The mystery might not even be completely solved then, have you confirmed that it's not performing some unwanted compression to get your data over a lower-bandwi

The mystery might not even be completely solved then, have you confirmed that it's not performing some unwanted compression to get your data over a lower-bandwidth cable? Maybe a good indication would be to check your CPU usage before and after using 5Gb/s and 10Gb/s cables. Even then it's not a great indication, many modern processors have dedicated modules on the CPU for video compression/decompression (given how large the use case is, it makes sense). And then you have other factors that may prevent you getting full speed: * Connection between the cable and the device on both sides (oxidization, scratches, etc) * Manufacturing flaws or damage during delivery/use in the cable (un-shielded areas, breakages in the cable, imperfections in metals, etc) * Temperature - if you're pulling 20W through this cable it's not entirely impossible you might start noticing some heat, especially if it wasn't designed to take the power draw (increased heat due to power pull = greater resistance = worse signal) My pointy is, as we turn up the requirements of these cables, just making sure they deliver on their promises will become more difficult. Most people won't have an oscilloscope that can even measure these cables anymore and just the act of putting the oscilloscope on the lines increases capacitance and changes how the signal looks. A while back I had issues with two Ethernet cables (CAT 7 I think), both straight out of the packet, one performed at half speed and the other at full. I was tearing my hair out for a while before I realized why the network was bottle necked.