OWC Opens Pre-Orders for New 12-Port Thunderbolt 2 Dock

[jayducharme]

Why oh why is there an audio out and audio in port?

I'm actually grateful for that. The Mac Pro has no audio in.

I posted another thread with my impressions.

https://forums.macrumors.com/threads/1835376/

So far I like it. The unit's HDMI port isn't great. The build quality is so-so. It doesn't mute the audio out when the computer shuts down (resulting in a loud thump from the speakers). But outside of that, it seems to be working fine. And it definitely has more ports than any other TBolt dock currently on the market.

 

[rockinrocker]

The unit's HDMI port isn't great.

Anything in particular make you say that?

I did notice that it takes longer for my monitor to come on when I use the HDMI (to DVI) cable rather than through one of the Thunderbolt ports....

 

[skippymac]

Why oh why is there an audio out and audio in port?

I just want a single port that's compatible with iPhone headphones.

The audio in will be one hell of a lot more useful for most people. If you're going to spend $249 on a dock surely you could spare a few bucks for a mic

If not I'm sure there's an adapter on amazon somewhere that would cost like $1.

 

[manu chao]

Isnt it a bit misleading to claim two Thunderbolt 2 "expansion ports"? One of them is an input from your computer isn't it? So there is no expansion of thunderbolt at all, just one port out for one port in (daisy chaining).

It's sloppy phrasing, it should have been written:

"It includes two Thunderbolt 2 ports and as expansion ports five USB 3.0 ports, 3.5mm audio in/out and one port each for Firewire 800 devices, Gigabit Ethernet, and HDMI 1.4b with support for 4K displays.

With a number of single port small thunderbolt drives out there, what I want is a thunderbolt hub, so I can connect two of such drives to my Mac. Technically this may not be possible to split a thunderbolt signal the way you can with USB. I don't know.

In the initial specs of TB, TB hubs were mentioned but I have not seen one.

----------

1TB + 1 HDMI. So two as long as one is an Apple TB since I don't know many other monitors that are pure TB and not just mini DP

Why would it need to be an Apple TB display? TB is designed to allow a standard DP display to be added to the end of every TB chain.

----------

No reason to think that this will be any different from other TB docks w.r.t. monitor support. One display per. Thunderbolt peripheral is a fundamental limitation of Thunderbolt. A second monitor would have to be a Thunderbolt display, or connected via a second Thunderbolt peripheral.

And what the heck is 'a TB display' exactly? Right, a display with a built-in TB hub, ie, a combined device that has a TB feed-through. Glue any TB hub to the back of a display and connect the display to video port of the hub (be it DP or HDMI) and you have 'a TB display'.

----------

Nope, that's what I thought too. Basically from what I understand, minDP is used up by the HDMI port, leaving on the second DP channel that's embedded into TB.

If you read their description, they kind of say that it only works with TB + HDMI, but not HDM + mDP or HDMI + DP, or HDMI + DVI, etc

People have successfully driven two external displays via a single TB cable using other docks. Why should this dock be different?

 

[theluggage]

Why would it need to be an Apple TB display? TB is designed to allow a standard DP display to be added to the end of every TB chain.

...but not if the last device in the chain includes a display or is driving a display via HDMI. You can't chain a non-TB display directly to an Apple Thunderbolt Display, for example.

Glue any TB hub to the back of a display and connect the display to video port of the hub (be it DP or HDMI) and you have 'a TB display'.

True, but if you already have a TB dock driving one HDMI display, you'd need a second TB dock glued to the back of your second display.

To explain: every TB peripheral contains a TB peripheral encoder. This breaks out (2?) PCIe lanes and one DisplayPort output from the thunderbolt signal. The DisplayPort signal can either drive a dedicated DisplayPort/DVI/HDMI/VGA connector or be fed to the 'Thunderbolt Through' port when a standard DP device is connected to it. It can't do both at the same time.

Attach a second thunderbolt device - e.g. a second TB display, a second dock or or even a disc drive with thunderbolt-through and you add a second thunderbolt controller, which breaks out a second DisplayPort signal.

In theory because Thunderbolt 2 has added support for DisplayPort 1.2, it ought to be possible to use DisplayPort 1.2's multi-display support to drive two displays from one Thunderbolt 2 port in DP mode. In this case, the first display would need to feature a second DisplayPort with multi-display support, and the second display would attach to that. However, that depends on OS X driver support for DisplayPort multi-display mode - it would be interesting to hear if anybody has got that working.

 

[manu chao]

To explain: every TB peripheral contains a TB peripheral encoder. This breaks out (2?) PCIe lanes and one DisplayPort output from the thunderbolt signal. The DisplayPort signal can either drive a dedicated DisplayPort/DVI/HDMI/VGA connector or be fed to the 'Thunderbolt Through' port when a standard DP device is connected to it. It can't do both at the same time.
Attach a second thunderbolt device - e.g. a second TB display, a second dock or or even a disc drive with thunderbolt-through and you add a second thunderbolt controller, which breaks out a second DisplayPort signal.

What is the difference between a TB peripheral encoder and a black box that contains two TB encoders in series? The second one is a more powerful TB encoder. Or said differently, it is a newer generation TB encoder. The point I am trying to make is that the limitation you describe applies to a given generation of TB encoders, to a given implementation of encoders, but not necessarily to TB encoders per se.

Until now, we did not have any dock that had video+Ethernet+FW (+USB & audio which all have). Doesn't this indicate that TB peripheral encoders have gained new capabilities?

 

[johnnyturbouk]

Still without eSATA. Not interested.

cant please everybody

I think this looks pretty awesome, however have decided to wait until i upgrade my imac to a skylake one with tb3, hence will wait for the updated version. furthermore, the caldigit TB2 dock looks pretty decent too, and cheaper too @$170, albeit, not as versatile.

 

[theluggage]

What is the difference between a TB peripheral encoder and a black box that contains two TB encoders in series?

Simple. One exists, the other doesn't.

(Oh, and it would use up 2 of your 6 devices per chain, although, to be fair, I don't think that's a limit that people have been knocking up against)

Yes its plausible that Intel could produce such a dual controller (or even just a single controller with dual DisplayPort outputs) but they don't, and probably won't.

Until now, we did not have any dock that had video+Ethernet+FW (+USB & audio which all have). Doesn't this indicate that TB peripheral encoders have gained new capabilities?

Actually, I think it has to do with limitations in Windows' support for Thunderbolt 1 devices being fixed for Thunderbolt 2: go to http://www.sonnettech.com/product/echo15prothunderboltdock.html, choose the Shipping Status Update and read the last bullet - although its not very clear.

 

[manu chao]

Simple. One exists, the other doesn't.

(Oh, and it would use up 2 of your 6 devices per chain, although, to be fair, I don't think that's a limit that people have been knocking up against)

Yes its plausible that Intel could produce such a dual controller (or even just a single controller with dual DisplayPort outputs) but they don't, and probably won't.

So, my initial hope when TB arrived on Macs that I finally could add two external displays to a MBP via the help of a dock is still unfulfilled. Sure, the 13" 2011 MBP (the first ones with TB) didn't have enough graphics power for it, but the 2012 ones do (and they have USB 3, one more reason to get them). All rMBPs have enough video ports but using them would break the one-cable idea. The Apple TB displays would of course have been a solution but they aren't cheap and I already have one nice monitor, thus the need to buy two to get a two-monitor setup wasn't very appealing. Moreover, they aren't wide-gamut and don't offer USB 3.

On the other hand, two TB1 docks (with both having a video port), could have given me two external displays already in 2012 (if there were any docks that shipped in 2012) but I somehow refused that idea, hoping a dock with two video ports (+FW) would be just around the corner. And the OWC dock, which I thought would allow me that needs yet another TB device. Though that could be the excuse to buy my first external TB storage. But then all new backup storage I buy is bare drives and I have not seen a two-port TB HDD dock.

Actually, I think it has to do with limitations in Windows' support for Thunderbolt 1 devices being fixed for Thunderbolt 2: go to http://www.sonnettech.com/product/echo15prothunderboltdock.html, choose the Shipping Status Update and read the last bullet - although its not very clear.

They imply that the two-PCI-controller limitation of TB on Windows doesn't exist anymore with TB2 but as you said it's not very clear. Their argument that their larger number of ports (or PCI controllers) requires TB2 though doesn't make much sense. TB2 is only faster because it aggregates two channels, if you have to serve multiple ports, than the two 10 Gbit/s channels of TB1 shouldn't be slower (on paper) than one 20 Gbit/s channel.

 

[theluggage]

Their argument that their larger number of ports (or PCI controllers) requires TB2 though doesn't make much sense.

I think what Sonnet are saying is that the Windows drivers for TB1 place an artificial limit of 2 PCIe controllers in any device, that the TB2 drivers remove this, but only if you are using a TB2 device.

Ok, so a TB Dock contains the TB controller, which provides the display output and an internal PCIe bus. Hooked to the PCIe bus are a number of "PCIe controllers" as in "PCIe to 4-port USB3" or "PCIe to Gigabit Ethernet". If you want another type of port, you either need another PCIe-to-whatever controller or (for something low bandwidth like audio) you can steal one of the USB ports.

So, you can make the typical 3xUSB3 + Ethernet + Audio dock using just 2 PCIe controllers, but if you want extras like SATA you need an extra controller or something has to come out. So a limit of 2 controllers would actually explain an awful lot about the limitations of current docks - and obviously be a problem for Sonnet's vapourware dock that promises everything but the kitchen sink.

OTOH, some docks manage to squeeze in Firewire (maybe there are dual USB/Firewire controllers - or maybe they don't work fully on windows).

This is all irrelevant to displays, which use the DisplayPort output from the TB chip, not PCIe.

 

[hfg]

OTOH, some docks manage to squeeze in Firewire (maybe there are dual USB/Firewire controllers - or maybe they don't work fully on windows).

There must be an engineering trick to extracting FireWire easily. The Apple Thunderbolt-FireWire800 dongle doesn't have a Thunderbolt controller within it, and thus is priced accordingly low.


EDIT: I stand corrected ... I somehow missed that interesting thread linked below ....

 

[theluggage]

The Apple Thunderbolt-FireWire800 dongle doesn't have a Thunderbolt controller within it, and thus is priced accordingly low.

That one was debunked: https://forums.macrumors.com/threads/1428004/

TLDNR: The TB controller is actually in the plug and the short 'cable' is actually a PCIe bus, not Thunderbolt.

It does save the cost of an active TB cable with cable controller chips at either end, but I think the bottom line is that Apple may not be making much profit on their 'cheap' Ethernet and Firewire adapters, but they were a necessity to allow Ethernet and Firewire to be dropped from the MBP.

 

[manu chao]

I think what Sonnet are saying is that the Windows drivers for TB1 place an artificial limit of 2 PCIe controllers in any device, that the TB2 drivers remove this, but only if you are using a TB2 device.

That part makes sense, what I was referring to was the second bullet point:

• Some workflows require Thunderbolt 2's 20Gbps bandwidth. With the Echo 15 docks' multiple options for connecting (internal SATA SSDs, eSATA, USB 3.0, FireWire 800, Gigabit Ethernet), transferring or editing files at full speed across multiple connections can potentially be throttled by the bandwidth of 10Gbps Thunderbolt.

 

[theluggage]

That part makes sense, what I was referring to was the second bullet point:

• Some workflows require Thunderbolt 2's 20Gbps bandwidth. With the Echo 15 docks' multiple options for connecting (internal SATA SSDs, eSATA, USB 3.0, FireWire 800, Gigabit Ethernet), transferring or editing files at full speed across multiple connections can potentially be throttled by the bandwidth of 10Gbps Thunderbolt.

All that is saying is that the Sonnet dock has so many ports that you could easily hang enough devices on it to use up the 10Gbps bandwidth of TB1, so TB2 is a good idea. Sadly, the way its going, by the time Sonnet actually releases this dock, Thunderbolt 3 will be looming and/or Apple will have released USB3.1-only laptops and Thunderbolt will be dying.

 

[manu chao]

All that is saying is that the Sonnet dock has so many ports that you could easily hang enough devices on it to use up the 10Gbps bandwidth of TB1, so TB2 is a good idea. Sadly, the way its going, by the time Sonnet actually releases this dock, Thunderbolt 3 will be looming and/or Apple will have released USB3.1-only laptops and Thunderbolt will be dying.

TB1 is only limited to 10 Gb/s per PCIe controller. And it can feed two 10 Gb/s streams to two PCIe controllers simultaneously . TB2 simply ups that to a single 20 Gb/s stream. This dock has five PCIe controllers, thus as long as no single controller is asked to provide more than 10 Gb/s, it should not be slower with TB1 than with TB2.

The dock has a total of four 6 Gb/s (2x eSATA, 2x internal SATA), one 3 Gb/s (internal optical drive), four 5 Gb/s (USB 3), and two ~1 Gb/s (FW800+Ethernet) ports. If they put the four SATA and USB 3 connections each onto two PCIe controllers and the optical drive, FW800 and Ethernet onto the fifth one, being limited to 10 Gb/s per controller should only slightly cap the speed for the SATA ports.

Having TB2 still makes sense as it offers 20 Gb/s to devices put behind the dock.

 

[deconstruct60]

TB1 is only limited to 10 Gb/s per PCIe controller.

No. A TB1 daisy chain is limited to 10 Gb/s. Two physical PCIe controllers in a single peripheral will get a subset of that as there is a pragmatic x2.5-x3 PCIe v2 bandwidth through a single TB controllers PCIe switch. There is even less if there is another high bandwidth demand up/downstream on the daisy chain.

There hand waving configs where there are two TB daisy chains coming off a host system where there is a flawed nothing that those are two additive 10Gb/s streams. Pragmatically it is not. If trying to feed from the host system then its TB has the same PCIe switch gating factor. Theoretically could get a PCIe to PCIe device transfer going between two peripherals to help fill the "gap" the host system can't fill with two TB streams. Not likely to happen in the real world.

TB2 simply ups that to a single 20 Gb/s stream.

That is tasked with channeling both DisplayPort and PCIe data. Again if there are DisplayPort data sinks up/down stream from a single peripheral not going to get any where near 20 Gb/s of encoded PCIe data traffic to any one peripheral let alone to any of the multiple PCIe controller data sinks inside of a single one.

This dock has five PCIe controllers, thus as long as no single controller is asked to provide more than 10 Gb/s, it should not be slower with TB1 than with TB2.

TB2 is better because it give about x3 PCIe v2 worth of throughput.

" ... Whereas most Thunderbolt storage devices top out at 800 - 900MB/s, Thunderbolt 2 should raise that to around 1500MB/s (overhead and PCIe limits will stop you from getting anywhere near the max spec). ... "
http://www.anandtech.com/show/7049/intel-thunderbolt-2-everything-you-need-to-know

That is conditional on no DisplayPort or other up/downstream controller tapping the same core bandwidth source.

The five controller will sit better a on with is effectively a x3 -> x1, x1, x1, x1 ,x1 switch ( 5:3 ratio ; 1.66 ) than on a x2 -> x1, x1, x1, x1, x1 switch ( 5:2 ratio ; 2.50 ) . Still run into a log jam if "light up" all of the ports with concurrent max effort data transfers, but it is better able to handle 2-3 concurrent at same time loads ( with 3-2 of the ports mostly idle).

Having TB2 still makes sense as it offers 20 Gb/s to devices put behind the dock.

 

[deconstruct60]

....
Actually, I think it has to do with limitations in Windows' support for Thunderbolt 1 devices being fixed for Thunderbolt 2: go to http://www.sonnettech.com/product/echo15prothunderboltdock.html, choose the Shipping Status Update and read the last bullet - although its not very clear.

That is pretty lame since the PCIe switch inside the TB controller is way more flexible than that.

"According to Intel, Cactus Ridge (2C/4C) can be configured in the following ways:

1 * x4: one device of four lanes
4 * x1: four devices of one lane each
2 * x2: two devices of two lanes each
1 * x2 + 2 * x1: One device of two lanes and two devices of one lane
.."
http://www.tomshardware.com/reviews/thunderbolt-performance-z77a-gd80,3205-4.html

That would mean that the Windows drives only supported the 1st and 3rd options only. The first is likely used by a external PCIe card enclosure. The card socket(s) hooked to a x4 switch if more than one. The second if wanted a x2 SATA and then split the other x2 among 2-3 x1 controllers with a x2 -> n * x1 switch.

The 4 * x1 lane is the natural mode that the vast majority of docking stations would use. 4 (or less ) different kinds of ports (e.g. USB , FW , Ethernet , Audio ) along with perhaps a video port (which is independent of some PCIe switching limitation).

I suspect Sonnet problems more so were grounded in putting a x4 -> 5 * x1 switch ( or something more esoteric than that if tried to use a x2 SATA controller).

 

[manu chao]

No. A TB1 daisy chain is limited to 10 Gb/s.

Anandtech writes in 2011 at the introduction of Thunderbolt 1:

"Thunderbolt is dual-channel, with each channel supporting 10 Gbps of bidirectional bandwidth. That’s a potential 20 Gbps of upstream and 20 Gbps of downstream bandwidth."

There hand waving configs where there are two TB daisy chains coming off a host system where there is a flawed nothing that those are two additive 10Gb/s streams. Pragmatically it is not. If trying to feed from the host system then its TB has the same PCIe switch gating factor.

In 2013 Anandtech wrote:

"Thunderbolt 2 moves to a new 20Gbps bi-directional channel that can handle both data and/or display."

So, next to upping things to 20 Gbps within one channel, it allows to mix PCIe and DP within a given channel.

 

[unplugme71]

I read this title as 12 TB2 ports. Not 12 mixed ports hosted by TB2.

----------

Anandtech writes in 2011 at the introduction of Thunderbolt 1:

"Thunderbolt is dual-channel, with each channel supporting 10 Gbps of bidirectional bandwidth. That’s a potential 20 Gbps of upstream and 20 Gbps of downstream bandwidth."



In 2013 Anandtech wrote:

"Thunderbolt 2 moves to a new 20Gbps bi-directional channel that can handle both data and/or display."

So, next to upping things to 20 Gbps within one channel, it allows to mix PCIe and DP within a given channel.

TB1 was 2 x 10Gb each direction, not 20Gb and 20Gb.

TB1 could not combine the channels whereas TB2 could.

TB3 will offer 40Gb (5GB/s) speeds, with support for dual 4K or dual 5K Apple Displays. Offer support for HDMI 2.x spec. Will be supported when Intel Skylake is released.

TB4 is rumored to double the 40Gb to 80Gb (10GB/s) speeds and drive four 4K displays or two 8K displays. No ETA or what chipset it will work with yet. The spec is still be written as TB3 hasn't even been released yet and is still testing in labs. This also allows additional devices on the daisy chain too which is a huge improvement over TB1 and 2.

 

[manu chao]

TB1 was 2 x 10Gb each direction, not 20Gb and 20Gb.

What Anandtech meant (and what I was referring to) was that two 10 Gbps channels could travel through one TB cable serving two separate 'customers' (PCIe controllers) and thus a combined amount of 20 Gbps of data could travel through a TB cable in one direction (and as I understand it simultaneously also in the opposite direction).

 

[bearda]

In theory because Thunderbolt 2 has added support for DisplayPort 1.2, it ought to be possible to use DisplayPort 1.2's multi-display support to drive two displays from one Thunderbolt 2 port in DP mode. In this case, the first display would need to feature a second DisplayPort with multi-display support, and the second display would attach to that. However, that depends on OS X driver support for DisplayPort multi-display mode - it would be interesting to hear if anybody has got that working.

Doesn't work on a late-model rMBP 15". I know the monitors are good for it since it works find on my Linux workstation in the same setup, but I had to switch from the DP daisychain setup when I connected the laptop up. I was shocked the Linux drivers supported it and the OS X ones didn't.

 

[jayducharme]

Anything in particular make you say that?

I did notice that it takes longer for my monitor to come on when I use the HDMI (to DVI) cable rather than through one of the Thunderbolt ports....

When I hooked up my monitor through the OWC HDMI port, the image quality was poor compared to the HDMI port on my Mac Pro. The image exhibited banding and other artifacts, as if the computer switched to a lower resolution.

 

[unplugme71]

What Anandtech meant (and what I was referring to) was that two 10 Gbps channels could travel through one TB cable serving two separate 'customers' (PCIe controllers) and thus a combined amount of 20 Gbps of data could travel through a TB cable in one direction (and as I understand it simultaneously also in the opposite direction).

yeah I read it wrong..

 

[deconstruct60]

Anandtech writes in 2011 at the introduction of Thunderbolt 1:

"Thunderbolt is dual-channel, with each channel supporting 10 Gbps of bidirectional bandwidth. That’s a potential 20 Gbps of upstream and 20 Gbps of downstream bandwidth."

You are playing "move the goal posts". In the context in which I was quoted, PCIe data, for TB v1 it is 10 Gb/s. The misconception you seem intent on throwing out there is that these two 10 Gb/s channels of TB v1 are flexible. They could be used either for PCIe or Display Port data. The truth was not literally what this drawing from the article shows for the TB cable.

One of those 10Gb/s paths is dedicated to TB encoded Display Port and one of those 10GB/s paths is dedicated to TB protocol encoded PCIe data. Even through the TB controller. The drawing is much closer to what literally is happening in TB 2 not TB 1.

In TB v1, it is two segregated paths on the cable. That is cheaper ( in terms of complexity and transistor count) and easier to implement. So that is what Intel went with to get it out the door.

In 2013 Anandtech wrote:

"Thunderbolt 2 moves to a new 20Gbps bi-directional channel that can handle both data and/or display."

Again this drawing is more a conceptual than reality. (the colors are more so up vs. down link direction that anything about the subtypes of data being moved. )

There are still 4 physical lanes in TB v2. Same as TB v1. There is logical ( virtual) bounding of the two pairs into one, more multiplexed data stream.

" ... By combining the channels together, Thunderbolt 2 enables two 20Gbps bi-direction channels instead of two sets of 10Gbps channels. ..."
http://www.anandtech.com/show/7049/intel-thunderbolt-2-everything-you-need-to-know

TB v2 has a larger transistor budget (and more modern process tech) so building a more complicated controller is possible at the same price points.

So, next to upping things to 20 Gbps within one channel, it allows to mix PCIe and DP within a given channel.

It isn't "a channel" as much as it is really enabling bonding of two channels into one virtual one. That's why the connectors and wiring is exactly the same and 100% backwards compatible.

Your misdirection side show does nothing to get around the fact that the PCIe switching is basically the same in TB v1 and TB v2.

same anandtech article as immediately above.
" ... Thunderbolt 2/Falcon Ridge still feed off of the same x4 PCIe 2.0 interface as the previous generation designs. Backwards compatibility is also maintained with existing Thunderbolt devices since the underlying architecture doesn't really change. ... "

TB v1 didn't push up to the switche's limits ( basically throttled at 10Gb/s due to the fixed data segration). TB v2 runs is running into a overhead wall ( so pragmatically get x3 while physically connected with x4 physical links). There is some increase in TB v2 but the bigger consumer enabled in 4K DP v1.2 video traffic, not PCIe data traffic.

Two PCIe controllers inside the same peripheral hooked to a single TB controller are not going to get 20 Gb/s. There is a difference between the bandwidth that TB encoded data travels at and what can get on/off the TB network at any one point. 100% at any one node isn't really possible for TB v1 or v2.

 

[manu chao]

You are playing "move the goal posts". In the context in which I was quoted, PCIe data, for TB v1 it is 10 Gb/s. The misconception you seem intent on throwing out there is that these two 10 Gb/s channels of TB v1 are flexible. They could be used either for PCIe or Display Port data.
One of those 10Gb/s paths is dedicated to TB encoded Display Port and one of those 10GB/s paths is dedicated to TB protocol encoded PCIe data.

I am not trying to throw out misconceptions, I am just describing how everybody I've have read so far is describing it. This conventional wisdom might be a misconception but I am not trying to spread something that I know is not true.

I might have missed it, but do you have a link pointing out that with TB 1, one of those two 'channels' had to be DP and the other PCIe? It is also an interesting question what the point of a bi-directional DP channel is? Yes, it could be used for video capture but then the computer/motherboard would need to be able to receive a DP signal and process it. I have not seen this mentioned anywhere in descriptions of Macs with TB. Thus, this 2x bi-directional channels 'marketing' would be a smokescreen if no computer actually was able to use the reverse DP channel.

In TB v1, it is two segregated paths on the cable. That is cheaper ( in terms of complexity and transistor count) and easier to implement. So that is what Intel went with to get it out the door.

There are still 4 physical lanes in TB v2. Same as TB v1. There is logical ( virtual) bounding of the two pairs into one, more multiplexed data stream.

TB v2 has a larger transistor budget (and more modern process tech) so building a more complicated controller is possible at the same price points.

It isn't "a channel" as much as it is really enabling bonding of two channels into one virtual one. That's why the connectors and wiring is exactly the same and 100% backwards compatible.
Your misdirection side show does nothing to get around the fact that the PCIe switching is basically the same in TB v1 and TB v2.

So, one aspect is that TB 2 can multiplex two physical lanes into one stream and thus this one stream can carry twice as much data? And the second aspect is that TB 2 can multiplex DP and PCIe and thus if needed it can dedicate the full 20 Gbps to either PCIe or DP?

And I object to the misdirection description, I am not trying to misdirect, I am merely explaining my current understanding of TB using sources which informed that understanding. And Anandtech is one of the most thorough general tech sites, which is why I quoted them.

same anandtech article as immediately above.
" ... Thunderbolt 2/Falcon Ridge still feed off of the same x4 PCIe 2.0 interface as the previous generation designs. Backwards compatibility is also maintained with existing Thunderbolt devices since the underlying architecture doesn't really change. ... "
TB v1 didn't push up to the switche's limits ( basically throttled at 10Gb/s due to the fixed data segration). TB v2 runs is running into a overhead wall ( so pragmatically get x3 while physically connected with x4 physical links). There is some increase in TB v2 but the bigger consumer enabled in 4K DP v1.2 video traffic, not PCIe data traffic.

That is good to know.

Two PCIe controllers inside the same peripheral hooked to a single TB controller are not going to get 20 Gb/s.

I don't think I ever implied that.

There is a difference between the bandwidth that TB encoded data travels at and what can get on/off the TB network at any one point. 100% at any one node isn't really possible for TB v1 or v2.

Of course, every protocol has overhead and implementation limitations.