Cache Locality

Cache locality refers to the degree to which a program's memory access patterns exploit the hierarchical structure of modern computer memory, particularly the cache levels that sit between the processor and main memory. Good locality means that when the CPU requests a piece of data, the adjacent data in memory is likely to be useful soon after — enabling the cache to prefetch and reuse data that has already been fetched, rather than waiting for slow main memory accesses. This principle is the foundation of data-oriented design and the primary reason that entity component systems outperform traditional object-oriented designs in simulation-heavy workloads: flat, contiguous arrays of homogeneous data maximize the probability that the next operation will find its data already in cache.

The significance of cache locality extends beyond performance optimization. It represents a fundamental constraint on software architecture: the physical layout of data in memory shapes the feasible structure of algorithms more profoundly than most programmers acknowledge. An algorithm with elegant asymptotic complexity can be pathologically slow if its memory access pattern resembles random walks through scattered allocations, while a conceptually simpler algorithm that respects the memory hierarchy can outpace it by orders of magnitude.