granularity

Granularity designates the level of detail at which a system is observed or represented. It is often described in terms of the size of the smallest distinguishable unit. A high-granularity representation has many fine units; a low-granularity representation uses coarser aggregates.

In data management and analytics, granularity describes the fineness of data items. High granularity means detailed records (for example, every transaction or event); low granularity means summarized values (daily totals, rollups). Granularity affects storage requirements, query performance, and analytical flexibility; finer granularity enables more precise analysis but requires more space and processing.

In computing, granularity refers to how a task is decomposed. Fine-grained parallelism exposes many small tasks with frequent synchronization, which can improve load balance but adds communication overhead. Coarse-grained parallelism uses larger tasks with less coordination, reducing overhead but potentially limiting scalability.

In measurement and modeling, granularity is related to resolution. It can be distinguished from precision: resolution is the smallest unit that can be discerned, while precision is the degree of exactness. In simulations and digital media, higher spatial or temporal granularity yields finer detail but larger data sizes and processing costs.

In other domains, granularity applies to imaging, geographic information systems, and database locking strategies, where spatial, temporal, or item-level detail influences performance and usability.

Choosing an appropriate granularity involves trade-offs among detail, storage, processing, and interpretability, and should align with the goals of a system or analysis.