Well, after listening to mine for a few days ( darn nice actually) I threw mine on my bench. Now, from the specs, this thing is better than the Focusrite 2i2 I have for my speaker measuring. The cheap IO box is no laboratory instrument! I am running 24 bit/ 96K for these tests using RightMark. Scarlett is in ASIO mode. Quick test with TrueRTA is similar, but not as fine a resolution and it only uses Windows Direct.
Asgard as the DAC, into the line in of the Scarlett, adjusting the gains carefully, I get a THD of the set @ .0020. Lower than the Scarlett in loopback. (.0031) Noise floor is around 105 dB combined which is lower by 5 dB from the loopback. Tells me it is a better DAC than the Focusrite, but darn is their A2D good for a $150 IO box!
So, this sample can clearly resolve more than 16 bits. What it's actual performance is I do not know as it is better than my bench with ONE EXCEPTION. 60, 120, 280 bleed through is @ around 99dB. It is below 110 for the Scarlett in loopback. ( USB powered) I think they should do a little better on this. I can. I have had no issues with USB stability, locking on or any such nuisance. Quite a "tick" on power down if it drops before the amp, a problem I am addressing elsewhere. So, unlike my experience with the Topping E30 and SMSL Scanscrit , I am keeping this bugger. Yea, it does run quite hot. Is it as good as separate Modi-3 and Asgard for the same price? Can't say.
DAC is audibly better than my old one. I asked my wife ( incredible hearing and sensitivity to distortion) if it was "OK" and she commented " better". If I can crank up trumpets to live levels and she does not run from the room, I got it done.