiming at folks with low budgets.


Apple stated at the Mac Studio intro that the more common configuration for the Mac Pro was a 16 core and W5700. That means the double W6800 duos and 28 core options are likely way outside the norm. Sell too few and Apple is likely not going to stay in that category.



I’d buy a MacPro the size of a wardrobe if it meant I could do all the things in a Mac environment.

Toward the beginning of your response you effectively said you would not. Lament the current price and then say will buy something that Apple would highly likely charge even more money for you would buy. Goes back to folks who wished that Apple sold a 'slot box" xMac for $1,599-2,399. They didn't with Intel. They even less likely to do that now ( the Mx , Mx Pro , Mx Max provision no substantive PCI-e lanes to do that at all. ).

The Mini , iMac, and lower 'half' of the Studio are being provisioned with laptop optimized SoCs. Apple has detached themselves from the general desktop PC market support for the R&D for a mainstream desktop So C (e.g., desktop Ryzen 7000 or desktop i9-i7 Gen 12 ) . over 70+ % of what Apple Macs sold are laptops so that is where the focus is. That leaves the Mac Pro way, way , way out on an island. It isn't going to be a 'low cost' island.

Apple's mid range is going to get covered by laptop optimized SoCs. That ship has sailed.

I doubt "half size" Mac Pro would drop the cost much at all. The highly custom , relatively very low volume Apple SoC that go into the upper half of the Mac Pro line up are going to be expensive. More likely Apple will crank up the minimal RAM and SSD storage capacities to keep the $5,999 (or higher ) base price on the next Mac Pro. That will help the perceived value proposition a bit. Apple price anchors the remaining Mac Pro users at $5,999 with an Intel model and then 2-4 years later come with M-series model at the same price point. They don't have to ask the remaining folks to adjust their spend much at all.
 
No, storage suddenly becomes the most important use case on the list. It’s the thing that, for me, makes multiple PCIe slots MOST likely.

Folks keep conflating 'most' with 'preference'. They are really too different concepts covered by two different adjectives.

There are non rigid , legacy PCI-e standard slot ways of adding storage without using up a full slot length x height x width worth of volume. M2. , U.2 , E1 and/or E3 EDSFF drive bay are mote volume efficient. Could probably do two E3 EDSFF drive bays in less space than the old "cheesegrater" Mac Pro did four 3.55 drive bays. And pull the internal drive bay out of the downstream slip stream of the SoC heat exhaust as a bonus.

A standard slot is most likely if they want to do something in addition to storage. If storage was the only thing that matter there are other paths. Picking the standard slot approach is so that don't have to pick a specific storage path. ( two E3 EDSFF slots would take a narrow range of options).

This issue for the Mac Pro is that just the base chassis is going to have very expensive. So going to attract folks looking to keep their $1,000-5,000 , 'sunk cost' cards going. For them, that will have higher utility value. A standard slot means not trying to hit a single use case utility value. Trying to hit several of that to kill multiple birds with one stone. The aggregate use case is what is most important.


The aggregate use case is likely also why the 3rd party GPU cards are getting dropped. Intel's driver purgatory with their new dGPU drivers is exactly what Apple does not need at all with the transition to the new Apple GPUs. The aggregate Mac market needs developers 100% focused on optimizing their apps for Apple GPUs for their macOS on M-series apps.

Similar for old abandoned card drivers that don't want to put effort into DriverKit transition. The base common kernel upon which iOs/macOS/iPadOS is being restructured and if they don't want to 'get with the program' then they are out. in part, Nvidia got kicked out for not getting on board with the roadmap ( Metal is #1 priority and the kernel is changing.)



P.S. a generic slot is also cheaper for Apple. Lower Apple bill of material cost , but charge a high mark up for it. ( the margins are higher than if they did one of the storage targeted options. )
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.