The top$()$ action operation in Spark RDD generates the top results in a way that it $...$

A$.$

avoids a full sort and runs only a partial sort of data to return the top results

B$.$

runs the complete sort of data and then gets the final top results.

C$.$

uses a priority queue in processing steps to avoid keeping the whole results in memory. It only keeps the top requested results in memory and avoids a final sort, so we can get the result in O$($n$)$ time.

D$.$

it uses a priority queue and sorts the priority queue at the end of the process.