Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
If Apple shows a new cinema dispay at WWDC I can almost guarantee they will show the Mac Pro along side it.

Not necessarily. An iMac Pro upgrade with the new screen would totally be in alignment with Apple's moves over the last 10+ years. Match the sales of iMac to the 'Monitor' to move more volume of the panels to get to economies of scale.

Even without a iMac Pro or Mac Pro preview, Apple could always just demo the new Display docking with the MBP 15" Vega option. Apple would drop the LG Ultrafine monitor from their Apple store line up and replace it with their own product. The current Mac Pro didn't drive major sales of the current LG screens. Neither likely would a new Mac Pro be principle driver of Apple's next Display Docking station.

That would be a huge selling point. Both sold side by side.

if one is for sale and the other is not .... the latter isn't very likely to significantly drive sales. If the docking station is ready to ship and the Mac Pro drifting much closer to ( or into ) 2020 then is zero rational reason to hold back the docking station.
[doublepost=1557238540][/doublepost]

While the Bloomberg article mentions the Mac Pro (and Display )..... it doesn't mention this "stackable" concept at all. Pull in plausible information and then lash it to something that is more flakey.

As much as folks are having a cow over Apple not shipping a new Mac Pro, the replacement docking station is even worse. Apple shipping the TB Display in 2011. That's two years prior to the Mac Pro 2013. ( creeping up on 8 years). They did a hand waved a discontinue/replace in 2016 with the LG Ultrafine. And it is 3 years after that intro. Same 2017 "dog ate my homework" meeting said they were working on a display. A display takes 2+ years to get out onto the market. Really?

Gutting an iMac could have gotten them to a replacement display back in 2014-16. Couldn't manage that. ( The Ultrafine looks like the base feature set was Apple so perhaps they they a "half" finished displays over the wall and asked LG to finish them. But as far as competency goes that is telling also. ) . As a primarily just a monitor ( with some integrated docking station abilities built in), Apple can't be rationally gyrating over what the basic design is. Industrial Design is conflicted over pedestal arm versus adjusting arm? The features and form is pretty much dictated by the singular, focused tasked assigned to the peripheral. (and it is a subset of what the iMac does so not new rocket science. )

If the display and the Mac Pro both slide for a protracted time that is indicative that they have coupled the connection technology to some "tail wags dog" technology. If the monitor is ready and the Mac Pro isn't, then Apple should ship what they have finished. That would go a long way to demonstrate that they were doing substantive work. ( it isn't quite the same Cirque de Soleil sizzle show , but it would be something to show and ship. ). If neither one is in the engineering verification stage by WWDC 2019, that is a bad sign.
[doublepost=1557239626][/doublepost]
There are not going to be stacking bloody boxes.

It's compatible with what I've been hearing.

That's a big deviation from " between Mac Pro 2013 and 2010-2012" models. Neither one of those was about walling off a subset into a different SKU.

The only subset that would make remote sense would be Apple augmenting what they thought was complete for most, but some did not. For example, if Apple was 'done' with SATA drives to put those in a "snap on" case. ( It isn't snap on but there a subsection of the Dell 7920 on the right hand side that is just for SATA drives , ODD , and inserts. ( the power supply is over there but if just dropped the subassembly (and attached a SSD to the motherboard) you'd still have a fully working workstation. Just smaller.

Stacking boxes for relatively low bandwidth peripherals may fit with that they are doing. ( a bit of a variation on what the Sonnet rack rigs that Mini/MacPro as tossed into now ). But major, essential components? That is very dubious.
 

Pretty much removed from tall of the "brain box" hype in the rumor mills. The whole computer is contained in one box. The only thing stacked here is a bigger speakers and auido , and a security plate to attach the small computer to the desk.

The latter ... yeah the Mac Pro 2013 punted on. ( keeping it securely closed physically Apple punted on. ). And folks did need to buy something to attach to secure the device to a desk/table. Apple doing something directly this time would not be pragmatic an introduction of the "brain box" meme.

Attaching audio. ? Yes, Apple has been dropping more esoteric audio connections. But still far removed from the "brain box" meme.


I don't think the new Mac Pro would be that small at all. ( Since operate standalone it would have CPU/RAM/GPU/Storage+T-series/Power Supply/essential ports/etc. ) If Apple is shooting for a mostly complete workstation it probably wouldn't be smaller than the Mac Pro 2013. If there is an empty PCI-e slot even less 'small'. The "stack" modules probably wouldn't nearly as big; as they are primarily incremental augments.

The stuff I can see Apple stacking would revolve around the debate of what should be dropped. Apple walking away from SATA is one. ( and perhaps the only one). [ where Apple sees this trend https://www.anandtech.com/show/14298/shipments-of-pc-hdds-predicted-to-halve-in-2019 and others are pointing to 'corner cases'. ]
 
Last edited:
The cylindrical Mac Pro was/is a great idea, let down by Apple's stupid marketing and pricing policy. An big/noisy ATX based tower that contains GPUs/mechanical disks/everything creates more problems than it solves.
No the trashcan wasn't/isn't a great idea. The thermals couldn't handle the hardware they shipped in it let alone more power hungry future hardware (past hardware now.) The non-upgradable dual gpus were a failure. The lack of internal storage options were a failure that created an ugly mess of cables and boxes in real life use. The trashcan design came out of Phil's ass and it stinks.

Apple made a mistake, it wasn't the first time they have made this mistake but they haven't yet come to terms with the market failure of the PowerMac G4 Cube. At least Apple continued to ship a proper PowerMac G4 tower along with their fail cube back in 2000, in 2013 they replaced the Mac Pro tower with the fail can with nothing else to fall back on. If Apple had continued to ship the tower Mac Pro alongside the trashcan then the trashcan sales figures would have been as dismal as the G4 cube's were.

All that is ancient history now though and it's time to move on. Hopefully Apple has learned its lesson this time.
 
No matter what Apple are going to do, they will always be on the second place.

Apple is used to enter (and conquer) a pre-existing market. The iPod came out when MP3 players existed already. Same for the Apple Watch, and even for the iPhone.

There is still hope that they will come out with something capable of establishing a new standard. (not a huge one, I know...)
 
Sorry, but are you serious about this?
Since when is expandability a problem? Are you a fan of sealed devices ?

As serious as it gets. Expandability can also be done through modularity. And if you want fault tolerance (standard) ATX is limiting too. I am fan of silence.


No the trashcan wasn't/isn't a great idea. The thermals couldn't handle the hardware they shipped in it let alone more power hungry future hardware (past hardware now.) The non-upgradable dual gpus were a failure. The lack of internal storage options were a failure that created an ugly mess of cables and boxes in real life use. The trashcan design came out of Phil's ass and it stinks.

Apple made a mistake, it wasn't the first time they have made this mistake but they haven't yet come to terms with the market failure of the PowerMac G4 Cube. At least Apple continued to ship a proper PowerMac G4 tower along with their fail cube back in 2000, in 2013 they replaced the Mac Pro tower with the fail can with nothing else to fall back on. If Apple had continued to ship the tower Mac Pro alongside the trashcan then the trashcan sales figures would have been as dismal as the G4 cube's were.

All that is ancient history now though and it's time to move on. Hopefully Apple has learned its lesson this time.

The thermals were nearly optimal for a single-fan, small-factor workstation. The (base) thing is even luggable. And since you brought that too, yes I was honoured to have a G4 Cube back in the day, one of the most interesting computers I have had (and I have used many). And of course not in the original price, but 2nd hand, when from a "failure" it became a much sought-after commodity by people from all platforms. People who, fed up with their loud or ugly machines, spent weeks on eBay trying to get their hands on one (for their desk or for their living room). Decades after, the supermarket-PC has died and the industry has forced laptops on everyone with just ordinary needs but the trend for better, quiet, yet powerful computers has been real in the computer world if you dare to check. Still, for the Apple world, many need something better than a MacMini; ideally something that could scale better money and performance -wise, up to a top workstation.

I talked in the past about mistakes on Apple's design/marketing/pricing plans. The thing should have been for a different audience, NOT less "pro" than the videographers who need TBs of storage. If I am a programmer or researcher, I want to remain sane, so there is no way I am having slow mechanical disks at the same room as my ears, which defies the reason of having this storage inside the main desktop in the first place. And don't get me started about PSUs. Ditto with GPUs. And if I wanted a large-scale rendering/number-crunching server then I would be better served with other solutions (which would still have external storage or be rackable and remotely accessible). Thankfully, it seems that Thunderbolt 3 and 10GBe can handle some of the tasks of the audience that Apple might/should be targeting.

Anyway, many problems with the cylindrical MP were related to the multiple penalty Apple/Intel/USA/AMD-NVidia taxes people had to pay because Apple was shortsighted, treated it as just a niche machine and only did things half-hearted (not just for the MacPro but for their whole workstation/pro/server ecosystem). E.g.

1) the decision to manufacture/assemble it in USA. Speaking for myself, I love quality cases and I can't stop from laughing when I see the heat=envy from PC-fanboys, when these have spent considerable time, effort and even more money for terrible custom case/cooling designs that do not outperform it. Nevertheless, Apple probably COULD choose to do everything in China without compensating on the quality.
2) the decision to go (just) for overpriced Xeons and ECC RAM, without giving users different (B2O or Apple-upgrading) options
3) the decision to limit GPU options to AMD when people wanted NVidia. Apple could have offered GPU boards.
4) the lack of support and upgrades while keeping the price the same, which was a punch in the face. At the very least they could have upgraded Thunderbolt, Ethernet, and USB.
 
Still pushing a two case design. "Brain" case might be similar in some respects to the 6,1 tcMP but able to cool as much TDP as you can stuff into it. "Brain" case would need some GPU resources, but they could be on die (Intel Iris xxxx). IMO, it would make sense to put a single PCIe slot in the "brain" case so some folks could get enough GPU grunt for their application without the "slotbox" case.
The "partner" case should have a single robust connection to the brain case along with PSU and serious cooling. Needs at least 24 PCIe4 lanes provide a powerful option for hard cores with no "rat's nest".

This concept would offer 3 approaches for various use cases:
1) brain case only, single PCIe slot, TB3 I/O - starts around $3,500 with lower end GPU
2) brain case only, single PCIe slot with extensive use of TB3 for eGPU, RAIDs, etc - $3,500 and up
3) both cases, partner (slotbox) case to house bomber GPUs, NVME storage, SSDs, etc - $3,500 and way up

Rather than stackable, I'm imagining something more like side by side towers designed to connect as close to the mobo as possible. Shape should allow for cases to NOT interfere with each others cooling.
 
Apple is used to enter (and conquer) a pre-existing market. The iPod came out when MP3 players existed already. Same for the Apple Watch, and even for the iPhone.

There is still hope that they will come out with something capable of establishing a new standard. (not a huge one, I know...)


Fair points .
But those are fairly simple devices compared to computers and the tasks they had or have to master .

Re. the MacPro, Apple will need to re-establish its position in the workstation market - the competition is ahead and well established, many users have switched, most bulk buyers are gone, 3rd party software and hardware support is at a low point .

Macs have never set a new standard - apart from the pre-Pro iMacs in their segment - and Apple has never been in a worse position to propose something new .
Not after the trashcan and the touchbar MBPs , and an OSX development policy that is erratic at best .

The ball is in their court .
 
  • Like
Reactions: Biped and AidenShaw
TB3 has limitations, full stop. That said, with enough headers/lanes it should support some pretty impressive eGPU/etc devices - at least in terms of available data path. Proper implementation is another thread.

The "daughter" case could have the same exterior as the main (brain) case to control costs. IAC, it's really just a dongle Apple can make money on that might encourage them to support mainstream cards from the largest vendors - read nVidia. The daughter slotbox with dedicated multilane PCIe extender/jack would have a modest performance delta, no rats nest - and - AppleCare for the whole shebang... I think that's a value add Apple can sell and make good margin on.

For the record, if it were my call, I'd have Apple make retrofit kits for the cheese graters. Gut 'em. Replace pretty much everything but the case framework to make it new school - just don't trash that over engineered Aluminum exterior case. Yes, that's perhaps the stupidest suggestion yet as a business case - but it's an environmental winner! All 4 of mine still look almost new 10 years on...
 
  • Like
Reactions: kipwheeler
Last edited:
  • Like
Reactions: Logix1 and Aldaris
Re. the MacPro, Apple will need to re-establish its position in the workstation market - the competition is ahead and well established, many users have switched, most bulk buyers are gone, 3rd party software and hardware support is at a low point .

I wonder if the software support issue is just the market contracting in general ( inevitable maturation ), and more hard felt because the macos pool was smaller to being with. It seems like the wealth of options that were available to me in windows land to do certain tasks are no longer as plentiful.

Ignoring the *nix universe where a new project pops up every 5 minutes on github to solve the same problems, and portable app stores which are flooded with an abundance of cheaply made apps that do nothing other than datamine or spam adverts. I'm curious if a contraction of software options/development is a managerial justification for a trend to 'sealed appliance' for end user compute ( ie tashcan being a FCP dongle ) ...
 
If there's no announcement on the Mac Pro at WWDC 2019, no pcie slots because modular crap and not being released until 2020 all from Tim Genius.......I'm moving on.....:mad:

There is nothing particularly concrete about it not being released until 2020. That seems to be all Appleinsider input. That isn't particularly 'sourced'. If the target is Q3-Q4 and not particularly doing a 'sneak peak' making a 2019 deadline is fully possible.

The Mac Pro has about zero inherent need to be "dog and pony" showed at WWDC. And doing a dog and pony show on something they can't actually manufacture at scale (e.g., AirPower ) would actually be worse at this point. Missing WWDC is not particularly more 'fatal' than them missing 'April'.

If there actually is no announcement than the "extreme modular" lego box notion won't be instantiated either.
[doublepost=1557264115][/doublepost]
The whole idea of a Mac Pro with stackable expansions seems really unlikely and silly to me. The Mac mini already covers this.

It doesn't cover this well. You can stack mulitple Minis but they pretty much would work the same if you didn't stack them. if there is no functional difference between stacked or not then really talking about how to use the footprint area.

That would be significantly different if say the top handles come off and can plug/attach a storage module onto the box (without adding any cables or wiring. ). There wouldn't necessarily be 'extra' cords ( e.g. handles come off revealing a SFF-8xxxx connector for data and a power coupling socket. ). Essentially the "top" comes off and get a new, taller "top".

The Mini's 7"x7" is fine for covering the footprint of a 3.5"/2.5" drive, but it won't particularly work all that well covering something like a Compute card, larger thermal loads than the Mini , and/or larger collections of drives (and vibrations).
 
December 2019 is close enough to call it 2020.:cool: No announcement at WWDC then I need to move remaining computers to PC's ....can't wait any longer:mad: It's been two years:mad: I've been on the MP platform since Quad950 (not a fanboy) and since 2012 the move towards PC's has been the logical position because of the piss poor decisions of Tim Genius.:eek:

Just a Phone company:(
 
  • Like
Reactions: Aldaris
Just a Phone company
a bad phone company since iOS 7,
since MacTrashCan Pro 6.1 since Pixar become a part of Walt Disney and NEXT Workstations and SGIs die ...

I had a dream

a little box with some USB2/3c, cinch/headphones, hdmi,miniDP/Thunderbold3/4
(like Mac mini) connected with big big BIG wire to a HUGE BOX with 4-6 300 Watt GPUs running CUDA to render new ILM stuff with Vray on MacOS...
connected to Retina Thunderbold 4/6K display ...


oh boy , well we, I have to travel to the past iray GPU rendering in Maya 2010 on Nvidia Quadro FX1800 with Stereo Shutter Glasses... with OS X.... Steve was alive and Pixar used renderman on xserve servers....

n o s t a l g i a .....

AND make microphone, and FaceTime iSight Camera PHYSICALLY possible to be DISCONNECTED
and sell some Bose speakers with this Monster as in Anniversary Mac...

hotplugging Drives and GPUs would be a GameChanger
virtualized MacOS instances with DirectAccess pci pass through GPU would be a Game...

forget it give us Snow White back and PowerPCs !!!

I do not need PowerVR graphics in a new MacPro without braking through render solution ....
I do not need secure boot with lightning cable connected

give us TargetDisk mode Target Display mode and a remote control back
even in the Apple Watch or sth. to suspend this workstation dreams ...

AirPlay 3 for Display with Pen support could be better then Snow White Design ....
 
Last edited:
a little box with some USB2/3c, cinch/headphones, hdmi,miniDP/Thunderbold3/4
(like Mac mini) connected with big big BIG wire to a HUGE BOX with 4-6 300 Watt GPUs running CUDA to render new ILM stuff with Vray on MacOS...
connected to Retina Thunderbold 4/6K display ...
Yes a BIG wire. Any modular solution that doesn't have a least 32 PCIe lanes to the GPU box will be DOA.

And if it doesn't have full Nvidia support - the whole system will be DOA.
 
  • Like
Reactions: aaronhead14
What Apple should do and what Apple will do are two different questions. I think most of us agree that Apple should provide more expandability than they're actually likely to...

What I think Apple should do:

Processors: Variant on just-released Xeon-SP processors that eliminates the very expensive multiprocessing hardware (or Xeon-W chips up to 28 cores) - that's really the same thing - the goal is chips with options up to well over 20 cores, while not paying $10,000 for the CPU. Intel has released several "SP" big-socket processors without multiprocessing, including a ~$3000 28-core, and they're reasonably priced for what they are. Threadripper (or Epyc) is another very good alternative.

RAM: 8-12 user accessible ECC RAM slots. Xeon-W or Threadripper would get 8 slots (4 channels, 2 slots per channel), Xeon-SP would hopefully get 12 (6 channels, 2 slots). Epyc would get 8 slots, but it's 8 channels - all the RAM would have to be replaced at once. A 4 or 6 slot configuration isn't necessarily enough, and 16 (Epyc at 2 slots/channel) is huge and probably unnecessary (although valuable to allow the user to upgrade RAM without removing all existing RAM).

Drives: Boot drive controlled by T2 plus 2-4 NVMe SSD slots.They aren't going to give up on the T2 and encrypted startup disks, but I'd love to see some internal SSD expansion in addition.

The proper place for spinning rust is outside the case. If you need one or two drives, the price differential between internal drives and external housings featuring the same drive is trivial (under $50, probably less than Apple would charge to make the Mac Pro case bigger to accommodate the drives). If you need multiple drives, there are so many choices ((4, 5, 6, 8,12 drives), (RAID 0,1,5,6,50,60), (inexpensive USB DAS, Thunderbolt DAS, NAS (10 GB or not))) that Apple could never design a solution that satisfies everyone. Better to provide all the requisite ports and let the rust spin externally - it can also go in a closet to reduce the noise.

Ports: 6 TB3/USB-C (3 TB3 buses), 4 USB-A (or even just 2 - it's mainly for flash drives and other "quick plug" items - put it in front, possibly with ), dual 10GB Ethernet (or 10GB plus 1GB), headphone jack... I left off a card reader, because there are so many useful types. SD would be cheap to include (the expense would probably literally be the hole in the case more than the circuit), but it's going to be replaced soon, especially in the video world. For photography, the standards are SD and XQD/CFExpress - video uses those plus CFast, SxS and a couple of different ways of mounting 1.8" and 2.5" SSDs. The really elegant solution would be a front bay with options ranging from a blank (or SD-only) module to REDMag, plus several audio-relevant options for people who may not have a camera, but have a room full of keyboards. The catch is whether they'll sell enough to make the modules viable (and at what price).

Graphics (this is going to be controversial): I don't think Apple should use standard PCI-e graphics. PC gaming cards run too close to the edge to be really stable, they're an old legacy form factor that could easily be replaced by something with better cooling (liquid cooling integrated with the rest of the system?) and NVidia can't or won't produce a stable Mac driver (Apple can't, won't or doesn't want to). They'll end up with a zillion GeForces of every shape and size (which they don't want to support) if people can shoehorn them in there. Either they should use a slightly modified PCIe (maybe as little as blank backplanes with an internal cable to redirect the signal out the TB3 ports) that AMD commits to supporting, or they should use a form factor that permits them to integrate the cooling with the rest of the system, but also permits graphics upgrades. One or two GPU (Vega/Navi 7nm) capability.

Slots: 1 PCIe x16 with 75 watts of power - for audio and video interfaces, not back door GeForces. It could also provide an interface for powerful or multiple eGPUs/machine learning coprocessors/etc.

And I think they should license the MacOS in a limited way: This Mac Pro will serve many needs, but some people will want dual processors, or four GPUs, or insist on 3.5" internal bays. Apple doesn't want to do that - fine... License the MacOS to HP, Puget Systems and one or two others. Make it $300/machine (like Windows for Workstations), so it's not useful for cheap commodity desktops Apple doesn't want out there (who cares about a $300 OS on a $25,000 Z8). Restrict the Hardware Compatibility List to keep stability high. I'd like to see license-built laptops, too - more powerful than the MBP 15" (if heavier).

My best guess as to what Apple will do:

Processors: I think they'll get this one "right" (similar to what I'd like to see) - I don't think they'll give AMD as long a look as they deserve, and we'll see some sort of single Intel Xeon with a range of core counts, topping out well over 20 cores. Their processor purchases are defined by laptops, and AMD doesn't offer either the ultra-low power chips Apple needs for the MacBook or the powerful 45-watt >4 core chips the 15" MBP takes.

RAM: I think they'll get this right, too - although I wonder about either gratuitous barriers to entry (RAM on the back of the motherboard or the like) OR only 6 RAM slots (I don't think they're stupid enough to give us only 4, and I hope not).

Drives: I'm almost sure I'm right on the boot drive - very fast, T2 (T3?) controlled. I hope they give us standard NVMe expansion, but I am concerned we might see some sort of Apple SSD Module expansion instead...

Ports: The modular bay is relatively unlikely - they'll probably go with external card readers only or a single SD-only slot. My only other worry about ports is whether they "forget" the standard USB-A up front.

Graphics: Even though I'm happy to see more restricted graphics than many others, I'm still concerned that Apple might go even more restricted (soldered GPU, or some form factor so proprietary that upgrades are unlikely).

Slots: I think they'll go external box only - at best, they'll have extra TB3 buses (4?, even 6?)and a feature that allows joining two TB3 buses together. Maybe they'll have a proprietary port that feeds PCIe x16 to an Apple (or close partner - Blackmagic?) external box.

Licensing: Unfortunately, I don't think they will.


[doublepost=1557273189][/doublepost]A lot of people don't need NVidia (most photographers, video editors who use software that prefers AMD, audio folks). Those of us who don't need it might not want it - GPU drivers are tricky to write and can destabilize systems. NVidia doesn't exactly have a track record of solid Mac drivers...

What I remember back to is Windows 2000 and XP. Windows 2000 was the most stable Windows (as stable as Windows gets) for a very long time. XP came out a year and a few months later, and was a lot less stable until the late service packs. What had changed was game support. Windows 2000 made no concessions to games. Games only ran if they were well-behaved Windows applications (and sometimes not even then), and Microsoft warned customers about that. XP was prettier, and it added support for games that did various odd things to the system (which had worked in DOS-based Windows 98, but in no NT-based Windows prior to XP) - but it was much less stable, even if you weren't playing games. It took at least Service Pack 2 and possibly Service Pack 3 before it was anything like as stable as 2000.

I'm concerned that adding NVidia support (especially for only one limited-market system) will be like Windows XP's game support - it might destabilize things for folks who are perfectly happy with AMD.

I wish Microsoft had continued to offer an "our most stable Windows, but games won't run" option...
 
  • Like
Reactions: nerdynerdynerdy
PC gaming cards run too close to the edge to be really stable, they're an old legacy form factor that could easily be replaced by something with better cooling (liquid cooling integrated with the rest of the system?) and NVidia can't or won't produce a stable Mac driver (Apple can't, won't or doesn't want to).
You are serious? Quadros are stable, and Titans are just as stable. Don't even try to make an argument that GPUs are unstable.

Few people put overclocked gaming cards in a $20k workstation. In fact, I recently upgraded one system with $24K of Quadro RTX 6000 cards,.
 
  • Like
Reactions: ETN3
Graphics (this is going to be controversial): I don't think Apple should use standard PCI-e graphics. PC gaming cards run too close to the edge to be really stable, they're an old legacy form factor that could easily be replaced by something with better cooling (liquid cooling integrated with the rest of the system?) and NVidia can't or won't produce a stable Mac driver (Apple can't, won't or doesn't want to). They'll end up with a zillion GeForces of every shape and size (which they don't want to support) if people can shoehorn them in there. Either they should use a slightly modified PCIe (maybe as little as blank backplanes with an internal cable to redirect the signal out the TB3 ports) that AMD commits to supporting, or they should use a form factor that permits them to integrate the cooling with the rest of the system, but also permits graphics upgrades. One or two GPU (Vega/Navi 7nm) capability.

Slots: 1 PCIe x16 with 75 watts of power - for audio and video interfaces, not back door GeForces. It could also provide an interface for powerful or multiple eGPUs/machine learning coprocessors/etc.
All of this is nonsense. Plenty of professionals use Nvidia Pascal founders edition (gaming cards) for 24/7 compute.

I'll agree that part of the delay is probably Apple trying to reinvent the wheel regarding cooling or packaging. But why the hell would you offer up a single x16 slot then try to dictate what users can populate it with??? Give users a standard tower, and let them use it as they will.

Your proposed box does nothing, and offers nothing that a Z series doesn't, with the possible exception of a bunch more TB3 expansion that it doesn't need because it has PCIe slots. I think some of you guys are developing Stockholm syndrome.
 
I never said Quadros weren't stable - that's why you pay a ton for them... Grade A components, a driver developed for the relatively small number of Quadro sales that doesn't do some of the tricks the gaming driver does... I'm sure NVidia is binning the heck out of their GPUs, and the best go in the Quadros (and Titans). I've never used a Quadro, but I have every reason to believe they're stable, and no reason at all to believe they aren't. I have no idea how stable Titans are, but again, no reason to believe they aren't.

The reason Apple won't provide a high-power PCIe slot (and a relatively inexpensive slotbox) has nothing to do with Quadros (except that they don't want to write a second graphics driver just for the Mac Pro - they have a good AMD driver shared across much of their line)- it's hot-rod GeForces they don't want to deal with. NVidia doesn't seem to want to provide a Quadro-only, Quadro-level Mac driver (makes sense, it's not a huge market) - their old web driver handled GeForces, and I'm not actually sure which Quadros it supported.

If they provided a $3000 Mac Pro with an open GPU slot and a NVidia driver (the web driver was never all that stable), they'd be dealing with a lot of gaming cards (and not just expensive, relatively conservative Titans, either). They'd also start getting badgered for a lot of less than stable, gaming-specific APIs...

Apple has a very stable, Apple-written driver for AMD graphics. They have a partnership with AMD that provides them GPUs they're happy with. They don't want to write another driver for one machine that'll be the slowest seller in their whole line. They don't want to deal with hot-running, insufficiently cooled PC gaming cards, either...
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.