So? Most mainstream PC designs - even the Mac Pro - don't have the bandwidth to simultaneously support full bandwidth cards in all of their internal PCIe slots. That doesn't stop them fitting the slots.
That's like saying the OS uses some small amount of swap space so why not go "Whole Hog" and take out more RAM so that use more swap space. "Using some why not use alot". That is serious misguided and dubious design criteria.
I never say there should be no switches what so ever. However, pouring PCI-s switches on the design like cheap ketchup isn't any better than pouring Thunderbolt controllers on the design like cheap ketchup. In fact, it largely pragmatically is the same thing.
That said, the mainstream PC design are more gear for folks posing as if they need high bandwidth. They are not for those who really need it. In those designs, "more" is pitched as being more powerful. ( 8 USB sockets instead of 2-4 , 6 PCI-e slots ... although only have bandwidth to support (oooh more slots gotta be better right? ) .
A "compact" Mac Pro would probably mean dropping 2-3 of the PCIe slots in favour of TB.
A 1/3 smaller (or more ) compact Mac Pro likely means changing CPU packages which would drive fewer PCI-e slots. The cooler, smaller processors only support smaller number of slots without PCI-e switching hocus-pocus ( which the design basically count on not particularly needed higher bandwidth. )
The current Mac Pro has two x4 PCI-e slots that are switched ( sharing bandwidth). Frankly, dropping them in favorite of Thunderbolt is , from a PCI-e perspective, just changing the flavor of the switch. Had x4 v2.0 bandwidth before and would stlll have it in new box only transparently routed to external boxes.
Eating into the two x16 slots is a gross mismatch for Thunderbolt. Doubling or tripling down on TB controllers isn't going to close the gap on what tossing there. TB can only add x4's back. It can't add back aggreated x8's or x16's. It just doesn't work that way.
Why? If you want to plug in 4 displays then you'll need a GPU that provides 4 signals anyway, and I'm sure its not beyond the whit of modern electronics to 'route' 2 or 3 video signals to the controllers using them.
For a box that has 1-2 x16 PCI-e slots this is exceedingly dubious approach. It is far more cost effective and easier just to plug the 4 displays into one (perhaps two if want to split the load) GPU PCI-e cards and
avoid Thunderbolt all together.
The current Mac Pro has no problem with connecting to this many displays. Applying Thunderbolt to this is a solution in search of problem.
Of course it doesn't work well, because it is mismatch of where to apply TB.
If a new Mac Pro still supports PCIe video cards, then you might not want to use the DP functionality of TB at a all.
Correct. So burying 4 GPU signals onto TB is soaking up more resources (additional cost and complexity) that the buyer likely wants thrown in that direction. The buyers are not going to see that as a "value add". That will make the Mac Pro harder to sell.
If Apple nukes all the PCI-e slots (and matches CPU appropriately) what have is essentially an iMac in terms of computation. Apple has an iMac in the line up. They don't need another one.
...but the point of having 4 x Thunderbolt sockets isn't to let you run 24 devices - the point is to deal with the situation where you have some ports 'hogged' by legacy DisplayPort devices
That is pretty bonehead design. If have a large user base with a large number of DP devices then it is far more simpler just to add 1-2 PCI-e slots so a "regular" GPU PCI-e card can be used to delivery that functionailty. If have a large DP problem use a large DP solution. Thunderbolt is
not that. By definition it is a very limited DP solution.
or TB devices with no pass-through,
That is not a host issue. That is a bought the wrong peripherals issue. Again throwing more TB controllers because the peripherals are jacked up is not address the core root cause of the system design problem.
or where you want a free port to plug in a portable drive without having to insert it in a chain.
Thunderbolt isn't USB. If treat it as USB then run into these sorts of "Problems". But they are largely invented corner cases; not canonical system configurations.
If you're going to use USB3 for portable drives etc. and keep TB for the RAID arrays
If the portable drives are single HDD units there is about zero advantage to TB over USB 3.0. Frankly, if it isn't really portable as much as the drives are being used for "sneaker net" data transport TB makes very little sense at all. USB 3.0 is going to be more ubiquitous than TB ever will be. If need to "sneaker net" to random machines USB 3.0 is going to work.
Thunderbolt is a huge mismatch when applied to a single drive. Even more a single protocol is a red flag. Thunderbolt's primary purpose is to aggregate traffic. If not aggregating traffic then really what trying to do is pound a round peg into a square hole. Injecting more round pegs into the system design isn't going to address the making those pegs more square.
It is not so much the RAID systems. It doesn't need to be RAID. It is the aggreation of multiple drive's traffic onto a single cable. That is where benefits of Thunderbolt come in.
Similarly if need 2-5 FW ports, then don't buy 5 FW dongles. The aligned move would be to buy one TB device that had 6 FW ports on it so that match the peripheral to the abnormal glut of FW ports needed.
Its a problem insofar as it costs money to keep a product 'alive' and harms reputation to have an out-of-date and overpriced product in the catalog.
Does Apple know they have a short term problem if they now plan to produce Mac Pro into the future. Sure they do. There is no indications whatsoever that Apple thinks they are running some "optimized" product management for the Mac Pro.
Most likely it was cancelled but Apple decided to give it another shot. Hence the bubble the Mac Pro is in right now. My comments were more aligned with the hand waving about how Apple "has to" keep selling the Mac Pro. They do not. I think Apple is going to give it another shot with a refined product targeting but if the customers don't buy Mac Pro's in the targeted numbers ( there is consistent year-over-year growth aligned with grow in rest of Mac line up) they will cancel it. Mac Pro's future is far more dependent at this point as to what customers do next rather than Apple.
Apple have another problem: what the Mac Pro customers really need is to be able to walk into their local PC builder and say 'I want that motherboard, that CPU, a pair of those video cards, 4 of those hard drives and a Blu Ray writer... and OS X' ...and if you bought an extended warranty and a Monster HDMI cable then maybe your local PC builder would make money on the deal.
That isn't a problem. The notion that Apple has to engage in a race to the bottom is not the issue. Those were not the customers Apple was targeting with the Mac Pro. If you are saying those are primarily the only folks left who want to buy "boxes with slots".... Again Apple has no problem . They just cancel the product. No buyers , no product , no problem.
If you think Bubba and his trusty screwdriver are going to out compete the Mini and iMac for customers in their targeted areas, you're fooling yourself. The PC wars are over. Apple doesn't have to become the top unit seller PC vendor. All Apple has to do is carve out the profitable, growing subset of the PC market. The rest can just bypass, because "biggest" is not the objective.
