Upgradable like a 2012, proprietary like a 2013.
That could primarily hinge upon the default display GPU. Thunderbolt enabling like 2013 ( so more than just PCI-e coming of the internal edge(s). ), but more discrete like the 2012 ( user plug/unplug ).
What I've heard is that it will be controversial just like the 2013.
For the folks who compose their workstation starting inside with the GPU card and work their way out .... that is all the controversy Apple would need to generate. The rest could be almost classic standard and there would be dozens of pages of folks moaning and groaning on these forums. ( just like the last 6 years. )
About the same level of controversy if they dumped SATA drives from the standard box.
So if I was a betting man, and I see a few people talking about a stackable Mac Pro...
It is possible, but doesn't seem probable. I wouldn't bet much.
You can even see the new logic with ARM coming up. They ship an Intel box now, and when you want to go to ARM, you just buy a new brain and leave the rest of the components as is.
I don't see much logic there at all. More likely if Apple goes 100% for the Mac line up, the Mac Pro is dead. Apple won't do an relatively super low volume workstation ARM CPU and the whole line up fall back more so toward mobile and stop somewhere around the mainstream iMac.
Separating the CPU and GPU in a workstation isn't particularly logical at all. ( a 3rd GPU perhaps , 4th even more so given common demographic usage. Mac Pro didn't particularly carter to that in 2012 model anyway. A second GPU is wishy washy. )
If the "attach box" is primarily a SATA drive cage what is the substantive savings of not just buying new sleds?
They could have a i9 brain and a Xeon brain and eventually an ARM brain.
If looking to roll out a range of Mac Minis perhaps. But in the workstation space that is mostly Cupertino kool-aid more so than logic (at least on historic Apple trends). That's looks like a company highly itching to get into the xMac space and crank up the fratricide on the iMacs. I wouldn't bet on that.
In their heads it probably all fits together. They're probably going "ok the problem people had with Thunderbolt was all the boxes, how can we do the same Thunderbolt thing but without all the separate boxes."
The 'additional' cables weren't the sole primary crux. The relative cost was too. If these "snap on" augments are at least as high, if not higher, than the Thunderbolt options then there is something in their heads ... drugs.
Folks are probably also going to want to consolidate not just the data cable but the power cable. Thunderbolt has a 100W limit which is good for a couple of drives ( several if careful about tiptoeing around initial spin up spikes and use very modern low power drives. ). GPUs would be a different story in size and power. ( if flip between radically different archs in 'brain box' why would the drivers for these sunk cost GPUs implicitly follow along with the switch? )
For extremely PCI-e centric solutions, a "Thunderbolt without the box" is just a slot ( classic x16 PCI-e slot , M.2 slot , U.2 , etc. ). So yeah I could see Apple doing a empty x16 slot and perhaps some highly focused SSD slots. Unless manically attached to building a "small as possible footprint", literal desktop that shouldn't be a major problem. ( and more like Mac 2012 ).
On top of that is the glaring problem is that Apple R&D can't even get monolithic Mac "boxes" on out a regular and reasonable amount of time. ( Mini comatose for years , iMac fell into rut, MacBook largely comatose in terms of design for 4 years, Mac Pro looooong rut ). Where is the logic in Apple not being able to do a small number of boxes on a regular base going to add a higher workload of different 'augment" boxes? Or they are suppose to be farming all of that work out ? If they are keeping to themselves that is highly likely a worse problem.
Apple sells
iMacs with a built in VESA Mount so it would a bit of a jump for them to sell a Mac that you attached to a 3rd party thing. The Mac Pro market though is has systemic problems with critical mass though. Some kind of Mac Pro only connector that 3rd parties will build boxes to snap onto.... possible but not very probable. ( even after some gallons of Cupertino kool-aid. .... going to have major problems if just clearly think ahead past the "won't this be slick/kool" phase. )
After all of the work Apple and Intel did to get Thunderbolt mainstreamed into the USB standard .... now when ecosystem is going to perhaps grow at levels expected initially ... Apple is going to introduce a new propriety 'fork' ? Put in lots of work and then walk away. I suppose there is some Cupertino kool-aid logic in that .
It would be nice if these were all modular components inside a tower, but I don't think that would rise to a controversial product.
Missing components would be (e.g., 3.5" SATA drives sleds superseded by M.2 slots ) or non-Rube-Goldberg TB enabling default GPU 'card" would be labeled "doom" by more than a few and still could be inside the tower.
There is even the "Xeon SP (and dual) or die" crowd that has occupied a sizeable chunk of this thread. Apple "stuck" at 18 cores (and single CPU package) would be controversial to those folks. 4 DIMM slot s rather than 8 ... some chatter will explode on that too. If Apple doesn't slavishly clone the other bigger players in workstation market their will be "controversy" by a "pitchforks and torches" crowd. Priced higher than $2999 price point..... ditto.
[doublepost=1557367052][/doublepost]
More thoughts on this, or are they fever dreams...?!?
So, why has Apple taken so long to replace the Mac Pro...?
Because it isn't strategic and doesn't make any significant, immediate impact to the overall financial picture of the Mac Product line up or to Apple overall. So it isn't allocated high priority resources and gets done in Apple's copious spare time.
That's the simple explaination that fits. Alien area 51 for some fancy Buck Rogers explanation is not "most likely" at all.
Some may say the Mac Pro was slated fade away into the distance, with the iMac Pro as its successor...
literal desktop Mac Pro 2013 --- successor ---> iMac Pro 2017 . Mostly yes.
The confusion is that Mac Pro is overloaded as a term. The path that the 2009-2012 model was on and the 2013 model are different. The iMac Pro is far more related to the latter than the former.
Many ask as to WHY it is taking so long, a mini-tower workstation with a few PCIe slots should be easy...
If don't assign anyone to work on it ... it can take forever.
My idea of the backplane with daughtercards modular Mac Pro cube still stands, but backplanes & daughtercards have been done by Apple before...
Actually the NeXT cube was done by NeXT; not Apple. Also a 12"x12"x12' cube probably wouldn't work well with Apple's typical literal desktop constraint objectives.
But then when one thinks of the smaller chassis size and the TDPs of current CPUs/GPUs, it seems like it would be another thermal corner for Apple to paint itself into...
There is no corner if add volume and fan(s). If primarily independently cooled 2-3 daughtercards wouldn't run into a "thermal corner" at all.
Let me introduce you to the new ARM-powered modular Mac Pro...!!!
ARM is simply just misdirection here. The root cause problem of "thermal corner" was highly coupled cooling. Not architecture.
Still the same SG Mac mini style PSU (same horizontal footprint, but the full height of the chassis), with all I/O to the lower portion of the rear panel (eight TB3 / USB-C ports, dual 10Gb Ethernet. & 3.5mm headphone jack), as well as the power input...
Each daughtercard has the following:
Four A13X Bionic APUs
Four independent computers isn't going to buy much but much higher NUMA impacts and need for something like (MPI ... which isn't a Apple core competency. )
