Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

MoonCakeTropics

macrumors member
Original poster
Feb 11, 2022
51
60

These articles talks about maybe there will be an M5Ultra if the M5 has an ultra fusion. The M5 will still be 3nm if the rumors are to be believed. The big step up would be the M6 at 2nm. It's not clear if going to 2nm would somehow help bring back the ultra fusion, or as the article speculates, there would need to be an all in one M5 ultra rather than via the fusion connector.

The article points out and apple confirms that making the ultra version of the chip just takes more time. Which explains their ass backwards introduction o the ultra chip last, rather than first. Which makes you never want to get an ultra machine because as soon as you get it, a couple of months later the next Max version of the chip eclipses the current ultra. If they started with Ultra and went down to the other versions of the chip it would make more sense.

The price difference between the M4max (which has better single core performance) to the M3max really highlights how much of a lame value proposition the ultra version of the chip is when it's introduced this late.

But apple has shown, it just doesnt care.

Which gets to my belabored point. I think the Mac Pro is dead. They didnt even bother slapping the M3ultra in the current body of the Mac Pro. I see no reason to believe why they would make an M5 or M6 ultra when they couldnt even be bothered to update the current lame duck M2ultra Mac Pro, which they oddly are still selling. Out, not with a bang, but with a withering whimper.

Apple could not be more mismanaged, lazy, thoughtful, and unproductive, and the complete lack of statement towards Mac Pro users is more evidence of that assertion. I hope Im wrong on that.

And for those of us that need "real" storage, the joke 16TB for $4800 is spitting in our eye, when we can get probably 2 30TB U.2 drives for that price now, and for those of us that need the extra SSD speed and capacity, the lunchbox glorified MacMini is not enough. Sadly, this feels like the nail in the coffin of the Mac Pro.

So truly sad.
 
Last edited:

These articles talks about maybe there will be an M5Ultra if the M5 has an ultra fusion. The M5 will still be 3nm if the rumors are to be believed. The big step up would be the M6 at 2nm. It's not clear if going to 2nm would somehow help bring back the ultra fusion, or as the article speculates, there would need to be an all in one M5 ultra rather than via the fusion connector.

The article points out and apple confirms that making the ultra version of the chip just takes more time. Which explains their ass backwards introduction o the ultra chip last, rather than first. Which makes you never want to get an ultra machine because as soon as you get it, a couple of months later the next Max version of the chip eclipses the current ultra. If they started with Ultra and went down to the other versions of the chip it would make more sense.

The price difference between the M4max (which has better single core performance) to the M3max really highlights how much of a lame value proposition the ultra version of the chip is when it's introduced this late.
The ultra isn’t about single core performance. It’s the same as the regular so if you need that, you wouldn’t buy the ultra. It would be nice if they can make an ultra with considerably faster single core performance.

But apple has shown, it just doesnt care.
Doesn’t care about what? They make products they feel people will need and buy. It’s not about caring. If there’s a market for a product, Apple is going to make it because they’re not going to leave money on the table.


Which gets to my belabored point. I think the Mac Pro is dead. They didnt even bother slapping the M3ultra in the current body of the Mac Pro. I see no reason to believe why they would make an M5 or M6 ultra when they couldnt even be bothered to update the current lame duck M2ultra Mac Pro, which they oddly are still selling. Out, not with a bang, but with a withering whimper.
Probably for the short time it is. The Mac Pro was a machine based in the Intel world. Graphics cards and huge cases aren’t really compatible with Apple Silicon. Is Apple working on some sort of revolutionary version of Apple Silicon for Mac Pro? Why would they put a Mac studio motherboard into a Mac Pro chassis just to make it bigger?

Apple could not be more mismanaged, lazy, thoughtful, and unproductive, and the complete lack of statement towards Mac Pro users is more evidence of that assertion. I hope Im wrong on that.
Well, for all of that, they seem to be doing pretty well. Do you have any specific ideas on what you think they should do? Realistically, what would you want Apple to do if you could say hey do this?


And for those of us that need "real" storage, the joke 16TB for $4800 is spitting in our eye, when we can get probably 2 30TB U.2 drives for that price now, and for those of us that need the extra SSD speed and capacity, the lunchbox glorified MacMini is not enough. Sadly, this feels like the nail in the coffin of the Mac Pro.

So truly sad.
The Apple charges too much for storage complaint. I’m not sure how long you’ve been buying Apple products but this is nothing new. I don’t think this is specific to the Mac Pro or any Apple product. Look at storage prices on the iPhone. In my opinion, this is how they subsidize the price of the entry-level model. People with more cash than they know to do with buy these maxed out products so Apple can sell the base model cheaper
 
Can we see out of the replies who actually has a Mac Pro (or numerous ones)?

The max ram level is an improvement but still behind the 2.0TB maximum of the top 2019 model (8x256GB DDR4-3200), but the storage price is ridiculous when for instance in previous Mac Pros we can just load up Sonnet cards with large amounts of fast NVME storage for less. And this doesn't require another box and cables to be sitting on the desk - it all fits in the computer.

2x 2019 Mac Pros here.
 
  • Like
Reactions: leifp
I disagree. I think they’ll come out with a super beefy Mac Pro sometimes maybe towards the end of the year to make all of us who are now skipping the m3ultra salivating again to get the much more expensive Mac Pro instead. Everything else would be complete and utter nonsense. Make us wait another two years for m5 ultra in the studio? Please god no!
 
I stood on that hill. I have the 7,1 Mac Pro with a ton of ram and PCIe slots almost full. A lot of internal (not Apple) and external storage space. I’m about to buy a new Mac Studio and use my 7,1 for networking-as well as my 5,1 2010 Mac Pro. It’s very sad. I got the 7,1 practically the day it came out. It’s dead.
 
Oh yeah. I have three Mac Pros. 7,1 and two 5,1 - 2010 and 2009. The latter recently died. I don’t thinking can recover it.
 
In addition to my signature i still have my 17" iMac intel core duo which I upgraded a few years ago to 2 meg's of ram. lol
 
  • Like
Reactions: psingh01
Aside from having storage internally in the existing Mac Pro 2023, the typical Mac Pro customer might as well just use a workstation like a Lenovo PX.

Unless Apple comes up with something very game changing - then the Mac Pro is probably on the way out.
I don’t think the Lenovo runs macOS natively. I’m sure you can do the Hacintosh thing, but that has its own set of problems that most macOS users don’t want to deal with.

I think the Mac studio is the Mac Pro we used to have. I don’t think it’s the high end one from 2019 though. I’m not sure if Apple has an answer to that one.
 
I don’t think the Lenovo runs macOS natively.

Oh I'm sure it won't with Nvidia 6000 ada GPUs.

Windows is different, but you get used to it. I have W11 Pro for Workstations on one machine here and it's fine, very reliable. Unless there is some software that is only on MacOS then that's a problem. Most of my software is Windows or Mac.
 
Can we see out of the replies who actually has a Mac Pro (or numerous ones)?

The max ram level is an improvement but still behind the 2.0TB maximum of the top 2019 model (8x256GB DDR4-3200), but the storage price is ridiculous when for instance in previous Mac Pros we can just load up Sonnet cards with large amounts of fast NVME storage for less. And this doesn't require another box and cables to be sitting on the desk - it all fits in the computer.

I'm not going to put that amount of money into something that has no headroom for reconfiguration if my needs change.

End of story.
 
Clearly there are still some customers not served by the Studio as a viable replacement bit the majority of Mac Pro sales will be bulk enterprise purchase where the IT technicians appreciate the expandability. How many enterprise customers (bar Pixar!) they actually sell them to though will dictate whrther it has a future. Statistically they sold maybe 180k units last quarter (3% of 6m Mac sales) which sounds like a lot but is it enough for the dev costs?
 
I suspect that, given Apple's comment that not all generations are going to get an Ultra variant, they will unveil a new processor tailored for high performance compute and more memory that will be primarily used by them on their data centers but will be offered in the consumer Mac Pro too. That way they can justify the investment and will also differentiate the Mac Pro from the Mac Studio further.
 
Probably for the short time it is. The Mac Pro was a machine based in the Intel world. Graphics cards and huge cases aren’t really compatible with Apple Silicon. Is Apple working on some sort of revolutionary version of Apple Silicon for Mac Pro? Why would they put a Mac studio motherboard into a Mac Pro chassis just to make it bigger?
I think that even the Intel Mac Pro was running into problems - mainly that it was only ever of interest to a niche of people locked into Mac-only pro apps, when such apps - or their equivalents - are increasingly available on Windows and/or Linux. At the end of the day, apart from running MacOS, it was just a Xeon-W/PCIe tower with a slightly neater way of connecting GPUs that needed extra power & thunderbolt plumbing. For GPU-bound tasks, its performance was limited by the same AMD chipsets that you could get on PC - and tough luck if you wanted NVIDIA.

With laptops and small-form factor devices aimed at mass-market users, Apple could wow people with cool designs, the nice UI and easy integration of iPhones etc. With a big box'o'slots that sits under a desk and spends its life doing one or two well-defined, specialised tasks, why not get a generic x86 system which is much easier to configure to your exact requirements by bolting together generic parts? Especially when AMD are offering workstation/server CPUs that thrash Xeon on price/performance. The 2019 MP had a short honeymoon when it was an early adopter of the latest Xeon-W chip with extra PCIe bandwidth and RAM support that hadn't shown up on other manufacturers menus yet, but there's a far wider choice of high-end x86 configurations - multiprocessor systems, dedicated GPU computing systems etc. than just Xeon-W.

Sure, for some truly pro users it was cheaper to shell out for a Mac Pro or six than to upset their workflow by shifting to PC - but that's going to be a rapidly evaporating pool. In fact, given 7 years of "wilderness" when the only Mac Pro was the Trashcan, then the 2019 attachment - then a mere few months later the announcement that Apple were abandoning x86... its a wonder that there is any business/professional market for a Mac Pro left at all...

Meanwhile, the whole concept of a high-end personal tower workstation - on whatever platform - is being eroded by increasingly capable laptop/small-form-factor systems at one end and flexible, hire-when-you-need-it cloud computing at the other. The Mac Pro is very much a personal workstation - turning it on its side and adding rack mounts doesn't make it a server or high-density-computing system.

Apple Silicon is perfect for laptops/SFF where its low power consumption/heat emission, and copious Thunderbolt bandwidth for connectivity pay dividends. That's where Apple makes its money. But - as you say - it's not the right tool for a big box'o'slots with discrete GPUs.Ultrafusion is a neat trick - but it's clearly not perfect (given the late appearance of the M3 Ultra and the non-appearance of the x4 option). The Mx Max simply doesn't have enough PCIe gen 4 lanes for a tower system - the M2 Ultra Mac Pro uses the second die's PCIe v4 lanes that would have been used for storage, but still wouldn't really be enough for a bunch of high end GPUs (even if they were supported). For some users who'd be happy with the (not inconsiderable) CPU performance of the Mx Max that's a very expensive way of getting fairly modest PCIe bandwidth and extra RAM.

There are rumours that Apple are working on an AI server chip (maybe they'll do something like NVIDIA's Grace/Hopper) which I guess could be the next Mac Pro - but I think it would be a very different product.

The Apple charges too much for storage complaint. I’m not sure how long you’ve been buying Apple products but this is nothing new. I don’t think this is specific to the Mac Pro or any Apple product.
Well, I've been buying Apple products since they used bog standard spinning rust hard drives which - whatever Apple charged for them - were easily upgraded with third party products.

What's new-ish is that even in the Studio and M4 Mini where the storage is potentially upgradeable, Apple are making it difficult... and what's the point of a small-form-factor system if the first thing you have to do is hang an external drive off it to get decent storage without selling a kidney?

Also, while gouging for BTO upgrades vs. commodity prices is par for the course, you'd still expect to get more for your money as the years passed & for the base spec to keep up with the times. Apple kinda parked their upgrade prices and base specs about 10 years ago, and while they improved their RAM options a bit with the M4, storage is still way behind the game. SSD may still be expensive c.f. spinning rust, but that's partly because spinning rust has continued to add capacity. Pretty sure 2025 bucks-per-gigabyte for SSD is way lower than 2012 bucks-per-gigabyte for spinning rust - but you wouldn't know that from Apple's specds and upgrade prices.
 
The article points out and apple confirms that making the ultra version of the chip just takes more time. Which explains their ass backwards introduction o the ultra chip last, rather than first. Which makes you never want to get an ultra machine because as soon as you get it, a couple of months later the next Max version of the chip eclipses the current ultra. If they started with Ultra and went down to the other versions of the chip it would make more sense.

"Ass backwards" would be to fabricate the most complicated chip first on a new process.

The reason we have an M3 Ultra and not an M4 Ultra today is because the M3 fabrication process has been ironed out and wafer yields and capacity are now high enough you can manufacture them for a similar cost structure to what M2 Ultra cost.

And we know these M3 Max die are not the same as those in a MacBook Pro since they have TB5 and new memory controllers that can access four times as much memory. I also would not be surprised if Apple announces M3 Ultra supports PCIe 5 compared to the PCIe 4 of M2.
 
I think that Apple is working on a server processor like the Xeon they had in the Intel Mac Pro. A good start would be using it to power AI.

Yes this has been reported. The original reports I believe said it would be based on the M2 Ultra, but I presume it will use M3 Ultra.
 
For some users who'd be happy with the (not inconsiderable) CPU performance of the Mx Max that's a very expensive way of getting fairly modest PCIe bandwidth and extra RAM.

"Fairly modest". I'll say - 16 lanes of PCIe 4.0. By comparison, my i5 has 20 lanes of PCIe 5.0. A Threadripper Pro has 128 of 5.0 off the CPU alone (plus chipset PCIe lanes).

There are rumours that Apple are working on an AI server chip (maybe they'll do something like NVIDIA's Grace/Hopper) which I guess could be the next Mac Pro - but I think it would be a very different product.

Even if Apple were interested in competing with Nvidia in that space, which is extremely unlikely, such a machine would be irrelevant to desktop Mac users.

What's new-ish is that even in the Studio and M4 Mini where the storage is potentially upgradeable, Apple are making it difficult...

The Studio and mini have flash storage on cards. Not only do these use a proprietary format, rather then the industry-standard (and similarly sized) NVMe, they won't even sell you them separately.

It's fortunate that various manufacturers have reverse-engineered them and are selling them at a reasonable price, but this could change for future machines (including potentially the new Studio) if Apple implement a hardware lock-out.
 
There is not enough popcorn in the world to watch the posthoc revisionism of how the M4 Ultra, which everyone everywhere was predicting as a certainty mere days ago isn't going to happen, but how the M3 Ultra is what makes perfect sense and reveals a grand strategy that the M4 Ultra lacked, and it doesn't matter if it's slower at single core than the M4 Max, it's better at big memory wide tasks... which somehow wasn't a good enough justification for Xeon machines.

Wake up. The Ultra is being fabricated on M3 because there's capacity for its established fabrication spare, now the M4 is the mainstream processor.

It's eating the whole buffalo, entrails and toenails, building from spares and scraps and slapping another zero on the pricetag. It's not a "strategy".
 
I think that even the Intel Mac Pro was running into problems - mainly that it was only ever of interest to a niche of people locked into Mac-only pro apps, when such apps - or their equivalents - are increasingly available on Windows and/or Linux. At the end of the day, apart from running MacOS, it was just a Xeon-W/PCIe tower with a slightly neater way of connecting GPUs that needed extra power & thunderbolt plumbing. For GPU-bound tasks, its performance was limited by the same AMD chipsets that you could get on PC - and tough luck if you wanted NVIDIA.

With laptops and small-form factor devices aimed at mass-market users, Apple could wow people with cool designs, the nice UI and easy integration of iPhones etc. With a big box'o'slots that sits under a desk and spends its life doing one or two well-defined, specialised tasks, why not get a generic x86 system which is much easier to configure to your exact requirements by bolting together generic parts? Especially when AMD are offering workstation/server CPUs that thrash Xeon on price/performance. The 2019 MP had a short honeymoon when it was an early adopter of the latest Xeon-W chip with extra PCIe bandwidth and RAM support that hadn't shown up on other manufacturers menus yet, but there's a far wider choice of high-end x86 configurations - multiprocessor systems, dedicated GPU computing systems etc. than just Xeon-W.

Sure, for some truly pro users it was cheaper to shell out for a Mac Pro or six than to upset their workflow by shifting to PC - but that's going to be a rapidly evaporating pool. In fact, given 7 years of "wilderness" when the only Mac Pro was the Trashcan, then the 2019 attachment - then a mere few months later the announcement that Apple were abandoning x86... its a wonder that there is any business/professional market for a Mac Pro left at all...
I think again all of this is based on the operating system. If there’s a need for macOS, then none of those AMD systems will work. You’re right though many have switched to Windows because Apple has not really supported the pro desktop market. The trashcan looked really cool, but wasn’t really a success. I think by the time they came out with the 2019 Mac Pro, it was too late. You probably had a few movie studios buy them along with a few rich kids that wanted to be cool and say they had a Mac Pro. I doubt the sales numbers made the R&D worth it. I could be wrong because I don’t work for Apple so I don’t know if any of this as fact.

Meanwhile, the whole concept of a high-end personal tower workstation - on whatever platform - is being eroded by increasingly capable laptop/small-form-factor systems at one end and flexible, hire-when-you-need-it cloud computing at the other. The Mac Pro is very much a personal workstation - turning it on its side and adding rack mounts doesn't make it a server or high-density-computing system.
On the personal workstation side I think they hit the nail on the head with the Mac Studio. I know you emphasize the word tower, but people that use these computer computers for work don’t care about the shape of the computer. If anything they would prefer the smaller for factor because it would fit easier on the desk. Again, I’m talking about people that are using them for work not tinkerers that want to pull out cards and spray can air in their computer every week. For that PC is the only option now.

Apple Silicon is perfect for laptops/SFF where its low power consumption/heat emission, and copious Thunderbolt bandwidth for connectivity pay dividends. That's where Apple makes its money. But - as you say - it's not the right tool for a big box'o'slots with discrete GPUs.Ultrafusion is a neat trick - but it's clearly not perfect (given the late appearance of the M3 Ultra and the non-appearance of the x4 option). The Mx Max simply doesn't have enough PCIe gen 4 lanes for a tower system - the M2 Ultra Mac Pro uses the second die's PCIe v4 lanes that would have been used for storage, but still wouldn't really be enough for a bunch of high end GPUs (even if they were supported). For some users who'd be happy with the (not inconsiderable) CPU performance of the Mx Max that's a very expensive way of getting fairly modest PCIe bandwidth and extra RAM.
I just don’t even know what an Apple Silicon tower would look like even if it was in an M18 Ultra Max Pro Uber Duber. They could have rack mounted storage, but I don’t think the external graphics is going to work with the Apple Silicon process. Everything is on that one chip and that’s what makes it so powerful.


There are rumours that Apple are working on an AI server chip (maybe they'll do something like NVIDIA's Grace/Hopper) which I guess could be the next Mac Pro - but I think it would be a very different product.
I would put my paycheck that they are working on some sort of AI computer or server. Whether this will actually come out in production, who knows. Everyone is working on this I’m sure.


Well, I've been buying Apple products since they used bog standard spinning rust hard drives which - whatever Apple charged for them - were easily upgraded with third party products.

What's new-ish is that even in the Studio and M4 Mini where the storage is potentially upgradeable, Apple are making it difficult... and what's the point of a small-form-factor system if the first thing you have to do is hang an external drive off it to get decent storage without selling a kidney?
The point is modularity. You can set up your system like you want on the go. You don’t need to pull your whole computer apart to change the SSD. If you’re talking about the beauty aspect of it a lot of people feel that having an external device looks ugly. You could potentially hide it somewhere, but to me that’s a bit silly

Also, while gouging for BTO upgrades vs. commodity prices is par for the course, you'd still expect to get more for your money as the years passed & for the base spec to keep up with the times. Apple kinda parked their upgrade prices and base specs about 10 years ago, and while they improved their RAM options a bit with the M4, storage is still way behind the game. SSD may still be expensive c.f. spinning rust, but that's partly because spinning rust has continued to add capacity. Pretty sure 2025 bucks-per-gigabyte for SSD is way lower than 2012 bucks-per-gigabyte for spinning rust - but you wouldn't know that from Apple's specds and upgrade prices.
I really think Apple is doing this to keep the base prices low. There are quite a few people that will needlessly upgrade their purchase. Is that an appropriate or ethical business practice? I don’t know.
 
The Ultra is being fabricated on M3 because there's capacity for its established fabrication spare, now the M4 is the mainstream processor.

Interesting. So you reckon there wasn't any particular difficulty with releasing an M3 Ultra last year, just that the capacity was was being used for MacBook Pros? And now those have moved on to M4 (and the M3's process has been refined), they can bring out an M3 Ultra? On the basis that this is an expensive yet niche processor, that has no competition in the high-end Mac space, so they may as well save costs?

Using the n-1 model for the Ultra could make sense going forward. The only issues being a) it's still very expensive to buy, and b) the n+1 Max comes out ~6 months later.

I guess almost any configuration will suit someone out there. There were people for whom the 2023 MP was 'ideal' (or at least acceptable). But the traditional workstation market is clearly a bad fit for Apple Silicon.
 
There is zero reason for the Mac Pro to exist ever again. I don't see why someone would want a computer with the same components in a much larger form factor. Let's face it, large form factor (Tower Computers) are a thing of the past and mini computers are the future.
 
  • Like
Reactions: Ruftzooi and assoc
Again, I’m talking about people that are using them for work not tinkerers that want to pull out cards and spray can air in their computer every week.

This is a rather pejorative description of tower use. The main benefit is value for money, as it reduces the manufacturer's leverage to rip you off. You can also take advantage of falling component prices over time, rather than being forced to pay top dollar upfront. Storage needs in particular only go up, whilst prices come down.

And don't knock the ability to dust out your computer; the Studio should be so lucky.

I really think Apple is doing this to keep the base prices low.

It's just price discrimination, no different to how an airline prices seats.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.