Based on theoretical analysis of the complexity of the serial algorithm and the parallel algorithm for matrix$-$vector multiplication, you have derived the following expected serial runtime $($T$\_$serial$)$ and parallel runtime $($T$\_$parallel$)$ for the multiplication on a n$\backslash$times n matrix using p processors:

T$\_$serial$=$n$^2+2$n

T$\_$parallel$=$n$^2/$p$+2$n$+100$log$\_2$ p

The parallel runtime is composed of the parallelized computation $($n$^2/$p$),$ the unparallelized computation $(2$n$),$ and the parallel overhead $(100$log$\_2$ p$).$