João Freitas

The following is an explanation of Amdahl’s Law and how its principles can be applied to achieve efficient parallel programs.

Amdahl’s Law is a formula that helps to estimate the maximum speedup that can be achieved by parallelizing a program. It’s intuitive and practical. The equation is fairly simple:

Speedup = 1 / ((1 - P) + (P / N))


What does it mean? Any improvement by parallelization (e.g., distributing work across multiple computers or processors) is limited by how much of the program can be parallelized.

#reads #matt rickard #theory #parallelization #distributed systems