Scala的期货如何在多核中安排

Question

How does Scala determine when to execute a future immediately, on a different CPU (core), and when to queue it for the current core to free? Or, if it uses a completely different method, please explain: How are Scala Futures scheduled? (Across CPUs, across Threads, across time)

Answer 1

It would be based on the ExecutionContext that the Future uses. When the future runs the task, or invokes the callback it uses the context to decide the policy.

Note that in a typical Scala app these are passed as implicits so you may not always see them. Ex. in Play

import scala.concurrent.ExecutionContext.Implicits.global

That's often used to bring in a default context, which actually comes from Akka.

You can specify your own thread pool like this:

  val ctx = ExecutionContext.fromExecutor(
  Executors.newFixedThreadPool(5))

And use it like this (or an implicit)

Future {do something}(ctx)

So to answer your question the threading policy is simply what you tell it (which is often an implicit made available to you)