I use a FireWire audio interface
* every time I record.
...
* Because USB audio interfaces depend on the CPU to move data across the bus in addition to encoding, etc. they are notorious for generating noise from the Rice Krispies triplets: Snap, Crackle, and Pop...
There is little evidence that USB 3.0 has that problem if use a modern multicore computer that isn't in the Atom/ARM market place. It was not the CPU that was the major issue. It was the lack of bandwidth to ship the low latency isochronous data traffic over.
USB 3.0 doesn't necessarily have that problem. (if force a revert to USB 2.0 circuitry it will.... but then you are strictly running USB 2.0 at that point. )
a. The SuperSpeed bus is a completely different set of wires. So if you plug in some pokey USB 1.1 keyboard and a chatty USB 2.0 web camera it has no impact at all on the SuperSpeed channel. (you might get very minor impact since share same PCI-e connection but that isn't likely to be a significant issue). Seperate sub-controllers run each different set of wires independently for the most part (some minor connection/metadata issues).
b. The SuperSpeed bus is duplex ( so that device and computer talk at same time). Again completely diffferent protocol than USB 2.0 where the computer's hub has to request each device to speak-only-when-spoken to.
If computer doesn't ask audio device for data updates often enough may loose data. Doesn't happen on SuperSpeed.
c. The bandwidth that can be reserved for isoschronous traffic is larger. Audio data (unless bound numerous tracks together ) isn't that much larger now than 3-4 years ago. That means more bandwidth than had before. Meaning less drop outs. Meaning cleaner data.
d. The increase in data traffic doesn't need anywhere near a 1:1 increase in CPU consumption. (in part because of b but also other factors incorporated into the new design. )
The primarily high end audio complaint left is when you stretch FW out to maximum distance ( dozens of meters ). LP will get much more traction there than USB 3.0 will.
The other self inflicted problem is where overload the computer's CPU generating audio (why? DSPs are more effective, but prehaps strapped for cash ) and run some
other software component at 90+% CPU utiliztaion levels while at same time capture some large music data stream. Besides if pressing keys and fiddling trackpad on the computer you are generating USB traffic (and CPU overhead ) too. The duration of the CPU overhead is also a major contributor. Again since faster the duration is shorter ( if isoloate SuperSpeed channel to primarily just audio capture. )
If plug a USB 3.0 capable audio device into a USB 2.0 socket ... then yeah you will get the same old USB 2.0 problems. However, if plug a USB 3.0 device into a USB 3.0 socket you get a new, not saddled with dubious design constraints and not limited by backward compatibiliy network between the two devices. It is brand new game. Trotting out the old USB 2.0 litany of complaints about the new 3.0 is lame.
