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.
What's Happening With the Mac Pro?
- Thread starter MacRumors
- Start date
- Sort by reaction score
What I’m proposing is the entire motherboard with SoC etc, the cost saving is reduced development time for Apple, the cost of the case, I/O, power supply, shipping. It’s an imperfect compromise. Given this option or the Mac Pro dying, this is more palatable.
If I remember correctly at the Mac Pro 2023 introduction Apple flashed a picture of a Mac Pro stuffed with six HDX cards.
AVID's test supported configuration maxes out at three.
Apple's extremely narrow corner case may work, but how many $100K consoles are going to get built. Apple having several of those folks 'parked' on the M2 Ultra version for 6-8 years also means they don't need to do anything either for a long stretch (because that niche isn't moving.)
The problem with these audio cards is that although a physical x4 PCI-e interface lots of them are electrically x1 PCI-e v1.1 in bandwidth. Latency is likely a more sensitive issue, but that probably push a new system development process forward faster.
[ finally found possible low level specs on HDX that suggest x4 PCI-e v1.1 is. It is not quite a x1 link but it is not a modern bandwidth consumer. PCI-e v4 is about eight times PCI-e v1.1 . If the base lanes on the next Mac Pro were upgraded to v5 the gap is even larger overkill. ]
That’s a good point - industry folks don’t mind running old hardware and know better than to run macOS updates that will knock out their config. So the life cycle is longer.
Workstations today are used in industries macOS does not play in (and in many cases, probably never played in) so even if Apple offered a Mac Pro that could take two of the latest ultra-high-core count Intel Xeons or AMD Threadrippers and a half-dozen RTX PRO Blackwell GPU cards with terrabytes of RAM, it would not be an option for said industries because the software runs on Windows and Linux.
Support is also an issue. While Aple does offer third-party on-site Enterprise support for large customers (so you don't have to lug your MP to the local Genius Bar or ship it off for a week or two to a repair center), compared to the support levels Dell, HPE and Lenovo offer it is more limited and with these maxed-out machines, time is literally money.
I understand the appeal of an "upgradeable / expandable" Mac like the 2006-1012 and 2019 Mac Pros where you spend 50-100% more at time of purchase, but save 500-1000% over the next decade by being able to upgrade and expand the machine over that period. But the general Mac market has not been interested in such a machine since the 2010s as Apple moved to "sealed boxes" with the iMac and MacBook Pro / MacBook Air and people bought more of them than ever before. Because most Mac owners are not "tech enthusiasts" nor are they "tinkerers" who want to hold on to their machine for a decade and enjoy upgrading them and have the technical knowledge to do so.
Even PCs are more and more becoming "sealed boxes". Microsoft solders the RAM on the Surface Pro and while you can tear open a Dell or HPE laptop to add memory or upgrade the SSD, it's often not the quick and easy process it was back with the original plastic MacBooks. My father has an older Dell Latitude with 4MB of RAM and a dog slow HDD, but I would have to tear through the keyboard and trackpad to get to the memory and storage bays with plenty of tiny screws and fragile ribbon cables in the way. Wasn't worth the risk, so I just bought him a new laptop with more RAM and an SSD.
Support is also an issue. While Aple does offer third-party on-site Enterprise support for large customers (so you don't have to lug your MP to the local Genius Bar or ship it off for a week or two to a repair center), compared to the support levels Dell, HPE and Lenovo offer it is more limited and with these maxed-out machines, time is literally money.
I understand the appeal of an "upgradeable / expandable" Mac like the 2006-1012 and 2019 Mac Pros where you spend 50-100% more at time of purchase, but save 500-1000% over the next decade by being able to upgrade and expand the machine over that period. But the general Mac market has not been interested in such a machine since the 2010s as Apple moved to "sealed boxes" with the iMac and MacBook Pro / MacBook Air and people bought more of them than ever before. Because most Mac owners are not "tech enthusiasts" nor are they "tinkerers" who want to hold on to their machine for a decade and enjoy upgrading them and have the technical knowledge to do so.
Even PCs are more and more becoming "sealed boxes". Microsoft solders the RAM on the Surface Pro and while you can tear open a Dell or HPE laptop to add memory or upgrade the SSD, it's often not the quick and easy process it was back with the original plastic MacBooks. My father has an older Dell Latitude with 4MB of RAM and a dog slow HDD, but I would have to tear through the keyboard and trackpad to get to the memory and storage bays with plenty of tiny screws and fragile ribbon cables in the way. Wasn't worth the risk, so I just bought him a new laptop with more RAM and an SSD.
Last edited:
When I look at the overall economics to Apple in offering such a solution, I do not see it happening for the following reasons:
- I do not believe enough existing customers would be willing to buy a new system board considering the cost will probably be over 50% of a new system;
- I do not believe enough existing customers have the technical skills (and will) to take apart their machine to perform such an upgrade and are willing to risk damaging their machine doing so;
- I do not believe Apple would see dedicating Genius Bar and mail-in repair center resources to offer such upgrades to customers willing to spend the money, but not comfortable in doing the work themselves, to be a better use of those resources compared to supporting other repairs.
I have almost saved enough money to buy wheels for my Mac Pro! a few more months...
Seriously, there are form factors that a Mac Pro could exist with high speed connectors and rack mount or stackable design. High power components could have a market for individual industries to custom configure. But Apple would have to have the will to innovate at that level. I think it would take a specific talent within the organization that I am not sure exists at present.
Seriously, there are form factors that a Mac Pro could exist with high speed connectors and rack mount or stackable design. High power components could have a market for individual industries to custom configure. But Apple would have to have the will to innovate at that level. I think it would take a specific talent within the organization that I am not sure exists at present.
With all due respect i totally disagree
1) getting the latest processor for 50% of what you paid before sounds like a steal. Give me that option on my macbook pro and i'd do it in a heartbeat.
2) the people buying Mac Pros are doing so typically to add pci elements, these people are technically savvy enough to do this.
3) Apple can offer this service at a cost that is advantageous to them; if customers don't want to spend, then the aftermarket will fill the void instantly.
Inserting a PCIe card isn’t much more difficult than plugging in a cable.
Inserting a logic board, SoC, and RAM is harder. Would this also entail applying thermal paste? Is the RAM on-package?
Probably way higher. Just think of the components you wouldn’t be replacing: the case, the PSU, fans. Some connectors. Those together, if really premium, would retail for, what, $800?
So I’d say more like 80%. Which makes this exercise pointless.
I thinK I alluded to such a solution this topic or eslewerhe.
Modular Blade of Blade Modules thinking for a Mac Pro unit. You find all kinds of similar proprietary solutions in the early years of computing, think cartridges for 8 bit machine and onwards to expand basic or ram, r just play a game!
If Apple nail the housing, and then sell on blade internals, in a way they have not limited themselves nor is the customer, with a housing system that might have 4 slots for each modular blade unit, or housing base units that come in larger array capacity, 4/6/8/12
Take a 4 slot base unit, come with (default/can not change):
1. Based unit - CPU/M-Chip, Ram, SSD
The base unit is where Apple can offer 1 or more Chips, need 4 x M5-Chips on the base array. No problem. Order it.
Now you have a fully working Mac and 3 empty Mac-Blade slots, and can purchase Apple blades as you wish, or not, but have built in expandability + you are not locked into the base unit either, that can also be swapped out if you want the new M7 Chip in an array of 8 CPUs in the base unit.
Need more RAM?
2. RAM Blade array - buy the RAM array blade from Apple pre-populated or empty ( add in your own DIMMS etc.)
3. SSD array - buy the SSD Mac-Blade array populated or push in your own SSD as required.
4. PCIe blade - by the PCI blade array (hey this thing can take 3 avg/ size larger card, 4 small one or 2 larger etc.)
Or, you could instead order 4 base units slots instead, for whatever reason (and they are deep connected to each other as user suggest in this post) and you have 4 Mac pro system in one box, deeply connected.
Need more then buy the 6 to 8 blade bay unit housing etc. etc.
The housing never has to change only the innards, Apple could easily nail the right housing + proprietary blade system, and if you have money, you can have all the power and configuration you need and easily upgrade on the fly and apply are producing Mac Pro Blades on demand.
Modular Blade of Blade Modules thinking for a Mac Pro unit. You find all kinds of similar proprietary solutions in the early years of computing, think cartridges for 8 bit machine and onwards to expand basic or ram, r just play a game!
If Apple nail the housing, and then sell on blade internals, in a way they have not limited themselves nor is the customer, with a housing system that might have 4 slots for each modular blade unit, or housing base units that come in larger array capacity, 4/6/8/12
Take a 4 slot base unit, come with (default/can not change):
1. Based unit - CPU/M-Chip, Ram, SSD
The base unit is where Apple can offer 1 or more Chips, need 4 x M5-Chips on the base array. No problem. Order it.
Now you have a fully working Mac and 3 empty Mac-Blade slots, and can purchase Apple blades as you wish, or not, but have built in expandability + you are not locked into the base unit either, that can also be swapped out if you want the new M7 Chip in an array of 8 CPUs in the base unit.
Need more RAM?
2. RAM Blade array - buy the RAM array blade from Apple pre-populated or empty ( add in your own DIMMS etc.)
3. SSD array - buy the SSD Mac-Blade array populated or push in your own SSD as required.
4. PCIe blade - by the PCI blade array (hey this thing can take 3 avg/ size larger card, 4 small one or 2 larger etc.)
Or, you could instead order 4 base units slots instead, for whatever reason (and they are deep connected to each other as user suggest in this post) and you have 4 Mac pro system in one box, deeply connected.
Need more then buy the 6 to 8 blade bay unit housing etc. etc.
The housing never has to change only the innards, Apple could easily nail the right housing + proprietary blade system, and if you have money, you can have all the power and configuration you need and easily upgrade on the fly and apply are producing Mac Pro Blades on demand.
Last edited:
I think Apple's answer to this is just stacking Mac Studios.
Sure, that comes with worse bandwidth and latency, but for many use cases, this is fine.
This presupposes an ARM CPU or ARM SoC from Apple that either does not include any RAM, or offers a hybrid solution where part of the RAM is on-package, and part is not. Which in turn raises complexity issues; does each process get to define whether it runs on the on-package RAM or the dedicated RAM? Can a process use both?
RAM on a blade would also have worse latency and bandwidth.
I think the Federighi/Ternus/Srouji answer to that is to avoid that complexity altogether and just have on-package RAM, for the same reason they just have on-package GPU.
This is just ridiculous. How difficult is it to (if nothing else) simply always put the newest chip in the Mac Pro and call it a day. This was the same with previous Mac Pro machines that instead got killed exactly by the same deliberate lack of simple upgrades. I would buy a Mac Pro over the Studio anytime, only if Apple simply upgraded the internals to be at least on par with the Studio.
The fact that this is not an option is one of the reasons Apple's virtue-signalling and greenwashing is so laughable. It contradicts their business model. Actions not words.
I honestly expect the significant majority of people buying Mac Pros for non-personal use (and therefore the largest customer market for this model) are not buying it first and foremost for the SoC's CPU and GPU performance, but instead because they have a collection of PCIe expansion cards that they either want to keep in a single box directly connected to the SoC and memory or because the cards work best when they do so and therefore putting them in an external chassis connected via ThunderBolt is not-desirable. And those cards are likely not dependent on the CPU and GPU speeds of the SoC.
We have actual commercial Mac Pro owners and users in this forum who note they use an M2 Mac Pro with audio capture cards and note the M2 Ultra is more than enough to handle literally three-digits worth of audio channel mixing. Others feed massive external storage arrays connected via high-speed fibre channel. Even a hypothetical M5 Ultra Mac Pro would gain them little to nothing to justify the expense of replacement.
As such, I believe there would bes a minimal generational upgrade cycle for the Mac Pro in general and therefore the staff time and money Apple would need to spend to research, design, test, certify, produce and distribute a new Mac Pro each generation would not be recovered.
It seems to me to make more sense to just let the chassis sit multiple generations to earn as much RoI as possible on each generation. So far, each generation of M has been an incremental improvement (some more significant than others) and eventually the aggregate improvements across generations will likely make it worth updating even if the latest generation is not exceptionally better than the direct previous generation.
I am honestly interested to know where does the Mac Pro for you provide the extra value for that extra cost?
Yeah, I think RDMA over TB5 and external PCIE TB5 chassis makes a lot of arguments against a very expensive tower. A 3rd party could solve the rat's nest with a chassis that it all sits in easily enough, similar to those chassis that take Mac minis into rack mount, but more towards hiding the snarl (obviously make it rack mountable as an option via bolt on ears).
Not sure why Apple has shown deliberate neglect for the Pro, like at least throw processor bumps in, a Mac Pro with M5 Ultra would be a bone at least to creatives to use, and require no special system design (not sure how difficult to processor bump between apple chip generations compared to PC's on x86) that alone seems a small engineering task without requiring some huge redesign, and it isn't really competing with the Studio given the massive price differential.
If you need a Fiber channel card and some complicated video/audio interface it's the card of choice, and not sure if an external TB5 chassis can compete speedwise with 8x or 16x PCIe cards, but hard to justify that choice compared to the horsepower you get with a Studio. The power draw is irrelevant (I can't imagine the guys who on the PC side have an NVIDIA 5090 and 512GB of Ram in a monster 9800X3D CPU tower with liquid cooling really care about power draw when rendering piles of 3D video. They care much more about raw speed and rock stable cooling.
My concern about apple truly ditching the creatives openly is the halo effect, yes the middle tier creatives are served just fine by current consumer level Macs (I edit video just fine on my M4Max MBP with 128GB of RAM) but the heavy weight guys (like say movie SFX studios) are going to switch over to PC workstations (and switching isn't as hard as it used to be), where all their PCIe cards work, they can upgrade their machines ad nauseam and that halo effect will lose. It's similar to why Chevy and Ford produce halo editions of their Corvette and Mustang that are way up there in price, in tiny volumes too. Sure they are a side hustle almost, but they are a halo to the brand.
It's not just the hardware, the software needs to come along and be accepted by the community. Apple could certainly boost their LLM performance, but the question is can they compete with NVIDIA on anything other than on a per-watt basis (what they typically show). Yes per watt matters but when I was doing some ML work (Not LLM, task specific imaging ML) on my ubuntu intel box with a big NVIDIA GPU, my system was close to using 1000W, but did I care? no, as long as the cooling could keep up, I wasn't trying to save the planet with my PC, per-watt mostly matters in a data center or if cooling can't be optimized. Hard to compete with a GPU card that is willing to burn 600+W inside that little case, although the shared memory model is an interesting advantage to the apple silicon since once you drop below the huge cards you quickly are VRAM limited. The Studio is cheap enough (compared to the monster cards) that Apple could get away with the soldered model if they price it around what the 5090 is since upgrading while possible is fantastically expensive on those high end cards (and yes you can repurpose/resell the prior generation as for most people a last-gen card/studio is powerful enough)
Nobody makes apple only PCI cards, so forcing a form factor that makes them unique would be a non-starter, the external chassis as cringy as it is would be the way to go. Of course TB5 suffers from the same connector issue that USB-C has and when it is your system bus essentially there has to be a locking mechanism to hold it in place. But that's a $0.10 clip to Make sure that the chassis and CPU can't disconnect from each other.. Then you can slap a M2 drive card to have massive on-bus storage, a GPU card, your RED card, etc... But not sure if enough people will do it to justify NVIDIA or AMD to produce drivers for their GPUs (if even for just AI compute rather than actual video) since MacOS doesn't natively ship with NVIDIA drivers (and getting PyTorch to play nice with the CUDA drivers is hard enough on Ubuntu let alone on a system that they don't support natively)
While a neat idea, given this could be solved by a single cable that has all the necessary connectors on it, making it less messy, that is a way easier solution that exposing the system bus on the feet or something. Remember each machine needs to connect to each other machine (hence the snarl) so stacking would need a hefty connector on top and bottom which seems a great opportunity to break the machine with an errant paperclip, but a custom cable would be an easy sell.
Third parties do (such as Sonnet) but it, like external drives end up being a snarl quickly on the desk. Now most video/audio studios are already a snarl (at least back when I was a sound engineer during the analog-digital transition era) and while TB5 is extremely performant and reliable, people like the security of a literally bolted in PCI card inside the chassis on the system bus compared to an external box)
Uh, as the gentleman above noted he has many such uses for PCIe SMPTE cards, and all the specialized gear for serious audio workstation use (and that they exceed TB5's bandwidth). Yes, most of us doing regular NLE audio work can get by just fine with a MBP over USB-C to an external audio interface (what I use) it's not even close to what the pros use. Sort of like sure I can shoot videos in my home studio with 3 cameras, and for videos for work I don't need anything other than the Final Cut's audio-based-sync between the multi-cam shots, but that isn't going to work for a real video shoot with 10 cameras, on 4K with external audio, all of which has to be dead on sync'ed via SMPTE, across all those devices, because HBO or whoever isn't accepting a 10msec delay between audio and video like I might in a training video. They also need REALLY fast storage, unlike me who shrugs if compressor takes an extra minute to render my video out, because it's just me, and I will go get a coffee. Those guys need M2.NVME arrays, plus for slinging 4K or god forbid 8K video around.
You're close to "getting it" with regards to what apple is doing now with the studio.
Don't think in terms of RAM slots or whatever, just think in terms of small/med/large lego bricks.
The studio is a lego brick
Pick the size of brick, pick the quantity and link them together.
We've been doing this with enterprise VM workloads for decades. I.e., you're not working with blade arrays or whatever. Just using lego bricks of self-contained compute/ram/storage in one module.
Depending on your workload the type of bricks you use may change, once you figure out your workload requirements, just scale with more of the same modules that contain cpu/gpu/ram/storage in the same brick.
In server land we call this hyper convergence.
This is where I see the "Mac Pro" going... being replaced by an array of studios or similar. I thought they should do it with a mainboard and blades as well (basically Mac Studio on a blade), but this way they can sell the studios individually and not change anything if thunderbolt is fast enough.
Are there compromises vs. a backplane with a bunch of blades? Sure. But its simpler and probably "good enough" to work and then speeding it up is just a question of making the thunderbolt ports faster with later generations.
The computing landscape is filled with previous bespoke solutions to problems that died because "simple and good enough" won.
Last edited:
As long as there is Trump you have double MAGA right now for a little over 3 more years.
If you get a third term Trump the US Mac Pro Apple factory is good for another 4 years at least after that.
It's only been under a year but so far Trump has been good for the Mac Pro in his second term.
Ross Perot part II
Last edited:
Primarily Base unit blade thinking, may cover this within the conceptual vernacular.
Base unit blade might allow at least 2 M-Chip systems ( 2 studio configs on one blade)
1 or 2 M-Chip systems, Pro starts with 1TB min unified 1/2/4/6/8/16TB RAM configs + SSD etc.
BTW when I say blade I mean a Apple proprietary system on slim (as the PSB could be, and if it had a housing).
Apple is then producing these proprietary Mac-Pro blades units various base and higher configurations, with an annual upgrade cycle or that is more achievable, because Apple is manly producing blades, not entire Mac Pro systems.
Who knows even PSU needs might stack on a blade by blade basis (regulatory things aside).
There is no reason why the Mac-Pro blade can not have epic spec. that entices the Pro market. Price sensitivity is different here, and this could be a better paradigm, even investment wise.
Based on some comments it seems better to have more systems than just one beefed up and maxed out box because it can't be used as efficiently as distributing across multiple systems.
If these systems were in a Mac-Pro housing that has I think what you call a backplane with virtually no latency you have a neat scalable system beyond how a Mac-Pro is currently conceptualised, but you can start with one and buildup later or max out from the start.
I could see people being happy upgrading their base system on an annual or bi-annual basis if the cost was 50% less than the initial investment, but means the initial investment is solid and performing.
I could see a healthy trade in used-Pro blades too, which helps reducing the upgrade cost ad allows others build up a system a bit cheaper.
The new Mac-pro can be an Apple system/server farm in a Mac-Pro looking box.
Last edited: