@AudioNarwhal: returning to this thread after quite a while, so insanely GOOD to have you here! I'll jump in, first with my current status (sorry for the long post), then some questions in a subsequent post...
=== Current status:
== @ home:
motu: 8M + 16A combo (AVB connected), currently running firmware 1.3.5+1434 on both (very briefly tried 1.4*, quickly went back as it seemed to have more problems but did not test for long). Each one is connected (USB2) to a computer running Linux.
distro: Fedora 36
kernel: currently 5.19.12 running PREEMPT=full - Fedora kernel built from .src.rpm with added patch so that it does not complain about duplicated endpoints
driver: drumfix usb avb driver (current git), loaded as a dkms module with samplerate=48000, midi=0, vendor=1
Results:
Running at @48KHz with a 256 sample buffer, jackd v2, 64 I/O channels over USB2 in so called "vendor mode" (using the motu endpoints, not the class compliant ones). The main software (four threads of SuperCollider supernova server audio processing) is running in four "isolated" cores of the eight available (small fanless Ryzen 8 core/16 thread computer based on an Asus PN51 nuc), other jack software in other threads (speaker processor, ambisonics decoder, etc).
After a reboot, or when coming back from sleep, I run the magic curl code that turns the Motu audio interface into vendor=1 mode (from the class compliant mode it switches to when the Linux driver connects).
Once I get the jackd server started it runs fine. Initial startup is tricky. The input channels usually start with an 8xn offset from where they should be, I think output channels are normally fine (I check inputs only initially). Restarting jackd several times eventually gets them right, and after that I do not see channel hopping (at least it has never happened while I have been performing with the system). Very very occasionally I have seen weird distortion (high pitched periodic noise) which is one of the failure modes I have seen. Again, never seen that while performing (lucky, I guess). I perform at home (through the internet), or take the whole system to a venue where I interface using AVB to the speaker array.
== @ ccrma (Stanford University)
This is complicated. I designed/maintain two big systems that use Motu hardware
== Stage: this is in our small concert hall and drives a quite complex 56.8 speaker array (in a full dome over the audience area). We have 9 Motu audio interfaces running the show and a computer that boots into the working system. The selection of Motu relied on being able to run 64 channels over USB using the class compliant driver in Linux, which appeared at some point with a firmware upgrade (tested and worked fine). I came to regret this as that option was taken away in a subsequent firmware upgrade.
The two audio interfaces that connect to Linux computers (main control computer and regular desktop workstation for user interaction) are running very ancient firmware (1.2.8, I would have to check, usually 16A hardware), class compliant drivers, and run at 48KHz/44.1KHz, 64 channels, 256 or 128 sample buffers. The other audio interfaces, including the ones where users can connect to the system through USB are mostly upgraded to the latest firmware (usually Mac laptops). Performance is "adequate". Sometimes the whole system can get into a bad state (symptom: periodic high frequency buzzing in some of the outputs, loud clicks sometimes), and the only way out is a reboot of the Motu hardware - sometimes a couple of interfaces is enough if it is localized, most times all Motu hardware, including the AVB switches, needs to be restarted. Depending on the problems sometimes a reboot of the control computer is enough.
== Listening Room: a specialized full 3d (speakers below a grid floor) 22.8 system in a medium sized studio. Migrated from an RME system to make it easier to connect user's computers (only USB needed with the proper Motu drivers, same as in the Stage above). Same characteristics as the Stage system, only four Motu audio interfaces. Same issues.
Plus:
== GRAIL: this is a "portable" speaker array we take to other venues to play concerts. Depending on the particular concert we can deploy up to 32 speakers with 8 subs. We use a mix of Motu audio interfaces, number depends on requirements (at least 6 in most cases). Startup of the system is "manual" and we still rely, as in the other systems, on Motu hardware that can run firmware 1.2.8 for the interfaces that connect to Linux computers (16A hardware)...
Long term very problematic, as new Motu hardware cannot be downgraded to 1.2.8...
So....
I have been using my home system as the "testing ground" for a solution that does not rely on the class compliant driver AND can run 64 channels over USB. Drumfix's solution comes quite close and I am using it all the time, but it is not something that I could migrate to at ccrma due to the startup problems I encounter. I guess I could find a way to script the startup process (ie: start, test loopback, retry if channels are wrong) but it would be a hack...
One more datapoint:
I encountered this in our Listening Room system. A user with a Mac laptop, updated Motu drivers, and bad problems with channel hopping and noises of all kinds. I had never witnessed this with someone running OSX. Other users (different computers) did not report problems. The exact same hardware had been running for a long time. Replacing the USB cables fixed it! So, this points, as outlined before in the thread, to problems in Motu's firmware where it cannot recover from USB error conditions on its own (ie: problems are not only experienced in Linux).