Concurrent

Concurrent describes systems or components that can progress during overlapping time intervals, allowing multiple tasks to be in flight at once. Concurrency enables responsiveness and efficient resource use by overlapping computation with waiting, I/O, or other work.

Concurrency differs from parallelism. Parallelism means performing multiple tasks simultaneously on multiple processors or cores, while concurrency concerns the ability to manage multiple tasks that progress in overlapping time, which may involve interleaving rather than true parallel execution.

Models of concurrent execution include shared-memory concurrency, where threads or processes access common data and synchronize through locks, mutexes, semaphores, and monitors; and message-passing or actor models, where components communicate by exchanging messages instead of sharing state. Asynchronous I/O and event-driven programming enable concurrency by non-blocking operations.

Common challenges include nondeterministic behavior, data races, deadlocks, livelocks, and starvation. Techniques to address these problems include disciplined synchronization, immutability, atomic operations, higher-level abstractions such as futures and promises, and design patterns like producer–consumer or readers–writers.

Programming ecosystems provide various concurrency constructs. Go offers lightweight goroutines and channels; Java provides the java.util.concurrent framework; C# exposes async/await and the Task Parallel Library; JavaScript uses an event loop with promises; and Erlang emphasizes the actor model for concurrent computation.

Applications of concurrency span servers, user interfaces, real-time systems, and any workload with parallelizable I/O or computation, enabling scalable throughput and responsive software on multi-core devices.