I would say the performance are equivalent, with maybe some overhead if there is some heavy thread spawning. I'll run some benchmarks at some point. One of the

I would say the performance are equivalent, with maybe some overhead if there is some heavy thread spawning. I'll run some benchmarks at some point. One of the biggest advantages is the fact this is not multi-threading, so context switches can only happen at precise points (yield from). This makes it easier to reason about the code.