What is amdahl's law?

Amdahl's Law is a formula used in computer architecture. It calculates the maximum potential speedup you can achieve when improving a part of a system. The key idea is that the overall speedup is limited by the fraction of the task that cannot be parallelized. For example, if 10% of a program is serial, you cannot speed it up beyond 10x, no matter how many processors you use. It highlights the importance of optimizing the sequential parts of any code for better performance gains.

Amdahl's Law is a principle that sets a realistic limit on how much faster a task can run when you add more processors. It’s a fundamental concept in parallel computing.

Think of it this way. Imagine you're painting a house. The job has two parts: painting the walls and waiting for the primer to dry.

You can speed up the wall painting by hiring more painters. One painter takes a long time. Four painters can do it in a quarter of the time. This is the "parallel" part of the job.

However, the primer needs two hours to dry. This is a fixed amount of time. You can't make it dry faster by hiring a hundred painters. This is the "sequential" part of the job.

Amdahl's Law mathematically describes this relationship. It shows that the total speedup you can achieve is always limited by the sequential portion of the task. If 10% of your program absolutely must run in a sequence on a single core, then even with an infinite number of processors, you can never achieve more than a 10x speedup.

In a nutshell, it's a critical reminder for developers. The true performance gains from multi-core processors depend entirely on how much of the software's workload can actually be split up and run in parallel. The sequential bottleneck always defines your speed limit.