"An important software design concept, especially in relation to components. The amount of detail or functionality - from fine to coarse - provided within a service component. One software component can do something quite simple, such as calculate a square root. Or, it can have a great deal of detail and functionality to represent a complex business rule or workflow. The first is fine grained, and the second coarse grained. Developers often aggregate fine-grained services into coarse-grained services to create a business service." (Judith Hurwitz et al, "Service Oriented Architecture For Dummies" 2nd Ed., 2009)
"Loosely defined as the amount of computation done before synchronization is needed. The longer the time between synchronizations, the coarser the granularity will be. Fine-grained parallelism runs the danger of not having enough work assigned to threads to overcome the overhead costs of using threads. Adding more threads, when the amount of computation doesn’t change, only exacerbates the problem. Coarse-grained parallelism has lower overhead costs and tends to be more readily scalable to an increase in the number of threads." (Clay Breshears, "The Art of Concurrency", 2009)
"A description, from "coarse" to "fine", of a computer activity or feature (such as screen resolution, searching and sorting, or time slice allocation) in terms of the size of the units it handles (pixels, sets of data, or time slices). The larger the pieces, the coarser the granularity." (Microsoft, "SQL Server 2012 Glossary", 2012)
"The amount of decomposition applied to the parallelization of an algorithm, and the grain size of that decomposition. If the granularity is too coarse, there are not enough parallel tasks to effectively make use of all parallel hardware units and hide latency. If the granularity is too fine, there are too many parallel tasks and overhead may dominate the computation." (Michael McCool et al, "Structured Parallel Programming", 2012)