Concurrency (computer science)

Concurrency is the structuring of multiple tasks so their execution overlaps in time, whether or not they run simultaneously. It concerns coordination and ordering among tasks that may make progress independently.

Thread (computing): A sequence of instructions that can be scheduled independently within a process. Threads let one process organize concurrent flows of execution.
Process (computing): An executing program with its own address space and operating-system resources. Processes are a common unit for organizing independent concurrent work.
Parallel computing: The simultaneous execution of computations on multiple processing units. Concurrency permits overlapping progress without requiring simultaneous execution.
Actor model: A concurrency model in which independent actors communicate through asynchronous messages. It structures concurrent components around isolated state and message exchange.
Amdahl's law: A formula estimating the maximum speedup from parallel execution given a computation's serial fraction. It shows why concurrency cannot accelerate work that remains inherently sequential.
Scheduler (computing): A system component that selects which runnable task receives processing time. Scheduling determines how concurrent tasks share a processor.
Shared memory: A memory region accessible to multiple processes or threads. Shared state enables communication but creates coordination hazards.
Sequential computing: A model in which instructions or tasks execute in a single ordered sequence. It lacks the overlapping task structure that defines concurrency.
Fork–join model: A computation pattern that splits work into concurrent tasks and later combines their results. It organizes parallelizable work into branches and synchronization points.
Linearizability: A correctness condition requiring concurrent operations to appear atomic and ordered in real time. It gives a precise standard for whether concurrent object behavior matches a sequential interface.

Linked from 6 pages