Apple Silicon Mac Pro Said to Feature Same Design as 2019 Model, No User-Upgradable RAM

[Unregistered 4U]

It isn't 1990 anymore. Single threaded apps went in the dustbin about 20 years ago.

And yet, every computer running today is running a lot of different single threads at a time. Unless, you believe things like “Notepad” and “minesweeper” are massively multithreaded.

If you want to compare Intel - a 13th gen i5 will outperform it.

Compare the performance of Apple’s lowest performing M1 to the lowest performing processors from anyone else. You’ll find that the M1 outpaces a wide majority of what AMD and Intel are currently shipping.

Blender? A laptop with a 3060 gpu is over 3 times as fast.

Final Cut Pro? The Mac completed the test. The PC was 0 times as fast.

 

[kingtj1971]

It's funny because the Mac Studio is the Trashcan 2.0: Lots of power that you expand by external Thunderbolt connections. Costs roughly the same as a Trashcan 1.0, too. Personally, I love both trashcans.

Yes, but the Mac Studio wasn't really marketed the same way. The original "Trashcan" Mac Pro was heralded as a worthy replacement for the previous Mac Pro towers. The "story" was, all the thunderbolt connectivity it provided would let you expand it as much as you'd ever need to, and it was supposed to support both plenty of RAM and video board upgrades. (The video board was, in fact, removable, but was very proprietary and nobody ever made third party options to swap into one. The thermal limitations imposed by the limited space for heat-sinks or additional cooling fans also made it impossible to really use it with most newer GPUs.)

I think Mac Studio owners understand that it's more of a "what you see is what you get" system.

 

[theluggage]

Final Cut Pro? The Mac completed the test. The PC was 0 times as fast.

...and that's the nail hit on the head. I don't think that Apple has been in direct competition with PC hardware in terms of raw power for some time now (maybe even since the 1990s) - because that would be a game they could never win without, basically, just making PC clones. The latest AMD space-heater GPU isn't going to run Blender significantly faster on a Mac Pro than a PC, but an ultra-portable M2 Air might run FCP with Apple ProRes usably where a PC ultrabook with integrated graphics running equivalent Windows software would struggle.

Problem with the 2019 Mac Pro is that - if you take MacOS out of the equation - it's just a Xeon-W tower with AMD GPUs in an over-engineered box and a slightly neater way of routing power and Thunderbolt video to PCIe GPUs - and PC tower users probably don't give a fig about using Thunderbolt for video. Continuing down that route is just a war of attrition as locked-in customers - who started with Macs way back when they used a different class of hardware and processor to PCs - gradually wean themselves off MacOS onto Windows, Linux & possibly power-on-demand cloud computing.

The M1/M2 Macs (coupled with MacOS and MacOS-optimised software) are providing something interesting and different in the small-form-factor market that does what it does well - but if I actually needed 2019 Mac Pro-scale power and scalability I'd probably go get a threadripper tower running Windows or Linux and I certainly wouldn't start down a MacOS-only road at that level if I wasn't already committed.

Don't get me wrong - I'd actually love a "Mac Studio tower" with a few (non-GPU) PCIe slots for internal storage and extra I/O cards.

 

[sunman42]

Indeed, I suspect that you'll be able to add storage after the fact with the Mac Pro, where as Apple artificially prohibits adding storage to the Studio

At the risk of appearing pedantic, the Studio has four (or six) ports that allow adding storage, just not internally — but at PCIe speeds.

 

[seek3r]

The quote from Gurmans letter is better stated on 9to5mac: ”
Gurman says the machine has two spare SSD slots, in addition to slots for graphics, media and networking.
”
So, if this is true (gpu!)the machine might be worth it. Personally I just need 128 ram and solid upgradable gpu that renders stuff in similar speeds as a PC.

I really do hope that there is an “extreme ” variant after all so even cpu bound loads that need more ram is catered for. 48 cores and 384 ram could be a world leading cpu perf workstation. A crown that would be cool to have for Apple. (Higher ST perf than all other high core count cpus)

Whether the GPU is upgradeable in any meaningful way will seriously depend on driver support. If it's only first party GPUs from Apple it'll be so niche that it'll be practically pointless. Think the situation with the MP 6,1s cards, but worse. If it at least supports AMD GPUs that makes things a lot better (and maybe would help get us back egpu support in future AS Macs)

 

[DavidSchaub]

Whether the GPU is upgradeable in any meaningful way will seriously depend on driver support. If it's only first party GPUs from Apple it'll be so niche that it'll be practically pointless. Think the situation with the MP 6,1s cards, but worse. If it at least supports AMD GPUs that makes things a lot better (and maybe would help get us back egpu support in future AS Macs)

... and the discrete GPUs *might* only be for "software acceleration" not display. :/

 

[seek3r]

And yet, every computer running today is running a lot of different single threads at a time. Unless, you believe things like “Notepad” and “minesweeper” are massively multithreaded.

FWIW Textedit has a combined 9 threads running on my MBP right now

 

[Boil]

Whether the GPU is upgradeable in any meaningful way will seriously depend on driver support. If it's only first party GPUs from Apple it'll be so niche that it'll be practically pointless. Think the situation with the MP 6,1s cards, but worse. If it at least supports AMD GPUs that makes things a lot better (and maybe would help get us back egpu support in future AS Macs)

First-party GPGPU would only seem pointless to those clinging to the non-Metal / non-TBDR / non-UMA way of handling graphics that Apple is obviously moved away from...

If one needs RTX / CUDA / Optix, then obviously a Mac is not the way to go, plenty of PCs out there; and hey, it will cut the winter heating bill at the same time...!

... and the discrete GPUs *might* only be for "software acceleration" not display. :/

Rumors have real-time ray-tracing as a goal for the Metal Blender viewport, so no worries there...

ASi GPGPUs handling compute/render on the backend, with the "iGPU" dedicated to display graphics seems like a non-issue to me; excellent working viewport performance with compute/render work happening in the background on the GPGPU(s)...

Full-tilt workflow without any slowdown because a render job is trying to work on the same GPU, how is that a bad thing...?

 

[cobaltlemon]

Apple came out of the gate swinging with the M1 Processor and its variants. However, I am unimpressed with Apple's performance improvement rate. If you plot the numbers of each new chip release on a line chart, you can see that the slope of Apple processors is much lower than AMD and Intel. This does not bode well for Apple. I can see a world in a year or so where Apple chips are more power efficient but are 50% of the performance of AMD and Intel. Picture a world where Apple apologists say that all you need is the performance of M2 or M3, downplaying the significance of competitors' performance gains. How will Apple users feel about that? That can't be good when you feel left behind because of this manufactured walled garden.

 

[Unregistered 4U]

I can see a world in a year or so where Apple chips are more power efficient but are 50% of the performance of AMD and Intel. Picture a world where Apple apologists say that all you need is the performance of M2 or M3, downplaying the significance of competitors' performance gains. How will Apple users feel about that? That can't be good when you feel left behind because of this manufactured walled garden.

I can, too. And guess what? It won’t even matter. Because, regardless of what AMD and Intel releases, they won’t run macOS or Mac applications anywhere near as well as Apple’s Silicon. That’s true starting today as, for example, neither AMD nor Intel will ever include ProRes hardware in their system where Apple will.

Apple users that desire to be in a waste water ejection distance contest with Intel/AMD will find themselves disappointed (starting back with Apple Silicon was released). The vast majority, however, will be using a system that far outpaces their day to day needs. So, the fact that a $700 chip that can’t run macOS outpaces their day to day needs even further… won’t even matter.

 

[DavidSchaub]

Apple came out of the gate swinging with the M1 Processor and its variants. However, I am unimpressed with Apple's performance improvement rate. If you plot the numbers of each new chip release on a line chart, you can see that the slope of Apple processors is much lower than AMD and Intel. This does not bode well for Apple. I can see a world in a year or so where Apple chips are more power efficient but are 50% of the performance of AMD and Intel. Picture a world where Apple apologists say that all you need is the performance of M2 or M3, downplaying the significance of competitors' performance gains. How will Apple users feel about that? That can't be good when you feel left behind because of this manufactured walled garden.

Apple set themselves up for "failure" (note: no failure has occurred) just due to expectation management.

The M1 had a 2 step process node advantage over the Intel chips they were previously using, and moved to an architecture that is mobile-first in design. Apple is still on the same process node and that means: we were never going to see as huge a leap. If the M2 came out 5 years after the M1... maybe

Now Intel's is trying to stay competitive with AMD by building CPUs that use a shocking amount of power.

Apple's apparent plan for the high-end haven't even shipped (or even rumoured to ship).

We'll see how Apple does over the next 3 years.

But as Unregistered 4U says: it doesn't really matter, because Apple is mostly competing with Apple.

 

[quarkysg]

Apple came out of the gate swinging with the M1 Processor and its variants. However, I am unimpressed with Apple's performance improvement rate. If you plot the numbers of each new chip release on a line chart, you can see that the slope of Apple processors is much lower than AMD and Intel. This does not bode well for Apple. I can see a world in a year or so where Apple chips are more power efficient but are 50% of the performance of AMD and Intel. Picture a world where Apple apologists say that all you need is the performance of M2 or M3, downplaying the significance of competitors' performance gains. How will Apple users feel about that? That can't be good when you feel left behind because of this manufactured walled garden.

Do you think it is possible that AMD and Intel have used every trick in the book to up their CPU's performance but Apple is just getting started?

Both AMD and Intel have to crank their CPU clock speed up to 5+ GHz to beat Apple's 3.5 GHz CPU, so that is food for thought. Do you think AMD and Intel can go higher and have that sustained?

If in the end Apple gets left behind because they couldn't compete, then they deserve it. I'm sure by then they will do another transition. Somehow I don't think this is going to happen anytime soon.

Let's see in 3-5 years how it pans out.

For now, let's celebrate Apple's achievement, because you know I'm sure AMD and Intel felt the Apple Silicon heat which resulted in those scorching products they have recently announced.

 

[seek3r]

First-party GPGPU would only seem pointless to those clinging to the non-Metal / non-TBDR / non-UMA way of handling graphics that Apple is obviously moved away from...

That's not why it would be useless. It would be useless as a slotted discrete card because it's unlikely to get many compatible upgrade options, which is one of the main selling points of a mac pro.

The 2013 MP used third party chips and the cards were slotted and still had this problem: the GPU's were custom designed for the machine, outside of the original 3 options (and the higher end 2 had heat death problems) there were never any new cards to fit into that slot. It was an upgrade dead end.

If Apple goes with a first party discrete GPU they need to commit to offering compatible upgrade options at somewhat reasonable prices down the road or it'll seriously discourage purchases

 

[DavidSchaub]

That's not why it would be useless. It would be useless as a slotted discrete card because it's unlikely to get many compatible upgrade options, which is one of the main selling points of a mac pro.

I would think the majority of Pro buyers buy hardware for a use case, and the rumoured Mac Pro will allow expanding the hardware to more use cases.

I doubt "upgradability" is really high on most Pros' list?

Get hardware that does the job. Next time, if the job is bigger, get bigger hardware.

 

[tstafford]

I would think the majority of Pro buyers buy hardware for a use case, and the rumoured Mac Pro will allow expanding the hardware to more use cases.

I doubt "upgradability" is really high on most Pros' list?

Get hardware that does the job. Next time, if the job is bigger, get bigger hardware.

I disagree. I think there is a significant segment of buyers that want to be able to modify the machine (RAM, SSD, video cards, PCIe cards, etc.) as their needs grow and as new tech is released.

If this wasn't the case the Mac Studio would be in essence a MacPro replacement. But it's not. And there are many, many folks who complain about the sealed nature of the Studio.

 

[Unregistered 4U]

If this wasn't the case the Mac Studio would be in essence a MacPro replacement. But it's not. And there are many, many folks who complain about the sealed nature of the Studio.

In essence, it is. All those mid-large companies that buy systems through the companies approved by their procurement agents get machines pre-loaded with what they’re expected to be configured with for their useful lifetime. And, any systems with problems are replaced as part of the agreement with the vendor. That’s where the majority of those systems are going.

The rest go to individuals and those, that are mostly risk averse, are also purchased similarly. Once they have something that works, they’re going to run that to the ground. There’s enough downtime that can occur due to SO many other things, most don’t want to risk causing their own downtime. There are definitely some of those individuals that would like to upgrade their system, but those are the cowboys. Anyone making a significant amount of money with their system on a day to day basis, humming along with a working machine that’s connecting to all their peripherals, storage, etc., and then, on a whim, deciding to take that well balanced, working system and make changes to it assuming that everything’s going to be alright after and that they won’t suffer any downtime… them AND the folks that just like the idea of upgrading and aren’t really losing money if they have a few days or weeks of downtime are the ones that upgrade. With the cost of this machine, I’d think that’s not a lot of them.

 

[chucker23n1]

I disagree. I think there is a significant segment of buyers that want to be able to modify the machine (RAM, SSD, video cards, PCIe cards, etc.) as their needs grow and as new tech is released.

Most corporate buyers can’t make use of it. They get assigned a machine every three years and that’s it. Then you rotate in a new machine. If they absolutely need higher specs sooner, IT gives the existing machine to another employee and gives you a new one.

That leaves the personal market. Most consumers aren’t interested in upgrades either. Of those who are, many are gamers and make demands Apple isn’t interested in anyway.

This leaves a small niche. It becomes even smaller when you realize this machine will be $5000+.

If this wasn't the case the Mac Studio would be in essence a MacPro replacement.

For many, it is.

But it's not. And there are many, many folks who complain about the sealed nature of the Studio.

Don’t extrapolate from what you read somewhere on the Internet to actual purchase behavior.

 

[tstafford]

Most corporate buyers can’t make use of it. They get assigned a machine every three years and that’s it. Then you rotate in a new machine. If they absolutely need higher specs sooner, IT gives the existing machine to another employee and gives you a new one.

That leaves the personal market. Most consumers aren’t interested in upgrades either. Of those who are, many are gamers and make demands Apple isn’t interested in anyway.

This leaves a small niche. It becomes even smaller when you realize this machine will be $5000+.



For many, it is.



Don’t extrapolate from what you read somewhere on the Internet to actual purchase behavior.

Ok. I hear you. And FWIW I'm a Studio owner, love the device and would never purchase a Pro.

Curious - if, as you suggest, few if any folks want the ability to upgrade the Pro what's the point of it in a world where one can buy a Studio with the Ultra chip? I'm asking this honestly and not to be argumentative.

 

[MarkC426]

As mentioned before, it's about Ram/SSD/GPU upgrades.
The one component I have NEVER upgraded in my cMP is the CPU....🤣
Everything else yes.
For me it's not about a 'super fast cpu', it's more about convenience of swapping and changing.

 

[DavidSchaub]

If this wasn't the case the Mac Studio would be in essence a MacPro replacement. But it's not. And there are many, many folks who complain about the sealed nature of the Studio.

I never said the Mac Studio was a sufficient replacement.

My point was that there are use cases where the Mac Studio is not sufficient. The rumoured Mac Pro sounds like it will be a pro machine because of its expansion capability, not it's upgrade capability. Having a system that can be expanded to do bigger, more varied, greater things... now that matters a lot to some Pros.

 

[tstafford]

I never said the Mac Studio was a sufficient replacement.

My point was that there are use cases where the Mac Studio is not sufficient. The rumoured Mac Pro sounds like it will be a pro machine because of its expansion capability, not it's upgrade capability. Having a system that can be expanded to do bigger, more varied, greater things... now that matters a lot to some Pros.

Didn't mean to offend - and I don't believe I misquoted you. But if I did, my apologies.

What's the substantive difference b/w expansion and upgrade? Both seem like they are cut from the same cloth and maybe this is just semantics.

 

[JazzyGB1]

Most corporate buyers can’t make use of it. They get assigned a machine every three years and that’s it. Then you rotate in a new machine. If they absolutely need higher specs sooner, IT gives the existing machine to another employee and gives you a new one.

That leaves the personal market. Most consumers aren’t interested in upgrades either. Of those who are, many are gamers and make demands Apple isn’t interested in anyway.

This leaves a small niche. It becomes even smaller when you realize this machine will be $5000+.



For many, it is.



Don’t extrapolate from what you read somewhere on the Internet to actual purchase behavior.

Apple towers were a staple of the product line up until 2012, so the form factor is very much is in demand by Apple users.
That the 2013 cylinder Mac Pro so spectacularly flopped would suggest your assertion that people simple spec up and replace every three years is unlikely, as obviously they would have done that in 2013 if it were true.
Also that so many PC towers are still in the market would also suggest that your assertion that people don't need expandability is without foundation.
So for those who need a tower Mac expandability obviously IS an issue.

 

[chucker23n1]

Ok. I hear you. And FWIW I'm a Studio owner, love the device and would never purchase a Pro.

Curious - if, as you suggest, few if any folks want the ability to upgrade the Pro what's the point of it in a world where one can buy a Studio with the Ultra chip? I'm asking this honestly and not to be argumentative.

Well, few isn’t none. Probably just tens of thousands (when the Mac as a while has about a 100 million users). But that’s not nothing.

If the ARM Mac Pro is like the 2019, that gives you 1.5 TiB RAM instead of a maximum of just 128 GiB. The M2 or M3 Studio will probably increase the maximum, but not by a factor of 12. And second, it does give you PCIe slots. Some use cases that can’t easily be accomplished with Thunderbolt do exist, most notably GPUs but also some niche hardware such as high-end audio and networking.

Trouble is, it’s not clear right now if that’s happening. Maybe Apple wants to stick to pure UMA; no memory slots then. And maybe Apple isn’t interested in third-party graphics drivers (opens up various cans of worms); no GPU then. That leaves maybe a higher-end SoC (e.g. a hypothetical M2 Extreme that’s essentially a 2x2 M2 Max), probably with a 384 GiB max memory. And PCIe slots purely for niche cases like fiber channel. Almost anyone would be better served by the Studio.

The other reason to do it is for prestige; to show how high-end the Mac, and Apple’s SoC design in general, can go.

 

[DavidSchaub]

Didn't mean to offend - and I don't believe I misquoted you. But if I did, my apologies.

What's the substantive difference b/w expansion and upgrade? Both seem like they are cut from the same cloth and maybe this is just semantics.

No offence taken.

I just wanted to make the distinction between "upgradable" and "expandable".

I see the distinction matters because Apple is rumoured to ship a machine where the ONLY advantage over the Mac Studio will be internal expansion of capabilities, with zero upgradability.

upgradable: The ability to take existing components of a computer and swap them out for bigger/better/faster ones later on. The most obvious case for this are: storage, CPU, RAM, and GPUs. You buy a machine with some set of components and then years down the line you replace some of those components with better ones.
For the most part, Apple seems to have killed this off.


expandable: The ability to buy a computer and, likely at the same time as the initial purpose, add additional capabilities and functionalities such that it can perform use cases that the original computer couldn't do.
So the original computer's CPU, RAM, boot storage, and video out GPU might never change but OTHER functionality could be added:
- additional fast storage,
- slower massive storage,
- hardware interface boards to very use case specific external hardware,
- digital/analog capture boards,
- various kinds of hardware acceleration cards (GPGPU, FPGAs, etc...),
- additional networking interfaces,
- more generic I/O,
- etc...

 

[chucker23n1]

Apple towers were a staple of the product line up until 2012,

They were already niche then. They were popular in the Power Mac G4 days.

so the form factor is very much is in demand by Apple users.
That the 2013 cylinder Mac Pro so spectacularly flopped

We don’t have any data on whether the 2013 “flopped” or whether the 2019 sold better.

would suggest your assertion that people simple spec up and replace every three years is unlikely, as obviously they would have done that in 2013 if it were true.
Also that so many PC towers are still in the market would also suggest that your assertion that people don't need expandability is without foundation.
So for those who need a tower Mac expandability obviously IS an issue.

Most computers sold are phones and laptops, not towers. That towers still exist is besides the point. Their peak was a quarter century ago.