MP 7,1 Mac Pro 8.1 - WHAT IF...

[mattspace]

That chart would look worse for Porsche. Would you as a shareholder ask them why they would put , forget even the 911, insane r&d money in to the 918 supercar and their next halo supercar when all they really sell are Cayenne SUVs? If they stop with the halo cars, the sales of their overpriced pimp SUVs dry up.

Oh sure, though Porsche has a somewhat unique makeup that its effectively veto controlled by the Porsche family.

And the 911 is what Porsche is primarily known for, but the Mac Pro isn't that for Apple - the iMac has historically been "the mac" in the post-NeXT era, although even then, arguably macbooks are "the mac" for most everyone, and the iPhone is the most Apple thing there is for most people - the Mac is a computer from the iPhone company.

Of course car world isn’t the same, but by pushing developments in heavy duty cpu they can help other parts of their business. Some parts through halo “association” like Porsche. But other parts are like maybe developing server farms. Or heavy graphics ability that eventually can be brought down to mobile devices.

Yeah, as a testbed for the future of lower-end devices it makes a certain sense, like the S-Class is for Mercedes (and frankly, the entire car industry). I guess I just wonder if the lessons of a workstation chip can be scaled down to a laptop chip in a way that makes sense, of if the laptop parts have their own evolution that is inherently not amenable to workstation trickledown.

Underestimate the power of halo at the company’s peril, imo.

Well yeah, the Mac Pro should definitely remain a product, and a halo product for the mac, I'm just wondering if it makes sense for it to be a halo product for Apple Silicon CPUs.

 

[ZombiePhysicist]

Oh sure, though Porsche has a somewhat unique makeup that its effectively veto controlled by the Porsche family.

And the 911 is what Porsche is primarily known for, but the Mac Pro isn't that for Apple - the iMac has historically been "the mac" in the post-NeXT era, although even then, arguably macbooks are "the mac" for most everyone, and the iPhone is the most Apple thing there is for most people - the Mac is a computer from the iPhone company.

Not really true. Ask the average Cayenne driver about 911s and they’re more likely to tell you it’s an emergency number. There is a pro/enthusiast class of buyer that love the 911 and brought the Suv to the attention of their family that don’t care about the heritage. The pro/enthusiasts drive the demand in people that don’t care. I think that does apply somewhat similarly in tech.

 

[mattspace]

Not really true. Ask the average Cyanne driver about 911s and they’re more likely to tell you it’s an emergency number. There is a pro/enthusiast class of buyer that love the 911 and brought the Suv to the attention of their family that don’t care about the heritage. The pro/enthusiasts drive the demand in people that don’t care. I think that does apply somewhat similarly in tech.

The SUVs definitely brought in people who don't care about the 911 - and that's the thing the 911 is so usable as a normal car when compared to other supercars - it's not like you needed a second vehicle so the 911 stays as a low mileage weekender. It's really only large families who needed the SUV as the second / "spouse car" amongst the 911 owner base.

But there's a huge percentage of the Apple userbase who don't care about the Mac Pro - I remember back in the days of even Gruber questioning whether Apple has lost it's ability to make a worstation around 2016, Manton Reece (micro.blog) was loudly proclaiming on his podcast with Daniel Jalkut (redsweater) that Apple shouldn't make a Mac Pro, that no one needed more than the iMac etc... the usual self-centric stuff we've heard a million times over "the (Apple Product) is good enough for me, so Apple's shouldn't spend effort on anything better".

I guess I keep coming back to this notion of an AS Mac Pro which isn't just the paradigm of a 2019 (user upgradable ram, gpus etc)... who is it for?

1. It's not going to be cheap enough for people who wouldn't have bought the 2019
2. It's not going to be flexi-performant enough for people who did / would buy the 2019

What's left - a refrigeatoaster nuts-and-gum-together-at-last that combines the price of a 2019 Mac Pro, with the expansion limits of the 2013 / a Mac Studio with a Thunderbolt Chassis.

I don't see what that product gets for Apple that they don't already cover with their existing products, while it loses them some significant user cases.

A halo machine that gets laughed out of town, like the 2013 was, can drag down the brand faster than it raises it.

Another generation of IA64 (and maybe waiting until Intel has transitioned the Xeon to a performance / efficiency core mix to match the way AS processors do it is a part of the process) with a stonking big AS-based Afterburner, maybe even something like a PCI storage board with m.2 storage-like AS modules just seems like a more logical way to progress that Machine. *shrug*.

 

[mode11]

This actually makes quite a bit of sense. We've mostly been assuming the AS transition will go exactly the same as the Intel transition - total, and as quickly as possible. Apple are well known for their reluctance to support legacy technologies, and we'd assume they'd want to rapidly move on. There would also be clear benefits to being on one architecture from top to bottom, over all their Mac / iOS products.

On the other hand, the last 40 pages of speculation have seen us tie ourselves in knots trying to figure out how a mobile-based SoC architecture with negligible PCIe expansion lanes can power a workstation in the mould of the Intel Mac Pro.

One school of thought is that it can't, and Apple will just cancel the Mac Pro. But this would be odd, since they've already said it's coming, and rumours from Amethyst and others suggest it's nearing production. In addition, it's only been a few years (2017) since Apple made a big song and dance about the need for a big, expandable Mac.

Another strand is that they will take their current highest-end SoC, the Ultra, and build a workstation around it. But it has hardly any PCIe. Using the few TB connections it has and fanning those out via PLX switches to a bunch of slots would make the per-slot PCIe bandwidth a joke. This isn't a possibility; it would be easier to just release a matching TB chassis for the Studio (or rely on others do so).

Another possibility is inserting some kind of bridge chip between two Max chips, connected on either side via their UltraFusion links. This would tap the super-high bandwidth connection to provide PCIe lanes. The problem with this is that aside from adding a lot of complexity, it would likely impact the latency between the SoCs, which needs to be essentially zero to allow them to work as a single large SoC. Also, it doesn't seem compatible with how Ultras are manufactured, which is presumably a bunch of Max's butted up against each other on a wafer, with perfectly functioning pairs becoming Ultras, and the rest getting cut up into single Max's or Pro's.

The last possibility is the all-new super-SoC, designed purely for the Mac Pro. Given this would be the most expensive chip they make, by far, being sold in their lowest-selling machine by far, this also seems very unlikely. It would have no application elsewhere in the range - and never could do. Everything else is mobile / efficiency-first, based on one of two chips (Mx or Mx Max) derived from the current iPhone SoC. The Mac Pro SoC would be performance / clock speed first, with little regard to power efficiency. What trickle-down could there be?

A hybrid of the previous two options is the (rumoured cancelled) 'Extreme' SoC, which logically-enough takes the Ultra concept and extends it to four Mx Max's. This would solve the 'unique SoC' costs, but would still have many of the remaining issues. 4x 'weak' GPU only becomes 'decent' GPU. 'Laughable' PCIe bandwidth only becomes 'mediocre' bandwidth. It's still not compatible with existing manufacturing, which is likely more akin to cutting Ultra's in half to make Max's, than 'stitching together' Max's to make Ultra's (despite the cool CGI animations in keynotes).

Given that sales of Macs are dwarfed by AirPods, it may actually make sense for the Mac Pro to just continue on Intel. It's so far out of the scope of everything else in the range, that trying to include it within the AS umbrella may be a fool's errand. I had assumed it would be a humiliating loss of face for Apple / AS to not have an AS Mac Pro - but would it really? How many of their customers are even aware of the Mac Pro? How many resources do Apple really want to devote to it? Judging by this thread, many actual Mac Pro customers would be delighted if Apple just released a new Intel Mac Pro every 3-4 years, with the latest chipsets and support for the latest GPUs.

 

[singhs.apps]

Another possibility is inserting some kind of bridge chip between two Max chips, connected on either side via their UltraFusion links. This would tap the super-high bandwidth connection to provide PCIe lanes. The problem with this is that aside from adding a lot of complexity, it would likely impact the latency between the SoCs, which needs to be essentially zero to allow them to work as a single large SoC. Also, it doesn't seem compatible with how Ultras are manufactured, which is presumably a bunch of Max's butted up against each other on a wafer, with perfectly functioning pairs becoming Ultras, and the rest getting cut up into single Max's or Pro's.

The last possibility is the all-new super-SoC, designed purely for the Mac Pro. Given this would be the most expensive chip they make, by far, being sold in their lowest-selling machine by far, this also seems very unlikely. It would have no application elsewhere in the range - and never could do. Everything else is mobile / efficiency-first, based on one of two chips (Mx or Mx Max) derived from the current iPhone SoC. The Mac Pro SoC would be performance / clock speed first, with little regard to power efficiency. What trickle-down could there be?

My bet is that the other half of the extreme Soc (that is supposedly no longer in existence) is a pure GPU/GPGPU connected to the ultra via ultra fusion (note : perhaps tellingly they didn’t name their interconnect as extreme fusion)

This means it isn’t tied to main Soc paradigm of Apple’s Mx chips. It will function through a far faster interlink than PCI-e (ultra-fusion)
It may be able to read and write to the SOC ram (instead of carrying its own …or carry its own that the SOC can see and access (like CLX)

There is no reason a future MacBook Pro cannot get its own variant of such a GPU to provide extra grunt.

 

[mode11]

I guess another option for the AS Mac Pro could be a reconfigured Max that is substantially the same (the cores would be identical), but has fewer CPU cores, more GPU cores, more PCIe lanes - and UltraFusion on both sides.

They could then be chained together, likely in the same package. They would start with two chips (better yields than one massive chip) and go up to four and six.

For that to work, though, the UltraFusion would need to be able to link arbritary SoCs together after manufacture. It's not clear if it can do that, or whether the UF is always there between adjacent pairs of Max dies on the wafer, and simply cut through to separate the dies where one side has a defect (or to suit demand i.e. if people are mostly buying Max-based Studios). If it's the latter, it would rely on there being up to four or six perfect dies next to each other, which is not much different than making one huge chip in terms of yields (though would provide more options for good pairs).

 

[mode11]

My bet is that the other half of the extreme Soc (that is supposedly no longer in existence) is a pure GPU/GPGPU connected to the ultra via ultra fusion (note : perhaps tellingly they didn’t name their interconnect as extreme fusion)

It couldn't connect to an existing Ultra like that - the Max only has one UltraFusion connector, and that's already being used to create the Ultra itself.

 

[avkills]

How do you even use octane on macpro2019 these days? Do you have to stay on pr14? With what host dcc?

This was answered; but I actually do not use Octane; my 3D work is so little that the cost for Octane isn't worth it. I am one of those stubborn Lightwave 3D guys, hunkering and crying in the corner over the Vzirt Newtek purchase. 😂

I am only on LW 2019, don't really see the point in upgrading since it looks to be dead.

But I did download the "Free" Octane from the App Store just to see how the W6800x Duo made out.

If and "if" Apple decides to do 7000 series MPX cards I would buy those instead of consumer cards because I would want the Thunderbolt integration to be there. But the chances are slim to none.

(I have not decided yet which direction I want to go in regards to 3D stuff. They all annoy me.)

 

[innerproduct]

This was answered; but I actually do not use Octane; my 3D work is so little that the cost for Octane isn't worth it. I am one of those stubborn Lightwave 3D guys, hunkering and crying in the corner over the Vzirt Newtek purchase. 😂

I am only on LW 2019, don't really see the point in upgrading since it looks to be dead.

But I did download the "Free" Octane from the App Store just to see how the W6800x Duo made out.

If and "if" Apple decides to do 7000 series MPX cards I would buy those instead of consumer cards because I would want the Thunderbolt integration to be there. But the chances are slim to none.

(I have not decided yet which direction I want to go in regards to 3D stuff. They all annoy me.)

Isn’t modo a natural fit for lightwave people? Other than that I know people who have migrated to Houdini but I guess it comes down to the work you do. I have a mixed bg. Started with c4d in the 90s (well, actually infini-d first) then maya on PC for a few years when that was all the rage. Also some Max. Then nothing for a long time before starting to play with Blender 2.7 on mac in 2017. When cycles sucked on opencl and also dropped support for Nvidia I returned to doing all 3d work on pc again but still hoping to get back on mac some day. Still hoping. And I am mostly a Houdini person and have done a few serious productions using it together with redshift and octane. Nothing serious on mac, just on the road experiments that doesn’t require rendering. I do not like to inflict pain on myself so rendering and serious work requires the best/fastest there is. In 2017 that was 2080ti:s then in 2020 3090:s. Now 4090:s.

 

[deconstruct60]

My bet is that the other half of the extreme Soc (that is supposedly no longer in existence) is a pure GPU/GPGPU connected to the ultra via ultra fusion (note : perhaps tellingly they didn’t name their interconnect as extreme fusion

It couldn't connect to an existing Ultra like that - the Max only has one UltraFusion connector, and that's already being used to create the Ultra itself.

If they are adding a new die not seen on the laptops (e.g., a pure GPU/GPGPU die) to the package then they are not really constrained to the same form factor or CPU/GPU ratio as the M1 Max. One new die versus two new dies.

For example if took a modified M2 Max baseline that had 8 P / 12 E CPU cores (if chasing max core count) plus the new GPU in one tile. And a Bigger GPU core only (strip out all the non Memory I/O , CPU cores , DisplayOutput , AV , etc.) that doubled up on GPU cores. And then sandwitch the GPGPU die in the middle.

            [ CPU+GPU ]
            <ultraFusion>
            [ GPGPU  ]
            <ultraFusion >
            [ CPU + GPU ]

On the CPU+GPU die the GPU is toward the bottom. If the top on is oriented with CPU cores on top and the bottom die is oriented with CPU dies on bottom then minimize the distance between all the GPU cores. (the GPU cores are likely much less tolerant of NUMA effects that the CPU cores are.)

The big problem that Apple has is not the single UltraFusion connector ( if it even exists on the M2 Max. It probably shouldn't if trying to do a decent chiplet disaggregation that scales. ). It is keeping up with the memory channel bandwidth increase demands. Apple could make a die that dumped more GPU cores onto a die , but the catch-22 is that they would need a die with sizably bigger edge space to hook up the additional memory channels to actually 'feed the beast' with enough data to take advantage of all of those additional ALU compute units. Apple would have a unified GPU with nice theoretical peak FP32 numbers but on non contrived workloads it probably wouldn't scale as well as the ratio they use now.


If attached two UltraFusion connectors to the building block die the 'problem' is where does the memory channel and other I/O controllers go? For example if tried.

          [ regular Max ] [regular Max ]
          <ultrafusion>.....<ultrafusion>
          [ really big GPGPU die or ... |
          |... two die with UF between ]

On that border between those 'Max' dies where does the memory fit? It doesn't. Can run highly asymmetrical RAM package traces off the package , but what does that buy you? Other problems.


All Apple needs a "Desktop" die that has Max-like core counts and ratios. It doesn't have to be limited to just one UltraFusion connector if step away from the Memory has to placed only on either left-right side (in the orientations listed above. Trying to over reused the laptop die isn't really an inherent UltraFusion problem. It is a 'have a chunky, not so good chiplet' problem. The solution to that is make a better chiplet ( don't dogmatically reuse the entirety laptop die. They can re-use exactly the same sub-building blocks, but far better disaggregation strategy. ).

 

[ZombiePhysicist]

The SUVs definitely brought in people who don't care about the 911 - and that's the thing the 911 is so usable as a normal car when compared to other supercars - it's not like you needed a second vehicle so the 911 stays as a low mileage weekender. It's really only large families who needed the SUV as the second / "spouse car" amongst the 911 owner base.

But there's a huge percentage of the Apple userbase who don't care about the Mac Pro - I remember back in the days of even Gruber questioning whether Apple has lost it's ability to make a worstation around 2016, Manton Reece (micro.blog) was loudly proclaiming on his podcast with Daniel Jalkut (redsweater) that Apple shouldn't make a Mac Pro, that no one needed more than the iMac etc... the usual self-centric stuff we've heard a million times over "the (Apple Product) is good enough for me, so Apple's shouldn't spend effort on anything better".

I guess I keep coming back to this notion of an AS Mac Pro which isn't just the paradigm of a 2019 (user upgradable ram, gpus etc)... who is it for?

1. It's not going to be cheap enough for people who wouldn't have bought the 2019
2. It's not going to be flexi-performant enough for people who did / would buy the 2019

What's left - a refrigeatoaster nuts-and-gum-together-at-last that combines the price of a 2019 Mac Pro, with the expansion limits of the 2013 / a Mac Studio with a Thunderbolt Chassis.

I don't see what that product gets for Apple that they don't already cover with their existing products, while it loses them some significant user cases.

A halo machine that gets laughed out of town, like the 2013 was, can drag down the brand faster than it raises it.

Another generation of IA64 (and maybe waiting until Intel has transitioned the Xeon to a performance / efficiency core mix to match the way AS processors do it is a part of the process) with a stonking big AS-based Afterburner, maybe even something like a PCI storage board with m.2 storage-like AS modules just seems like a more logical way to progress that Machine. *shrug*.

The Point is if they stop making the 911, all the pro.enthusiasts go somewhere else and won’t push the suv on their families that don’t care. They push the Ferrari suv or something else. Eventually the Porsche suv fades and people stop buying it because there was no real will to buy it, the will came from recommendations from the pro.enthusiasts who were there for the 911.

So Porsche will, keep making a good 911 for at least 2 reasons. To keep that halo effect in place to sell their stupid SUVs, but the more import reason. ENTHUSIASM!

What is that? Well it turns out Porsche engineers like making super amazing sports cars more than their crap SUVs, because they are real car people. they like pushing the limit and beyond the value of some tech trickling down to their stupid SUVs, they really like making sports cars. So much so they see making the stupid suv as a necessary evil to finance what they really like doing more, making super sports cars.

if you work at apple and are not jazzed with the prospect of making their super pro Mac. you suck. And you shouldn’t be working there. If you look at the job of making the most powerful Mac ever, and feel dread and boredom, you suck, and you shouldn’t work in the computer space. That job is the BEST job at apple if you’re making a computer or a chip for it. People there should be clawing and scraping and doing whatever it takes to push technology to the breaking edge. What boundaries they push there today WILL have an effect on the rest of the line tomorrow. It is their halo supercar.

Everyone here is such a beaten housewife they act like the job of making the Mac Pro is some kind of readheaded stepchild afterthought drudgery. If I were at apple, it would, for me, be the pinnacle design for the pinnacle of their line. Maybe everyone here is right and apple treats it as a drudgery, but if so, then the company is already dead to me.

TLDR, they need to spend the money to keep the halo in place. Bad things are in store for them as a brand if they don’t. The halo product matters for Porsche and it matters for apple, whether they get that or not, the consequences will come, and they won’t be good if they get it wrong.

 

[ZombiePhysicist]

Why the Mac Pro lacks upgradable RAM and support for eGPUs

The just-launched 2023 Mac Pro lacks features considered critical for workstations: upgradable RAM and support for eGPUs. Here’s why.

www.cultofmac.com

 

[ZombiePhysicist]

Report: New Mac Studio unlikely due to Mac Pro similarities

While the Mac Studio has been well-received since its launch last spring, Apple may not launch an upgraded model in...

9to5mac.com

 

[innerproduct]

these rumors sound to me like the worst possible direction. A big freaking box for something that’s just a mac studio with slots for nothing. 99% of people buying mac pros gotta do it for gpu perf. It sound so greedy and dumb that it probably is true 😓

 

[AdamBuker]

I just had the thought that there could be two Ultra SOCs on one daughter card module with the heatsink already attached. The Mac Pro would have slots for four daughter cards total for a maximum of eight Ultra SOCs. This would be how CPU, GPU, and RAM are upgraded. This could be user upgradable with DFU restoration or done at an Apple Store.

This would make a machine with the following max specs:

• 1.5 TB RAM
• 192 CPU cores (128p/64e)
• 608 GPU cores
• 16 media encoders
• 256 NE cores
• 8 PCIe slots (only 4 maximum are active if only 1 SOC daughter card is installed)

This would give Apple a massively upgradable machine without sacrificing the benefits of Apple silicon. The major challenge would be the massive amount of engineering needed to pull this solution off, but if they could, I think it would be the right solution for Apple. Most people would be happy with the base mode with one SOC daughter card and 4 PCI slots. The bandwidth can be distributed amongst the slots via control panel if needed similar to the current 7.1 machine.

 

[deconstruct60]

Any chance we got this all wrong and the new MacPro will be Intel based? After all, Tim did say they had an Intel Mac coming and the new Xeons are set for the end of Feb.

What Apple said was

"...
Apple will continue to support and release new versions of macOS for Intel-based Macs for years to come, and has exciting new Intel-based Macs in development. ..."

Apple announces Mac transition to Apple silicon

Apple today announced it will transition the Mac to its custom silicon to deliver industry-leading performance and powerful new technologies.

www.apple.com

That was June of 2020. In August of 2020, Apple introduced the Intel iMac 27" ... finally getting the iMac off the non T2 (non transition, 'T' ) status. Did Apple ship multiple new configurations of Intel iMacs in that 2020 update? Yes.
Apple didn't explicitly say "new Intel-based Mac products in development".


Technically Apple satisfied the multiple Mac units being shipped before the M1 units made it out the door. 10,000 iMacs is an entirely valid usage of the plural word 'Macs'. ( The developer kit shipped before the new iMacs; so technically shipping 'Apple-Silicon' actually had already started. Some folks hand wave at the iMacs release as before the Apple Silicon era. That is a lot of mental gymnastics. It was the Developer Transition Kit ... if you were not running macOS on Arm what was it running? And the Intel transition kit wasn't really running Intel huh? Not. )


. The twist of 'new Intel-based Macs' is that it is talking about multiple Model model numbers as apposed to multiple actual Mac units. It is probably the case that Apple was talking about timely Intel follow ons.

Exclusive: Intel ‘Sapphire Rapids’ Xeon W3400, W2400 And W790 Chipset Launch Date And More — Wccftech

Okay so I have a short and sweet update on Intel's upcoming Sapphire Rapids-SP, Sapphire Rapids-112L and Sapphire Rapids-64L platforms. Officially dubbed the 4th Generation Intel Xeon Scalable Xeon W-3400 and Xeon W-2400 CPUs and powered by the company's W790 platform, Intel is gearing up for...

apple.news

It is very probably not these. The last, May 2020 , Intel MBP shipped with IceLake ( mainstream Gen 10 ) processors in them. If Intel's 2017-2018 roadmaps had kept to schedule the Xeon W-6300 (IceLake) CPUs would have coming in late 2020. If those CPU packages had arrived on time and on thermal budget then they were a much, much better much with the 6000 series MPX updates that Apple also had lined up ( provision PCI-e v4 and GPUs take PCI-e v4. no mismatch. )

The W-3300 didn't 'launch' until almost mid 2021. They were much higher on thermals. The performance was middling if didn't use the new AVX-512 ( which Apple was completely ignoring in Rosetta) and AI/ML augments. Single threaded it wasn't much of a bump at all. The 38 core model bumped into macOS 64 thread limitation.
About the main upside it would have had for Mac Pro would have been slightly more competitive pricing versus AMD's line up. (not huge but Intel knew they were in trouble).


The W-3300 series got skipped by Dell/HP/Lenovo. (largely because even though it was paper launched Intel never put tons of die allocation to the line up. Too busy trying to save the rest of the Xeon SP line up). So Apple dropping that project wouldn't have been a big leap. ( several other vendors dropped it too. )

It would also have been a leap from one dead-end socket to another dead-end socket. Not that Apple would likely go past.

When the W-3300 collapsed on itself as a timely solution, Apple's 'Plan B' was far more likely just release the 6000 MPX modules by themselves on the 2019 platform they already had. That is the much more inexpensive route.
There is a small performance knock by not having matched PCI-e v4 backhaul, but it was far better than nothing.


If the W-3300 had been better than Intel said with higher all around performance with zero Thermal overhead increase and arrived ahead of schedule. If Intel was inclined to attached unusally long term support to these that might have helped also. Apple probably would have used it as a much better 'gap filler' early 2021 product.


The 2018-2019 road maps for W-3400 were into 2022. The Xeon SP Gen 4 was suppose to be late 2021 on those roadmaps and the W series typically came much ( 1-2 Quarters) later. So Apple had a plan in 2018-2019 to launch a new Intel Mac Pro in 2H 2022 ... right when they were projecting to finish the Intel transition? Probably not. Remember Intel already had screwed up lots of roadmap deadlines by 2018-2019... if Apple added in the "probably not telling the truth" fudge factor for those 'mid-2022' arrival times for W-3400 they'd end up with 2023. Which is the year after they were suppose to be done. While Apple said they were going to support macOS on Intel for years , they were not in anyway suggesting that they were trying to maximize those number of years. After 2022, it was probably going to be on the clock toward retirement.


I suspect some folks are thinking Apple is in some kind of desperation mode if the quad-die/"Extreme" package fell through and will quickly whip together a W-3400 update to stall for more time. That is unlikey. It took 2017-2019 to get to the current Mac Pro. Apple whipping something together hyper fast? Probably not. If it wasn't on the road map 12-18 months ago it is probably not happening.


The other huge problem is that the Intel Mini just got dropped. The iMac 27" has been dropped for almost a year. There is not even any other Intel Mac product 'limping along' in 'low key' mode in the background. All the other Intel "No Mac Pro" Macs are actively on the Vintage/Obsolete countdown clock. That huge inertia is going to take macOS on Intel down. No way the Mac Pro is going to 'swim upstream' in that roaring current heading the other direction at this point.

Apple is already committed to several years of macOS on Intel support for the Mac Pro 2019. At best, if they needed a cheaper "plan b" to extend gap time they would do the same thing they did in 2021 ... toss out one or two GPU cards. Even that is a stretch given the current environment of 'dead end' 3rd party GPU support. However, the overall system relation to operating system lifecycle suppor term probably wouldn't change. (new GPU isn't going to extend the overall macOS on Intel countdown clock much at all.)

 

[mode11]

if you work at apple and are not jazzed with the prospect of making their super pro Mac. you suck. And you shouldn’t be working there. If you look at the job of making the most powerful Mac ever, and feel dread and boredom, you suck, and you shouldn’t work in the computer space.

Lol, hear hear.

Everyone here is such a beaten housewife they act like the job of making the Mac Pro is some kind of readheaded stepchild afterthought drudgery. If I were at apple, it would, for me, be the pinnacle design for the pinnacle of their line. Maybe everyone here is right and apple treats it as a drudgery, but if so, then the company is already dead to me.

It's expecting Apple to make the Mac Pro we want that feels more like being a loyal creature that keeps coming back, despite being kicked in the face every time (satisfied 2019 customers excepted (for now)).

 

[deconstruct60]

these rumors sound to me like the worst possible direction. A big freaking box for something that’s just a mac studio with slots for nothing. 99% of people buying mac pros gotta do it for gpu perf. It sound so greedy and dumb that it probably is true 😓

Apple's own statements about what they sell as Mac configurations outline that is highly likely not true.
The bigger seller GPU is the W5700. Not the Vega II or 6800/6900 MPX units. Lower than that.

If there is a normal distribution of GPU cards with the W5700 at the center then there is about as much 'below' as there is above.

Apple also kept the 580X in the line up for about two years. Some folks probably did rip it out and install their own off-the-self (with zero MPX modules). But substantially many probably kept it in there. If need 3-5 audio DAW , interface (some double wide), and storage cards , then a Quad wide GPU isn't helping much at all. Apple probably relatively sells lots of Mac Pros into the audio space. Higher than the small "1% or less" range.
Similarly folks completely tossing all MPX modules , but buying XDR monitors. Probably didn't happen much at all. (So the Apple MPX stats have some significant standing. )

 

[deconstruct60]

Report: New Mac Studio unlikely due to Mac Pro similarities

While the Mac Studio has been well-received since its launch last spring, Apple may not launch an upgraded model in...

9to5mac.com

Gurman has been ‘doom and gloom’ about the Mac Studio for a while. Before it was the iMac Pro was going to come back and shove it aside .

Gurman: Apple Still Working on 'Pro' iMac With Larger Screen

Apple is still working on an iMac with a larger display and a more powerful chip aimed at the professional market, according to Bloomberg's Mark...

www.macrumors.com

His

“…I wouldn’t anticipate the introduction of a Mac Studio in the near future. … “ is a ‘sky is blue’ update. Given Apple hasn’t updated any M1 system in a rapid fashion and the Mac Studio is still less than a year old .

And the notion of Apple can’t get Mac product x out because consumed with getting Mac product y out isn’t a newsflash either.

 

[mattspace]

Apple's own statements about what they sell as Mac configurations outline that is highly likely not true.
The bigger seller GPU is the W5700. Not the Vega II or 6800/6900 MPX units. Lower than that.

1. Apple lies.
2. Apple twists the truth, to mean the opposite of what was intended "people asked us for better dual GPU support", becomes "people wanted dual non-upgradable GPUs".
3. Apple lies.

 

[mode11]

Given Apple hasn’t updated any M1 system in a rapid fashion and the Mac Studio is still less than a year old .

True, though Gurman is saying no update until M3, which is a bit different.

Obviously no Mac can get updated to the next M-chip before the chip exists, but once it does, updating the logic boards can’t be that hard - especially with so much commonality between the different machines.

Once one machine gets e.g. an M2 update, people looking to buy a model that hasn’t will be inclined to put off their purchase, particularly if it’s a higher-end / more expensive machine. With the MBP, Air, mini and the new mini Pro now on M2, would you want to buy a Studio now? You’d be a bit gutted if the M2 version came out a couple of months later - it would seem rather ‘obvious’ in retrospect.

Also, M1 was the first AS chip, the Studio was a brand new model, and the last couple of years have had widely-publicised supply chain issues (GPUs were notoriously affected). Weren’t the M2 laptops intended to have come out in October? It may be too early to identify the ‘typical’ cadence of Apple releases in the AS era.

 

[innerproduct]

Since m1 ultra gpu scales so bad and m2 seems to fix it, I guess only an uneducated fool would buy the ultra right now. Or if there really is no other option for a specific problem/task. If cpu bound it doesn’t suck as bad oc

 

[AndreeOnline]

Apple lies.

If this is how you feel... and the same goes for all the likes under your statement: how come you guys spend time here?

It's an honest, meant-to-not-be-inflammatory question.

Across many of the pages in this thread, but others too, I get the feeling I'm reading the opinions and thoughts of people who don't appreciate Apple very much. So why spend time here?

 

[mode11]

I love macOS. I think Apple makes excellent laptops, smartphones and tablets (though would grumble about iPadOS). iMacs and minis can be great solutions. Their products hold their resale value. I trust their attitude to privacy, if only because their business model is product rather than advertising based. In general, I respect the thought that Apple put into design. If I were primarily a MBP user, I'd likely be dancing a jig right now. Ditto if macOS could be installed (reliably) on a PC tower. It's just a shame you have to choose between macOS and expandable hardware, as Apple doesn't seem inclined to sell the latter.

 

[innerproduct]

If this is how you feel... and the same goes for all the likes under your statement: how come you guys spend time here?

It's an honest, meant-to-not-be-inflammatory question.

Across many of the pages in this thread, but others too, I get the feeling I'm reading the opinions and thoughts of people who don't appreciate Apple very much. So why spend time here?

I guess many of us have been apple users since very long (80s, 90s) and since 2012 or so things have continued down hill for us “classic” users: creative pros that are some kind of enthusiasts that have a deep love for what we believe was Apple and NeXT. When the 2017 iMac Pro was released, apple started to price things ludicrous. The entry level version never was good enough so realistically the machine people needed was the 10 core/64gig/vega64 was about 8000$ (right? Don’t remember exactly)
At that time opencl had been fading for some time and sw was not well optimized. It was at the time quite a bad buy but the best Apple would offer us. I passed. I waited the for the new modular mp that they promised. Once announced it was both exactly what we wanted but at an insane price. Tower mac pros don’t need super special miniaturization like the laptops so it should not be that expensive to make. I mean this is basic stuff that has been solved since time immemorial. The entry level mp2019 had worse performance than an imac. So again, the machine that was what people was looking for was something like the 12 or 16 core with a 5700 at least. And let’s remember that the 5700 was quite bad. Just about the same perf as the old vega64 but worse perf in double (fp64)
I specced that out and planned to use my vega64s that I user in egpus. But again, sw was not ready and I was waiting for octane or redshift to be ready using Houdini. That didn’t happen until after Apple announced the ASi move. Making the mac pro 2019 and obviously bad choice with questionable longevity. I passed and had to continue using PCs for my 3d needs.
And the years passes by and the ASi systems are not yet powerful for competitive 3d work except for people that have no issue spending 5x the money that is actually needed. If I work in Houdini all day. There is almost no benefit to use Mac tbh so it is not a problem to use PC. But it would nice since all other software I use runs good enough on my MBP16.
So, i guess we feel used and mistreated. Apple should be happy that there are people like us who wants to spend extra for the benefit of niceness and the love for the platform. A machine that is at the same (real world) power in 3d work as the (actual, real, needed) competition should be priced just like the MBP: at a premium but offering a premium. So the new pro desktops this years should be: mac studio with m2ultras and same prices as now. Maybe a price drop for some upgrades. (Since the m1ultra scales gpu horribly, apple should be quick to fix this with the upgrade if they had som pride in their products)
And then finally, a mac pro with cpu power at up to 2x a studio while allowing Gpu to scale at least by 4. Preferably using replaceable parts. Priced at 2-3 times a studio ultra or so, maxing out at maybe 20000$ for an insane spec.
But now we hear about no update for the studio and a macpro that will be a studio with slots. Mac pro in name only.
That is why I think we are grumpy. Our old love has been cheating for a decade and we can’t let her go. She keeps taunting us and mistreating us and still we hope for the love to return.