Yeah, I've had more than a few chipsets or periphs that worked on Windows, and worked on Linux but were.... quirky, especially when dealing with stuff like suspend states etc.

For USB3 in particular, I've found many storage devices or adaptors like to drop out partway through an longer copy process on Linux (like they'll be fine for copying a smaller amount of data, but the controller or device would reset during longer ones). This didn't seem to occur in Windows, but I'm pretty sure the copy process was also slower so guessing it's some sort of buffer or heat quirk that 'nix didn't account for in the more generic driver