Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
I might be in the minority on that, because I thought that was a great, practical solution to the options available at the time - quick recharge for a couple of minutes plug time, without needing any adapters or cables etc. I think a lot of people thought you were supposed to use that all the time to charge the pencil, when it's just an emergency option.

I definitely use it as my only way to charge the pencil, how else would you do it more efficiently ? plug into USB adapter and stick it out a wall? The new way to charge is great though! I basically leave my iPad Pro on my desk ad use it like a laptop, so some way to charge from the top edge for me would've been the best solution
 
I was also the thinking the stackable module idea makes no sense. It's been tried by several vendors in the last decades and it always failed. There's a number of reasons for it:
  • When you design a proprietary interface, you have the maintain it. Suppose it's 'just' a more rugged version of TB3. Then when TB4 comes out, you have to update your proprietary system, making sure that it's backward compatible. Companies underestimate how much effort goes into that.
  • You launch a 'brain' and two modules. So you test brain alone, brain+A, brain+B and brain+A+B. When you add module C, you have four more possibilities. The number of combinations effectively doubles each time you add one module. That's called exponential. It's completely impossible to test all combinations as soon as you have four or five. And that's where support hell starts: I have brain + this combination and it doesn't work! It only gets worse when you daisy-chain them and the order is important: “You should always plug your GPU module in first, then add storage after that” (or vice versa).
  • Instead of having 1 machine to keep up-to-date, you have as many as you have modules. And then there's choice: you may want the fanciest super duper GPU, but for me, middle of the road may do. Is Apple going to make a wide range of modules and keep them all up-to-date?
With eGPU over Thunderbolt, Apple has an easy way out of this hassle. They just support the interconnect and let other vendors take care of the enclosure and the GPU cards. I cannot believe they would lock themselves into a completely proprietary ecosystem with support hell like this. Rather, stackable components connected by Thunderbolt. I see some merit in that idea. But it would involve quite a lot of cables at the back.

And no, I don't think an19” rack model is likely, either. Apple is a consumer company, not an IT company.
Thunderbolt is to slow for real video use and an real pro system may need like 6-10 TB 3.0 buses.
 
Thunderbolt is to slow for real video use and an real pro system may need like 6-10 TB 3.0 buses.

"Real" video use? Most modern Thunderbolt SSDs have enough bandwidth for ProRes and DNG RAW 4K60.

This is what bugs me about these discussions. Everyone arbitrarily deciding their professional use is the one that should matter most, and if someone is met with other needs then they aren't true pros.
 
I definitely use it as my only way to charge the pencil, how else would you do it more efficiently ? plug into USB adapter and stick it out a wall? The new way to charge is great though! I basically leave my iPad Pro on my desk ad use it like a laptop, so some way to charge from the top edge for me would've been the best solution

Just get a standard USB power adapter with a lightning cable, and a cable adapter for the end to go from male lighting, to female lightning. I thought the iPad Pros actually came with one?
 
  • Like
Reactions: ekwipt
Thunderbolt is to slow for real video use and an real pro system may need like 6-10 TB 3.0 buses.
I work at a tv station. We haven't got a single Thunderbolt port in sight. It all goes over USB and even SD cards. We're mainly using 4 year old Xeons with 32GB of RAM. Suits most people fine, except for larger programs (> 30 minutes). But I guess tv isn't 'real' video?
 
I work at a tv station. We haven't got a single Thunderbolt port in sight. It all goes over USB and even SD cards. We're mainly using 4 year old Xeons with 32GB of RAM. Suits most people fine, except for larger programs (> 30 minutes). But I guess tv isn't 'real' video?

Especially in broadcast there's a huge schism between what cameras can output and what you're actually using. Some national news orgs only switched over to 1080p in the last few years. Nothing is actually getting delivered in 8K; even most Hollywood films don't have their effects mastered at higher than 2/2.5K. I'm still producing 1080p deliverables for conferences because most setups don't have 4K screens or 4K needs.

(Which is why I don't get why so many people are buying these cameras based on resolution. Unless you're a YouTuber who's doing 1-camera punch in/out and doesn't want to sacrifice resolution, you're buying a camera that's often inferior to "lower-spec" cameras with superior optics.)
 
  • Like
Reactions: barmann
I work at a tv station. We haven't got a single Thunderbolt port in sight. It all goes over USB and even SD cards. We're mainly using 4 year old Xeons with 32GB of RAM. Suits most people fine, except for larger programs (> 30 minutes). But I guess tv isn't 'real' video?

I was questioning the 6 thunderbolt ports as well, we actually have a Aja T-tap to plug into our client monitor on our i7 iMacs. I doubt most people would need more than 4 with at the daisy chaining possible
 
Come on guys the Mac Pro isn’t supposed to be a rack server
I know it was unintentional, but thanks, I got a really good chuckle - that you would even have to say something so extraordinarily obvious is such a perfect encapsulation of how far most people here have jumped the shark.
 
I was also the thinking the stackable module idea makes no sense. It's been tried by several vendors in the last decades and it always failed. There's a number of reasons for it:
  • When you design a proprietary interface, you have the maintain it. Suppose it's 'just' a more rugged version of TB3. Then when TB4 comes out, you have to update your proprietary system, making sure that it's backward compatible. Companies underestimate how much effort goes into that.
  • You launch a 'brain' and two modules. So you test brain alone, brain+A, brain+B and brain+A+B. When you add module C, you have four more possibilities. The number of combinations effectively doubles each time you add one module. That's called exponential. It's completely impossible to test all combinations as soon as you have four or five. And that's where support hell starts: I have brain + this combination and it doesn't work! It only gets worse when you daisy-chain them and the order is important: “You should always plug your GPU module in first, then add storage after that” (or vice versa).
  • Instead of having 1 machine to keep up-to-date, you have as many as you have modules. And then there's choice: you may want the fanciest super duper GPU, but for me, middle of the road may do. Is Apple going to make a wide range of modules and keep them all up-to-date?
With eGPU over Thunderbolt, Apple has an easy way out of this hassle. They just support the interconnect and let other vendors take care of the enclosure and the GPU cards. I cannot believe they would lock themselves into a completely proprietary ecosystem with support hell like this. Rather, stackable components connected by Thunderbolt. I see some merit in that idea. But it would involve quite a lot of cables at the back.

And no, I don't think an19” rack model is likely, either. Apple is a consumer company, not an IT company.

thunderbolt is retro compatible BECAUSE it is pcie protocol...

if you expend the principle to a connector that can move 16lane of gen 3 .0 pcie you will still move 12gb/s either way... so even if they do tunderbolt 4 wit pcie 4.0, pcie switch can still put that bandwith thru a 8xgen4 pcie
we do this all day long with 2 4x pcie m2 ssd gen 3 to 16x gen 2 on the macpro 2009...

this is the beauty of PCIE lanes splitting and agrégation...
basically you could have only one connector with the 64 pice gen3 of the new xeon to any kind of split box, and plug any pcie combination of card of your will...

my macpro is curently running everything outside on pcie expender
and one of them have 4 gpu on a single 16x pcie switch...

and believe it or not, all 2010 and 2009 mac pro have a internal pcie switch on port 3 & 4 .... so apple is familiar with this... and no one ever noticed until we had pciex4 ssd that the bandwidth of port 3&4 was actually shared on the same 4 pcie lane...

i dont know what you work on but if black magic cant saturate a pcie gen3 x8 link capturing 8k raw footage (16 full hd raw stream) I highly doubt a 16x pcie gen 3.0 link will be obsolete anytime soon...

on top of that you forget that quick path is pcie, and if you woul have a external 16x conector you could built a cluster of macpro... interconnected with pcie 16x
 
Last edited:
Yeah, with pure PCIe or Thunderbolt, it works exactly as you describe. Good to know Apple has experience with this sort of setup.

What I'm afraid of, though, is proprietary connectors. Most modular “add a box by clicking it into the previous one” sort of setups used proprietary connectors or proprietary backplanes. Even if this proprietary connector is just a straight PCIe or Thunderbolt with a different connector, it will take converters and stuff to make it work with anything that isn't part of the original, Apple offering. That's obviously bad for customer choice, but also for Apple because they'll always get demand for more modules than they can produce (are they really going to produce SAS, SDI and so on boxes?). Also, the more work they put into adapting the internals to fit into a proprietary design, the more work it is to develop and maintain such modules. In other words: I’d prefer it if they just give us “container” boxes that connect over Thunderbolt and that have a standard PCIe inside. It will make our lives and Apple's much simpler. But, knowing what Apple did with the Retina MacBook Pros and their 'standard' PCIe SSDs, I am guessing we're going to see something proprietary despite all the history lessons that plead against it...
 
  • Like
Reactions: askunk
thunderbolt is retro compatible BECAUSE it is pcie protocol...

Thunderbolt is not PCI-e. It is like saying what is inside the telephone system is sound. There is no compressed/modulated air in the telephone system. There is a representation of the sound, but there is no sound. Same with Thunderbolt. There is a representation of PCI-e data that is being transported to another subsystem but this is not "raw" PCI-e itself. Thunderbolt transports PCI-e so can couple do distance PCI-e devices together. Thunderbolt also can transport DisplayPort data protocol streams by also encoding/decoding them for transport and delivering them at an external destination.

Thunderbolt is not "external PCI-e". The PCI-e standard has somewhat of a complete standard for that.


my macpro is curently running everything outside on pcie expender
and one of them have 4 gpu on a single 16x pcie switch...

Typically most x16 expansion boxes come coupled to a card/cable to the box. The box will be certified with its card and not random card 23.


The two solutions are aimed at somewhat overlapping but different problems. Thunderbolt is more robust range of connections that should "just work". It carries more protocols and typcially for more power ( more aimed at single cable solution spaces ).

Neither one is a "solves everything" ketchup to be poured on because it always works.
 
Thunderbolt is not PCI-e. It is like saying what is inside the telephone system is sound. There is no compressed/modulated air in the telephone system. There is a representation of the sound, but there is no sound. Same with Thunderbolt. There is a representation of PCI-e data that is being transported to another subsystem but this is not "raw" PCI-e itself. Thunderbolt transports PCI-e so can couple do distance PCI-e devices together. Thunderbolt also can transport DisplayPort data protocol streams by also encoding/decoding them for transport and delivering them at an external destination.

Thunderbolt is not "external PCI-e". The PCI-e standard has somewhat of a complete standard for that.




Typically most x16 expansion boxes come coupled to a card/cable to the box. The box will be certified with its card and not random card 23.


The two solutions are aimed at somewhat overlapping but different problems. Thunderbolt is more robust range of connections that should "just work". It carries more protocols and typcially for more power ( more aimed at single cable solution spaces ).

Neither one is a "solves everything" ketchup to be poured on because it always works.

look on the thunderbolt add on card that people use on macpro, from gigabyte, and they tunel pcie via thunderbolt no brainer... it is just a software issue...

it is totally possible to pass 16 lane pcie lanes thru 4 thunderbolt on 4 cards, not that it would make any sense, but it is a proof of concept a thunderbolt 3 cable is indeed a x4 pcie cable...
 
Again...

Mostly we know about 2016(hopefully) nMP 7,1 are personal assumptions and some "anonymous leaks".

Most rumors point to Introductions at WWDC with availability not earlier than Q4'16

  • Should be Based on Intel C612 Server or X99 Workstation/ProSumer Chipset.

  • GPUs x2 most leaks points to AMD GPU (Polaris and Vega) very few forum speculations still believe possible an nVidia based nMP.

  • Thunderbolt 3 upgrade it's a logical evolution, as USB-C and HDMI2.

  • CPUs Xeon E5v4 family most likely few rumors account on AMD Zen.

  • Storage, it's also assumed as logical evolution either 1x NVMe on PCIe3 or 2xNVMe on PCIe2 2.5 GBps total throughput...

  • RAM upto 256 GB possible as supported by C612 on 4 RDIMM DDR4 ECC Slot.

  • Performance as Xeon E5v4 should be utpo 630GFlops compute FP64 (CPU only E5-2699v4).

  • Performance on Upcoming GPUs speculatively should be (on each GPU) among 5.5Tflop FP32 (Polaris) to 9 TFlop FP32 (Vega) FP64 performance should be from 500GFlop to 4-5 TFlop.

The purpose of this thread
is to discuss on possible configurations also share news and discussion on the probable nMP 7,1 components.

Everybody is welcome.
[doublepost=1551392106][/doublepost]Here is a prediction...based on a pattern.

The 2019 Mac Pro will come with a T2 chip, which is likely to contribute to temporarily bricking the OS if the hardware configuration is changed outside of the hardware parameters that Apple specifies.

In other words, my prediction is that gone are the days of the flexibility of the Mac Pro models 5,1 and earlier, when you could put in a multitude of different graphics and I/O cards, memory, and even upgrade the CPU.

My evidence for this is: the 2018 Mac Mini, iMac Pro, Mac Book Air, and Mac Book Pro - and the very soon to be released 2019 iMac - all contain T2 security chips, and an Apple employee has confirmed that all models from here on will contain the T2 chip. In other words, each new model / refreshment of a model line that Apple has released has included the T2 chip, and it's highly likely that it is coming to the new Mac Pro too.

For those who want to upgrade their Macs, it is possible that Apple will place considerable limits on what you can upgrade.
 
  • Like
Reactions: H2SO4
That would be a surefire way of getting people to jump ship to Windows boxes, either through mainstream OEM workstation builders or indie ones like Puget or Storm something or another.
 
  • Like
Reactions: ssgbryan
Doubling the price on the high end SP with next to no performance improvements, ouch.
 
[doublepost=1551392106][/doublepost]Here is a prediction...based on a pattern.

The 2019 Mac Pro will come with a T2 chip, which is likely to contribute to temporarily bricking the OS if the hardware configuration is changed outside of the hardware parameters that Apple specifies.

In other words, my prediction is that gone are the days of the flexibility of the Mac Pro models 5,1 and earlier, when you could put in a multitude of different graphics and I/O cards, memory, and even upgrade the CPU.

My evidence for this is: the 2018 Mac Mini, iMac Pro, Mac Book Air, and Mac Book Pro - and the very soon to be released 2019 iMac - all contain T2 security chips, and an Apple employee has confirmed that all models from here on will contain the T2 chip. In other words, each new model / refreshment of a model line that Apple has released has included the T2 chip, and it's highly likely that it is coming to the new Mac Pro too.

For those who want to upgrade their Macs, it is possible that Apple will place considerable limits on what you can upgrade.

The T2 can be deactivated: https://eclecticlight.co/2018/11/21/welcome-to-your-new-mac-living-with-the-t2-chip/
 
  • Like
Reactions: filmak
Even if this proprietary connector is just a straight PCIe or Thunderbolt with a different connector, it will take converters and stuff to make it work with anything that isn't part of the original, Apple offering. That's obviously bad for customer choice, but also for Apple because they'll always get demand for more modules than they can produce (are they really going to produce SAS, SDI and so on boxes?). Also, the more work they put into adapting the internals to fit into a proprietary design, the more work it is to develop and maintain such modules. In other words: I’d prefer it if they just give us “container” boxes that connect over Thunderbolt and that have a standard PCIe inside.

It's not going to be a series of boxes. We've been through this. A bunch of idiots on youtube like that nutty professor guy are pushing this for their own ends.

Jony Ive's designs deify neatness and minimalism. The Mac Mini stack or some variant of that would go against everything that Apple has been producing for the last decade.

Every computer needs a processor. Every computer needs ram, a graphics card, storage. The idea that each of these things would be separately installed in a separately powered and cabled module is utter madness. Just think of the supply chain nightmare of packaging, storing, dispatching and tracking module after module.

It is not going to happen.
 
Prepare to be disappointed....

I didn't say that I was excited about a Jony Ive minimalist workstation. I don't really care about how it looks, but I'm sure it'll look nice.

I'm telling you to look at every product Apple sells, and then explain how a fiddly, multi-case jumble makes any sense.

Sure they designed a workstation that lots of nerds didn't like. But it's bizarre logic to then pick a random design that nerds also don't like, and then extrapolate that Apple must be designing one of those.

The amount of analysis going into the Mac Mini Stack on this forum is really quite something to see.
 
Last edited:
That would be a surefire way of getting people to jump ship to Windows boxes, either through mainstream OEM workstation builders or indie ones like Puget or Storm something or another.
I personally see this as a positive relief, in the context of this thread’s demography. It is like having a crazy wife, finally does something stupid enough to justify divorce, and nobody will think it’s your fault.
 
I personally see this as a positive relief, in the context of this thread’s demography. It is like having a crazy wife, finally does something stupid enough to justify divorce, and nobody will think it’s your fault.
Certainly deserves a ‘Ha’
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.