what the ADAU1701 processing is based on? sample or block?

if the processing is based on block , then how many are the less samples grouping one block? thanks!