In practice, we$$ve already seen that we need to do granularity control when the problem size is small.$$The scan algorithm above also uses this idea. You can view it as doing granularity control $$manually$.$ In$$particular$,$ we are splitting the input into n$/$k chunks, each with k elements, and try to process the chunks$$in parallel. Inside each chunk, the elements will be processed sequentially. Answer the following questions.