MP 7,1 New mac pro in december 2021 ?

[Mac3Duser]

What you are saying is very interesting.
I think Apple communicates poorly with pro-users.
Buying a mac pro is a big budget, like buying a car. Even studios need long years of reliable support and spare parts beyond apple care.
If you buy a machine between 6000 and 18000 $, it is ok if it is scalable. With the right card drivers and being able to change components easily. The mechanic must have the parts for your supercar, or for your locomotive.
I see the AS mac pro as a super mac mini but not as a workstation. Or all the studios will go to Dell and HP. And Apple will only have the pro videos and youtubers market, with optimized apps for AS.
Already, those who make the effort to buy a mac pro do so mainly because they have a love for mac os (and beautiful objects).
The mpx modules are interesting for thunderbolt.
We can consider taking an average graphics card by apple, and then buying later a better AMD graphics card but not pro, only for rendering, not for display, in addition to the base card that would be used for the display.
(The 6.1 already worked like this with two cards, one for display and the other for rendering)

 

[avkills]

What you are saying is very interesting.
I think Apple communicates poorly with pro-users.
Buying a mac pro is a big budget, like buying a car. Even studios need long years of reliable support and spare parts beyond apple care.
If you buy a machine between 6000 and 18000 $, it is ok if it is scalable. With the right card drivers and being able to change components easily. The mechanic must have the parts for your supercar, or for your locomotive.
I see the AS mac pro as a super mac mini but not as a workstation. Or all the studios will go to Dell and HP. And Apple will only have the pro videos and youtubers market, with optimized apps for AS.
Already, those who make the effort to buy a mac pro do so mainly because they have a love for mac os (and beautiful objects).
The mpx modules are interesting for thunderbolt.
We can consider taking an average graphics card by apple, and then buying later a better AMD graphics card but not pro, only for rendering, not for display, in addition to the base card that would be used for the display.
(The 6.1 already worked like this with two cards, one for display and the other for rendering)

Being able to assign different GPUs for different tasks is wholly on the developer of the app. You can do this in Resolve on any Mac with multiple GPUs.

 

[fuchsdh]

My guess is that the Mac Pro will be the last model to be replaced with Apple Silicon - that much has been predictable since the initial announcement since:

(a) The Mac Pro - along with its ecosystem of custom MPX GPUs & Afterburner - was only launched in late 2019 so discontinuing it too soon won't look good.

(b) Mac Pro users will probably have the most dependence on obscure third-party apps, plugins, libraries etc. some of which will never make it to ASi while others won't start being updated until at least the "pro" M1X MBPs and iMacs are out. (The M1 is a laptop/tablet chip, folks, that just happens to be good enough for low-end desktops).

...and we're now only a little more than halfway through an "about two years" transition period that has very likely been further delayed by world events, so it's a bit early to start reading too much into the non-appearance of an ASi Mac Pro.

However, question is whether it makes economic sense for Apple to make a "Xeon-killer" monolithic Apple Silicon CPU with dozens of cores that would probably have to concede the advantages of integrated/on-package GPU and RAM in order to offer the same sort of specs as the Mac Pro...

It would seem to make more sense to do something more radical that played to the advantages of Apple Silicon: maybe (as people have already suggested) multiple M1Xs with RAM and GPU power distributed between them. Trouble is, that's going to depend very much on properly optimised software, so it might make sense to keep the Intel Mac Pro on the books indefinitely to cater for people committed to a more traditional workflow, while releasing something more "courageous" in parallel.


Good question - and/or do they actually understand why people want them?

Only Apple know how much money they've made from Mac Pro sales. If Apple have been making good money shifting $20k workstations then they might want to continue. Also, they might see value in having a "flagship" high-end workstation in the same way that (say) VW sees value in making Bugattis. If it's helping keep that Apple logo on the end credits of major movies & TV shows then, maybe.

However, who the heck is still buying Mac Pros? Let's recap:

1. Classic "Cheesegrater" Mac Pro - pretty much abandoned after ~2010 (The EU thing you mentioned was the lack of an internal fan guard and should have been trivial to fix) and it wasn't getting any substantial upgrades in the US either.

2. After the Cheesegrater had been left to wither for a couple of years until it was almost a non-starter (and discontinued in the EU), it was dropped for a radically different design - the trashcan - totally reliant on Thunderbolt for expansion with almost no internal upgradeability, totally reliant on using dual GPUs (and no CUDA) for performance.

3. Trashcan then get's zero updates (apart from rotating out the entry-level model) between 2013 and 2019 and was probably going to be replaced by the iMac Pro in 2017 until that was rejected by key customers/partners resulting in that unprecedented press conference where Apple announced the (then) mythical modular Mac Pro.

4. iMac Pro gets launched in 2017, is never significantly updated (apart from a new GPU option?) and dropped in 2021... by which time the low end models were out-performed by the top-end iMac but if anybody had committed to the all-in-one format and needed ECC ram or the better cooling they were out-of-luck.

5. New Mac Pro gets announced in 2019, ships in early 2020 (along with a 2x price hike c.f. previous Mac Pros that simply threw lower-end Mac pro users, who just wanted some PCIe slots and internal storage, under a bus) and... six months later Apple announces that they're moving everything to Apple Silicon by ~2022 while not breathing a word, publicly, about any sort of roadmap for how they will cater for Mac Pro users (they haven't even clearly said that they won't support discrete/PCIe GPUs - that's just a, not unreasonable, extrapolation from comments in developer docs).

The Mac Pro is sold at "serious callers only" prices, yet I can't see why any "serious caller" would want to buy in to a line with such an appalling track record. Sure, if you buy a Mac Pro, Apple will be sure to have replacement parts etc. for the next 3-5 years and "real pros don't upgrade mid-life" - but what if you add an employee mid-term and need to kit them out with a new machine? What about the specialist PCIe cards that cost more than the machine that you might want to keep more than 3 years? Buy Apple and the chances are, next time your accountant tells you to buy new computers now for tax reasons, you'll have a major workflow change forced on you.

It seems that the 2019 Mac Pro was aimed at people who are so irrevocably committed to some Mac-only piece of software or file format that they will pay any price to keep working. That certainly applied in some cases, but that is going to be a shrinking pool, and the Mac Pro is doing nothing to attract new customers to Apple - outside MacOS devotees it's just the butt of jokes about $700 wheels and $1000 display stands.

Apart from being licensed to run MacOS, it really is just an Intel tower that mainly just took advantage of being an early adopter of the latest Xeon-W chip to make sure it won a game of Top Trumps when it came to PCIe slots and max. RAM.

...on the bright side, Apple Silicon might be Apple's chance to offer something a bit more distinctive that doesn't invite like-for-like comparisons with commodity workstation/server hardware.

You should always buy the Mac that exists right now, instead of hoping for the Mac around the corner. That's always been the case. If you want a Mac Pro and can afford it, you get it. I agree that they should be more transparent, since that would help, but I think the transparency is kind of irrelevant compared to just actually shipping products. Most people aren't religiously reading forums or reading leaked Intel slides to try and find the perfect time to upgrade. They just buy what's there and buy the new thing that's there the next time. Apple's problem for years now has been there wasn't a new thing.

What throws me is that the 2019 Mac Pro was something Apple hadn't made for arguably decades—a high-end tower workstation. Certainly the enthusiast crowd was looking for the cheesegrater redux—a mid-end computer with a few extra slots—and instead Apple went on the high end (which is great if you need the expandability, if not you're grossly overpaying for capability you'll never use.) To a degree this made sense, because the iMac Pro did fill, if not elegantly, a lot of the same niche as the older models (higher performance, ECC RAM, high-end internal graphics), just in an all-in-one format (and another $1K more expensive than it should have been.) But it was clearly (I don't think you have to guess, considering how the roundtable went) supposed to be the be-all, end-all of Apple's high-end until they changed their mind quite late and added on the Mac Pro.

Now, with Apple Silicon, you're left with a weird world where a lot of the same assumptions can't be made, especially since the 7,1 Mac Pro is such a different machine from previous pro Macs. The half-size model does sound like the tube Mac Pro in lots of ways, and considering that the 6,1 would have been much more popular if they just updated the damn thing, it doesn't seem to me a bad plan, especially if it leverages a lot of the chip work being done in other products (thus making it a much cheaper high-end model that fits better into their consumer focus.) But by the time Apple decided they were going to make a new, more grand Mac Pro... they absolutely knew they were switching to Apple Silicon. So the idea that they'd revive the brand and just make another one-off Intel Mac Pro in this novel form factor they'd previously abandoned, before dumping it when their own processor machines came out seems really weird to me.

 

[TrevorR90]

You should always buy the Mac that exists right now, instead of hoping for the Mac around the corner. That's always been the case. If you want a Mac Pro and can afford it, you get it. I agree that they should be more transparent, since that would help, but I think the transparency is kind of irrelevant compared to just actually shipping products. Most people aren't religiously reading forums or reading leaked Intel slides to try and find the perfect time to upgrade. They just buy what's there and buy the new thing that's there the next time. Apple's problem for years now has been there wasn't a new thing.

What throws me is that the 2019 Mac Pro was something Apple hadn't made for arguably decades—a high-end tower workstation. Certainly the enthusiast crowd was looking for the cheesegrater redux—a mid-end computer with a few extra slots—and instead Apple went on the high end (which is great if you need the expandability, if not you're grossly overpaying for capability you'll never use.) To a degree this made sense, because the iMac Pro did fill, if not elegantly, a lot of the same niche as the older models (higher performance, ECC RAM, high-end internal graphics), just in an all-in-one format (and another $1K more expensive than it should have been.) But it was clearly (I don't think you have to guess, considering how the roundtable went) supposed to be the be-all, end-all of Apple's high-end until they changed their mind quite late and added on the Mac Pro.

Now, with Apple Silicon, you're left with a weird world where a lot of the same assumptions can't be made, especially since the 7,1 Mac Pro is such a different machine from previous pro Macs. The half-size model does sound like the tube Mac Pro in lots of ways, and considering that the 6,1 would have been much more popular if they just updated the damn thing, it doesn't seem to me a bad plan, especially if it leverages a lot of the chip work being done in other products (thus making it a much cheaper high-end model that fits better into their consumer focus.) But by the time Apple decided they were going to make a new, more grand Mac Pro... they absolutely knew they were switching to Apple Silicon. So the idea that they'd revive the brand and just make another one-off Intel Mac Pro in this novel form factor they'd previously abandoned, before dumping it when their own processor machines came out seems really weird to me.

Agreed, I never understand why people say "just wait." Otherwise, its just an endless "just wait" game.

With any technology, there is no such thing as future proofing. Get what you need now and not what you think you need in the future.

 

[deconstruct60]

What you are saying is very interesting.
I think Apple communicates poorly with pro-users.
Buying a mac pro is a big budget, like buying a car. Even studios need long years of reliable support and spare parts beyond apple care.
If you buy a machine between 6000 and 18000 $, it is ok if it is scalable.

The problem is a two way street. Many folks don't do a good job of listening. Apple has been quite clear about support for decades.

"... Products are considered vintage when Apple stopped distributing them for sale more than 5 and less than 7 years ago. ... "

Obtaining service for your Apple product after an expired warranty - Apple Support

Learn about your options for getting service and parts for Apple devices that are past their warranty period.

support.apple.com

Apple starts the countdown clock when a product is discontinued (not when it starts shipping) and that countdown clock only goes 7 years.

To get to 10 years of coverage you would have to talking about a product that had at least 3 years of active sales and then go the full measure 7 years of countdown.


There is zero points in Apple's support pledge where money comes in as a factor. You can't buy more time (extended support ***). And a $600 Mac Mini has the exact same terms as a $60,000 Mac Pro. This is where the bad listening comes in. Apple pragmatically says money doesn't matter. Numerous posts pop on these and other mac forums saying money has to matter. That is just disconnected.


The notion of " oh that is only hardware .. the OS is muddled" is largely misdirection. Most large enterprise companys don't support software on obsolete equipment for free ( if hand over a check with 5-6 digits of funds on it perhaps). MacOS generally comes out in October. The beta gets finalized around May . Mac products get release at various times around the calendar year. The support pledge is in units of years. So there are countdown mismatches because the "clocks" are counting from different zones of the year. There are quirks were product look like goes past 7 but if you just add another year to clean up synchronization differences the move onto the obsolete list is extremely highly correlated to OS desupport. [ And Apple gives themselves flexibilty to cut short when they do a OS platform change. Personally, I think the inertia is too high on this cut over to only M-series in a much shorter window. Last time there were issues in Rosetta not really being Apple's but a licensed solution they had to pay for. Second the platform leaving didn't have large number of other system vendors. So it was going to collapse behind them. That is far from the present context. Windows isn't going to collapse; it is 9x bigger than macOS and Apple wrote their own translation tool this time that also leaves some subsets of apps behind (so more limited in scope ). ]

What got the Mac Pro 5,1 more time was that it got a two year extension. Released in 2010 there was also a 2012 model ( with same model number. And the 4,1 was in many ways a firmware tweak and firmware hack away from presenting as a 5,1. Basically the exact same motherboard. ). However, that was a dual edge sword in that it also didn't get any substantive improvements either.

It wasn't the money customer paid, but the lack of updates that extended the service life. The Mac Pro 2013 may get an even longer window , but again this is primarily driven by a lack of upgrade priority on Apple's part. ( won't be surprised if the software gets cut off "early" there before the hardware. And in last 1-2 years Apple just gives away money because parts are gone. )

Mac Pro 7,1 if replaced (discontinue sales ) this year .... 9 year horizon.

I wouldn't bet on the 8,1 selling past 2-2.5 years. It has better pricing value, PCI-e v4 , more cores , but it is also runs very hot. Even Intel has quickly moved their other product line up past this 10nm node ( Gen 11 Tiger Lake is on SuperFin. ). Presuming Intel doesn't dump the product quickly, if enough folks buy it then it could be a " Rip van Winkle" product like the MP 2013. But that is highly doubtful. Being the only Intel orphan at 2 years out will increasingly be a problem. And the "half sized" Mac Pro is probably designed to eat into a hefty chunk of current Mac Pro sales.


*** Apple has a somewhat new policy that if they happen to have enough spare parts in their inventory they may choose to extend things. That isn't something to bet the farm on. Apple works hard to have just enough inventory as necessary to do the job. ( So if things go to their planned schedule there is next to nothing there at the end. If they have some product defect exchange and have to do some initially planned buys later then probably a better chance of left overs. )

This still isn't even a "buy more support" issue... It is a "happen to get the lucky draw" extended support.

With the right card drivers and being able to change components easily. The mechanic must have the parts for your supercar, or for your locomotive

The extremely critical but often overlooked part there is "right card drivers". The fast pit crew parts changes doesn't matter if no software to run the hardware.

Going back to listening issues, Apple has never stated they were "all in" on the generic add-in card market.
At best it has been a relatively narrow set of cards that all have relatively obvious tie-ins with other Mac products besides the Mac Pro or some Apple sanctioned and sold card.

I see the AS mac pro as a super mac mini but not as a workstation. Or all the studios will go to Dell and HP. And Apple will only have the pro videos and youtubers market, with optimized apps for AS.
Already, those who make the effort to buy a mac pro do so mainly because they have a love for mac os (and beautiful objects).

Hmm, no. Mac Pro buyers often have serious requirements about non-GPU cards. More than a few have large sunk costs and/or inertia into audio cards ( HDX Protools ) , video cards ( declink to reference monitors ) , very high bandwidth I/O ( 10 , 20 GbE cards, direct attached double digit TB storage pools through SAS/SATA , etc. ) , and more than one storage drive ( which Macs only do one these days because APFS is a single SSD device focused file and volume management system. ) so need M.2 or 2.5 or 3.5 HDDs.

It is not just solely about additional core counts but also about having I/O inside the box.

Pretty good chance that a "half sized" Mac Pro still ends up with at least a MPX sized bay with more than two slots . Might not come with AUX power and MPX connector slots, but a "retreat" back to 3-4 slots would be a pretty high overlap with the number of slots the MP 5,1 had ( four ... but 3 after default GPU card filled slot. ).


A more affordable (e.g. $4K instead of $6K ) , "half sized" Mac Pro with 3 or 4 x8 PCI-e v4 slots (and same/similar J2 HDD bracket ) would probably get some decent traction with folks not primarily focused on GPU cards. The "more affordable" part being a big issue for the folks left behind by the 100% entry price increase of the Mac Pro 2019 over the 2013 levels.

If sold alongside an improved Intel one the folks with the GPU focus can just buy that one for next couple of years.

The mpx modules are interesting for thunderbolt.

Apple's SoC will likely have a iGPU and Thunderbolt built in. If Apple wanted to significantly simplify the motherboard of a half sized Mac Pro they could just drop the MPX connector if banning all GPUs other than Apple's.
( the DisplayPort provisioning lanes don't need as much. Removing the need for aribrarty power wires and two x4 to provision the Thunderbolt controllers could still be there. But it going to get increasing easier to pull two x4 PCI-e streams out of x16 PCI-e v4 and v5 bundles with less impact on the GPU chip bandwidth. )



Also the overlooked driver/firmware/software issue again. M-series Macs don't boot with UEFI. The folks who are enamored with going do with the local computer parts store and picked up some generic GPU card off the shelf and tossing it into a M-series ... that will likely have boot issues. Certainly will have "is Apple deeply participating or not" issues.

We can consider taking an average graphics card by apple, and then buying later a better AMD graphics card but not pro, only for rendering, not for display, in addition to the base card that would be used for the display.
(The 6.1 already worked like this with two cards, one for display and the other for rendering)

The software interface problem there is Apple has entangled "GPU compute" into "GPU display" with Metal. However, yes Apple do an update where non Apple GPU cards could contribute as "compute engines" in like a "half sized" Mac Pro even if not primarily driving the displays. That has limited usages. Some worksloads have stronger requirements for display and compute to be on the same card. ( why 6,1 was very useful for some , but didn't work as super broad baseline. )


That primarily rides on the assumption dealing with a UEFI boot environment when trying to dip into the generic off the shelf market. That should still work with a 8,1 Intel Mac Pro that is using the current T2 chip to validate a UEFI firmware image and boot the Intel chip.

But that is basically a dead end for the M-series running macOS. No UEFI. Apple as made that clear in communications.

The usual "average graphics cards by Apple" are over the last decade or so mostly embedded GPUs. Not discrete ones. If all the AMD embedded ones stop there is likely going to be a substantive issue. Mac Pro 2019 (7,1) GPU options:

AMD 580 -- used in iMacs previously
AMD 5500 -- used in iMacs and MBP also
AMD 5700 -- used in iMacs also
Vega II (Vega20) -- Vega (Vega10) used in iMac Pro

6800/6800duo/6900 -- don't have embedded matches though. (Navi1 and Navi2 aren't a big jump though at the baseline interface.)


If think there are shops that have questions and Apple isn't answering them. That isn't "miscommunication". Apple has a clearly stated policy of not talking about future products. That is a clear communication; not miscommunication. ( that is like going to a strictly kosher deli and asking for a pulled pork sandwich. They don't do pork so kind of a waste of time to ask for pork). That those shops would prefer long distance roadmaps is a mismatch in desired expectations.

 

[deconstruct60]

One thing I haven't seen address that seems a wee bit important... M1 is not UEFI. It's back to a proprietary Apple firmware.

That means all the option ROMs won't work. Graphics cards won't have boot screens again. Storage cards may not boot. This could be a big reason that the Intel Mac Pro is continuing to get updates. M1 or M1X would not be compatible with PCIe cards like an x86 UEFI machine would. Even if the card slots were physically there.

There may be work around for that. Although I don't think Apple wants to go that way.

First, since every M-series SoC probably has an iGPU in it then if really need really early boot screen diagnostics then just plug into that GPU's output. It would be awkward, but it is a workaround. However, it is probably very much behind why Apple isn't pressed about this ( the now ubiquity of their GPU. ).

The "one true recovery" (1TR) that powers the interactive tools screen if launch recovery now appears to be just a greatly trimmed down version of macOS. In the past, folks have gotten non-boot-screen GPUs to work by loading drivers after the basics of the OS are up and running. macOS tells the GPU card how to get going. Klunky but it could work of the card doesn't need any long term UEFI facilities; just some passed some initial data on how to set up. The downside would be that Apple is trying to keep 1TR very small in footprint. It is bigger than UEFI, but dozens of GPU drivers it doesn't need if stick with just Apple GPUs. Plus wold only probably only work for 'backwards looking" GPU coverage ( something stable enough to be included into 1TR).


Another coupled software issue though is Apple's path to getting rid of kernel extensions. I won't be surprised if Apple is pushing 3rd party GPUs solution past the point where they pragmatically kill off kext .

It's also possibly one part of why eGPU isn't present on M1. The firmware isn't compatible with UEFI GPUs.

Internal ( add-in-card ) or external slot in a external box probably doesn't make much of a difference. But yes the issues aren't confined just to Mac Pro ... would be across the whole product line up.

I think Thunderbolt qualifications have a "get out of jail free card' on certification if the OS doesn't support eGPU. Probably, gets Apple an 'out' if they just say macOS on M-series just does support it all.

Maybe someday Apple supports UEFI again. UEFI does work with ARM. Seems like it might be needed to boot Windows someday.

I wouldn't bet on that at all. UEFI is probably toast except in virtual machines that some folks layer on top of Apple's Virtualization framework. Pretty good chance that Apple is going to punt UEFI firmware to someone else that they 'box' in a virtual context.

UEFI is too much entangled into the boot security and Apple has their own path for that. Same issue for 3rd party GPU cards. Not only firmware on the cards , but the likely security priority to validate the integrity of that firmware.

Similarly, Windows11 has more trusted key/security requirements that I'm not sure Apple's VM provides on M-series. As for windows 10 why would Apple put tons of effort into something that is going away in 2025. ( Parallels and VMWare get the "extra" work of making that work for only intermediate term gain. )

But it's not as simple as just sticking a PCIe port on an M1 board. Apple's M1 firmware isn't generically compatible with third party PCIe cards like the x86 Mac Pro is.

On the other hand, poking around Sonnet site there seems to be "Thunderbolt on M1" certifications for some cards in external Thunderbolt PCI-e enclosures. May not be boot capable but once running these work. ( Apple isn't 100% hostile to 3rd party PCI-e cards. ) "Thicker/taller/complicated" driver stack probably makes a difference. it is easier to be a standard device with a generic interface.

How much of that is just a "war" on 3rd party GPUs and how much is wide spread to a wide variety of cards is up in the air. [ checked AVID's page and ProTools HDX says "Not yet supported" on the M1 column.
https://avid.secure.force.com/pkb/articles/en_US/Compatibility/macOS-Big-Sur-Support

It doesn't say "doomed" or never going to work.

poked around on Blackmagic site and several TBv3 devices are working.
]

With all that in mind, it seems realistic that Apple wouldn't include PCIe slots on an AS Mac Pro Mini. Too many cards would be incompatible.

Given there are TB connected cards that work now that doesn't really stand up all that well. There is more to life than just GPU cards. And that is why a AS Mac pro should have at least a few slots; 2-4 . A storage card, an audio card, and a video link/capture card.

Leaning on Thunderbolt for all the additional storage doesn't make sense. ( even Apple said so back in 2017). Folks who have multiple TB of stuff to store usually a good idea to stack all of that up on one and only one drive. Doing everything in one big pile doesn't scale. APFS is rather narrow minded there but there is no reason the Mac Pro should be limited to the lowest common denominator APFS constraints.

At best, we'd be back to the PPC days where you needed to buy special cards with special firmware. Or like the cMP days with GPUs.

Unlike the PPC days storage drives are bigger and much, much faster (at least the Mac OS ones which are required to get the system going. Can boot off the external drive but have to have the internal Apple SSD functioning for that to work. )

So the boot firmware does really have to be minimized to the same levels. As long as the card can be idenified attaching a driver to it can be done. Doesn't necessarily have to come from the card itself.

 

[goMac]

Given there are TB connected cards that work now that doesn't really stand up all that well. There is more to life than just GPU cards. And that is why a AS Mac pro should have at least a few slots; 2-4 . A storage card, an audio card, and a video link/capture card.

So again, a few issues. Even non-GPU devices, like storage, need compatible option ROMs. At least for boot. That may make a lot of PCIe storage cards non-bootable. (I haven't seen any reports of people booting from Thunderbolt attached PCIe storage yet on M1, I'd assume it probably doesn't work.)

At best, we're back to the PPC days where PCIe card options were restricted down to companies like Sonnet on things like storage. That already makes an AS Mac Pro quite a bit less attractive.

But, even going down that route... size becomes a problem. A half height AS Mac Pro would barely have enough room for an MPX Slot. It's definitely not going to have a whole bunch of slots just based on a case layout alone. And height isn't the only problem. Length becomes another problem. You'd end up with a Mac Pro that looks like a giant sausage if it's half height but full length.

And is Apple willing to ship a Mac Pro without PCIe slots? Yes! They did with the 2013. It's completely reasonable, given all the issues and everything described, that Apple could ship a Mac Pro without PCIe slots. Especially given the last time they optimized for size they cut all the PCIe slots.

 

[grouch]

The 2013 mac pro is absolutely fantastic. It was a radical experiment that yielded results that will help them down the road. The bad part though is I'm stuck with a device with old GPUs and the way it was designed, the second GPU is useless anyway.

Going forward I'm hoping for the traditional pro machine for people who need lots of horses, then a smaller, less expensive pro machine for people who want more expandability than the mac mini provides.

I am not optimistic at all about 2021 for either such options - Based on what we've seen Apple do for years, I believe that the 27" iMac, a second generation of mac minis, and the full line of laptops must be finalized before any mac pro refresh will happen.

Oh, don't forget that refreshing the iPhone, iPad and AirPods about 50 times is much more important than all of this.

 

[jscipione]

I wouldn’t wait for the Apple Silicon Mac Pro unless you are prepared to accept the compromises of the 2013 Mac Pro. If you’re like me, and you want the flexibility that the current Mac Pro offers then this is the one to get. But if appreciate instead a unique, powerful but non-modular Mac then the Apple Silicon one is for you. I imagine the Apple Silicon Mac Pro base price will also be cheaper.

 

[jscipione]

I wouldn’t wait for the Apple Silicon Mac Pro unless you are prepared to accept the compromises of the 2013 Mac Pro. If you’re like me, and you want the flexibility that the current Mac Pro offers then this is the one to get. But if you appreciate instead a unique, powerful but non-modular Mac then the Apple Silicon one is for you. I imagine the Apple Silicon Mac Pro will also be cheaper.

 

[deconstruct60]

So again, a few issues. Even non-GPU devices, like storage, need compatible option ROMs. At least for boot. That may make a lot of PCIe storage cards non-bootable. (I haven't seen any reports of people booting from Thunderbolt attached PCIe storage yet on M1, I'd assume it probably doesn't work.)

That really doesn't line up. Here is a link to the Sonnet Technogiy PCI-e card compatibility matrix for their Thunderbolt PCI-e enclosure boxes. It is a PDF file but I'll loop in a summary count for each of the categories of cards . The M1 gets its own column so it ia pretty widespread issue. These are counts of "Yes" compatible driver among the options and then the total number of cards ( followed by the percentage).

Category Yes Total %
Audio Pro Cards 0 24 0
Ethernet Adapter 1GbE 0 5 0
Ethernet Adapter 10GbE 2 12 17 ( two more Q1 33% )
Ethernet Adapter 25GbE 0 6 0 ( but none of these work on Intel Macs either. No drivers )
Fiber Channel 0 9 0
FireWire 2 5 40
GPU Cards 0 4 0 ( these aren't the GPU focused boxes )
PCI-e Bus Extender 0 4 0
SAS/SATA (RAID) 2 17 12 ( only two are Sonnet Tech )
SSD 9 15 60 ( if drop the ones were no coverage on Intel mac percentage 75% )
Signal Acquisition 0 4 0 ( no Intel Mac qualify either; no new issue here)
Video Capture Process 13 134 9 ( first 65 cards ran on Intel Macs at some point)
USB 3 Adatper 5 5 100 ​

SSD drives work reasonably well. Enabling more external USB devices works just fine (so the boot drives there ... no problem). Even FireWire if wanted to boot relatively slow would be mostly OK.

Booting off of "real RAID" cards is a problem, but is that really a firmware problem or a don't have a System Extension replacement for their Kernel Extension (kext) problem? If those RAID card pretended to be a NVMe SSD or just a plain SATA drive , then it probably would work as well as the SSD category. Won't be surprising if the non-disk features ( metadata interace to RAID management ) was the bigger problematic issue here.

1GbE works good enough on USB 3 adapters ( I suspect this more 'old' , funky controllers assumptions than something to do with boot firmware). If there are two 10GbE adapters on an updated Mac Pro who has a deep need a 1GbE card anyway? The 25GbE clearly illustrates that macOS has a driver issue in general ( nothing really new about the M1 ).

Booting off of Fibre Channel probably isn't a show stopper for most folks.

So, as far as keyboard, mouse , storage getting going on the boot process isn't a huge problem if mainly depending upon an attached drive. Probably if toss out the older cards that have controllers that have been dropped by Apple and/or have abandoned/obsolete drivers there is already decent coverage.


GPUs are harder to tell because Apple just simply doesn't cover them at all and the Apple GPU is ubiquitous. Can get a boot screen in plug into the Apple GPU port on all M-series.


Video processor aren't necessary at boot. Drivers initializing the card post boot should be possible if the vendor is up to modernizing the drivers. That is a large category for Apple to completely walk away from. That is a bigger set than several of the other categories combined.

Audio is suggestive that it is in part simply drivers. There is link the matrix for the "No" for Rednet Dante card.

https://focusritepro.zendesk.com/hc/en-gb/articles/360017461919-Focusrite-Pro-product-compatibility-with-macOS-11-Big-Sur-on-Apple-Silicon-based-systems

The PCI-e card is a bright red NO there. The dedicated TB device is a "Yes". ( if digging into the article on M1 thunderbolt drivers find out they are stll kext ; not System extensions. )

Apple's push to revamp the "extensions" system from kext to system extensions is probably going to be a bit of a "garbage collection" on cards and devices that have about zero driver R&D budget for upkeep and maintenance. That isn't really an M1 only problem since Apple is on path to dumping kext on Intel side over an extended period of time also.


The M1 intensifies Apple's 'war' on old, vintage kext drivers, but it is an issue that is in motion regardless of platform. There probably are a sizable number of board controllers that have BIOS and odd ball x86 instructional mode quirky incantations that have been in zombie support mode for a long while that will drop out. However, that is mostly a tangent to strict UEFI going away. ( UEFI has been used as a crutch to keep old BIOS cruft danging around. Apple not committed to BIOS and multiple decade old stuff shouldn't be a shocker. )



PCI-e has some standards augments that could help. There is a push to get PCI=e devices to authenticate themselves and their firmware metadata.

PCIe® Component Authentication | PCI-SIG

pcisig.com

Intel also has a draft on PCI-e security enhancements. (NOTE: link to a PDF file)
The metadata about the card and its firmware is to be verfied before load the drivers/bootstramp code from the PCI-e device. That "Vendor ID , device ID , firmware version" data is all the System Extenstion dispatch should need to play match maker between a still unitilized ( or hotplugged) device in the PCI-e device tree. Even if don't use the card's bootstrap code the singature of that code can be used as a "matchmaker" for a driver . That driver though would have to do the initial incantation to get the PCI-e device running.

There are tons of old cards that are not going to meet those security standards. but slots in a M-series Mac Pro would be just as much about new cards in the future.

In short, lack of UEFI should not be a show stopper for the post boot environment. There is a giant stack of not written yet software though.

At best, we're back to the PPC days where PCIe card options were restricted down to companies like Sonnet on things like storage. That already makes an AS Mac Pro quite a bit less attractive.

The days were a significantly large chunk of card vendors didn't write macOS drivers really didn't change with Intel.
The macOS driver model isn't the windows driver model. That means work which some folks will punt on. MacOS on both Intel and M-series means there is twice as much work for those that have drivers with significantly large chunks of handwritten assembly in them.

Did macOS Intel have a large broach ecosystem of Infiniband cards avaialble (that were on Linux) ? No.
100GbE cards ? No.
> 10GbE cards ? No ( outlined that above already. )

MacOS just being > 10% of the market is a factor here at least as much as pre-boot roms.

But, even going down that route... size becomes a problem. A half height AS Mac Pro would barely have enough room for an MPX Slot. It's definitely not going to have a whole bunch of slots just based on a case layout alone. And height isn't the only problem. Length becomes another problem. You'd end up with a Mac Pro that looks like a giant sausage if it's half height but full length.

I have been though this before, but if just look at the core 'box' ignore the feet/handle , then if drop MPX bay 2 and combine top CPU and "bottom" MPX thermal zones then have dumped 1/3 of that height. So only 1/6 short of getting to a full overall height drop by a 1/2. So nuke the feet(substitute small fraction of an inch non-stracth pads and place on desktop ). That is a couple of inches there. So now down to about a 1/12 (or less) of getting to "half as tall". That's if just stick dogmatically to the "space frame" model. If you get rid of the patruding handles, feet, twist off handle on top ( some kind of more normal latch closed case with integrated handles ). That is the rest of the half.

That is just leaving the CPU thermal zone alone. If willing to get with two different sized fans on front then could possibly take the 1/6 off of that (and still keep some feet height ) .

The notion that is super hard to get to half spatially is only if looking for an excuse to get rid of the slots; not that it is necessary.

The fact that it isn't some perfect cube or whatever OCD geometric shape Apple industrial design might want doesn't really matter. The length would be serving a purpose. The holes in the grill serve a purpose. If magical oompa loompas could move the heat of the Mac Pro without moving air then they get rid of those holes too.

And is Apple willing to ship a Mac Pro without PCIe slots? Yes! They did with the 2013. It's completely reasonable, given all the issues and everything described, that Apple could ship a Mac Pro without PCIe slots. Especially given the last time they optimized for size they cut all the PCIe slots.

Let's say they bring back a 7" x 7" x 7" cube and put 6-12 Thunderbolt sockets on the back and they would have really solved a whole lot of nothing if do not have the driver creation problem sorted out. The number of solution suppliers collapses down to a smaller number and the solutions available greatly decrease then really haven't solved the issue ( if non pre-boot was a core issue).

The the thunderbolt driver issue gets solved then it that would help internal PCI-e slot also ( to the cards Thunderbolt is really largely transparent once get past having to deal with hot plug/unplug. ) Hot plug/unplug means card initalization isn't necessarily at overall system boot time.

All they would have done is balloon squeeze the PCI-e cards out to a more expensive Thunderbolt PCI-e slot external enclosures. As a net "save desktop footprint consumption" solution that is a spatial diaster. Now have the total desktop area consumption consists of the "cube" plus the TB eclosure which is going to include all of the "evil length" avoided by punting to another box.

If the Apple "Mac Pro" SoC can't provision a decent number of slot then that would be a rational path to dump them. For example only had 12 TB and 16-20 x1 PCI-v4 lanes coming out as general PCI-e related I/O then sure skip the slots. The widest lane bundle in that context is just x4 PCI-e v3. If that is the best they got and are trying to compete in the 2022 workstation SoC marketplace then they really do need some serious misdirection. "Don't look behind the curtain we are the all powerful Oz" is really all they got to work with.

If they leave enough space to put an user optional 2.5" SSD or a couple of M.2 SSD then they wouldn't have very tightly painted themselves back into the same corner that the Mac Pro 2013 did. There would be enough room to declare themselves "king of that small kingdom" but much of the same problems would be the same.
Minimally they'd need more than one storage device even if there are no legacy standard PCI-e slots. That one, and only one, storage device restriction was explicitly mentioned by Apple as being a mistake for the Mac Pro. Since the RAM is likely soldered down they'd need something users could insert to label this as part of the "Mac Pro" line up.


Otherwise this would be more of a "Mini Plus" . With the current Mini enclosure there is pretty good chance they could get the Jade2C ( 20 cores and 64 GPUs) into that box with some small board changes. Get some "front to back" reasonable speed airflow through there by making it taller they could get 40 cores and 128 GPUs could be possible if minimized any other additions.

However, what would be the point of a "Mac Pro" that largely just covered the same space as the Mini. Apple could through the "Mac Pro" name at it because the price tag was quite high for a Mini , but it pragmatically would just be a Mini.

if Apple only wants to do 2-3 M-series product lines then yeah the role that the Mac Pro has filled probably dry up. That would be almost entirely Apple's choice though, not precipitated by shift from UEFI.

 

[widEyed]

That really doesn't line up. Here is a link to the Sonnet Technogiy PCI-e card compatibility matrix for their Thunderbolt PCI-e enclosure boxes. It is a PDF file but I'll loop in a summary count for each of the categories of cards . The M1 gets its own column so it ia pretty widespread issue. These are counts of "Yes" compatible driver among the options and then the total number of cards ( followed by the percentage).
[…]

Can anybody confirm using any video encoding PCIe cards with M-series Macs (Blackmagic Design Decklink 4K for example) in external PCIe boxes? Has Apple made any public comment around this (expecting not because it would distract from the marketing bonanza of (Pro/Max speeds they have going ATM)?

From the linked list above I pulled out the Blackmagic Design cards (see PDF for more exact details).