Benefits of cooperative multitasking

[vns]

@isaac-weisberg

Is there any way to prove that there are cases where we will see a performance increase from migrating from GCD to Swift concurrency?

If you have a carefully well-designed GCD code, I wouldn't expect migrating to Swift Concurrency increase performance just by this fact. The major benefits of migration I see is

1. Cleaner code.
2. Static compiler checks.

This can unveil some issues you might be not aware of and make it simpler to reason about the code. This can end-up with some performance gains, especially if during migration you review design aspects of the solution, maybe got some blocking code to go away, etc. – but, at least to my knowledge, there is no guarantee that you actually will have determined performance impact.

@Panajev

Cooperative vs preemptive if we are talking about OS managed threads or worse processes sure, but I think when people think cooperative these days what they picture is allocating a fixed number of threads based on the CPU cores available (virtual or physical where it matters) and each thread managing a series of jobs cooperatively (software manager scheduling).

Yeah, that's the question of what we compare to what. I just understood initial question in terms of Swift Concurrency vs (largely) GCD. Yet on a larger scale there are also questions to consider – like the one discussed in this thread a while ago.