It is likely when using scheduler, the stack sizes for coroutines don't change often, thus we can reuse the allocated stacks. Similarly the closure for dynamic thread pool is unlikely to change that often.
Ideally if stack sizes never change, the performance should be something similar to using std.heap.MemoryPool for unmanaged stacks.
It is likely when using scheduler, the stack sizes for coroutines don't change often, thus we can reuse the allocated stacks. Similarly the closure for dynamic thread pool is unlikely to change that often.
Ideally if stack sizes never change, the performance should be something similar to using
std.heap.MemoryPoolfor unmanaged stacks.