Simple Distinction Between Concurrency and Parallelism

When I first started learning about concurrency and parallelism, I always got a bit confused. Now I'm finally taking the time to properly clarify the relationship between the two.

First, the word for concurrency is "Concurrency". The root "con-" means "together", and the root "current-" means "current", so the overall meaning is "things happening together".

The word for parallelism is "Parallelism". The root "parallel-" means parallel or side by side, so the overall meaning is "things happening in parallel", which can also be said as "things happening simultaneously".

Here's a vivid example mentioned by a Zhihu answerer:

You're eating a meal when the phone rings. You wait until you finish eating before answering. This means you support neither concurrency nor parallelism.

You're eating a meal when the phone rings. You stop to answer the phone, then continue eating after the call. This means you support concurrency.

You're eating a meal when the phone rings. You answer the phone while continuing to eat. This means you support parallelism.

The key to concurrency is having the ability to handle multiple tasks, not necessarily simultaneously. The key to parallelism is having the ability to handle multiple tasks simultaneously. So I think the most crucial point is: whether it's simultaneous.

From this Zhihu answerer, we can actually understand that: parallelism is a special case of concurrency. Concurrency only requires the ability to handle multiple tasks, while parallelism requires the ability to handle multiple tasks simultaneously.

With this in mind, let's clarify the concurrency and parallelism of multi-core and single-core CPUs.

First, for single-core CPUs: A single-core CPU is very clear - it can only do one thing at a time, so it cannot have parallelism, but it may have concurrency (if this CPU supports the ability to handle multiple tasks).

Second, for multi-core CPUs: Multi-core CPUs are composed of multiple single-core CPUs, so handling multiple tasks at the same time becomes effortless, thus having parallelism (having parallelism means definitely having concurrency, as we've already said parallelism is a special case of concurrency).