Agree. I can't see any scenario of Apple not allowing a top chip for the Mac Studio unless it's long into the future where the did a redesign of everything and the Mac Studio becomes the Mac Pro with the Studio going away type of scenario.
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.
Mac Studio Speculation [Merged]
- Thread starter darkus
- Start date
- Sort by reaction score
Agree. I can't see any scenario of Apple not allowing a top chip for the Mac Studio unless it's long into the future where the did a redesign of everything and the Mac Studio becomes the Mac Pro with the Studio going away type of scenario.
Graphene heat sink...?
I would love to see a Mn Extreme Mac Pro Cube...
I concur it’s an unlikely scenario. BUT the Ultra Studio does overlap significantly with the Mac Pro and I would think Apple would want to increase the differentiation between the two products.
Assuming the data attached is still close to accurate (is there better or more current data?), the Mac Pro is a larger share of the market than the Mac Studio (I know it just doesn’t seem right). So hurting Studio sales for the benefit of Mac Pro (by limiting the Full M3 Ultra only to the Mac Pro) might still be a possibility.
Most popular Mac: Here's Apple's best selling computer - 9to5Mac
A new study from CIRP looks at the mix of Apple's Mac sales and gives interesting insight into the most popular Mac and more.
9to5mac.com
As that data was from Q3 2022 (ending in September), those Mac Pro sales are Intel models and considering how long "pro power users" waited for an expandable Intel Mac Pro, along with the knowledge the model was probably not going to on offer much longer, I am not surprised it sold well compared to the (at the time) relatively new Mac Studio.
I expect Apple Silicon Mac Pro sales represent a significantly smaller percentage, now.
Last edited:
Those numbers are literally impossible. An obsolescent desktop that STARTS at $6000 selling 1/3 the numbers of MacBook Airs? Apple sells 5-7 million Macs per quarter. Does anyone believe upwards of half a million Mac Pros sold in a single quarter? Or that the Mac Pro outsells the Mac mini by 10:1? Remember that thousands of Mac minis end up in data centers... Relatively inexpensive electronic devices always outsell relatively expensive ones, generally by orders of magnitude...
At that specific point in time? It's not beyond the realm of possibility. I mean the numbers are guesses (though likely educated ones) since Apple no longer releases model sales. The M2 MacBook Air was released in June 2022, so the M1 would have been some 18 months old at that time and there were probably plenty of rumors an M2 MacBook Air was due by WWDC so I could see M1 MBA sales stalling out in that reported quarter as people waited for the M2 model.
Yes, considering the eventual (at the time) Apple Silicon model was widely-believed to not support external GPUs and there was even concerns it would not support any internal PCIe expansion. We'd also been hearing Apple was working on a "four-way MAX" SoC, but that Apple was having major issues. And the fears that it would not have anywhere near the memory capacity of the Intel model (which proved to be true). So, again, I could see anyone even remotely interested in a Mac Pro pulling the trigger on an Intel one around that time, knowing that at least it would be supported for upwards of a decade.
Apple's executives implied that the Mac Pro outsells the Mac mini during their "Pro Mea Culpa Event" in 2017. The mini always went the longest between updates during the Intel era, in part because Intel only made the specific "B-model" CPU it needed every other processor generation and in part because Apple sold so few of them it was not worth spending the money updating the systemboard to support the new "B-spec" CPU when Intel did release one.
So yes, I can believe the Intel Mac Pro outsold the Intel and M1 Mac mini, both in general and especially in Q4 2022.
Last edited:
What about desktops being 1/3 of TOTAL Mac sales? Every time I walk into a university library or a Starbucks (I admittedly live in an area with a lot of Mac-toting hipsters - Cambridge/Somerville, MA), I see 20 MacBooks between the Pro and the Air. I rarely HEAR of the desktops, and I'm a photographer who writes for a photography magazine. I have one friend and colleague who owns a Mac Studio, and a few who have 27" iMacs (the replacements are mostly going to be MBP or Mac Studios, maybe 50/50).
I certainly don't exist in the world where Intel Mac Pros stuffed full of GPUs are most common. Still photography stuff runs REALLY well on Apple Silicon for the most part and many of us like laptops. The rest of us are using older iMacs and eyeing the Mac Studio - we hate the 24" iMac. The other half of my career is in the academic world where Mac Pros are both heavy to get to the library and coffee shop (where much academic writing is done) and unnecessary. I just don't believe the 3D/video/music world is big enough to accommodate half a million Mac Pros in a SINGLE QUARTER (even when you add in that some scientists also like them as research machines).
Pixar has under 1300 employees (no, it's not the whole industry, not by a long shot) - but absorbing 500,000 Mac Pros in a quarter would require close to 1,000 times that many people. I can believe that Pixar is 5% of employees in the high-end animation industry, but not 0.1%, and I'd guess Mac Pro penetration is lower at companies that Steve Jobs DIDN'T found...
According to the US Bureau of Labor Statistics, the ENTIRE movie and music business PUT TOGETHER employs half a million people (I believe that includes that group within the ad industry, too). Did they all buy Mac Pros in the same quarter? Does every 65 year old Best Boy Electrical have a Mac Pro? By reputation, they don't use computers and the director and DP had better accommodate their flip phones and handheld radios, or many kilowatts of lighting won't turn on.
As for the uses in scientific research, the American Association for the Advancement of Science has about 120,000 members - and a huge percentage of scientists are members. Yes, some scientists have more than one Mac Pro (as do some animators), but there aren't enough of them to buy half a million of the things in a quarter (by a very long shot).
I know a little bit about annual camera sales, since I write in the photography space. The total market (mostly among four players) for expensive, pro-level cameras isn't a million bodies per year (it's probably close, but under). The total interchangeable-lens camera market is about seven million per year, mostly the kind that sell at Best Buy for under $1000, and most estimates put the pro stuff at around 10% of the market. A single pro camera that sells 100,000 units a year is a huge, massive, Camera that Saved the Company kind of hit, like the Nikon Z9, which nobody could find in stock for a year.
It's not strictly comparable, but for the Mac Pro to sell in a quarter almost the number of units all pro cameras combined sell in a year doesn't pass the sniff test. I don't know the sales of pro video and audio gear anywhere near as well as I do cameras, but I can't imagine they're hugely greater.
For the Mac Pro to sell what those numbers suggest, EVERY purchase of a professional camera, video camera or music interface/keyboard/mixing board would be accompanied by a Mac Pro. Nobody uses Windows, nobody uses laptops, nobody uses iMacs or Mac Studios. The photographers I know are close to half Windows and 60-70% laptop... All Mac desktops combined are lucky to be owned by a quarter of professional and artistic photographers in my sample (and most of that is iMacs, now migrating to Mac Studios).
50,000 Mac Pros in a quarter, or 100,000 in a full year, would be a lot...
I certainly don't exist in the world where Intel Mac Pros stuffed full of GPUs are most common. Still photography stuff runs REALLY well on Apple Silicon for the most part and many of us like laptops. The rest of us are using older iMacs and eyeing the Mac Studio - we hate the 24" iMac. The other half of my career is in the academic world where Mac Pros are both heavy to get to the library and coffee shop (where much academic writing is done) and unnecessary. I just don't believe the 3D/video/music world is big enough to accommodate half a million Mac Pros in a SINGLE QUARTER (even when you add in that some scientists also like them as research machines).
Pixar has under 1300 employees (no, it's not the whole industry, not by a long shot) - but absorbing 500,000 Mac Pros in a quarter would require close to 1,000 times that many people. I can believe that Pixar is 5% of employees in the high-end animation industry, but not 0.1%, and I'd guess Mac Pro penetration is lower at companies that Steve Jobs DIDN'T found...
According to the US Bureau of Labor Statistics, the ENTIRE movie and music business PUT TOGETHER employs half a million people (I believe that includes that group within the ad industry, too). Did they all buy Mac Pros in the same quarter? Does every 65 year old Best Boy Electrical have a Mac Pro? By reputation, they don't use computers and the director and DP had better accommodate their flip phones and handheld radios, or many kilowatts of lighting won't turn on.
As for the uses in scientific research, the American Association for the Advancement of Science has about 120,000 members - and a huge percentage of scientists are members. Yes, some scientists have more than one Mac Pro (as do some animators), but there aren't enough of them to buy half a million of the things in a quarter (by a very long shot).
I know a little bit about annual camera sales, since I write in the photography space. The total market (mostly among four players) for expensive, pro-level cameras isn't a million bodies per year (it's probably close, but under). The total interchangeable-lens camera market is about seven million per year, mostly the kind that sell at Best Buy for under $1000, and most estimates put the pro stuff at around 10% of the market. A single pro camera that sells 100,000 units a year is a huge, massive, Camera that Saved the Company kind of hit, like the Nikon Z9, which nobody could find in stock for a year.
It's not strictly comparable, but for the Mac Pro to sell in a quarter almost the number of units all pro cameras combined sell in a year doesn't pass the sniff test. I don't know the sales of pro video and audio gear anywhere near as well as I do cameras, but I can't imagine they're hugely greater.
For the Mac Pro to sell what those numbers suggest, EVERY purchase of a professional camera, video camera or music interface/keyboard/mixing board would be accompanied by a Mac Pro. Nobody uses Windows, nobody uses laptops, nobody uses iMacs or Mac Studios. The photographers I know are close to half Windows and 60-70% laptop... All Mac desktops combined are lucky to be owned by a quarter of professional and artistic photographers in my sample (and most of that is iMacs, now migrating to Mac Studios).
50,000 Mac Pros in a quarter, or 100,000 in a full year, would be a lot...
The general rule of thumb was laptops made up ~80% of Mac sales and desktops ~20%, but Intel desktops did get some love in 2019-2020 and then we had the M1 mini in late 2020 and the M1 iMac and M1 Mac Studio in 2022 so I could see demand for Mac desktops starting to ramp through Q4 2022. In Q4 2022 laptops were all well into the M1 cycle and M2 did not drop with the Air and 13.3" MBP until WWDC and the 14" and 16" did not move to M2 until January 2023 so that could have depressed laptop demand for that quarter (especially with all the front-loading that happened in Q1-Q3 2022 thanks to COVID).
Yes, university libraries & coffee shops, these are places I haul my entire desktop rig to on the regular...
And being a photographer who writes for a photography magazine does not really scream "I have my finger on the pulse of the desktop computer market...!"
I was being sarcastic about hauling desktop rigs - but pointing out that this is an important use case for a ton of Macs, and that a lot of people prefer the laptops because work gets done in those places, at least among the two communities I know anything about - academics and photographers...
I'm in contact with a ton of people who use pro-level Macs, and I do a good deal of research about them, but all on the still photo side (so my knowledge of Mac Pros in video and music is more general). I personally use a laptop and a lot of external storage and monitors (currently a last-generation Intel, but my M3 Max order will be going in soon - I got over four years out of this machine, and am hoping for more from my next one), but I know people who go both ways on that one... The photo world loves both the Apple Silicon MBP (with the M3 Max being very eagerly embraced) and the Mac Studio.
Even though I'm a 40 year Mac user (yes, my family had a 128K when I was in middle school, and I've had almost exclusively Macs since), I keep track of the Wintel market as it relates to photography (mobile workstations and "creative pro" laptops. and to a lesser extent, desktop workstations and "dual-purpose" gaming laptops like Razer). I know nothing about the gaming desktop market, but gamers have never been a big part of the Mac Pro market, both because of the cost and due to the "no nVIDIA" restriction.
I'm in contact with a ton of people who use pro-level Macs, and I do a good deal of research about them, but all on the still photo side (so my knowledge of Mac Pros in video and music is more general). I personally use a laptop and a lot of external storage and monitors (currently a last-generation Intel, but my M3 Max order will be going in soon - I got over four years out of this machine, and am hoping for more from my next one), but I know people who go both ways on that one... The photo world loves both the Apple Silicon MBP (with the M3 Max being very eagerly embraced) and the Mac Studio.
Even though I'm a 40 year Mac user (yes, my family had a 128K when I was in middle school, and I've had almost exclusively Macs since), I keep track of the Wintel market as it relates to photography (mobile workstations and "creative pro" laptops. and to a lesser extent, desktop workstations and "dual-purpose" gaming laptops like Razer). I know nothing about the gaming desktop market, but gamers have never been a big part of the Mac Pro market, both because of the cost and due to the "no nVIDIA" restriction.
80/20 is what I've always heard, too. Apart from that 11% (of total Mac sales) Mac Pro number, nothing else looks ridiculous. The only other desktop numbers I've seen are older - late Intel era. At that time, desktops were 20% of the total, but could go up to 25% for a quarter easily enough (if the iMac had just been refreshed, and the MBP was overdue, for example)...
Over 75% of desktop sales were iMacs (including a small number of iMac Pros). Apple was pushing iMacs by making everything else unattractive. The Mac Pro had been a bigger factor back around 2009- 2012, before it got wildly expensive, and when it had unfettered expansion. Even then, I'd have been shocked to see it around 40% of desktop sales (it might have been 25% of desktops in its heyday in 2009???). The Mac mini was usually a 13" Air in a desktop case, and was too low-powered to sell well. Occasional flirtations with more powerful Minis never stuck around for long - people would get used to the idea of a Mini with a 65 watt chip, and its replacement would be right back to a 15 watt chip.
The 27" iMac was the ONLY Mac that offered any speed advantage from being a desktop and was available at anything like a reasonable price in the late Intel era. The entry-level 2019 Mac Pro benchmarked almost exactly like the top (non-Pro) iMac, and it cost twice as much. There were, of course, much faster 2019 Mac Pros, but they were extremely expensive. I don't think it was ever more than a few percent of desktop sales, nor more than 1% of total Mac sales?
By late 2022, the contours of the desktop market had changed... The powerful 27" iMac barely hung on as an old Intel machine in an Apple Silicon world. The Mac mini was low-end only at the time (the M2 Pro model wasn't out yet - there was only a base M1 and an outdated Intel). The only newer iMac was the base M1 "iPad on a stick" that replaced the 21.5", but not the 27". The Mac Studio was around. It was VERY clear that Apple Silicon was the future, so I can't imagine the very expensive Intel Mac Pro selling in any significant numbers (a few in highly specialized applications, sure)... Apart from the Mac Studio and the two most expensive ($11,999 and $12,999) versions of the Mac Pro, a laptop was faster than any of the other desktops.
The paucity of M3 Studio Ultra rumors makes me doubt there will be anything announced before WWDC '24. Seems odd seeing as how the M3 Max in the MBP has equivalent performance compared to the M2 Studio Ultra. That situation can't be good for M2 Studio Ultra sales. Make me wonder if there are technical issues with slapping 2 M3 Max chips together to form the Ultra.
It's probably more an issue of initial yields (Apple is using all the M3 MAX SoCs TSMC can make for the MacBook Pro) and the fact that the Mac Studio was upgraded only six months ago with the M2 family and as good as the M3 MAX is and the ULTRA will be, I expect most of those customers are not going to drop another $2000-8000 that soon.
One possible technical issue (and I don't know if the chips need to be adjacent on the wafer, or if the interposer is added during packaging) is getting two adjacent perfect chips if needed. I'm pretty sure they can cut apart an Ultra if one side is bad, selling the good side as a Max - but I'm not sure if an Ultra has to start out as an Ultra, or if an Ultra is two Maxes and a separately produced interposer added at the packaging stage. If they have to get a double-sized, 184 billion transistor (plus interposer) chip perfect, that's going to be tough on a low-yield process.
M1 Ultra & M2 Ultra are two Mn Max dies coupled via the UltraFusion connectors, M3 Ultra should be the same...
Well, I recall this was talked about in tech circles around the M1 Ultra reveal.
That the tolerances for such a connector to work was so low that really, as what @danwells is getting at, an M1/M2 Ultra is two Max-chips that were produced adjacent to each other on the actual wafer. If true, yeah, getting an M3 Ultra must be very hard with the reported yield.
That wouldn't mean there isn't an interconnect. Just that the "look at us taking two separate chips and connecting them" marketing video was kinda untruthful about the process.
HOWEVER I don't really recall an authority confirming this is actually the case. I want to say with some degree of certainty that I heard it from ATP and that one of them had talked to someone, but my google-fu is failing me and I can't come up with a good source for the claim.
As I understand it, the wafers that make the MAX/ULTRA have the interconnection present so when TSMC goes to wafer cut, they can either make a single ULTRA or two MAX from each pair, the latter being done by cutting it at the interconnector.
There is some confusion as to whether or not an M1/M2 PRO is an M2 MAX with the lower-half of the GPU die space sliced off or if it is a custom SoC. The M3 PRO has to be a custom SOC based on the difference in Efficiency and Performance cores between the PRO and MAX.
There is some confusion as to whether or not an M1/M2 PRO is an M2 MAX with the lower-half of the GPU die space sliced off or if it is a custom SoC. The M3 PRO has to be a custom SOC based on the difference in Efficiency and Performance cores between the PRO and MAX.
M1/2 Ultra consist of two separate Max die that are joined together with a silicon interposer. They can come from anywhere on the wafer, they do not have to be adjacent. I have no idea where @danwells got that idea from, it's completely untrue.
It wasn't. Why would they lie? What purpose does it serve to invent weird conspiracy theories about this?
You won't find confirmation from an authority on silicon technology. Anyone in the silicon design world who's been paying attention to modern packaging techniques knows that it is perfectly reasonable for Apple to join two separate die together, that the M1 Ultra marketing video clearly and specifically references that technology, and that the alternative (relying on two adjacent die being perfect) is a terrible idea. Thus, none of us would ever tell you that this outlandish theory is likely to be true.
By the way, ATP is not an authority on things like this. They're software guys.
You understand completely wrong. There are no wires bridging adjacent Max die on the wafers which sometimes get cut through and sometimes do not. That simply is not how this packaging technology works. They singulate 100% of the Max die on every wafer. Some get chosen to become Ultras, and these are assembled into Ultra packages using further processing steps.
The reason there may be some confusion here is that you are spreading it.
M1/M2 Pro are not M1/M2 Max with half the GPU physically sliced off, they are M1/M2 Max with the lithography artwork cut to remove half the GPU and half the memory controllers. (And also cleaned up so there's no dangling wires anywhere, buses are terminated, etc.) They still made a separate set of lithography masks for M1/M2 Pro chips, and run completely different wafers for Pro and Max.
That's what makes the economics of Pro work, whether M1 M2 or M3. It always costs about the same to run one wafer through the fab, so if you are designing a family of chips and you want the midrange one to be substantially cheaper to make than the high end, you must make sure the midrange chip has a smaller die so you can fit significantly more of them onto each wafer.
Yes, with M3, Apple has begun to spend more engineering budget on creating a truly different Pro die layout in which the artwork isn't simply a crop+cleanup of the Max artwork. This just means that they're spending more money in order to differentiate Pro and Max more, and does not imply that Apple ever made M1/M2 Pro chips by physically slicing away part of a Max.
With Mr. Roboto's insights on how the interposer works (that it happens at the packaging stage, and can be any two Max chips), how much more complex would it be to connect FOUR chips together for an Extreme? How much design is there at the side of the chip to connect the interposer? There's clearly some - you can't just bond any two sides with Magic Supercomputer Glue and have it work...
The current Ultras connect the same side of the two chips - let's arbitrarily call it North to North. If they connected North to South, you could add more chips easily enough (at the cost of an increasing hop count), but since they're North to North, the only connector side is used when connecting two chips.
I have no idea of the difficulty of adding another connector? Either on the South side OR on the East or West sides, allowing a square layout of four chips. A square is more efficient (fewer hops) than a line of four, but it's not further expandable without even more connectors (a square would use North to North and East to East bonds, and would not expose any potential connection sites, while a line would expose the North connector on the northernmost chip and the South connector on the southernmost).
Four should be enough for almost any practical purpose, though. Projected performance should exceed the largest Intel and AMD chips at most tasks (with the possible exception of a few server chips with ~100 cores that could beat it at highly parallel tasks). If it were M3 Max based, it would wind up with 48 P-cores (plus 16 e-cores) and 160 GPU cores. VERY roughly, it should benchmark about 2.5-2.75x as fast as the M3 Max, more on GPU tasks, which tend to be highly parallel. That's enough to be faster than even a Threadripper Pro or a big Xeon, and four M3 Max GPUs should beat pretty much any single Nvidia card, although not the $25,000+ AI accelerators.
Such a chip gives the Mac Pro a reason to exist - it would be a VERY tight fit in the Mac Studio chassis, both physically and in terms of power, but it would be easy to fit in the Mac Pro.
The current Ultras connect the same side of the two chips - let's arbitrarily call it North to North. If they connected North to South, you could add more chips easily enough (at the cost of an increasing hop count), but since they're North to North, the only connector side is used when connecting two chips.
I have no idea of the difficulty of adding another connector? Either on the South side OR on the East or West sides, allowing a square layout of four chips. A square is more efficient (fewer hops) than a line of four, but it's not further expandable without even more connectors (a square would use North to North and East to East bonds, and would not expose any potential connection sites, while a line would expose the North connector on the northernmost chip and the South connector on the southernmost).
Four should be enough for almost any practical purpose, though. Projected performance should exceed the largest Intel and AMD chips at most tasks (with the possible exception of a few server chips with ~100 cores that could beat it at highly parallel tasks). If it were M3 Max based, it would wind up with 48 P-cores (plus 16 e-cores) and 160 GPU cores. VERY roughly, it should benchmark about 2.5-2.75x as fast as the M3 Max, more on GPU tasks, which tend to be highly parallel. That's enough to be faster than even a Threadripper Pro or a big Xeon, and four M3 Max GPUs should beat pretty much any single Nvidia card, although not the $25,000+ AI accelerators.
Such a chip gives the Mac Pro a reason to exist - it would be a VERY tight fit in the Mac Studio chassis, both physically and in terms of power, but it would be easy to fit in the Mac Pro.
Majin Bu posted shots of what they claimed were both the "horizontal" interposer used to connect two MAC into one ULTRA as well as a "vertical" interposer that would have connected two ULTRA side-to-side to make the the "M1 EXTREME" that was supposedly planned for an M1 Mac Pro.
Per claims by Mark Gurman, Apple concluded this "M1 EXTREME" was so expensive that it would have been a multi-thousand dollar BTO upgrade and they felt it was not worth pursuing so the program was cancelled and Mac Pro went forward with just the (M2) ULTRA.
Last edited:
The Ultra itself is a multi-thousand dollar upgrade, so the Extreme would almost have to be...
Teasing out the cost of the M2 Ultra by itself, it seems to be between $1300 and $1400 as an upgrade. The Mac Studio costs $2000 extra with the smaller-GPU version of the Ultra, but it also comes with a $400 RAM upgrade, a $200 storage upgrade and two Thunderbolt ports replacing USB. The Thunderbolt ports are not for sale separately, but are probably not terribly valuable on a machine that already has four (I'm putting them between $0 and $100, but that is a complete guess).
The full-power M3 Ultra is very expensive ($1000 extra over the low-GPU version). That is interesting because the difference on a Max is only $200, suggesting that an Ultra should be around $400. That's either profiteering, extra yield issues or both.
If we interpolate for the M3 line, there's not a ton of evidence, but it's not going to be cheap. The base M3 Max is $200 more expensive than the M2 Max (minus a little for the difference between 32 and 36 GB of RAM - would that be a $50 upgrade if it existed???).
I'd guess that base M3 Max to base M3 Ultra will be a $1500 -$1800 upgrade - a little more than the $1300-$1400 in the M2 generation.
What's much costlier in the M3 generation than the M2 generation is the upgrade from a base Max to a loaded Max ($500 instead of $200). The differences are much more important in the new generation - it's a bigger change in GPU cores, adds two P-cores to the CPU, and changes the memory bus. I'd suspect the difference from base Ultra to loaded Ultra will be somewhere between $1500 and $2000 (the old $1000 difference would be in line with what the Max costs, but it hasn't been - there's been a significant premium for upgrading an Ultra as opposed to upgrading two Maxes). If that is a legitimate yield issue, an Ultra isn't "just two Maxes from anywhere", and a loaded M3 Ultra (let alone a hypothetical M3 Extreme) could be very difficult indeed to manufacture - but it's possible that Apple is just profiteering.
We're already looking at a $3000-$4000 total upgrade from a base M3 Max to a loaded M3 Ultra, and the upgrade to Extreme would be AT least that much again, plus the cost of a base M3 Max (somewhere around $600). If there are yield issues, it could be more. At a guess, it will be about $8000 to go from a base M3 Max to a loaded M3 Extreme (maybe $5000 if you settle for a 40 P-core with 120 graphics cores, instead of 48 P-cores and 160 graphics cores).
The Extreme would probably only fit in a Mac Pro, which comes with a base Ultra for $6999. How about $9999 for a base Extreme and $12,499 for a loaded Extreme (before RAM and storage options)? It might even be a little more because of obligatory RAM upgrades...
Oddly, that's a deal... The Extreme should compete pretty easily with the really big Xeon and Threadripper Pro chips (which are $10,000 for the chip alone - and that's before your system builder marks it up - they cost $12,000-$13000 just for the chip if you get them installed) The M3 Extreme may not be as fast in weirdly parallel workloads, since it's a 56 or 64 core chip (and 16 of those are e-cores), while the Threadripper Pros are as much as 96 cores, all "big" On the other hand, the Threadripper cores are quite a bit slower than Apple's, and even if they are 2/3 the speed, trying to schedule them will eat up the difference in theoretical power (with the possible exception of very parallel jobs, like lots of small database queries). The Threadripper Pros and big Xeons are server chips adapted to workstations, and they're optimized for server work.
A Threadripper Pro 7995 WX machine is going to be significantly more expensive than the Mac Pro - around $20,000 from the first few vendors to hit the market for a pre-built machine with Apple's presumed pathetic base specs (128 GB of RAM and 2 TB of storage). That includes a single RTX 4090 GPU, which may or may not be as fast as the Mac's 160-core GPU. The RAM upgrade to 512 GB (Apple's probable top spec) is somewhat cheaper than Apple charges - ~$3000 on the Threadripper, and I bet Apple will try to get $6000 (judging from the usual RAM prices on Macs), which eats up some of the difference, but not all of it. Yes, you could get the Threadripper to 512 GB for significantly less than your system builder would charge by upgrading the RAM yourself, but, realistically, owners of $20,000 workstations don't do that - those machines are under service contracts. Apple will also overcharge for storage - but there is probably no reason to let them on a tower with multiple PCIe and Thunderbolt storage options. The only catch might be things the operating system won't let you put on another drive - and I don't know how MacOS will handle other internal drives, since the Mac Pro is the ONLY Mac that can have internal drives other than the boot drive, and I don't have one around.
The one real advantage of the Threadripper is that it can scale beyond where the Mac Pro can. Threadripper Pro CPUs are limited to single-CPU installations - but closely related Epyc chips aren't. Similarly, Xeons that support up to 8 CPUs per system are available. You'll pay even MORE per CPU for those, but if you want hundreds of CPU cores in one box, you can have them. Similarly, two and even four GPUs per system are possible, and you can even use eight by using much more expensive workstation GPUs (you'll pay more than twice as much per FLOPS, but you can use more per system).
The sky is literally the limit - a modern supercomputer is made up of a LOT of AMD or Intel CPUs with AMD or Nvidia GPUs, and if you call up Cray and ask "how many can you do in a box", they'll reply "how many are you willing to pay for?" and/or "how much electricity and cooling do you have?" The only way to do that with M-series chips is to cluster a bunch of individual computers. I wouldn't be terribly surprised if we see some M3 Ultra (etc.) clusters - several "supercomputers made out of Macs" were built in the early G5 era, when the PowerPC G5 significantly outperformed Intel chips for a couple of years, but wasn't available in dedicated supercomputing boards. The performance per watt of the M-series is enough better than the competition that I wouldn't be surprised if some university's computer science department is contemplating a big cluster (homemade supercomputers are almost always academic).
Teasing out the cost of the M2 Ultra by itself, it seems to be between $1300 and $1400 as an upgrade. The Mac Studio costs $2000 extra with the smaller-GPU version of the Ultra, but it also comes with a $400 RAM upgrade, a $200 storage upgrade and two Thunderbolt ports replacing USB. The Thunderbolt ports are not for sale separately, but are probably not terribly valuable on a machine that already has four (I'm putting them between $0 and $100, but that is a complete guess).
The full-power M3 Ultra is very expensive ($1000 extra over the low-GPU version). That is interesting because the difference on a Max is only $200, suggesting that an Ultra should be around $400. That's either profiteering, extra yield issues or both.
If we interpolate for the M3 line, there's not a ton of evidence, but it's not going to be cheap. The base M3 Max is $200 more expensive than the M2 Max (minus a little for the difference between 32 and 36 GB of RAM - would that be a $50 upgrade if it existed???).
I'd guess that base M3 Max to base M3 Ultra will be a $1500 -$1800 upgrade - a little more than the $1300-$1400 in the M2 generation.
What's much costlier in the M3 generation than the M2 generation is the upgrade from a base Max to a loaded Max ($500 instead of $200). The differences are much more important in the new generation - it's a bigger change in GPU cores, adds two P-cores to the CPU, and changes the memory bus. I'd suspect the difference from base Ultra to loaded Ultra will be somewhere between $1500 and $2000 (the old $1000 difference would be in line with what the Max costs, but it hasn't been - there's been a significant premium for upgrading an Ultra as opposed to upgrading two Maxes). If that is a legitimate yield issue, an Ultra isn't "just two Maxes from anywhere", and a loaded M3 Ultra (let alone a hypothetical M3 Extreme) could be very difficult indeed to manufacture - but it's possible that Apple is just profiteering.
We're already looking at a $3000-$4000 total upgrade from a base M3 Max to a loaded M3 Ultra, and the upgrade to Extreme would be AT least that much again, plus the cost of a base M3 Max (somewhere around $600). If there are yield issues, it could be more. At a guess, it will be about $8000 to go from a base M3 Max to a loaded M3 Extreme (maybe $5000 if you settle for a 40 P-core with 120 graphics cores, instead of 48 P-cores and 160 graphics cores).
The Extreme would probably only fit in a Mac Pro, which comes with a base Ultra for $6999. How about $9999 for a base Extreme and $12,499 for a loaded Extreme (before RAM and storage options)? It might even be a little more because of obligatory RAM upgrades...
Oddly, that's a deal... The Extreme should compete pretty easily with the really big Xeon and Threadripper Pro chips (which are $10,000 for the chip alone - and that's before your system builder marks it up - they cost $12,000-$13000 just for the chip if you get them installed) The M3 Extreme may not be as fast in weirdly parallel workloads, since it's a 56 or 64 core chip (and 16 of those are e-cores), while the Threadripper Pros are as much as 96 cores, all "big" On the other hand, the Threadripper cores are quite a bit slower than Apple's, and even if they are 2/3 the speed, trying to schedule them will eat up the difference in theoretical power (with the possible exception of very parallel jobs, like lots of small database queries). The Threadripper Pros and big Xeons are server chips adapted to workstations, and they're optimized for server work.
A Threadripper Pro 7995 WX machine is going to be significantly more expensive than the Mac Pro - around $20,000 from the first few vendors to hit the market for a pre-built machine with Apple's presumed pathetic base specs (128 GB of RAM and 2 TB of storage). That includes a single RTX 4090 GPU, which may or may not be as fast as the Mac's 160-core GPU. The RAM upgrade to 512 GB (Apple's probable top spec) is somewhat cheaper than Apple charges - ~$3000 on the Threadripper, and I bet Apple will try to get $6000 (judging from the usual RAM prices on Macs), which eats up some of the difference, but not all of it. Yes, you could get the Threadripper to 512 GB for significantly less than your system builder would charge by upgrading the RAM yourself, but, realistically, owners of $20,000 workstations don't do that - those machines are under service contracts. Apple will also overcharge for storage - but there is probably no reason to let them on a tower with multiple PCIe and Thunderbolt storage options. The only catch might be things the operating system won't let you put on another drive - and I don't know how MacOS will handle other internal drives, since the Mac Pro is the ONLY Mac that can have internal drives other than the boot drive, and I don't have one around.
The one real advantage of the Threadripper is that it can scale beyond where the Mac Pro can. Threadripper Pro CPUs are limited to single-CPU installations - but closely related Epyc chips aren't. Similarly, Xeons that support up to 8 CPUs per system are available. You'll pay even MORE per CPU for those, but if you want hundreds of CPU cores in one box, you can have them. Similarly, two and even four GPUs per system are possible, and you can even use eight by using much more expensive workstation GPUs (you'll pay more than twice as much per FLOPS, but you can use more per system).
The sky is literally the limit - a modern supercomputer is made up of a LOT of AMD or Intel CPUs with AMD or Nvidia GPUs, and if you call up Cray and ask "how many can you do in a box", they'll reply "how many are you willing to pay for?" and/or "how much electricity and cooling do you have?" The only way to do that with M-series chips is to cluster a bunch of individual computers. I wouldn't be terribly surprised if we see some M3 Ultra (etc.) clusters - several "supercomputers made out of Macs" were built in the early G5 era, when the PowerPC G5 significantly outperformed Intel chips for a couple of years, but wasn't available in dedicated supercomputing boards. The performance per watt of the M-series is enough better than the competition that I wouldn't be surprised if some university's computer science department is contemplating a big cluster (homemade supercomputers are almost always academic).
Yes because if I do remember correctly M2 Max MBP's all shipped with 1TB base, but the M2 Max Mac Studios started at 512GB [which is crap lol, insta-up to 1TB].
Hopefully apple just puts 1TB as the bare minimum for studio's least i'll have one thing worry less.
im looking at that too... but that said i do have the ultra basic m3 max 16" 14/30 and 36GB/1TB and its a ridiculous beast for laptop... shocked me; [trial unit from apple store, im getting into macs] - i am impressed with the laptops performance but it was not cheap [nearly 6000aud here] for that... and ive never put 6k on a laptop before until now...
only thing im particularly concerned is thermals given the beast of M3 Max... which is why i'm eyeing the mac studio too for ultimate thermals and wide selection of ports and i rarely carry laptop outside. plus have screen/keyboard/mouse already so just need a box or so.
Gurman wrote today:
"Apple is planning to release M3 MacBook Airs in just a few months. [...]
The move means there’s only a couple Macs left that don’t have the latest processor generation: the Mac Studio and Mac Pro. But they probably won’t get upgraded until the end of 2024 at the earliest, if not 2025."
If there is any reality in that, and there well could be, it is not good news for those of us who have been waiting... and waiting.