Amdahl's Law Calculator

Amdahl's law shows how the serial part of a program limits parallel speedup.

How the Math Works

Amdahl's Law reveals the theoretical maximum speedup when parallelizing a task, calculated as speedup = 1 / ((1 - p) + p / s). Here, 'p' represents the fraction of the program that can be parallelized, and 's' is the speedup factor of the parallel portion. The formula works by recognizing that the unparallelizable portion (1-p) limits overall improvement, while the parallel portion's speedup is constrained by how much faster those parts become. For example, if 80% of a task can be parallelized with 4x speedup, the maximum improvement is only 2.5x - not 4x - because the remaining 20% still runs sequentially.

Practical Applications

When optimizing code or designing systems, use this calculator to determine if parallelization efforts are worthwhile. Enter the parallelizable fraction (p) and expected parallel speedup (s) to see the diminishing returns you'll actually achieve. If your parallel portion only yields 2x speedup and represents 50% of the workload, you'll only get 2x total improvement - helping you decide whether to invest in more cores or optimize the serial portion instead.

Day-to-Day Use

This principle helps you make smarter decisions about upgrading technology and managing time. When deciding between a faster computer or learning a more efficient workflow, Amdahl's Law shows that improving the slow parts often matters more than speeding up the fast parts. It explains why adding more browser tabs slows your computer, why some apps feel sluggish despite powerful hardware, and why breaking big projects into smaller parallel tasks can dramatically reduce completion time.

Worked example

90% parallel on 8 cores gives about 4.7x; the ceiling is 10x.

FAQ