Even if they only sold ten M2 extreme mac pros the bragging and marketing rights would have been worth it. Just giving them away to Spielberg et al.... I miss those days when cachet was everything.
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.
Gurman: All-New Mac Pro Still in Testing, But 'M2 Extreme' Chip Likely Canceled
- Thread starter MacRumors
- Start date
- Sort by reaction score
Yes, the “halo effect” of the “supercar” Mac Pro has value for the whole platform. Gives developers a good reason to port high-powered software to Mac OS, and proves Apple is not just a brand for ultralight laptops and phones.
Since the Ultra is currently only found in one Mac model, the high end config of the Mac Studio, the Mac Pro is a logical flagship for the Ultra CPU. It beats the current Xeon CPUs at almost every level, Apple can truthfully call it a CPU upgrade. (Of course 2x Ultras - the Jade 4C Die -would be even better!)
But Apple couldn’t truthfully call the M2 Ultra’s GPU performance an upgrade from the current Mac Pro’s MPX GPU configurations - up to 2x Radeon Duo cards - quad beastly GPUs. So they couldn’t call an Apple Silicon Mac Pro an improvement on the current Intel Mac Pro unless it can be spec’d with a lot more GPU than the Ultra has.
Whether it’s the rumored “Lifuka” Apple GPU card, or support for AMD MPX GPUs, one way or another they simply can’t go backwards in GPU performance, for this product to have any interest to its market.
The Mac Pro buyer above all is looking for raw power on Mac OS, and isn’t going to buy an M2 Ultra Mac Pro just to get Apple Silicon under the hood, if the performance doesn’t truly exceed the Mac Pro they already have.
(And in fact many Mac Pro customers would probably prefer Intel CPUs for compatibility with their existing, niche, slowly updated work software.) So the AS Mac Pro is going to have to be very compelling to succeed at all. It just has to be.
Since the Ultra is currently only found in one Mac model, the high end config of the Mac Studio, the Mac Pro is a logical flagship for the Ultra CPU. It beats the current Xeon CPUs at almost every level, Apple can truthfully call it a CPU upgrade. (Of course 2x Ultras - the Jade 4C Die -would be even better!)
But Apple couldn’t truthfully call the M2 Ultra’s GPU performance an upgrade from the current Mac Pro’s MPX GPU configurations - up to 2x Radeon Duo cards - quad beastly GPUs. So they couldn’t call an Apple Silicon Mac Pro an improvement on the current Intel Mac Pro unless it can be spec’d with a lot more GPU than the Ultra has.
Whether it’s the rumored “Lifuka” Apple GPU card, or support for AMD MPX GPUs, one way or another they simply can’t go backwards in GPU performance, for this product to have any interest to its market.
The Mac Pro buyer above all is looking for raw power on Mac OS, and isn’t going to buy an M2 Ultra Mac Pro just to get Apple Silicon under the hood, if the performance doesn’t truly exceed the Mac Pro they already have.
(And in fact many Mac Pro customers would probably prefer Intel CPUs for compatibility with their existing, niche, slowly updated work software.) So the AS Mac Pro is going to have to be very compelling to succeed at all. It just has to be.
How so? There’s not even an Apple Silicon Mac Pro on the market and they’re selling MBAir’s at a decent clip. Along with their other laptops, iPads, and iPhones. Having an Apple Silicon Mac Pro for sale wouldn’t materially shift any of those numbers and the sales of the Pro wouldn’t even be a blip on the quarterly call.
For the small number of folks that would actually own the machines? Developers won’t port for the 20 million PLUS machines that Apple are selling now, they’re not going to port software to a market that numbers under 1 million a year (which is what any high end desktop Mac they release would sell).
Apple’s revenue mostly comes from things that aren’t Macs, but, of the Macs, the revenue comes primarily from mobile systems. For folks that are buying these mobile systems, there’s nothing that a large desktop system that starts at over $5,000 is going to do to entice them. The Apple Silicon Mac Pro really doesn’t bring a huge amount of value (Just a small amount for the few they’d sell) and that’s witnessed by the current non-existence of an Apple Silicon Mac Pro during the existence of good sales for their mobile Apple Silicon systems.
I’m sorry but this seems insane. Scrapping the ‘extreme’ chip because it would cost a lot to the end user? Is Gurman aware that the MacPro with a 24 core Intel CPU costs an extra 6k? Or that studios have bought MacPros for 50k?
With the current Mac Pro, you can choose to spend your extra $6K configuring the machine in any number of ways. An M2 Extreme chip would be extremely limited for the price (if you need GPU cores on the M1 machines you have to buy more CPU cores and ML cores as well.) It doesn't make sense to have high-end, low-yield chips which don't actually benefit buyers as they spend more money.
Developing silicon is some of the most expensive processes in modern manufacturing. Cost/benefit has to play into it at some point. But ofc I hope it lands on the benefit end
Last edited:
PPC -> Intel -> ARM in 14 years. Yes I know, 14 years is an eon in the world of computing but BIG software takes BIG resources and development time. The depreciation of OpenGL for YEARS, lackluster hardware, removing or drastically altering various parts of the OS that software may rely on, and no NVIDIA means Apple essentially dug itself its "Mac Pro grave".
The only two pieces of software I can possibly see anyone needing specifically a Mac for nowadays is FCPX and Logic. There isn't anything else BIG that you can't get on Linux or Windows. Unlike the Mac you can expect to use your software for years to come instead of just arbitrarily guessing with MacOS.
I don’t disagree in general with your points, but all of this was also true in 2019 when they announced the new Intel Mac Pro, and explicitly stated how committed they were to this niche market.
And they also proudly announced how many developers of high-end content creation tools would be supporting the new Mac Pro and Metal 2, several for the first time.
Pro app developers react to the new Mac Pro and Pro Display XDR (Apple)
They could have released the Mac Studio back in 2019 and called it the new Mac Pro, it would have been a natural successor to the 2013 Mac Pro, but instead they took a dramatic reversal towards a big case with many expansion slots, mega GPUs and RAM capacity, just eye-watering specs (and prices.) They did this at a time when they knew Apple Silicon was about to be announced. They clearly thought they’d be able to sustain this kind of niche high-end product into the future, and that there was a real value in doing so.
I don’t think the digging of the Mac Pro grave was unintentional. Like the users of the old FCP7, folks in that realm have a very specific set of wants, one of the highest of which is backwards compatibility (which one will find plentiful on Linux or Windows). While that’s good for many, it doesn’t allow Apple the flexibility to focus on “today’s” potential users, the large majority of which have never owned any Mac at all and might find the implementation of some new tech, like Apple Silicon, desirable enough to spend money on. So, while I believe Apple has lost hundreds of thousands of potential Mac Pro customers, they’ve gained millions of MBAir customers. Customers that likely don’t care much for BIG software in the first place. A Mac Pro (that sells in the tens of thousands a year) will never be a draw for developers whose bottom line is always profit or for customers that just aren’t interested in anything they can’t easily take with them anywhere.
I honestly think that the notion of grave digging of the Mac Pro are misguided. Nothing has suggested that Apple has wavered off their course of supporting the professional market. The only rumour is that an already-rumoured chip, the M2 Extreme, is 'likely' cancelled. This doesn't mean Apple isn't committed to the pro desktop, nor does it necessarily mean the chip we do get for the Apple Silicon Mac Pro isn't something unique in and of itself.
Well, it IS true that the professional market is almost entirely x86 based systems, and Apple’s wavered pretty far off from that. It’s not so much about “supporting the professional market” (as long as there are “professionals” some will find Apple’s solutions suitable) but more about choosing which specific parts of the professional market Apple WANTS to support. They’re very tightly focused on satisfying the portion of the professional market that depends on macOS and will continue to make the fastest macOS systems in the world for that small group of folks.
Certainly true. I have a hunch that Apple will continue to try to push into more facets of the pro market over time. I do agree, that there a subset of professionals that just 'prefer' Mac OS, and will use whatever is the most powerful Mac to do so, even if it might be done faster on a different platform.
I would love to see Apple dig further into the likes of Afterburner cards, but for other purposes. Imagine dedicated hardware that accelerates Photoshop across the board? I guess it's about identifying where the pros need more speed, and finding a way to provide it. Dare I say, there isn't an industry out of reach that Apple could 'accelerate' using custom hardware, even on-die custom silicon.
I can imagine a time in the near future where a general purpose CPU is left largely for managing the overall machine, while numerous, dedicated, focused chips are used for each specific task, all contained within the main silicon.
I actually think there will be more pulling than pushing
There are professionals now that understand how Apple Silicon is different from other desktop solutions and are modifying current libraries to take advantage of the fact that, on some systems, the GPU has access to over 40 gigs of contiguous fast main memory much less, memory SHARED with the CPU. That’s something that is and will be a competitive advantage for certain use cases and Apple may find itself being pulled into currently unforeseen areas by these professionals. Not because of any fondness for Apple, but because they’ve defined some AS-only use case that provides them a competitive advantage.Afterburner was required just because they couldn’t build it into the CPU when the CPU was Intel’s. These days, if Apple decides they want to accelerate drawing functions, they can build that into a future core, no further need for external solutions. And, I agree that’s likely where Apple’s headed. There is no winning a performance race when one side can just look at what you’ve released and then goose the processor another 80 watts or so.
They CAN ensure that they continue to build performance modules into the SoC such that, even though Apple will be losing a lot of cross platform benchmark races by a few percentage points, those developers coding specifically for AS will be enabling features for their customers that won’t be easy to replicate elsewhere.
One of the selling points of these ARM chips is shared/pooled RAM. Adding RAM slots is obviously great for modularity, but wouldn’t it be a worse product than the Studio? In other words, since the RAM is not shared on the same chip, it’d be slower.
I presume Apple could design the Mac Pro to have both a shared pool of fastest on-package RAM (256GB or less) and then a larger pool (up to 1TB+) of slower off-package RAM accessed via an external memory controller.
If Apple adds end-user replaceable RAM, it would most likely take the place of the onboard SSD for swap when needed; it would be interesting to see how this works with different levels of the on-package UMA RAM; a bunch of UMA RAM and a bit of tier-two DIMMS, or a bit of UMA RAM and a bunch of tier-two DIMMS...?
The Mac Pro would have up to 192GB (courtesy of the M2 Ultra and the new higher density Samsung LPDDR5X chips, that's why an M2 Air now goes up to 24GB instead of 16GB) of near/fast RAM and up to (let's say) 1.5TB of far/slow RAM.
The Studio has up to 128GB of near/fast RAM (and it won't get 192GB until it's updated to the M2 Ultra, which could be for a while if they update it as often as a MacMini) and 0TB of far/slow RAM.
How would this make the Mac Pro a worse product?
It would beat the current Mac Studio both in the near memory pool (192GB vs 128GB) and in the far memory pool (1.5TB vs 0TB). Let Apple, macOS and app developers figure out how to make the best of leveraging the two pools, it's not so unusual or unfathomable, actually it's been this way for decades: L1 cache, L2 cache, (L3 cache), near RAM, far RAM, RAM disk, (Optane while it lasted), fast SSD, slow SSD, HDD. These are all increasingly slower (and higher latency) memory pools. The Mac Pro would just add the "far RAM" layer. Which incidentally is also going to become a thing outside of the Apple bubble (look up "CXL Memory expansion", which will be supported by Intel "Sapphire Rapids" Xeons and AMD "Genoa" EPYC CPUs).
Last edited:
Apple switched from LPDDR4X SDRAM with the M1 SoC to LPDDR5 SDRAM with the M2 SoC; they are not using LPDDR5X SDRAM yet, if it was the UMA bandwidth would be more than 100GB/s for the M2 SoC...
If Apple does decide to go with LPDDR5X SDRAM the theoretical maximum would be 1TB for a Mn Extreme SoC utilizing the 64GB chips from Samsung, with a UMA bandwidth of 2.13TB/s...
Why does Apple need to target this group?
Is targeting this group profitable for Apple? Is targeting this group require additional resources to maintain, develop new processors, frameworks? Are these requirements adaptable to other segments in the market?
What is the growth predictions for this segment? Is this going to be sustainable for the next 5-15 years to invest resources here?