Speedup is a way to quantify the advantage of using a parallel algorithm over a sequential algorithm. The speedup $S$ is defined as

$$ S = \frac{R}{P} $$ Where $R$ is the running time of the best available sequential algorithm and $P$ is the running time of the parallel algorithm.

Ideally, on a system with $N$ processors, the speedup for any algorithm would be $N$ Amdahl's law deals with the speedup in more realistic situations.