Color me skeptical about the Mac Mini vs. other computing devices thing, as well as the usb cable, but I enough respect for Danny's ability to set that aside here. It's not relevant to the comparison data anyway. The equipment he used is what it was. As a speculative thought as to reasons, does the mini perhaps have discreet graphics? That would keep the graphics core off the pci bus where the sound card resides. Even laptops with discreet graphics will frequently use a chipset that actually has integrated graphics on the northbridge even if they're not in use.
When I heard the PS Audio room at RMAF last year it was really dry. As mentioned above, it did a lot right, and had all the notes in place, but there wasn't enough spacial information. Perhaps it was over-damped in the name of reducing the noise floor or something, but it sort of felt like trying to have a conversation in an anechoic chamber. Just a little unnatural and uncomfortable. Of course I don't know which component's fault it was, but it sounds like Dave is describing something similar.
Just out of curiosity, since I have a special interest in the Onix DAC-25, was the upsampling function used at all, and at what setting? Also, did you try any of these out using their digital (non-usb) inputs fed by some other transport? If so, the S/PDIF input on the Onix should have an advantage due to a Cirrus Logic receiver/buffer. I'm a little curious as to whether that made a difference you could hear - even as compared to the optical input on the same dac in the absence of another dac without that feature. In my own testing it sounded better that way (fed by the S/PDIF outputs on the Onix CD-10 or my RME Fireface 800) than when fed directly by the pc via usb although admittedly my pc, while acoustically very quiet, may have held it back more than your Mac Mini.