I tested it with a Behringer Uphoria UMC202HD sound card and with that 64 sample buffer I get completely stable operation. UFX+ does not work properly with a buffer of less than 512 samples. Is its Class compliant that bad? Or is the device broken?

EDIT:
In addition to tests, 802 devices. The behavior is similar to UFX+. However, I got to a buffer of 256 samples with it. But it's not stable either. It's amazing that Behriger beats RME equipment in this matter. That Behriger is a cheap device under 100€.

So what do you propose as a solution. If I install Windows on this machine? Or is this AMD configuration unsuitable for audio use?
The problem also appears on another machine, based on Intel i9.

As said in another thread:

Maybe then they are intended to work only with Apple devices in CC mode. I don't know any better. But at least this doesn't work properly with Linux, so maybe you're right.

@waedi: CC mode is explained in every manual of devices supporting it. For the UFX+ see chapter 31ff.
In short: allows for driverless operation (iPad, Linux).
Even on Windows, CC mode is supported, but who requires this if you have an RME ASIO driver and TotalMix FX..

@sjzstudio: Sorry to hear about such nasty problems.

I think you know this already, but just to mention it. Only the UFX III supports USB3 CC mode.
The UFX+ only supports CC mode with USB2 ("30ch interface without MADI").

But this doesn't seem to work, for unknown reasons. I would do the following.

1. get a separate USB3 card to isolate the UFX+ from the rest of the USB infrastructure of your PC.
This is always a good basis strategy. Best a card which is known to work very well under Windows,
like the Sonnet USB3 PCIe card with FL1100 chipset.
https://www.sonnettech.com/product/alle … 4port.html
It's a PCIe 2.0 x1 card so it has only very "low" bandwidth requirements in terms of PCIe version and number of PCIe lanes.
Best is using a PCIe socket on your mainboard, where the PCIe lanes are not shared with other PCIe slots.

Maybe this chipset is also beneficial for operation in CC mode and USB2.
Each USB3 chipset has backward compatibility to USB2 and each USB3 plug has dedicared signals/pins for USB2 fully separate from USB3.

[ If this doesn't solve the issue then you need to invest more work. The only solution is to perform a parallel installation under Windows, best on a separate/dedicated SSD for that. ]

Then you have the following possibilities:

2. You can install the ASIO driver and use the driver setting window to check for USB transport / CRC errors

3. You have the usual tools like LatencyMon to validate, whether stable audio operation is possible with the system.
I know, under Windows you have other driver compared to Linux. But sometimes audio issues can also be caused by bad BIOS or main board design. Therefore I think it is interesting to check whether its possible to get it to work under Windows.

If it works well with the Sonnet card under Windows, then try exactly the same setup using Linux, same USB port, same cable, only difference here: CC-mode and USB2.

If everything fails and it works only reliable under Windows, well then you have now a parallel Windows installation in addition to Linux.

USB and Windows always use a port that runs directly via the CPU and not via the chipset. Yes, there are differences o)
UEFI assigns PCI lanes chaotically, after each boot they are distributed differently, so always use the USB port of the CPU for such things.
Linux and UFX+ (with current firmware) are also currently messed up for me, Linux is currently a chaos construction site in this area.
UFX+ now only provides me with inputs under Mint, no outputs, and I also had these noises there, only much more intense. o)

4 PCI4 lanes chipset to CPU . If there is a lot of memory in the PC, it quickly becomes tight o)
These are serial connections, they can't all be used at the same time, so interruptions are needed.
With my AMD B550, the UFX+ definitely works better when connected to the CPU.1 level better latencies jump out there.
My keyboard, for example, does not work properly when connected to USB (chipset) and console. Connected to the CPU, it does its job perfectly everywhere.
Crazy, just IT technology o)))
PCI lanes, the hardware sets the limits, but the allocation to OS and software is dynamic.
MS will also work more cleanly with PCI slots, the lane allocation should be objective there. But with USB AMD chipsets the situation is unfortunately very different, chaos. I suspect that AMD itself will soon no longer be able to see through this. o((

I think under " PCI Info" you can see where what is connected and how...  o))

http://rh-software.com/