I don't think you quite understand what's going on here.
I'm going to have to respectably say it is you who doesn't understand.
What you call a "pair" is really "not that". But I feel like my words will not convince you, so here it is from someone else:
https://9to5mac.com/2016/06/04/lg-27ud88-w-usb-c-4k-monitor-macbook-review/
Pair. Two wires using differential signaling to transmit a single bit wide data signal.
A USB-C connector has 24 pins. Eight are power and ground, four are for housekeeping, two for the USB2 fallback when the cable is right side up, and two for the USB2 fallback when the cable is upside down. The remaining eight are grouped into four pairs. These are the interesting pins. When used for USB3, one of the four pairs is used for transmit, one for receive, and the remaining two are unused. Which two pairs are used and which two are unused depends on the cable orientation.
What you are suggesting is that USB-C connects to separate USB controller and DisplayPort controller with separate "data lanes" (or "pair" as you put it), but that's not the case here.
The data lanes I am speaking about are on the USB-C connector, not internal signals on the logic board.
(Of course, in today's computers many of the internal signals also use differential signaling on pairs of wires. But this doesn't matter in our discussion of the USB-C connector.)
The fact is that Intel Core M systems simply have just one USB-C controller that manages all of the data lanes all at once, and it allocates the data lanes as necessary depending on what "mode" it has it on. It's not a "big knife switch". It's just "one big knife".
The Core M chip knows nothing about USB-C. The chip has a USB2 controller with multiple ports. It also has a USB3 controller with multiple ports. The USB2 and 3 controllers are entirely separate. It is up to the device driver in the OS to pair them into what the user perceives as a single port with both protocols. The Core M also has DisplayPort outputs. Finally, the Core M also has PCIe lanes, but those don't come into play in the rMB.
In a mass market laptop without USB-C, the USB 2/3 ports would be connected in sets to USB-A connectors. One DisplayPort output would connect to the internal LCD, while another may connect to a DisplayPort connector.
The "knife switch" doesn't really "switch between" anything. It's "one big knife" being used for two different purposes all at once. Otherwise you can't have both USB and DisplayPort at the same time, or otherwise why would USB speeds be limited?
The "knife switch" in the 2016 rMB is a Parade Technologies PS8741A. As iFixit puts it, this is "likely an iteration of the PS8740 USB Type-C Redriving Switch." In the 2015 rMB, Apple used several jelly bean components to implement the same function.
One "side" of the PS8741A connects to the USB-C connector. The other "side" connects to the Core M's USB3 and DisplayPort signals. Under software control, the PS8741A routes the interesting USB-C pins to various combinations of USB3 and DisplayPort, i.e. it is a crossbar switch, euphemistically a knife switch. It likely also handles cable inversion. "Redriving" means that it cleans up the signals to the proper voltage and timing specs so that they don't suffer cumulative degradation.
Presently it appears that the PS8741 supports the following combinations:
* USB3 in the right side up position and two unused lanes
* USB3 in the upside down position and two unused lanes
* USB3 in the right side up position and a half width (two lane) DisplayPort signal
* USB3 in the upside down position and a half width (two lane) DisplayPort signal
* A full width (four lane) DisplayPort signal for a right side up cable
* A full width (four lane) DisplayPort signal for an upside down cable
In all cases, USB2 is also available.
In the 2016 MBP's, instead of the Parade chip, Apple uses an Intel Alpine Ridge chip. This adds the processor PCIe lanes into the mix of signals that can go on USB-C (with some protocol changes.) Interestingly, Alpine Ridge also has an internal USB3.1 gen 2 controller. This allows the MBP's to support gen 2 even though Skylake does not have gen 2!
Alpine Ridge adds these combinations:
* Four lane (two in, two out) ThunderBolt protocol for a right side up cable
* Four lane (two in, two out) ThunderBolt protocol for an upside down cable
ThunderBolt protocol encapsulates PCIe and DisplayPort.
However, for whatever reason (most likely power consumption?), Intel has "blocked" the USB-C controller from letting USB mode use the other data lanes, and DisplayPort (likely HDMI as well) mode has access to all of the lanes.
Once again, Core M knows nothing about USB-C. There is no limiting going on. USB3 is a protocol that uses four wires for data communication, two outbound, and two inbound. In a USB-A plug, it has just these pins plus the original USB2 connections (and an additional ground.) USB-C has more pins available, but USB3 cannot make use of them. Perhaps in the future there will be a USB4 that uses eight wires? Or maybe USB-C ports will start carrying two USB3 ports, but that would have issues with USB2 fallback?
[doublepost=1488029503][/doublepost]
DisplayPort (likely HDMI as well) mode has access to all of the lanes.
The rMB does not support an HDMI "alternate mode". That spec is too new. Instead, the multi media adapter converts DisplayPort into HDMI.
If a future model does support HDMI alternate mode, that will be limited to HDMI 1.4 due to the number of pins available. An active conversion from DisplayPort will still be needed for HDMI 2.
Screen Shot 2017-02-23 at 11.42.14.png
Screen Shot 2017-02-23 at 11.42.26.png
Screen Shot 2017-02-23 at 11.42.37.png
Screen Shot 2017-02-23 at 11.42.44.png
Screen Shot 2017-02-23 at 11.42.51.png
Screen Shot 2017-02-23 at 11.42.57.png
Screen Shot 2017-02-23 at 11.43.07.png
Screen Shot 2017-02-23 at 11.44.46.png
