Overview
Originally from queuing theory, Little's Law is widely used in computer architecture to analyze the performance of pipelines, buffers, and memory systems.
The Formula
L = λW
- L: Average number of items in the system (e.g., instructions in a pipeline).
- λ: Average arrival rate (e.g., instructions per cycle).
- W: Average time an item spends in the system (e.g., pipeline latency).
Application
It helps designers determine how large a buffer or queue needs to be to prevent stalls. For example, to hide a 100-cycle memory latency at a rate of 2 requests per cycle, a system needs a buffer that can hold at least 200 outstanding requests.