Here's the thing. Apple could take off-the-shelf tech and put together a box just like every other makers' Xeon box. That's kinda what they did with the iMac Pro, innovative thermal venting notwithstanding.
The iMac Pro is not hardly an "off the shelf" system. Apple has 'custom' Xeon W for the CPUs. The Vega GPUs are soldered onto the custom motherboard ( unlike generic Vega PCI-e slot cards ). The singular drive is implemented via the T2 series (which no other vendor has ). It is a SSD only workstation. And the data NANDs cards are detached from the SSD controller. None of that is really "off the shelf".
The 10GbE as a base configuration. That isn't standard in the same general price range workstation either. The component controller is pretty off the shelf though. It is the usage that is out of current "norms".
That's not usually what Apple does. When each of the the G3, G4, G5 and later Xeon Mac Pros came out, they were pushing the edge, even adopting things that wouldn't be standard for a few years.
The iMac Pro pushed norms. The Mac Pro 2013 certainly pushed norms.
The basic thing is the PCI bus. PCIe 4.0, which isn't even really out yet, is already being leapfrogged by PCIe 5.0, but that's not out either, though solutions for manufacturers are now on the way (
https://www.electronicdesign.com/industrial-automation/pci-express-gen-5-homestretch).
The Mac Pro primarily waiting for that would be a colossal blunder. The Mac Pro highly likely has not been waiting some major base technology breakthrough. Far more likely it has been waiting for resources to be assigned to the product. A new Mac Pro doesn't desperately need PCI-e v5.0 . From referenced article...
"... Initially, PCI Express Gen 5 will show up in servers to handle 400-Gb/s Ethernet and dual-channel 200-Gb/s InfiniBand. ..."
Mac workstations are barely getting one 10GbE socket; let alone some 400GbE sockets. There aren't any Infiniband solution on the Mac at all last I checked. None. So the 200Gb/s option increment on zero will probably still be zero. The Mac Pro isn't going to be some central nexus point in several major data centers any time soon. It isn't aimed at that market at all. The Mac Pro and iMac Pro helping to push 10GbE into the widespread client status will help the ecosystem far more than trying to win some max bandwidth, "mine is bigger than yours" contest. Servers will need PCI-e v5 more so if more clients have 10GbE.
3-4 years down the road will Mac Pro, find some high utility in picking up PCI-e v5 ... perhaps . However, that is far, far, far away from being a critical path issue now.
I would have a hard time believing that Apple wouldn't try to equip their top-of-the-line unit with radically wider and faster buses - from CPUs to cache and main memory, and between PCIe devices and Thunderbolt.
Apple does
not use all of the PCI-e bandwidth of the CPU provisioned lanes in the iMac Pro now. There are x20 PCI-e v3 lanes in the iMac Pro that go completely
unused. That don't necessarily need to jump to v4 (or v5) if not using all of the v3 lanes have now.
Another standard PCI-e x16 slot and perhaps another standard x4 slot (along with some M.2 drive slot(s)) . Or less likely a pair that could optionally split x16 into two x8 and a x4 would be plenty sufficient to differentiate the iMac Pro and the Mac Pro.
So there isn't a huge need to double up on even larger CPU packages to get a even higher lane budget when aren't even using the full range of the single CPU solution they have now. Apple can easily make the Mac Pro "have more" simply by steeping around the constaints that the iMac Pro has on it that are similar to the constraints that the MP 2013 had (that were "designed into a corner" limiting. ). More system power and optional second GPU. Done.
A broader thermal envelope would allow the Intel W a bigger operating range with the caches they already have. Clock not cache size difference.
Very similar issue with the DIMM slots. iMac Pro is limited to 4 so-DIMM slots. Mac Pro could relatively easily have 8 "normal" DIMM slots. The max Capacity of the two systems would be substantively different.
Sure, you can then throw all the current buzzwords at it; it'll likely have your favourite tech if not something better.
The Mac Pro doesn't need buzzwords or uber/ultimate, extremely expensive tech porn. It should be more than good enough for a substantial number of people. It shouldn't be some "technology concept" demo or hobby project. Nor the most expensive tech possible be selected just because it "has to be" the most expensive base priced Mac ever.
For a pro workflow, bandwidth and latency is often more of a bottleneck than CPU power. Whatever CPU they choose to ship with (pending availability) will be fast, but they need to have a next-generation subsystem to really make it sing. More / faster lanes, and also, they need to deal with potential threats like the Spectre bug.
They aren't fully unitilizing the current Intel W ..... they don't need any Area 51, Buck Rogers tech CPU.
An iteration with the Spectre/Meltdown fixes, those are on the immediate horizon but those are primarily fixes to the current micro-architecture. Not some Buck Rogers future version.
"... The takeaway message from our discussions with Intel is that there are some hardware mitigations in the new Whiskey Lake processors. In fact, there are almost as many as the upcoming Cascade Lake enterprise parts. ..."
https://www.anandtech.com/show/1330...e-intel-clarifies-whiskey-lake-and-amber-lake
The
whole mac line should be trying to pick up those fixes as they trickle out; not just the Mac Pro.
I'm sure Apple has mobo prototypes in the lab but hardware is tricky, and it needs to be rock solid, they need to get their supply lines set up and their cost structure settled... if they are going to be first to market with PCIe 5.0 then they need to at least have a GPU partner that can supply a cutting-edge card, demos have to be coded, drivers and firmware written, etc etc.
That is about exactly what they should not be doing. They should have made a derivate of what they already had finished ( iMac Pro) without the limitations they put on that first cut at a solution. Intel W with more DIMM slots and Vega on a removable card, go with two (instead of just one Ethernet ports: two 10 or a 10 and 1 ), same Thunderbolt, but with some internal alternatives ( some standard PCI-e slots ).
They were/are/will be grossly late with this product. Shooting at some pie-in-sky new tech porn feature wasn't the Mac Pro's primary problem. Apple needed to get the product restarted without engaging in maximizing unnecessary risk. Even ramping up with they had could run into delays ( Cascade lake is probably off the schedule Intel floated over a year ago. ) .
The T2 firmware (and BridgeOS) has needed work on the iMac Pro and MBP 2018 models so that isn't in the super stable zone either. Yet again pointing to why the Mac Pro would be better if merged into to some of the other tech moves that Apple has already started up, instead of an even bigger jump.
If Apple took that approach a new Mac Pro in first half of 2019 is possible. If they were shooting at power point tech that Intel showed them in 2017 then they could easily slip into 2020 ( which would be pretty close to a disaster now that they pegged it as a 2019 product. At some point they are going to loose critical mass. 2 years to get their stuff together is plenty of time even for a product that was complete dead in the water as far as resources go at the start. )
[doublepost=1536599050][/doublepost]
Longer lifecycles due to lack of suitable replacement hardware. Even if you bought into the 6,1 Mac Pro philosophy that model has been stagnant for almost five years. Meanwhile one is limited in the upgrades they can perform with memory being the only official aspect which can be upgraded. Is it any surprise people are holding on to their cMP's?
They aren't solely do to lack of replacements. While I put some mac specific examples in my second, that is isn't a Mac specific issue. That trend had popped up across broad swaths of the general PC marketplace. Lots of folks, even including some pros, are in the the "fast enough" current core system zone.
Mac 6,1 folks who absolutely needed something else could have moved off to Windows ( which is the primary premise being pushed in the initial post . "They are gone" ) or moved to other Mac systems MBP , iMac , iMac Pro. The primary issue of my point is that even with options available, more folks than in previous decades are moving at all. The "why" has many facets, not just 'new' hardware upgrades. Workflow complexity growth(or non growth) , budget allocations , applicability of "lower" products (modern laptop will do to split load) , etc. are all just about equally relevant factors in that growing pool of folks that haven't moved.
For a significant chunk of the Mac Pro 2013 (6,1) users the iMac Pro is and will be Apple's answer. A quite substantive number of folks are buying them. Some are not. A big "chunk" of that not group is what Apple is probably aiming at but the real issue is that of those left, they aren't all bolting at a extremely fast pace either. Many folks can just pull major purchases out of the air (which is why Apple told folks for their fiscal planning purposes that Mac Pro wasn't coming this year; not a 2018 purchase option. )
If Apple were to release a new Mac Pro similar to the cMP I suspect we'd see a lot of interest in it.
A base that almost 100% is dependent upon the remnants of the cMP base will have have problems long term. That is a much smaller base than the Mac Pro had in 2008-2009 when Apple drifted out of a high priority Mac Pro work. In flow from Windows and other Mac products is probably more critical now than it was then. The folks sitting on the 2013 mode are probably a contributing factor to the equation for the new Mac Pro.
If Apple "pulls" in a huge fraction of the remnants of the cMP base all at once and those are primarily the only folks they are trying to target, then in year 2-3 there is nothing left for those year's Mac Pro to survive on. A huge, rapid boom bust cycle is not likely healthy for the Mac Pro. Once again will enter the context of low resource prioritization because "nobody is buying". There has to be some folks other than the die hard, "hold out" group to make the Mac Pro viable.
