It's the same USB interface (E-MU ,the ones DIYAudio uses) loopback with the exact same settings through 2 different PC's.
No voltage dividers,same cabling,same everything.
Of course there are analog steps between but its the same ADC-DAC measured a million times with the same repeatable results to the last db (you can see it on Multitone thread,hundreds of measurements).
Only PC changes and noise floor with it.
Both connected to power grid,not laptops.
The three possibilities I see:
1 - You have a ground loop but with very low coupling of noise, different between the two PC's. I find this unlikely. Ground loops normally couple specific frequencies, such as mains frequency, or the frequencies associated with switching power supplies or graphics hardware, from the magnetic fields these noise sources create. These would show up as spikes in your plots rather than a shift in the wideband noise floor.
2 - There is wideband noise coming from the USB connection (possibly through the power connections), that is coupling into the analogue electronics of the DAC.
3 - The different motherboards output slightly different power supply voltages. The dac noise floor level is influenced a tiny amount by power supply voltage.
Possibly other mechanisms but I can't think of any right now.
I think the important point is neither of your PC's are worsening the noise floor in any audible way. I don't know if either have an "audiophile" motherboard - but even if one were used, it would make no audible improvement, in exactly the same way there is no audible improvement going from a Dac that can achieve 110dB Sinad to one with 120dB.
This just confirms the basis for my statement "So no, audiophile motherboards will do nothing for USB connections"