Use the usual matrix multiplication algorithm, but perform every row-column product in parallel using $n$ processors. This product is actually just a summation of $n$ terms; hence, we can use the parallel summation algorithm described in the book.