Although USB is in the market for very long there are here and there still issues in the implementation of USB in "communication chips" and computer "chipsets" in terms of reliable audio transfers.

On the recording interface side everything is under RME's control because USB communication is being performed by the FGPA and can be flashed with a new firmware in case of any error.

I am really waiting for RME to produce their own USB3 and USB2 compatible FPGA based PCIe card to have things even more under their control also the computer driver. The code for USB and PCIe communication has already been written, is already part of RME products.

It would IMHO even not hurt if only those USB transfer modes and features would be implemented, that are required to reliably operate an RME interface (and nothing else) to make things not too complicated...

Rationale: even if e.g. some USB3 cards/chipsets have been tested by RME to work with certain RME products like UFX+, then this can only be regarded as a snapshot,  as chipsets and drivers are being updated and this is something which is out of control of RME/customer and especially driver updates are something which can happen quite often and can cause issues.

I personally think that this would definitively be another milestone to achieve even higher quality and predictability of audio transport, if communication up to the PCIe socket of a computer is under RME's hood.

Especially for studios / professionals this could bring more stability directly after purchase and in case of driver updates because then its RMEs own HW, firmware and driver. As studio time is a precious resource, but at the end everybody would benefit from it.

Maybe this could even be the foundation for RME's own more advanced transfer protocols optimized for stability and realtime demands of audio transport.

Using standard hardware and communication protocols like USB and Firewire are needed for sure for interoperability and customer acceptance.

My idea goes one step further which might/should guarantee hassle free communication plus even better scalability in terms of the number of channels (if the power of the FPGA would allow for a little headroom).

After that there can only issues remain that the mainboard is complete crap in terms of interrupt handling / pcie communication which I would regard as less likely.

Another advantage might be that RME could maybe tweak then it's own driver to run tad a bit longer to be able to work on the audio workload more efficient and maybe to be able to compensate by that the bad behaviour of other drivers (nVidia, network, ...) doing the same.

Just a few thoughts, hope you like them..