What is Concurrent Programming?The answer is obvious: executing multiple tasks (or work units) in parallel. While true parallelism doesn’t occur on a single core computer (instead, the CPU switches between tasks), on systems with multi-CPUs (and multi-cores in each CPU), true parallelism is achieved. Nor is parallel computing confined to utilizing the cores on one physical machine. Distributed computing is a form of parallel computing, in which work units are distributed across numerous machines. However, distributed computing adds additional requirements to task management (namely task distribution), and is not discussed here.

The issues of task communication are more broadly categorized as state and memory sharing issues. A typical solution to the synchronization issues involved with shared state and memory is to use locks, monitors, semaphores, and other techniques to block threads from altering state while one single thread makes changes. Synchronization is hard to test, and this is where bugs and performance problems develop. In Erlang (a functional language), synchronization is not even an issue. This is achieved by treating variables as immutable, and by messaging between threads, where the message is a copy of the sender thread’s variables.