Apple's Mac Pro Refresh: What We Know So Far

[deconstruct60]

With Mac Studio it seemed to me that the UltraFusion Interconnect was used in that case to combine 2x M1 Maxes into the Ultra, but the actual power of UFI is giving customers ability to customise what kind of compute power the machine will be focussed on, even within the SoC architecture, rather continuing to double everything with "extreme".

The Ultra Fusion Interface is primarily a hyper local inter-die communication interface. It is contructed by putting two bigger dies on top of another small silicon die. ( InFO_LSI on the right below. UFI is plays the role of 'LSI' in that diagram )

3DFabric: The Home for TSMC’s 2.5D and 3D Stacking Roadmap

www.anandtech.com

It is not very likely that Apple is going to run off and create 3-4 different dies to mix-and-match with other dies.
Perhaps there is a small scale die that Apple 'spins out' that is PCI-e provision and just maybe does some

However, it is pretty doubtful they are going to 'rip' the cores they normally co-locate on die ( CPU , NPU , GPU , Impage process) out into an independent , homogenous die. If push too much memory traffic over the UFI connection it won't as well. By putting 'mega' memory bandwidth on each of the two medium-large dies in the Utlra package they cut down on the traffic that has to traverse UFI.


What is being over looked is that the two 'Max' dies cover up all ~10,000 connectors on the UFI LSI die. Pointing off to well other stuff is going to get hooked up is from where? Out the 'bottom' of the LSI off to other Package? Or just an indirect way of doing what InFS does on the left above diagram and run the connections via the package substrate ( at a higher cost of power consumption and higher latency ) .


Folks are treating UFI like it is AMD's Infinity Fabric or Nvidia's NVLINK2 (or gen 3 ) and they may not be necessarily true. Fairly decent chance that UFI is 'fabulously' low power because it is super short range. Perhaps there is some re-drivers/signal boosters that could communicate to a shared PCI-e and/or secondary memory controller where they take the speed/latency hit , but probably not for huge direct memory bandwidth consumers like Appe's main types of cores.

Using the plain Mx to span MBA , MBP 13" , iMac , Mini it is pretty apparent that Apple is looking to do as few dies as they can get away with. The Ultra package using two 'Max' dies where the second die has a redundant secure enclave , some spare TB controllers , system management , etc. To do quad-die it probably would be effective to come up with a non exact matching 'twin' , but still 85+ % overlapping die design elements and layout. ( e.g, sub out secure element and some TB controllers for PCI-e lane provision). Then use one TB focus and one PCI-e focused die to get a Ultra like 'pair' ( could use that in a studio and just not provision out the PCI-e lanes internally) . An then two of those pairs to get to quad ). It they 're-use' most of the Max design then the R&D costs get spread out over more SoCs in more products. ( that has been Apple's modus operandi for their silicon SoCs for years now. Do a few types and spread them out over as many products as they can. )

Mac Pro probably gets something incrementally different. But probably won't be radically different.

So say every Mac Pro was build to order with an M1 Ultra as the base, but the UFI is extensible in a way that allows you to fill up to (say) 8 extra slots with chips that are some combination of dedicated CPU, GPU, RAM or neural engine modules, all connected to the M1 Ultra via UFI. If the performance is there, there might not be too much grumbling about lack of support for third party components.

Whatever is directly coupled to UFI is going to be "glued' to the package. I doubt Apple is going to do 'build to order' chip packages.

Apple probably provision some PCI-e so folks can put in cards into a Mac Pro. But their cores ... those will be embedded in the package. UFI doesn't open that up at all. It is meant for dies to be 'glued' on top.

 

[deconstruct60]

So with WWDC on Monday, we will hopefully see the Apple Silicon Mac Pro announced (not released).

Or not.

So looks like Apple did at least develop an "M1 Extreme" which is effectively four M1 Max using the UltraFusion Chip Interconnect. Code-named "Redfern", this is the 40 CPU core (32P/8E) + 128 GPU core model first leaked by Mark Gurman as "Jade4C-Die".

With the M2 announced... pretty decent change that the Mac Pro would at least a M2 foundation ( and an opportunity to add in 4 die interupt support. As well as a couple of "ultra fusion 2" (or something a bit different for a different packaging technique ) links on a "max sized" size (e.g., swap out one of the memory controller arrays on the side going to do a 'dense pack' with neighbor on another side. ) . For the Mac Pro level , "new" M1 tech at this point doesn't make much sense. That won't satisfy the impatient but need a different die features and new cores at this point.

Now even though Apple did develop it, there is some belief that it might never see the light of day because Apple said M1 Ultra was the last M1 SoC, so there can be no "M1 Extreme".

Pretty good chance. The first iteration may not have worked all that well in addition to Covid impeded roll outs.
M2 seems like it is 'late' from feature sets presented. ( as in it might have originally (back in 2017-18) suppose to go 1-3 quarters before now. )

Some believe it will be a new generation SoC with 40/128 cores based on M2/A15 (so Avalanche/Blizzard). Sami Fathi said it won't be called "M" at all, but instead something new like X1 or Z1.

If M2 is just rolling out then that SoC probably wouldn't be ready real soon. Hence , no sense in doing a "sneak peak" if more than 6 months out until have high volume production of finished "half sized" Mac Pro's to sell. If Apple is putting the M2 into its two highest volume selling Macs then there are not going to be tons of "excess" spare wafers laying around to do quad large die packages with. If they were waiting to do highest volume later than perhaps can do "big" , low unit volume stuff first. But highest volume at the same time biggest wafer 'hogs' (lots of big dies) there is not that kind of fallow wafer start bandwidth out there these days.

 

[CWallace]

Well with Kuo saying Mac Pro is a 2023 product, Apple might have just kicked it down the road a year and will announce it at WWDC 2023.

That should allow it to have the 40-core M3 that William Ma said Apple was developing in his The Information article outlining what he believes the M2 and M3 families will entail.

 

[deconstruct60]

Well with Kuo saying Mac Pro is a 2023 product, Apple might have just kicked it down the road a year and will announce it at WWDC 2023.

The way that reads Kuo says blah so Apple kicked it down the road is likely backward. More like Apple kicked it down the road so Kuo supplier connections said their contracts were kicked down the road , so Kuo says 2023.

There is little good reason to wait for WWDC 2023. If they have to because component production issues (e.g., TSMC N3 has even more major hiccups. Of the packaging process is problematical due to size and complexity. Natural disaster hits some single point of failure factory in the supply chain . etc. ) then they'd have to do it. But wanting to go that late and squat on something ready to go because some magic in syncing it up with WWDC. That is just empty.



Kuo said 2023 several months ago before kind of obvious that the large screen iMac was disappering ( at least for a long while)

Kuo: New Mac Pro, iMac Pro, and Mac mini all delayed until 2023

New insight from analyst Ming-Chi Kuo suggests new Macs we thought were debuting later this year may not arrive until 2023 instead.

www.imore.com

(or slightly earlier without the Mini.

Kuo: iMac Pro Won't Launch Until 2023, 27-inch External Display Without Mini-LED Coming This Year

Apple will not launch an iMac Pro or a Mac Pro until 2023, while it still plans to release a 27-inch external display without mini-LED sometime this...

www.macrumors.com

)

Why the Mini would completely slide into 2023 is a serious mystery. One of the variations of the Mini slide into 2023 maybe. But slapping an M2 into a Mini-like enclosure shouldn't be a "moon landing" kind of complex project.

If the Mac Pro with a N3 based SoC won't ship until March-April 2023 , then Apple could still do a "sneak peak" in October 2022 (along with some other Fall Mac products). Apple could use some 'at risk' production engineering sample to do a limited demo . Or like in 2013, just as static display for the press as to what it "looks like". If the industrial design running way behind schedule that would be a problem , but probably could shovel some hand built static display out the door by December for smaller show.

If they are substantively changing the form factor again (e.g., cut the height in half) then they can "sneak peek" it as a 'new' product 6 months in advance. ( Just like the last two times. )

That should allow it to have the 40-core M3 that William Ma said Apple was developing in his The Information article outlining what he believes the M2 and M3 families will entail.

This?

Future Apple Silicon Macs Will Reportedly Use 3nm Chips With Up to 40 Cores

The Information's Wayne Ma today shared alleged details about future Apple silicon chips that will succeed the first-generation M1, M1 Pro, and...

www.macrumors.com

M2 will be Ok for laptop battles on N5P but hard core desktop workloads in late 2022 is kind of dubious. Which points to why Apple would want N3. A M2 Pro and Max probably won't show much gains. Just incremental better cores with same counts and better memory capacity and that is about it. And that Apple pour most of their effort into M3 (in 'opposite' order; biggest to smallest. ).

If the Mac Pro is going to leverage TSMC N3 then that probably means a push into 2023. If go back to roadmaps from several years ago there was a margin chance that N3 would have started HVM in July-August 2022 and maybe a couple thousand new Mac Pro's could have hit December 2022 as a release date with completed SoC packages and other internals. TSMC gave themselves a broad window to hit; 2H 2022. So if they don't start HVM until October they are still "on time", but the products that are dependent upon that silicon will not be. Systems don't use the stuff at the wafer start. They use the fully fabricated , cut , tested , packaged packages that come out of that long manufacturing chain. If Apple bet the whole Mac Pro farm on TSMC N3, then it will probably slide into 2023. There is a delay with N3, but it isn't huge.

Until WWDC 2023? Errrr probably no. Probably a quarter or 1.5 quarter slide. If the slide for HVM start was siding all the way into very late Q4 '22 then TSMC wouldn't be saying they expected N3 revenue recognition in Q1 2023. Probably it slid toward the middle of 2H 2022 so that the customers don't pay until just after exit of Q4. If it was far to end of Q4 then customers wouldn't be paying until Q2. That is not what they said. Mac Pro is going to need a month or two of full parts availabiity to build up a product launch inventory. It is just every easy to get knocked out 2022 when long term planning has launch in December in the first place.

If using N3, WWDC 2022 was never are realistic ship date even back when this was reported; as N3 had no hope of starting high volume in Q2 '22 .

If Apple is going to try to throw the Mac Pro SoC up against AMD 7000 (RDNA3) and Nvidia 4000 ( Ada ) I'm not sure why they would even want to mess around with kneecapping themselves with TSMC N5 ( even N5P).

 

[CWallace]

Why the Mini would completely slide into 2023 is a serious mystery. One of the variations of the Mini slide into 2023 maybe. But slapping an M2 into a Mini-like enclosure shouldn't be a "moon landing" kind of complex project.

The only compelling reason I can see for the Mac mini (and, by extension, the 24" iMac) waiting until 2023 to add an M2 option would be due to lack of available SoCs. Apple has said the Air is the top-selling Mac with the 13" MacBook Pro being the second-best so if demand for them takes all of the currently available production volume, then Apple might want to wait until supplies are not so constrained before they expand the family.

M2 will be Ok for laptop battles on N5P but hard core desktop workloads in late 2022 is kind of dubious. Which points to why Apple would want N3. A M2 Pro and Max probably won't show much gains. Just incremental better cores with same counts and better memory capacity and that is about it. And that Apple pour most of their effort into M3 (in 'opposite' order; biggest to smallest. ).

I was ready to accept M2 as being only a consumer model SoC and Apple just going from M1 Pro/Max/Ultra direct to M3 Pro/Max/Ultra in 2023.

But there are leaked codenames and core configurations for an M2 Pro, M2 Max and M2 Ultra as well as an "M2 Duo" which sounds like two M2s using the Ultra Fusion interconnect (8 performance cores / 8 efficiency cores / 18-20 GPU cores). So it sounds like Apple will refresh the MacBook Pro 14/16 and Mac Studio with M2-class SoCs.

 

[deconstruct60]

The only compelling reason I can see for the Mac mini (and, by extension, the 24" iMac) waiting until 2023 to add an M2 option would be due to lack of available SoCs. Apple has said the Air is the top-selling Mac with the 13" MacBook Pro being the second-best so if demand for them takes all of the currently available production volume, then Apple might want to wait until supplies are not so constrained before they expand the family.

The initial demand bubble for those two should die off in about a Quarter ( maybe a 1.5 Quarters. ). That would still leave November-December to get the Mini and iMac out. Scarcity of wafer is more likely that the non-Pro iPhone 14 is still on A15 and also soaking up N5P production. That's why the M2 release should have started in the Spring to get to the 'opposite' side of the calendar than the Fall. Apple packing lots of Macs into the same quarter where the iPhones soak up the maximum amount of wafers doesn't make sense. Apple got away with that in 2020 (kind of) but to plan that as the 'norm' is asking for trouble eventually.

If somewhat the A16 is also still stuck on N5P then it would be an even bigger SNAFU. Apple herding all there high demand SoC onto an older node might work if there were not a ton of other folks trying to pile into that node also at the same time.

I was ready to accept M2 as being only a consumer model SoC and Apple just going from M1 Pro/Max/Ultra direct to M3 Pro/Max/Ultra in 2023.

If Apple is going to toss the "Pro" sized die at a "current sized" Mini it would make sense to include it into the M2 generation also. And if doing that then a Pro/Max offering would allow them to clock bump the MBP line up (just to keep the pressure on the competitors). But yeah, since M2 is going to float into the mid-range iPad also. They can get volume.

Decent chance an Ultra won't work is bloat the Max die larger. The packaging techology they are using only scaled to 1x reticle limit. They are already close that now with Max die size. Even if could squeak in even more narrowly under the limits that probably does not help yields.

But there are leaked codenames and core configurations for an M2 Pro, M2 Max and M2 Ultra as well as an "M2 Duo" which sounds like two M2s using the Ultra Fusion interconnect (8 performance cores / 8 efficiency cores / 18-20 GPU cores). So it sounds like Apple will refresh the MacBook Pro 14/16 and Mac Studio with M2-class SoCs.

Same design rules for N5P as N4 so Pro/Max/Ultra could be tagged M2 but not necessarily same fab process. ( A10 and A10X not the same) . Unclear how to get to "Ultra" if bloating up the Max dies. Apple could do the same UltraFusion connector on the end of a Pro die and get a "duo" that is much smaller than an "Ultra" package ( about a Max package size) but has a denser mix of CPU cores than a Max does. It would help with folks who wanted to buy more CPU cores (and RAM) , but didn't need more GPU cores.

The edges of the M1/M2 are mostly used up with I/O already. Not really a vast empty void to put a UltraFusion connector. Also the UltraFusion connector isn't exactly "small" when compared to a M2/M1 sized die. it fits comfortably on a Max die because that is close to 4x as big. Can say the core count is so much smaller can toss out most of the pins (and bandwidth) for the UltraFusion connector. However, now have a different LSI die to do connections with. The connector edge on a Pro and a Max should be about the same. The is same edge use to 'extend' the Pro into Max size ( or viewed another way, chop the Max design to get to a Pro ).

 

[CWallace]

If Apple is going to toss the "Pro" sized die at a "current sized" Mini it would make sense to include it into the M2 generation also.

I am skeptical we'll see an M2 Pro in a Mac mini due to price overlap with the Mac Studio. Apple is going to want at least $100 more for an M2 Mac mini and I could see the base price going up $200 to cover the new case. Apple would then tack at least another $200 on that for an M2 Pro. And going from 8GB RAM to the minimum 16GB RAM will be another $200. And it makes sense the M2 Pro Mac mini would start with 512GB of storage, so that is another $200.

So now we're at $1499 for M2 Pro with 16GB RAM and 512GB SSD. Make that $1699 for 24GB and $1899 for 32GB.

For $100 more, you can get a Mac Studio with an M1 Max and more ports and better (if louder) cooling. Don't really see many not going for the Studio, instead, unless they really need single core performance.

Offering 24GB of RAM on the M2 Mac mini will cover a not-insignificant number of folks feeling memory pressure at 16GB. And an M2 mini with 24GB of RAM and 512GB of SSD in the new case with more ports for $1499 would still leave a $500 gap to the Studio which is large enough to encourage customer differentiation.

 

[deconstruct60]

I am skeptical we'll see an M2 Pro in a Mac mini due to price overlap with the Mac Studio. Apple is going to want at least $100 more for an M2 Mac mini and I could see the base price going up $200 to cover the new case.

$100 more for a smaller case? Why? On top of that why would the M2 Pro Mini have to use a new case at all. It could use the current case ( The Intel one with actually 4 TB sockets. ). Call it Mini Pro. The R&D costs on that basic chasiss used since 2018 should be very , very low. Will need some tweaks but should be no major drama.

As for perhaps the M2 only "new" case ... $100 cost increase how? Rumors are that it is thinner and uses more plastic to promote better RF ( WiFi and Bluetooth). That should not be a major cost increase (it is less aluminum and cheaper plastic. Paint is relatively cheap. ) . If trying to leverage f the M2 MBA cost increases that is probably dubious. The screen on the new MBA is a major shift from what it was. (bigger , notched , brighter and full 10-bit color backlight. ) That is likely a substantive contributor to higher costs ( and why keeping around the M1 model). MageSafe (with colored cords ). Higher bill-of-materials (BOM and inventory overhead ). 1080p camera. Higher BOM. The M2 is a bigger die will lead too is higher BOM, but the previous stuff are far more likely major contributors.

Likewise look at the MBP 13" M2. Same case. keyboard, touchbar , screen, and no MagSafe ... old container and non-binned M2 ( get the extra 2 GPU cores). Same price. Zero cost increase. Apple 'ate' the BOM increase for the M2 on that system.


A M2 Pro Mini with a cost controlled case (i.e., 'old') should be about $300 over a M1 with 16GB RAM and 512GB SSD. Around $1099 + 300-400 --> $1,399-1,499. Large chunk of that $300 is extra margin for Apple (not BOM costs increase). That is $500 off of a Studio Max price. That is a substantive gap to use for market segmentation.

Apple would then tack at least another $200 on that for an M2 Pro. And going from 8GB RAM to the minimum 16GB RAM will be another $200. And it makes sense the M2 Pro Mac mini would start with 512GB of storage, so that is another $200.

So now we're at $1499 for M2 Pro with 16GB RAM and 512GB SSD.

Again which is $500 away from an Ultra.

Make that $1699 for 24GB and $1899 for 32GB.

That's not particularly material. If Add ram to the Studio, then it too moves up away from the $1,999 entry price.

If saying the BTO range of the "Mini Pro" can't overlap at all with the Studio ... that is unfounded. M2 MBA BTO range overlaps with MBP 14" and Apple is just fine with that. The more BTO options stack on top the more warp and distort trying to market segmentation issues. If Apple uses consistent increases on all systems then they all move up/down when apply those and the segmentation gap you started off with stays constant.

For $100 more, you can get a Mac Studio with an M1 Max and more ports and better (if louder) cooling. Don't really see many not going for the Studio, instead, unless they really need single core performance.

That is a sales marketing trap that Apple lays out to suck another $100 out of people's pockets when they already probably over their original budget (already spent an extra $400 what is another $100 ) . Reasonable folks can keep that $100 and spend it on something better. ( like perhaps a new backup drive).

The gap between the "Mini Pro" and the Studio is $500. This $100 thing is just misdirection.

Offering 24GB of RAM on the M2 Mac mini will cover a not-insignificant number of folks feeling memory pressure at 16GB.

It is not going to cover folks who have +8 GPU core and 100GB/s bandwidth more workload. (M2 is only going to give 2 core if upgrade).

And an M2 mini with 24GB of RAM and 512GB of SSD in the new case with more ports for $1499 would still leave a $500 gap to the Studio which is large enough to encourage customer differentiation.

Err. the rumors were that it was a thinner case and the ports were fewer; not more. The M2 is still stuck with two TB ports and less than TBv4 qualifying video out. Can't drive more video out. So just more USB A ports? ( probably not). If the "thinness politburo" gets to the M2 mini like they got to the M1 iMac , then "more" is probably not on the table.

There have been 2-3 iMacs . Still are 4 different laptops and three "MacBook Pro" models. The Mini could have two also and the sky wouldn't fall down. And still have a gap between them and the Studio.

There are lots of contexts that have racked up or stacked up Minis already for the last 8+ years. "Mini Pro" as a physical swap out / swap in would make lots of sense and have substantively large and very viable market. The Studio is "too big" and the very good chance Apple is going to muck around with the new M2 Mini case and make it too small. And will have a roughly similar same segmentation that the MBP models have; only not primarily on screen size and refresh rate.

 

[CWallace]

$100 more for a smaller case? Why?

Because "Apple"? New design so new, higher, price to allow them to continue to sell the M1 Mac mini for $699+.

That's not particularly material. If Add ram to the Studio, then it too moves up away from the $1,999 entry price.

The Mac Studio includes 32GB for $1999.

If saying the BTO range of the "Mini Pro" can't overlap at all with the Studio ... that is unfounded.

The more BTO options stack on top the more warp and distort trying to market segmentation issues. If Apple uses consistent increases on all systems then they all move up/down when apply those and the segmentation gap you started off with stays constant.

Apple's goal is to get customers to spend "just a bit more". Hence why they offer binned CPUs and GPUs and then have up-charges for the "full core" versions.

If the price of a BTO Mac mini with a "full core" M2 Pro, 32GB of RAM and 512GB of SSD is within $100-200 of the price of a base Mac Studio with an M1 Max, 32GB of RAM and 512GB SSD, I would not be surprised if a significant portion of customers choose to spend the extra money on the Mac Studio since it offers more GPU cores and two more USB4 ports.

(24GB of RAM) is not going to cover folks who have +8 GPU core and 100GB/s bandwidth more workload. (M2 is only going to give 2 core if upgrade).

And for that, there is Mac Studio.

Err. the rumors were that it was a thinner case and the ports were fewer; not more. The M2 is still stuck with two TB ports and less than TBv4 qualifying video out.

The mockups of the "M2 Mac mini" case show 4 USB-C, 2 USB-A, Ethernet and HDMI ports. So figure two of those USB-C ports are TB4 and two are USB4. (And if Apple does release an M2 Pro or M2 Max in that Mac mini case, then all four become TB4 ports.)

There have been 2-3 iMacs . Still are 4 different laptops and three "MacBook Pro" models. The Mini could have two also and the sky wouldn't fall down. And still have a gap between them and the Studio.

It is looking more and more like the 24" iMac 4.5K might be the only iMac model going forward (since it bridges the gap between the 21.5" 4K and 27" 5K models). If Apple offers it with an M2 Pro and 24/32GB/48GB of RAM, that would help it close more use cases and then there would be the Mac Studio + Apple Studio Display for the workloads that the top-end iMac 5Ks were used for.

There are lots of contexts that have racked up or stacked up Minis already for the last 8+ years. "Mini Pro" as a physical swap out / swap in would make lots of sense and have substantively large and very viable market. The Studio is "too big" and the very good chance Apple is going to muck around with the new M2 Mini case and make it too small.

This presumes Apple still cares about supporting those types of configurations. Or that those configurations are even appropriate for massive scale Apple Silicon at this time (as in if they are Intel-centric solutions).

 

[Snahbah]

Well with Kuo saying Mac Pro is a 2023 product, Apple might have just kicked it down the road a year and will announce it at WWDC 2023.

That should allow it to have the 40-core M3 that William Ma said Apple was developing in his The Information article outlining what he believes the M2 and M3 families will entail.

I thought he just said iMac Pro was a 2023 product?

 

[NeoSe7en]

 

[CWallace]

I thought he just said iMac Pro was a 2023 product?

He mentioned both the iMac Pro and Mac Pro as 2023 products:

https://twitter.com/x/status/1500507564148400128

 

[theorist9]

If the price of a BTO Mac mini with a "full core" M2 Pro, 32GB of RAM and 512GB of SSD is within $100-200 of the price of a base Mac Studio with an M1 Max, 32GB of RAM and 512GB SSD, I would not be surprised if a significant portion of customers choose to spend the extra money on the Mac Studio since it offers more GPU cores and two more USB4 ports.

The price of a base 14" M1 Pro MBP is not far from that of an an equivalently-spec'd (same SSD and RAM sizes) M2 Air, and Apple has clearly allowed that.

Plus the fact that Apple's not yet discontinued the Intel Mini means they don't see either the AS Mini or AS Studio as adequate replacements. The former lacks capability, and the latter is too big and expensive. What would serve as a replacement for the Intel Mini, which can drive three external displays and has up to 64 GB RAM, is an M2 Pro Mini. That would drive at least 3 externals (that's what the M1 Pro can do), and probably offer up to 48 GB RAM. That's the Goldilocks middle.

Apple acknowledges that, for laptop users, offering only the M1 and M1 Max would leave too big a gap. That's why they have the M1 Pro. Desktop users do the same things with their computers, and thus also need that gap filled.

Of course, I suppose you could argue for an M1 Pro Studio instead of an M1 Pro Mini. That might work, and would certainly handle the M2 Pro's thermals better, though their Mini server farm customers (Mac Stadium and, more recently, AWS) might not approve.

 

[CWallace]

The price of a base 14" M1 Pro MBP is not far from that of an an equivalently-spec'd (same SSD and RAM sizes) M2 Air, and Apple has clearly allowed that.

Yes, you can run an Air up to $1699. But laptops are different beasts from desktops since the former is meant to be portable and the Air's lighter weight can be an important decision influencer if you tote it around 8+ hours a day. Desktops sit on a desk and if you have the footprint for a Mini on your desk, you have it for a Studio.

Plus the fact that Apple's not yet discontinued the Intel Mini means they don't see either the AS Mini or AS Studio as adequate replacements. The former lacks capability, and the latter is too big and expensive.

The Space Grey Intel Mac mini is the same $1999 as the Mac Studio when configured with the i7, 32GB of RAM, 512GB of SSD and 10GBe. And the M1 Max will destroy an i7 on CPU and (especially) GPU benchmarks so the only reason I can see Apple keeping the Intel mini around is for folks who really need x86 support for macOS and Windows.

The M1 mini does indeed not have the expansion of the Space Grey model, but we have been seeing CAD drawings of a new mini design with more ports. Apple could release this as the base M2 Mac mini at, say, $799 and keep the M1 model in the lineup with a price drop to, say, $599. Then with a 24GB RAM option, that would address much of the memory pressure issues people with the M1 are experiencing.

What would serve as a replacement for the Intel Mini, which can drive three external displays and has up to 64 GB RAM, is an M2 Pro Mini. That would drive at least 3 externals (that's what the M1 Pro can do), and probably offer up to 48 GB RAM.

It would, but so would a Mac Studio. And as Apple's strategy is to try and upsell people, if the M2 Mac mini in the new case does not offer enough RAM or cores or displays for you, then the Mac Studio certainly will. And when you take into account Apple BTO upgrade prices for RAM, you'd be paying as much for 48GB on an M2 mini Pro as you would for 64GB on a Mac Studio so the better value is in the Studio.

Apple acknowledges that, for laptop users, offering only the M1 and M1 Max would leave too big a gap. That's why they have the M1 Pro. Desktop users do the same things with their computers, and thus also need that gap filled.

The trick is, with the Mac Studio, Apple gave desktop users M1 Max performance at effectively an M1 Pro price. They easily could have put a 10/16 core M1 Pro into the Mac Studio for $1999 and then offered Max as $200 and $400 upgrades, but instead started with the 10/24 Max and offered the 10/32 model as the BTO upgrade (and then the hyper-expensive Ultra options).

Laptops also have thermal and battery life issues that desktops do not have. M1 Pro draws less power and makes less heat than M1 Max and that's important in a laptop. So the only reason for Apple to have offered M1 Pro in the Mac Studio was to make the base price lower.

 

[Sean bh]

I was going to start a new thread on this, but about to buy a new 2019 Macpro 16 core and was only considering the stock Radeon W5500X, but one of the MAIN reasons I am upgrading is speed of rendering NR plugs, freezing tracks and making Quicktime ref for clients. I heard that it's worth getting the +$400 W5700X 16G since it can help speed up making quicktimes in Protools?

Is there much truth to this and if so, how tangible is the speed of offline QT bounce?

 

[theorist9]

I was going to start a new thread on this, but about to buy a new 2019 Macpro 16 core and was only considering the stock Radeon W5500X, but one of the MAIN reasons I am upgrading is speed of rendering NR plugs, freezing tracks and making Quicktime ref for clients. I heard that it's worth getting the +$400 W5700X 16G since it can help speed up making quicktimes in Protools?

Is there much truth to this and if so, how tangible is the speed of offline QT bounce?

Since you're specifically interested in the Intel Mac Pro, I think you'd probably do better asking this on the current Mac Pro forum. There you'll find a concentration of members who actually use that machine (as well as earlier versions).

Mac Pro

Apple's professional workstation

forums.macrumors.com

Since this thread is about the future AS Mac Pro, I suspect you'll instead mostly get questions from those (like me), who will be curious to hear why you're considering an Intel Mac Pro rather than a Mac Studio, or waiting for the new Mac Pro. I'm sure you have valid reasons—it would just be interesting to hear what they are during the presumable twilight of the Intel machine.

 

[CWallace]

You can also take a look at https://barefeats.com as they did a number of performance tests with various 2019 Mac Pro video card configurations and relevant data might be present.

 

[theorist9]

I didn't agree with all your points, but gave you a thumbs-up because I thought it was well-argued. Often people only give thumbs-ups based on whether the other poster echoes their views, a practice I disagree with.

Yes, you can run an Air up to $1699. But laptops are different beasts from desktops since the former is meant to be portable and the Air's lighter weight can be an important decision influencer if you tote it around 8+ hours a day. Desktops sit on a desk and if you have the footprint for a Mini on your desk, you have it for a Studio.

That's a fair point, and one I hadn't considered—portability is a differentiator that exists for laptops but not desktops.

The M1 mini does indeed not have the expansion of the Space Grey model, but we have been seeing CAD drawings of a new mini design with more ports. Apple could release this as the base M2 Mac mini at, say, $799 and keep the M1 model in the lineup with a price drop to, say, $599. Then with a 24GB RAM option, that would address much of the memory pressure issues people with the M1 are experiencing.

Though this doesn't address the difference in monitor support between the M# and M# Pro chips

It would, but so would a Mac Studio. And as Apple's strategy is to try and upsell people, if the M2 Mac mini in the new case does not offer enough RAM or cores or displays for you, then the Mac Studio certainly will. And when you take into account Apple BTO upgrade prices for RAM, you'd be paying as much for 48GB on an M2 mini Pro as you would for 64GB on a Mac Studio so the better value is in the Studio.

In evaluating BTO RAM prices, you're comparing a future M2-gen Mini to the current M1-gen Studio. You should instead compare apples-to-apples—a future M2-gen Mini to a future M2-gen Studio.

The trick is, with the Mac Studio, Apple gave desktop users M1 Max performance at effectively an M1 Pro price. They easily could have put a 10/16 core M1 Pro into the Mac Studio for $1999 and then offered Max as $200 and $400 upgrades, but instead started with the 10/24 Max and offered the 10/32 model as the BTO upgrade (and then the hyper-expensive Ultra options).

What's your basis for saying the M1 Max Studio is under-priced? It's $800 less than an identically-spec'd 14" M1 Max MBP. Are you saying the extra cost for the MBP version should be less than that?

Also, for the same SSD and RAM, the Pro is $500 or $400 less than the Max if you compare the min-spec’d and max-spec’d (CPU/GPU) models respectively. So let’s call it ~$450. Plus with the Pro you have the option of getting 16 GB RAM, which is a further $400 reduction.

Laptops also have thermal and battery life issues that desktops do not have. M1 Pro draws less power and makes less heat than M1 Max and that's important in a laptop. So the only reason for Apple to have offered M1 Pro in the Mac Studio was to make the base price lower.

I don't think that's a valid argument against the Pro—base price is important. If it weren't, you could just as well say there was no reason for Apple to offer the Studio in anything other than the Ultra version, since all the Max would do is to make the base price lower. And that is the Max's only benefit. Being a desktop, there's essentially no downside to the Ultra except for the price.

Your upsell argument has the same issue—why not offer the Studio with just the Ultra, to upsell people to that? The problem, of course, is you lose sales volume, and the Ultra is a big jump. But for those that might want a Pro Studio instead of a Max Studio, a jump of several hundred dollars might be a big filter as well.

As far as whether it would be better to put a Pro chip in a Mini case or a Studio case, that's an interesting question. If they make the Mini smaller then it probably should go in the Studio. But let's consider the current Mini. It has 94% of the volume of 16" MBP, and effectively a much greater free volume, since it doesn't have a display or battery. Plus the lack of a a display somewhat reduces its TDP. Based on just those factors, the Mini's thermal capacity should be larger than the 16" MBP's, and thus have no problem handling a Pro.

OTOH, the MBP's surface area is nearly twice as large as the Mini's. And based on the M1 vs. the M2, the M2 Pro will likely have a higher TDP than the M1 Pro. So if I were to choose, I might prefer to see an M2 Pro chip in the Studio instead of the Mini—IF Apple's engineers leverage the Studio's TDP to make this hypothetical Pro Studio essentially silent.

 

[CWallace]

Though this doesn't address the difference in monitor support between the M# and M# Pro chips.

True, but Apple may feel that if you need more than a two monitors on an Apple Silicon desktop, you should be using a Mac Studio which can support five of them.

In evaluating BTO RAM prices, you're comparing a future M2-gen Mini to the current M1-gen Studio. You should instead compare apples-to-apples—a future M2-gen Mini to a future M2-gen Studio.

Apple's memory BTO pricing is pretty standardized, so the pricing should stay the same across SoC generations. Apple wants $400 to go from 8GB to 24GB on the M2 MacBook Air/MacBook Pro so they should want that on the M2 Mac mini, as well.

If Apple offers an M2 Pro on a Mac mini, that would come with 16GB as standard, as opposed to 8GB. But Apple has consistently charged $400 for a 16GB to 32GB upgrade, so that would be the price to upgrade an M2 Pro Mac mini from 16GB to 32GB. And the M2 Pro upgrade itself would likely be $400 (or even more) to cover the cost+margin of both the Pro SoC and the 16GB base RAM.

What's your basis for saying the M1 Max Studio is under-priced? It's $800 less than an identically-spec'd 14" M1 Max MBP. Are you saying the extra cost for the MBP version should be less than that?

No, I am saying that Apple could have launched the Mac Studio with an M1 Pro SoC and 16GB in the base model instead of an M1 Max and dropped the base price at $1399 (as the difference between a 10c/16g Pro and 10c/24G Max is $600 because it also requires a mandatory 16GB to 32GB upgrade which adds $400). That would have put the price delta to a Mac mini with M1/16GB/512GB at $300 ($1099 vs. $1399) and $300 is narrow enough there would be no pricing headroom for an M1 Pro Mac mini to slot in between them.

 

[Sean bh]

Since you're specifically interested in the Intel Mac Pro, I think you'd probably do better asking this on the current Mac Pro forum. There you'll find a concentration of members who actually use that machine (as well as earlier versions).

Mac Pro

Apple's professional workstation

forums.macrumors.com

Since this thread is about the future AS Mac Pro, I suspect you'll instead mostly get questions from those (like me), who will be curious to hear why you're considering an Intel Mac Pro rather than a Mac Studio, or waiting for the new Mac Pro. I'm sure you have valid reasons—it would just be interesting to hear what they are during the presumable twilight of the Intel machine.

Thanks,
I work in audio post production and because of a lot of evaluation and research there is no certainty on how long that little bit wait will be for the new M1 desktop Especially till it becomes rocksolid with ProTools, video and everything all the third-party plug-ins on the market today. Could be eight months or could be almost 2 1/2 to 3 years from now for audio postproduction mixing 5.1 features.
Second it has become extremely tiresome and time-consuming working 12 hour shifts but only getting roughly 8 hours of work done because of constantly waiting on freezing (And then unfreezing for a quick fix just to wait to freeze again) tracks and rendering and waiting some more to make quick times for clients same as source and then converting that into a smaller file from 20 GB to 50 GB something smaller using QuickTime Pro that’s manageable for emailing clients.
For the smaller shows that have extremely quick turnaround times and networks waiting on the final product they after day week after week. It’s not worth it to jump on something much more powerful that I can get increased to productivity.

Others might have different situation but this is mine.

I figured if I’m spending this kind of money mine as well go all out if it would behoove me timewise if the 5700 GPU would help out vs the stock 5500

 

[theorist9]

Thanks,
I work in audio post production and because of a lot of evaluation and research there is no certainty on how long that little bit wait will be for the new M1 desktop Especially till it becomes rocksolid with ProTools, video and everything all the third-party plug-ins on the market today. Could be eight months or could be almost 2 1/2 to 3 years from now for audio postproduction mixing 5.1 features.
Second it has become extremely tiresome and time-consuming working 12 hour shifts but only getting roughly 8 hours of work done because of constantly waiting on freezing (And then unfreezing for a quick fix just to wait to freeze again) tracks and rendering and waiting some more to make quick times for clients same as source and then converting that into a smaller file from 20 GB to 50 GB something smaller using QuickTime Pro that’s manageable for emailing clients.
For the smaller shows that have extremely quick turnaround times and networks waiting on the final product they after day week after week. It’s not worth it to jump on something much more powerful that I can get increased to productivity.

Others might have different situation but this is mine.

I figured if I’m spending this kind of money mine as well go all out if it would behoove me timewise if the 5700 GPU would help out vs the stock 5500

There's been speculation that the new AS Mac will be smaller, because it won't have the TDP of the Intel Mac. But I'm wondering you and your colleagues would like it to stay the same size regardless, so you can cleanly fit everthing you need inside the machine:

E.g., be able to do this:

Instead of this:

Source:

 

[Sean bh]

There's been speculation that the new AS Mac will be smaller, because it won't have the TDP of the Intel Mac. But I'm wondering you and yourcolleagues would like it to stay the same size regardless, so you can cleanly fit everthing you need inside the machine:

E.g., be able to do this:

View attachment 2034534

Instead of this:

View attachment 2034535
Source:

Actually, that is the ONLY reason way I use desktops. For a Protools post rig with various Avid cards, separate video cards (not GPU), etc. All in one and the ability to easily upgrade almost everything, even the CPUs.

When working in audio, extra noise, fans, heat, wires, cables, space, etc is a dealbreaker which is why I never bought the trashcan or the macmini.

My current 2012 5,1 is maxed out and can't put any more cards and it's been an issue over the years with work arounds. The new 2019 is another game changer with the extra PCIE and ram capabilities.

 

[mdriftmeyer]

Actually, that is the ONLY reason way I use desktops. For a Protools post rig with various Avid cards, separate video cards (not GPU), etc. All in one and the ability to easily upgrade almost everything, even the CPUs.

When working in audio, extra noise, fans, heat, wires, cables, space, etc is a dealbreaker which is why I never bought the trashcan or the macmini.

My current 2012 5,1 is maxed out and can't put any more cards and it's been an issue over the years with work arounds. The new 2019 is another game changer with the extra PCIE and ram capabilities.

The other route is realizing you would benefit from Neve 1U/2U and desktop centerpiece 5060 gear with the Orbit 5057 and Satellite 5059 Summing along with the 5254 Diode Bridge Compressor, the Master Buss Converter and the Portico II Master Buss Processor, not to mention the Shelford Channel for full vintage sonic tone, allowing for the true console Neve sound w/o relying on digital hardware that become obsolete every five years.

It's the direction I'm moving already bought the MBC to pair with my Avalon Vt-737sp with the Apogee Digital Ensemble that will need replacing when moving to Atmos. Next purchase is the Orbit & Portico II Source Enhancer, then 5254 Bridge Diode Compressor, followed by the Portico II Master Buss Processor. Last on the list will be the Satellite 5059 followed by the Centerpiece 5060 and possibly 500 series options. Relying on Plugins and daughter cards inside hardware that becomes obsolete in 5-7 years is a waste of investment. The other cost is in instruments, Neumann mics and Neumann Atmos speaker configurations while making sure the rooms are properly treated.

A mixed In/Out of the Box is the only way to go. No SoC will ever duplicate what this man has provided and continues to provide.

Rupert Neve Designs – The Most Trusted Name in Sound.

Rupert Neve Designs is the culmination of over six decades of Mr. Neve’s legendary work. With Rupert as our founder and mentor, our team of engineers preserve the revered elements of his most prized vintage designs, while thoughtfully advancing and refining their feature sets to meet the needs of th

rupertneve.com

If you don't have the money for it I understand. Then again if you're going to invest > $10k into a future Mac Pro you probably do have the resources for the gear.

 

[skippermonkey]

My worry is that I'm sure this new Mac Pro will be more powerful than my 24-core Intel – by a decent margin, but not any order of magnitude. However, If I can't slot in any dedicated GPUs, how good are the on-chip systems? Because a pair of W6800X Duos is a formidable amount of rendering power. If it can't be matched by the Apple Silicon, there's no point upgrading for me. Not yet anyway.

 

[Deepshade]

At this rate you wont even be able to add the wheels!