07/05/2017, 8:20 AM
@elizarov Hi Roman, I have an easy question for you regarding Why is the coarse-grained confinement strategy so much better than fine-grained here? is this because no switching between coroutine contexts is necessary? Also, I'm wondering whether coroutines executed in a single thread give any benefit in terms of parallelism? is this any different from a sequential computation in a loop? In this example the overload introduced is too expensive I think