What does it mean?
Small blocks create shorter deadlines and can reduce part of the latency. Larger blocks give more processing time but last longer. A stable size depends on the device and rig.
Example256/48000 × 1000 = 5.333 ms per block, not the total round-trip latency.