Amdahl's Law describes the theoretical maximum potential speedup that can be achieved in a system by using parallel processing.
The law is based on the idea that not all parts of a program or task can be parallelized. In any computation, there are some portions that must be executed sequentially, and only certain portions can be executed in parallel on multiple processors or cores.
Basically what Amdahl's Law is stating is if 50% of the code cannot be parallelized, then parallelizing the code will always leave at least 50% of the execution time. Because 50% of the code execution time is unaffected by parallelization.
See for more information: