I wouldn't be surprised if prototypes have been developers hands for 6 months now.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
New Mac Pro and Mac mini HAVE to be announced this year
- Thread starter rm5
- Start date
- Sort by reaction score
No. Rosetta 2 has explicit areas of exclusion where it does not apply.
" ...
What Can’t Be Translated?
Rosetta can translate most Intel-based apps, including apps that contain just-in-time (JIT) compilers. However, Rosetta doesn’t translate the following executables:- Kernel extensions
- Virtual Machine apps that virtualize x86_64 computer platforms
..."
About the Rosetta Translation Environment | Apple Developer Documentation
Learn how Rosetta translates executables, and understand what Rosetta can’t translate.
Rosetta2 doesn't allow 'hodge podge" mixes of x86_64 and ARM binaries either. (yet another reason why kernel extensions are out). Rosetta2 only does a defined subset of x86_64 code that is largely targeted to just covering simple user level applications. Boot some 1990's version of DOS , boot Windows 7 , etc.... Apple threw that out the window. (and for very justifiable reasons )
Apple deprecated the classic x86 kernel extensions (drivers ) back in 2019. There is a whole new driver framework cards/devices are suppose to be moving to which is called DriverKit. In the new system, drivers do not run inside the kernel. They run in a special "in between" state where they get a rigid, more secure , narrow slice of shared kernel space that is only good for their device. So even x86_64 drivers need rewrites for the longer term toward the end of macOS on Intel lifetime.
Kernel extensions depended on a now deprecated framework called IOKit. IOKIt has abstraction classes for both PCI-e cards and Graphics Cards. DriverKit has a PCI-e card abstraction (there are already dozens of cards that work. With the new iPadOS also work with Mx powered iPads too. That was the major new 'feature' at WWDC 2022). Back in 2019 Apple said they were going to roll out all the replacement classes over time. WWDC 2020, 2021 , 2022 came and went and still only have the PCI-e card abstraction. Period.
There is no Graphics Card abstraction in DriverKit. Three years ... nothing. Some claim that "nothing" means there is still a chance. That's doubtful. It would be somewhat challenging to put that Graphics driver into that "in between" zone. With very good IOMMU support it should be doable. But there is less "drama" if stays inside the kernel. But Apple is making the kernel a 3rd party code exclusion zone. There is no mismatch of putting Apple GPU code into Apple only kernel. Apple GPUs are also critically dependent upon having extremely good drivers. So any distractions and sideshows really aren't going to help that objective.
If Apple Metal compute API wasn't as intertwined with Metal GPU ( distinct APIs like OpenGL and OpenCL , or Vulkan and CUDA ) then the PCI-e Driver kit abstraction with a "compute API" library would be enough to get a GPGPU "compute" card into the frameworks that Apple provided. Apple has deprecated OpenCL so that is messy also.
For non super early boot process provisions cards ( GUI for boot screen , or specialized storage for boot disk (software RAID) ) there are problems too. For generic cards that are not providing GUI windows or boot stuff there is a documented , open path to having Apple Silicon Drivers on DriverKit that work. There is even small, 'stop gap" kludge of having kext that will "happen to work for now" work. Although the latter are "doomed" from the start has Apple already said before the transition even formally started that kext were going away. Nobody should be counting on that for very long term stability. For card vendors who want to port their code just once and have it live on a long term evolutionary path, DriverKit is a more cost effective path over the long term if going to stick with the Mac ecosystem over the long term.
Why would Sonnet Technology create a rack mount Mac Studio with slots if 'nothing' was going to be able to go into the slots?
xMac Studio - SONNETTECH
3U enclosure to install and secure one Mac Studio with optional PCIe card expansion in a standard 19-inch rack.
Do some homework. [ Sonnet Tech has borked the tech spec link to the supported PCI-e cards. The follow is from the Echo III which is effectively what it built-into one of the xmac studio configurations. ] Here is a PDF file.
In short, the "M1" column already has >50 cards there. And for straightfoard M2. carriers add-in cards with mainstream standard chips, it is really not exhaustive at all ( more so skewed to highlight what Sonnet Tech themselves have to offer. )
[ Some "glass half empty" folks are going to point at there are lots more than 50 cars listed there. Chop off the ones that don't have macOS on Intel either. Chop off the ones that are pragmatically dead (no new driver support in years), chop off the GPUs. Chop off those that requirement more than x8 PCI-e v3 bandwidth. It isn't as long as that skewed viewpoint tries to make it out to be. ]
"GPU cards are the only cards that matter" is a large pile of male cow poo-poo. There are far more users that have real needs for more than one internal drive ( increased capacity , multiple boot drives, and/or faster drive throughput ) than some blazing high end GPU card. NVMe standard drivers should cover cards that present a standard NVMe interface. There is no need for constantly mutating drivers to cover some quirk some game developer threw into their app and is making the GPU vendor chase down a proper fix.
Apple has done little to nothing in the M-series transition Macs so far to solve the "one and only one" internal drive problem/issues. APFS is largely written for SSDs. The easist path for Apple to provisioning more SSDs internally is a generic PCI-e slot. Folks can add in whatever specialized SSD interface they want M.2 , U.2 , E1/E2 EDSFF , etc.
The PCI-e cards that are not going off-the-charts with power consumption don't really have a major disconnect with Apple. The bigger issue is that Apple has a bigger issue with implementations that have insatiable power consumption increase objectives. The new PCI-e v5 aux power standards are shooting up to 600W
PCIe Gen5 "12VHPWR" power connector to feature 150W, 300W, 450W and 600W settings - VideoCardz.com
Intel’s ATX 3.0 specs for PCIe Gen5 power cable It appears that official specs of the PCI-Express 5.0 power cable for next-gen graphics cards have been leaked. Officially going by a name of “12VHPWR” is the upcoming standard for PCIe Gen5 graphics cards. The data provided by Intel, who are...
Every single M-series SoC introduction Apple gets up and preaches a sermon on Pref/Watt. Hard charing toward 600W is not what they are trying to do. Cards that are 75W bus powered (no aux cable flapping around) ... Apple has zero problem with. In a previous post above there are > 50 cards that work right now with Apple Silicon. Are anything of them huge power hogs? Nope. Is that entirely a coincidence ? Probably not. Not purely coupled , but not purely decoupled either. Apple also doesn't want to do everything for everybody. So they aren't going to provide even I/O option under the Sun straight from the SoC ( Apple isn't trying to maximize the number of I/O pads/pins coming off their SoCs either. ).
Afterburner was a $2K prototype toward getting the bugs out in a deployed environment before stuff the functionality into the SoC. T2 was a transition-step 2 chip. MPX set a 500W limit but was step zero on going down on GPU power consumption, not up. MPX was also assisting on integrating Thunderbolt more naturally into the system ( what Intel was already doing in 2020 with their mobile SoCs. ) .
Last edited:
Why would Sonnet Technology create a rack mount Mac Studio with slots if 'nothing' was going to be able to go into the slots?
xMac Studio - SONNETTECH
3U enclosure to install and secure one Mac Studio with optional PCIe card expansion in a standard 19-inch rack.www.sonnettech.com
Do some homework. [ Sonnet Tech has borked the tech spec link to the supported PCI-e cards. The follow is from the Echo III which is effectively what it built-into one of the xmac studio configurations. ] Here is a PDF file.
In short, the "M1" column already has >50 cards there. And for straightfoard M2. carriers add-in cards with mainstream standard chips, it is really not exhaustive at all ( more so skewed to highlight what Sonnet Tech themselves have to offer. )
[ Some "glass half empty" folks are going to point at there are lots more than 50 cars listed there. Chop off the ones that don't have macOS on Intel either. Chop off the ones that are pragmatically dead (no new driver support in years), chop off the GPUs. Chop off those that requirement more than x8 PCI-e v3 bandwidth. It isn't as long as that skewed viewpoint tries to make it out to be. ]
"GPU cards are the only cards that matter" is a large pile of male cow poo-poo. There are far more users that have real needs for more than one internal drive ( increased capacity , multiple boot drives, and/or faster drive throughput ) than some blazing high end GPU card. NVMe standard drivers should cover cards that present a standard NVMe interface. There is no need for constantly mutating drivers to cover some quirk some game developer threw into their app and is making the GPU vendor chase down a proper fix.
I had no idea this existed.
It's good to see that all these cards can be used via Thunderbolt. Of the ~50 cards working on Apple Silicon, about half are SSD adaptors or USB 3 cards. There are already tons of Thunderbolt to SSD and Thunderbolt to USB adaptors.
I never said "GPU cards are the only cards that matter". There are plenty of options available for additional storage via USB or Thunderbolt. This enclosure essentially offers up the same.
I don't know how many of these are newly developed drivers or they just happen to work by coincidence. In the PDF, some discontinued items are listed as compatible such as the RME HDSPe AIO.
Apple has done little to nothing in the M-series transition Macs so far to solve the "one and only one" internal drive problem/issues. APFS is largely written for SSDs. The easist path for Apple to provisioning more SSDs internally is a generic PCI-e slot. Folks can add in whatever specialized SSD interface they want M.2 , U.2 , E1/E2 EDSFF , etc.
The PCI-e cards that are not going off-the-charts with power consumption don't really have a major disconnect with Apple. The bigger issue is that Apple has a bigger issue with implementations that have insatiable power consumption increase objectives. The new PCI-e v5 aux power standards are shooting up to 600W
PCIe Gen5 "12VHPWR" power connector to feature 150W, 300W, 450W and 600W settings - VideoCardz.com
Intel’s ATX 3.0 specs for PCIe Gen5 power cable It appears that official specs of the PCI-Express 5.0 power cable for next-gen graphics cards have been leaked. Officially going by a name of “12VHPWR” is the upcoming standard for PCIe Gen5 graphics cards. The data provided by Intel, who are...videocardz.com
Every single M-series SoC introduction Apple gets up and preaches a sermon on Pref/Watt. Hard charing toward 600W is not what they are trying to do. Cards that are 75W bus powered (no aux cable flapping around) ... Apple has zero problem with. In a previous post above there are > 50 cards that work right now with Apple Silicon. Are anything of them huge power hogs? Nope. Is that entirely a coincidence ? Probably not. Not purely coupled , but not purely decoupled either. Apple also doesn't want to do everything for everybody. So they aren't going to provide even I/O option under the Sun straight from the SoC ( Apple isn't trying to maximize the number of I/O pads/pins coming off their SoCs either. ).
Afterburner was a $2K prototype toward getting the bugs out in a deployed environment before stuff the functionality into the SoC. T2 was a transition-step 2 chip. MPX set a 500W limit but was step 0 on going down on GPU power consumption, not up.
That has been their M.O. for a long time. I'm not saying I like it. I'm just saying that's the reality. Do I wish there were slots for additional NVMe SSDs in all the Macs? Sure. But it's not happening.
What makes you believe this thinking by Apple will change with the next Mac Pro?
It is their M.O. for laptops. They have done it to the low end Macs. The real tension for the Mac Pro is whether Apple wants to cast it into the literal desktop category or are willing to let it stand freely desk-side or under desk. ( and how much they want a setup that is not rack hostile that they want to do themselves. ).
Apple seems to have some "has to fit inside of a 7x7" inches square footprint" rule for literal desktops. If they try to 'round peg into square hole" the Mac Pro into that then they run into problems. Don't paint themselves into a corner there. then there is really no huge problem with multiple internal drives. The big change at this point is they have the Studio that already fits into that footprint. So why would they need another literal desktop machine? They do not.
Apple openly admitted that having only having one internal drive for very high end systems was a problem back in 2017. Even at half the volume of the 2019 chassis there is still plenty of space for more than one drive.
There are folks demoing PCI-e v5 x4 SSD at this point. Thunderbolt isn't going to salve that problem. Or even long term if put four (or more ) of those on a single add-in card.
If you can't get bulk data to the SoC fast enough what is the good of newer fancy future cores? With no data can't finish compute statements. If building every hungrier cores for data then need to provide the data also to have a balanced system. Apple isn't really out to create greatly unbalanced system. That isn't their M.O. either. If going to build ginormous packages with gobs of cores they are also going to need to feed the dragon. A single SSD isn't going to cut it for a wide range of high end workloads.
Because it isn't rigid dogma. The single drive constraint is largely driven by overall system size and they already have a literal desktop system at the high end that has to labor under that constraint.
The need for fast storage can be addressed internally. Apple just has to RAID a few SSD modules together.
It's not like Apple included a bunch of drive bays in the MacPro7,1. Yes, there were SATA ports but mounting brackets had to be purchased separately.
It is internal 'capacity' at least as much as 'speed' that is the issue. For folks with 10, 15 , 80+ TB of data to storage a single SSD drive isn't going to work.
Apple's SSD controller can deal with dual NAND 'SSD' modules already. Unlikely Apple is going to make a SSD controller broader than that only shipped in a Mac Pro. If Apple sticks with simply just reusing laptop dies to make multiple 'chiplet' packages then modifying the SSD controller so that could have muliple controllers active in a package is possible. But seems likely that will be more 'drama' than it is worth (where the encryption keys are kept and tagging one as 'primary' is extra complexity overhead. ). Relatively very few end users will be happy with the capacity pricing for that. ( less customers interested , high internal chip design costs .... probably not a good path on what is probably already 'scaring' off numerous customers with higher modularity priorities. )
Don't need a single bigger SSD ( trying to push 'RAID" as a big capacity drive maker vector). There are enough CPU cycles around that folks can layer post boot software RAID if they need it.
Pragmatically they did. M.2 carrier cards in standard PCI-e v3 slots. But legacy drive bays? No. Apple isn't going to stock and sell HDDs off the build-to-order (BTO) page. They aren't going to sell other folks 2.5" or M.2 SSD either off the BTO page.
If pick the Mac Pro in the accessories section of the Apple online store.
Storage - Mac Accessories
Backup your Mac with a portable or desktop hard drive. Safekeep all of your music, photographs, movies and more. Buy online with fast, free shipping.
www.apple.com
They are more external drives than internal. That is mainly because those drives work just as well with rest of Mac line up. So no good reason to exclude them for the Mac Pro. The two promise solutions J2i and R4i.
The R4i is likely a dead ender. MPX is probably not coming to the next Mac Pro. It is indicative yet again of a solution that is drawing far more power than a modern , focused alternative would need.
If Apple sliced the Mac Pro 2019 main box about a 1/3 less tall (one of the fans and behind disappears) , then a J2i could probably still fit on a shorter space frame the same way it does now. Apple isn't a big fan of HDDs , but being able to store 20TB on a single drive so the J2i could cover 40TB of storage. An order of magnitude more than than there 'common' primary SSD configuration is going to leave a big gap in internal storage coverage if they leave it out.
Provisioning the SATA lanes off the C620 chips set was "free" (almost since Apple had to buy chipset anyway). They'd have to add a discrete SATA controller to the logic board, but they also don't have to buy a C620 chip package anymore so they have that extra $60-80 to allocate toward buying one. If still charging around $6K for an entry system, that would be a serious Scrooge McDuck move to leave it out.
The option to do so was/is the salient factor. If want to near-line archive dozens of hours of ProRes RAW footage then SSDs are a slippery slope of being cost effective. Apple priced SSD capacity is even less close to being cost effective.
Apple wants to be directly done with HDDs (and cheaper SSD ) storage, but not all workloads are. Even Apple is still pushing TimeMachine (which is a bit more entrenched in the legacy Mac Pro user base that tapped into "internal TM" drive as one part of a common backup strategy. ). 'Ultimate speed' isn't really the primary issue as opposed to 'having a duplicate' issue. ( APFS has snapshots, but it is not quite as resilient as having a duplicate on a separate physical drive).
It is internal 'capacity' at least as much as 'speed' that is the issue. For folks with 10, 15 , 80+ TB of data to storage a single SSD drive isn't going to work.
Apple's SSD controller can deal with dual NAND 'SSD' modules already. Unlikely Apple is going to make a SSD controller broader than that only shipped in a Mac Pro. If Apple sticks with simply just reusing laptop dies to make multiple 'chiplet' packages then modifying the SSD controller so that could have muliple controllers active in a package is possible. But seems likely that will be more 'drama' than it is worth (where the encryption keys are kept and tagging one as 'primary' is extra complexity overhead. ). Relatively very few end users will be happy with the capacity pricing for that. ( less customers interested , high internal chip design costs .... probably not a good path on what is probably already 'scaring' off numerous customers with higher modularity priorities. )
Don't need a single bigger SSD ( trying to push 'RAID" as a big capacity drive maker vector). There are enough CPU cycles around that folks can layer post boot software RAID if they need it.
Pragmatically they did. M.2 carrier cards in standard PCI-e v3 slots. But legacy drive bays? No. Apple isn't going to stock and sell HDDs off the build-to-order (BTO) page. They aren't going to sell other folks 2.5" or M.2 SSD either off the BTO page.
If pick the Mac Pro in the accessories section of the Apple online store.
Storage - Mac Accessories
Backup your Mac with a portable or desktop hard drive. Safekeep all of your music, photographs, movies and more. Buy online with fast, free shipping.www.apple.com
They are more external drives than internal. That is mainly because those drives work just as well with rest of Mac line up. So no good reason to exclude them for the Mac Pro. The two promise solutions J2i and R4i.
The R4i is likely a dead ender. MPX is probably not coming to the next Mac Pro. It is indicative yet again of a solution that is drawing far more power than a modern , focused alternative would need.
If Apple sliced the Mac Pro 2019 main box about a 1/3 less tall (one of the fans and behind disappears) , then a J2i could probably still fit on a shorter space frame the same way it does now. Apple isn't a big fan of HDDs , but being able to store 20TB on a single drive so the J2i could cover 40TB of storage. An order of magnitude more than than there 'common' primary SSD configuration is going to leave a big gap in internal storage coverage if they leave it out.
Provisioning the SATA lanes off the C620 chips set was "free" (almost since Apple had to buy chipset anyway). They'd have to add a discrete SATA controller to the logic board, but they also don't have to buy a C620 chip package anymore so they have that extra $60-80 to allocate toward buying one. If still charging around $6K for an entry system, that would be a serious Scrooge McDuck move to leave it out.
The option to do so was/is the salient factor. If want to near-line archive dozens of hours of ProRes RAW footage then SSDs are a slippery slope of being cost effective. Apple priced SSD capacity is even less close to being cost effective.
Apple wants to be directly done with HDDs (and cheaper SSD ) storage, but not all workloads are. Even Apple is still pushing TimeMachine (which is a bit more entrenched in the legacy Mac Pro user base that tapped into "internal TM" drive as one part of a common backup strategy. ). 'Ultimate speed' isn't really the primary issue as opposed to 'having a duplicate' issue. ( APFS has snapshots, but it is not quite as resilient as having a duplicate on a separate physical drive).
I'm not debating the need and/or desire for many people to have more mass storage. But there are so many USB, Thunderbolt, and NAS solutions out there that I don't see it as much of an issue at all. We are talking about mechanical 2.5" and 3.5" drives. USB, Thunderbolt, and 10GBase-T have more than enough bandwidth for large storage.
When I replaced my hackintosh with a Mac Studio, I took the 3.5" drives from the hackintosh and threw them all in to a NAS and access them over 10GbE. Everything has been working swimmingly.
I configured my Mac Studio with more than enough internal storage for my needs. If/when I ever need more, I can always look to Thunderbolt solutions. I've seen enclosures for up to four NVMe SSDs. While they may still not perform as well as the internal SSD, it's still extremely fast.
Back to the Sonnet Tech enclosure... Besides the storage related and USB PCI-e cards. There are a handful of Blackmagic cards and Audio interface cards marked as compatible. Both Blackmagic and RME already have Thunderbolt and/or USB equivalents to those cards. If anyone must use the PCI-e cards instead, Sonnet Tech proved that just using a Thunderbolt to PCI-e bridge will get the job done. So, what incentive or reason does Apple have to include a bunch of PCI-e slots in the next Mac Pro? Again, PCI-e slots are the antithesis of Apple's M.O. to build "closed boxes".
I've seen a lot of people saying that the Apple Silicon Mac Pro will be released in 2023, but I feel like this conflicts with the whole idea of the Apple Silicon transition taking two years. But the key word is released, so that means it could still be announced before the two-year mark is hit this November. Whether or not the announcement constitutes as completing the transition is up to Apple.
Based on how the previous Mac Pros were released around six months after being announced, if the Apple Silicon Mac Pro is announced this month, October 2022, then it could be scheduled to release in April 2023.
Side note:
Building the RAM into the SoC would require each M2 Max to have 384 GB of RAM, which would be a huge excess if the standalone M2 Max's found in the MacBook Pro and Mac Studio would be the same as the ones that get UltraFused™ together to make M2 Ultra's and M2 Extreme's, which is currently the case with the M1 lineup.
Besides, having more M2 Max SKUs with differing amounts of RAM may not be as practical as just building DIMM slot support into every M2 Max and only making use of it on the Mac Pro.
Based on how the previous Mac Pros were released around six months after being announced, if the Apple Silicon Mac Pro is announced this month, October 2022, then it could be scheduled to release in April 2023.
Side note:
I feel like Apple could get the same 1.5 TB of RAM in the Apple Silicon Mac Pro by just allowing the M2 Ultra/Extreme chips to use DDR5 DIMMs, the same way they could allow them to use PCIe slots.
Building the RAM into the SoC would require each M2 Max to have 384 GB of RAM, which would be a huge excess if the standalone M2 Max's found in the MacBook Pro and Mac Studio would be the same as the ones that get UltraFused™ together to make M2 Ultra's and M2 Extreme's, which is currently the case with the M1 lineup.
Besides, having more M2 Max SKUs with differing amounts of RAM may not be as practical as just building DIMM slot support into every M2 Max and only making use of it on the Mac Pro.
Technically, the M1 Max was never "fused" to make the Ultra. Ultras were already being burned in at the time that the Max was first released – Maxes are simply splits of Ultras.
They tried that in 2013 and it back-fired on them so badly they had to publicly apologize for it and then invested a not-insignificant amount in remaking the 2006-2012 model for 2019 to bring PCI-e back.
Plus they already have the Mac Studio for "PCI-e free desktop power computing".
But the important takeaway is that they actually included space to mount internal drives. And, of course, the inclusion of PCI-e slots allowed one to add significant storage (via NVMe cards).
They tried that in 2013 and it back-fired on them so badly they had to publicly apologize for it and then invested a not-insignificant amount in remaking the 2006-2012 model for 2019 to bring PCI-e back.
Plus they already have the Mac Studio for "PCI-e free desktop power computing".
But the important takeaway is that they actually included space to mount internal drives. And, of course, the inclusion of PCI-e slots allowed one to add significant storage (via NVMe cards).
Yup. Two 2.5" drives.
Apologized for the Mac Pro? Kind of. I know Mac Pro want to take that session and claim it was only about the Mac Pro. It really was not. It is included, but there was no deep undying 'love' thrown a out there for hypermodularity ( or high prioritization).
An answer to a question about the Mac Pro and whether Apple was trying to move every Mac user to iPad Pros. :
"... The Mac has an important long future at Apple. Apple cares deeply about the Mac. We have every intention to keep going and investing in the Mac. It’s important to us, it’s important to our customers — including Mac Pro users, all pro users, including Mac Pro. And if we’ve had a pause in upgrades and updates on that, we’re sorry for that, what happened with the Mac Pro, and we’re going to come out with something great to replace it. And that’s our intention. We care about our Pro users who use MacBook Pros, who use iMacs and who use Mac Pros, who use modular systems as well as all-in-one systems, who use the pro software we make. It’s all important to us and we’re invested in that and we see a long future with that stuff. ..."
https://techcrunch.com/2017/04/06/t...-john-ternus-on-the-state-of-apples-pro-macs/
The Mac Pro is weaved in there, but at no point are they really letting go of less modular ( modular -- largely means in their transcript largely "screen less" ) . However, in 2017 Apple was "messing up" several Mac products. It wasn't just the Mac Pro. The Mini was also comatose; not as bad as the Mac Pro but comatose. Most of the laptop line up had butterfly keyboards. The Thunderbolt docking station was stock in Thunderbolt v1 zone (another comatose product) and the outsourced LG Ultrafines looked like Apple treated them as a distraction.
Apple has dribbled out MPX module updates to the Mac Pro 2019 chassis in 2020 , 2021 , and 2022. It really hasn't been a radical change where the Mac Pro system got updates on same cadence at MacBook Pro laptops. The whole M1 (first generation) went by and no Mac Pro. Pretty good chance going to get a gap just very similar to the 2010 -> 2013 gap for an up ( a bit longer). ( and not like 2006 , 2008 , 2009 , 2010 like cadence. ).
The bottom of the Mac line up is running a "iPad Pro" processor , but the software stack shift to move the "Mac" over. More so a cross pollination than an ending for the Mac line. ( iPad Pro can now run modern DriverKit drivers that macOS uses. It is a cross mixing. )
Apple took another two years to deliver a Mac Pro. The iMac Pro and Mini 2018 arrived well before ( with higher resource prioritization).
Apple mentioned doing stuff for users that require very high bandwidth. Well, UltraFusion does provision very high bandwidth. Many customers wanted a bigger , single GPU. Physically going to be a very large GPU when use UltraFusion to combine multiple dies together.
2017 eventually brought the iMac Pro and the T2 ( transition chip) . The iMac Pro was very much about what Apple discussed in that session. Following trend lines of what Pro users were buying.
And? They raised the entry price of the Mac Pro up 100%. They spent more and charged more. And took two years. [ That wasn't entirely level of difficulty , but finishing the iMac Pro first and then doing the Mac Pro. ] The notion that there is some huge sunk cost that Apple is struggling toward break even to so they "have to" continue selling the Mac Pro 2019 chassis is delusional. That is more so how does Apple fund a "hobby product" more so than a core strategic Mac Product. ( e.g., AppleTV at about 2x the price of the 'competitor' streaming devices. )
Plus they already have the Mac Studio for "PCI-e free desktop power computing".
MP 2013 --> iMac Pro --> Mac Studio
Technically not "PCI-e Free" , but leaning "too hard" on Thunderbolt is a limitation. There are cards with x8 (or more) PCI-e v3 that actually need to use that much bandwidth for some jobs.
That is pretty likely a happy accident. What to do with the empty space that was there after they finished the rest. Placing your HDDs downstream from where they consumed air heated by the CPU heat sink is not an optimal place to put those drives. In a canonical tower the drives are placed upstream from the CPU in the airstream; not behind it. [ same for 1U-5U rack servers. Drives then CPUs in front to back airflow. ]
The required PCH chipset from Intel provisioned SATA and USB anyway. It is just a route and place on the board; Apple has already bought the functionality. Apple doesn't install any bracket themselves.
Again Apple acknowledge the weakness of the "one and only one internal drive" design rule in the session but only that it had some upper limits if "leaned on that mindset too hard".
Will the next Mac Pro get some minimalistic discrete SATA controller they can hang off of one PCI-e v4 lane? Perhaps. Is it going to be equivalent of the RAID capable controller in most x86_64 ( Xeon W-6000 / Threadripper) chipsets? Probably not. If there is 'extra' space in the chassis and extra space on the logic board, then they'll throw it in . But it likely will be an inexpensive gap filler feature.
Two 3.5" drives.
Promise Pegasus J2i 8TB Internal Storage Enclosure for Mac Pro
Promise Pegasus J2i two-drive internal storage enclosure delivers 8TB of SATA storage for your Mac Pro. This module can be installed in your Mac Pro and connects to the main logic board with included cables. You also have space to add a second 3.5-inch SATA drive to your system.
www.apple.com
I've seen a 3d printed bracket for two 2.5" that takes less space. Apple really didn't formally endorse that.
They did and I was one of the people here who questioned the logic in having a macpro that was not expandable. Many people mentioned NAS and cloud based storage options and it (drive bays) wasn't needed back in 2013.
Funnily enough, now in 2022, I basically echoed that sentiment. For professional workstations (speaking out of ignorance), do people really need drive bays?
I can see the logic in using PCIe in the current Mac Pro, but would apple do that for their ARM based logic board? Technically, yes, its doable, but Apple has been marching towards more proprietary implementations, and everything they sell now is a sealed system - other then the current Mac Pro of course.
Drive bays may soon be obsolete, but NVMe slots are still extremely useful. I'd expect (an option for) 8 slots in addition to the system drive in a decent midrange workstation, and 4 slots in an entry-level model.
Some data-intensive applications require plenty of fast storage for temporary data. The storage has to be internal, because external ports like Thunderbolt are slower than SSD. And you need many slots, because individual SSD drives are quite small.
A hissy fit.....that'll be cool......seeing someone get angry and annoyed and Apple not doing anything about it. This should be quality comedy!!
and what about the needs for big storage that does not need to be SSD fast or NVME fast.
the needs for data only TB (not shared with video out?)
The need for big multi screen set ups?
So you and I disagreed strongly about whether the AS Mac Pro would feature PCIe slots. I strongly believed it would, and you strongly believed it would not. Ultimately, after realizing the discussion was just going round in circles, I said we should wait to see what Apple does and revist this then.
As you can see, Apple did include PCIe slots with the AS Mac Pro.