Part 2 of the install.

https://docs.rme-audio.com/shared/

It took me a good while figuring out the configuration.
I should have read the (full) manual more closely. I advise everyone to do the same. Because the clocking configuration wasn't working. So I ended up hours trying to figure out what I was doing wrong. Swapping PCI-e slots and all that jazz.

By default the SPF expansion modules workclock is set to output. So my connected wordclock didn't do anything until I used the 'input' option of the configuration settings. DOH! The manual doesn't say that if you have two expansion modules that both become inputs by enabling the input option. But I guess that is assumed.

If you have more than 1 interfaces you can sync them with the additional cable that comes with the interface. If you have 3 cards you can daisy chain that (I think).

As for the Dante connectivity. You can connect both cards to a router or switch. You still need a (in my case dedicated) networking card of your computer to be connected to the same router or switch (or be on the same subnet), It helps to have a router of switch that has support for Dante or AVB in general. (That is another topic in itself.)
You can download the Dante controller software from Audinate paths the IO of your Dante devices. (You need to be registered for that.)

Total mix only can show 256 channels while for ASIO you have 512 available. In the Aox configuration you can disable Totalmix but it doesn't seem to work. (Or I'm doing something wrong)
But if you want to do some routing you can use Totalmix but have to keep in mind you have to to be creative with the setup routing if you want to mix and mingle Madi and Dante.

DIGICheck NG also is limited and cannot monitor all 512 channels. It's limited to 200. RME peeps said this was a known issue and hopefully it will be address. It's really handy to quickly see which inputs are active.

Asio buffersize is default 256. I'm hoping lower sizes also will work but that remains to be seen.

Since with audio interfaces the ASIO settings for your DAWs must be re-configured for IO.
I keep a spreadsheet with all devices and their connections. It really helps to have that at hand when filling the names. It's a luxury problem for sure. You also can rename the channels in Totalmix but that only works for the first 256 channels.

Hi!

Thanks for taking the time to write this up! We're always trying to make our products as robust and resilient as possible. But both, audio networks and computer interfaces, are complex technologies and it's impossible to check every possible combination of hardware in our labs.
We are aware that some PCs have a hard time to successfully flash the card and we are still investigating why that is. As Murphy dictates, it works just fine on all PCs and Macs that we have available. In this regard it's extremely valuable that you have tested two systems where one failed and one worked just fine, this might be the key for us to reproduce this issue and finally resolve it.

As for the warning you are seeing, "Received ctrl data not valid", it might have happened during sampling rate changes. However, there is a checksum along with every command that is sent internally, so non-valid data is just filtered out. We decided to log it anyways in order to provide better support in cases where this happens due to hardware issues or incompatibilities with the system. As written above, we strive to continuously improve our products, and getting rid of those warnings that only happen in transition phases is already on our list.

Best regards
Marc

With the latest update for the new Aox-D firmware and drivers the flash tool does work on my AMD Threadripper Pro system. So that's good news.

Also the AVB messages in the windows logging seem to be gone.

The defaults settings for the audio buffer are set to 1024. But it works fine at 64 for me. I'll try to see what 32 will do.

There is an issue with Wavelab 13 taking 17 minutes to communicated with the RME drivers. But I've informed the Steinberg that I won't be an intermediary between them and RME.

For the Aox-D the WDM devices are 0 since for generic Windows applications I'm using the WDM of the UFXIII which acts as my monitor controller.

I'm hoping that RME and the WL engineers will get together and figure this out. But for a single edge-case that may be not worth their while.

Can you confirm that the time needed to start is shorter with 128 or 256 channel limit (in Settings dialog)? I recall we have had that with Cubase/Nuendo a while ago and it had been fixed with an update from Steinberg.