DVI dual link

Modern graphics chipsets may actually expect the driver to set this stuff up.

It might be as simple as flipping a bit, or it may be more complicated. There might be a
provided firmware controlled part of the chip which does it for you, or there might not. As
far as I remember there was a distinct piece of work for supporting things like dual-link in
the radeonhd driver.

So yes, Linux (or rather, the X driver) does know what sort of transport is being used, and
indeed for some types of hardware it knows the type of connector (this information is in a
firmware lookup table e.g. for the various DVI-like connectors) used, how the user is supposed
to use it (USB's HID standard specifies things like "thumb control" even though obviously a
hardware driver doesn't per se need to know that level of detail) even the color (Intel's HDA
tells you this, though it's not always very accurate)

The monitor definitely knows the difference, a display isn't required to implement dual-link
DVI if it doesn't need it and so they don't. The pins for the second TMDS link can be omitted
from the design altogether.