With no facts how can it make sense? All this is doing is seeing how much imagination/trust we have with Mark Gurman comments.
If you say lots I would say you're sleep deprived.
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.
Apple Silicon Mac Pro Reportedly Not Coming at WWDC, Mac Studio Refresh Likely Delayed Until M3
- Thread starter MacRumors
- Start date
- Sort by reaction score
If you say lots I would say you're sleep deprived.
Because the only use ever for any sort of expansion slot has been for discrete GPUs only...
Imagine a pro machine that couldn't do generative AI in close to real time sitting on the market in 3 years. Like you say it would be pointless.If they abandon "Pro" niche more power to them - but we are discussing a potential "Mac Pro" workstation in a "Mac Pro" threat so let's stick to that. So, it's either a Mac Pro workstation or a subject to ridicule, which it will be if they deliver it in closed architecture form factor. I don't see them opening it and embracing 3rd party GPUs - it would be against everything else they do at this point. If they do that I will be positively surprised and happy to buy it. Other option is creating their own standalone GPUs but lets face it - noone in the right mind would go that route for a niche product that constitutes 0.02% total product sale of the company
I think we can agree that without real GPU power, this Mac Pro would not be well suited for any visual work in 2-3 years - not only because of no 3D capability but also because of weak support of all the upcoming AI tools we will get in Adobe Creative Suite etc (there is NO other route of development for software companies - they have to embrace AI tools or become obsolete, and I'm not talking about stable diffusion but more about generative, procedural AI tools like the nVidia tools for game devs, AI up-resing, denoising, all the AI stuff you can do with motion picture which is a separate matter, etc etc).
So to sum up there is only one way - 2024 Mac Studio M3. And that should be it. No real R&D cost on Apple side, its sufficient for *some* Pro applications and all the rest will have to switch to Intel/nVidia combos.
It is always possible Apple's plans have been scuttled because they didnt anticipate the need for giant amounts of VRAM to enable near real time AI processing and they know a product that cannot do it would be instantly obsolete.
Maybe you missed the part where I said the Gurman rumors are stupid?With no facts how can it make sense? All this is doing is seeing how much imagination/trust we have with Mark Gurman comments.
If you say lots I would say you're sleep deprived.
Apple will either surprise us all with something much, much more capable than a Mac Studio, or they will retire the Mac Pro. They won't release a Mac Studio inside a cheesegrater case that no one will buy, like Mark Gurman is claiming.
Yeah, nice try, but that wasn't what we were talking about.
You asserted that, if you need PCIe cards, it was "time to time to either ditch Apple or more realistically, figure out a different workflow." I replied it makes no sense to abandon Apple because you need PCIe slots, since a Mac Pro will most likely have those. Nothing in your reply has anything to do with that.
If you can't counter what I said, you've only got two honest choices: (A) acknowledge you got it wrong; or (B) say nothing. Please don't give me a response that pretends to reply to my comment but doesn't.
Apple's AI cores may already have sufficient for client side processing, and Apple is ahead of the competition for VRAM capacity as well, so I just fail to see how Apple is screwed here?
The only significant use for PCIe slots in 2023 is a massively powerful GPU. Aside from that, you can achieve most things though thunderbolt.
This discussion got me wondering that maybe the Mac Studio was really supposed to be the Mac Pro. And the reason for its exsitence and the delay to the Pro may have to do with something contractual that's still in place between Apple and Intel. It does not make sense to me that Apple would create a brand new product category powerful enough to outperform the Mac Pro line and still continue to sell both computers. Also after the release of the Studio, I wonder how much demand is still left for the Mac Pro. I know there's still some but is it enough to have both the Studio and Pro to coexist.
The new Mac Pro... might... might ...be using a new "system design"... The smaller rumored Mac Pro could have been an adequate solution until this new Mac Pro is ready... but delays caused it to get chopped up... and stuffed into a large Mac mini case (Mac Studio).
For the new Mac Pro... The main CPU... Apple M2 based... is used to run the day to day OS and "generic apps"... and handle the traffic (this is important) to the real power... the compute modules.
These modules could be PCIe 5.0 based cards that hold "special Apple Silicon" chips... one... or two chips per card... and up to 3 cards in each Mac Pro. Something like a SUPER Afterburner cards (Afterburner might have been an experiment)... these compute modules could also have their own dedicated memory per chip... thats why there will be no upgradable memory.
Why we didn't get this new Mac Pro sooner... maybe 🦠 and/or 💰
For the new Mac Pro... The main CPU... Apple M2 based... is used to run the day to day OS and "generic apps"... and handle the traffic (this is important) to the real power... the compute modules.
These modules could be PCIe 5.0 based cards that hold "special Apple Silicon" chips... one... or two chips per card... and up to 3 cards in each Mac Pro. Something like a SUPER Afterburner cards (Afterburner might have been an experiment)... these compute modules could also have their own dedicated memory per chip... thats why there will be no upgradable memory.
Why we didn't get this new Mac Pro sooner... maybe 🦠 and/or 💰
Yup, it's either obsolete on release day or completely different than any rumors we've heard so far - which is very unlikely due to many (economic/ideological) reasons. That's why I think it got scraped and we're getting only Mac Studio updates, while Apple pivots - they know sooner or later they need good hardware AI support even for non-pro applications, which means they need to step up their GPU game - and that's the time for a Mac Pro release. They might be thinking: "hey, we've already f*cked up with Mac Pro by delaying it so much anyway - they can wait 2 or 3 more years."
It would seem weird to scrap the Mac Pro after saying it was coming at a “later date”, but Apple is not above “fake it till you make it”, as AirPower demonstrated. They continued to promote AirPower on product packaging and in software updates up until as late as the month it was canceled. They clearly thought it would eventually work and be released, despite the problems. It was a gamble and they lost the bet. I wonder if the same thing is happening with the Mac Pro…
Because no one uses M.2 RAID cards, or 8K video I/O cards, or assorted networking/fiber channel cards, or anything that goes beyond the bandwidth that TB3/TB4 can support...
Here we go again. It's not that no one uses them, it's that there aren't enough people that care about them for Apple to make a whole new computer that no one will buy.
AI, just as 3D, needs quick, multiple parallel computation capability which requires very high memory bandwidth. M silicon uses unified memory, which in case of M2 Max has a cap of 200 Gb/s bandwidth. Geforce 4090 has a GDDR6X memory with a bandwidth of... 1 Tb/s. FIVE times faster memory. Not to mention their GPU is made for these computations. It's not like "unified memory" is a holy grail of design. There's a reason why the memory was not shared / integrated in Windows PC, it's not like all the other companies are just dumb. Dedicated/separate, graphic card VRAM memory is just faster. And you don't need that big memory bandwidth for other, non GPU related stuff, so separate RAM and VRAM makes all the sense in the world. "Unified memory" rolls of the tongue nicely though
As for AI Neural core mumbo-jumbo - I've no idea what the hell is that thing, but it definitely is not fast with the GPU computation, so I doubt it will be fast with AI applications either, considering the memory bottleneck.
Here's my utterly uninformed take:
- Apple found out that their SoC architecture didn't scale as well as they thought it would and needed to make significant changes in order to get what they wanted out of the "Extreme" / "Quadra". This necessitated waiting for the M3 as M2 was already too far along for such major changes.
- MBA and Mini will end up a generation behind the MacBook Pros. So 15" Air will be M2 with M3 coming to the MacBook Pros in late 2023 / early 2024.
- Studio seems to be out of whack. I'd have thought they'd follow the Pros so they'll get M3 later in 2024 once pent up demand for M3 MacBook Pros has dissipated. Possibly they skipped M2 due to lack of demand.
All big tech companies slept on AI - most likely because they didnt anticipate how quickly it would improve and develop. Not even the developers of it predicted it I think.. I think Mac Pro will be released after they pivot a bit / make up for the shortcomings that current M silicon has. I dont know if this means much better integrated GPU cores with higher bandwidth memory, maybe separate (still integrated) VRAM on the chip for GPU... anyway, they need to change something and I think Mac Pro will be created once they figure this stuff out. All the companies have some catching up to do with AI, on different fronts. That includes Alphabet/Google and Apple. Microsoft's good after buying OpenAI.
If the rumors are true, i reckon the "One More Thing" will be their MR headsest
That comparison is almost irrelevant. There are plenty of wireless charging devices out there. Heck my Phillips razor offers wireless changer that works too for iPhones. Maybe thats the reason apple abandoned that. Essentially making something below any useable profit margin.
But talking about a Mac Pro computer that is suppose to way more high tech and very likely needs a large number of M2 Ultra SoC's in addition to the ones needing to be produced for Mac Studios does take time.
Last edited:
That comparison is almost irrelevant. There are plenty of wireless charging devices out there. Heck my Phillips razor offers wireless changer that works too for iPhones.
But talking about a Mac Pro computer that is suppose to way more high tech and very likely needs a large number of M2 Ultra SoC's in addition to the ones needing to be produced for Mac Studios does take time.
Well, the point of the comparison was the way in which Apple announced a product assuming that it would come together in the end, and it didn't. No one knows if that's happening with the Mac Pro, but I don't think it's unreasonable to conclude that it is taking longer than they anticipated.
Disney, LucasFilm, Pixar, these are the customers that Apple has a special relationship with and listens to with regard to the Mac Pro. So, it is not a stretch to ask what does Apple need to do to satisfy these core customers. Apple was distracted with ski goggles and missed the boat on generative AI. We will not see Apple's design changes as a result of this mistake until M6 at the earliest assuming they maintain a yearly cadence. That water is long gone under the bridge. This unfortunate circumstance will not scuttle the imminent MP. Release is targeted to those customers Apple has a special relationship with.
They will not waste One More Thing on the ski goggles. It will only be used for a Halo product announcement.
Nope, you didn't explain why it made sense to assert pros will need to abandon Apple just because they need PCIe slots:
....particularly since there's no reason the MP won't have them.
Yes, if someone needs more GPU power or RAM than the new Mac Pro offers, then they'd need to buy a different machine (or alter their workflow to run those particular calculations on, say, a cluster).
By contrast, if (as is expected) the Mac Pro has PCIe slots, the idea of not buying the new Mac Pro because you need PCIe slots simply defies logic. It's like saying "If you like to drive red cars, it's time to ditch Ferrari".
Last edited: