With M1 and Zen3 both offering significant gains in low buffer performance, I think it would be great to see a 44.1/48 16smps option and a 88.2/96k 32smps option.  On my 5950x rig, messing with a presonus quantum, there is very little difference/reduction in performance between 32 samples and 16...

Looking at reported latencies, it seems RME has extra output buffering?  I assume this is at least partly why we get the better performance (I tested and it is better than quantum, especially on Windows).  But, it would be nice to have an option to reduce the output buffers for the super low latency nuts.

I know that in practice, these kind of latency differences are likely trivial, but running a place geared towards freelancers, I find myself accommodating a variety of (ahem) even-less-than-trivial differences.

The 5950x is a bad MF'er in low latency performance...new era of native audio.