Hacker News Reader

Would it be that difficult to just queue up those threads on a single 32 bit processor instead of sending them off to separate cores?

It wouldn't be optimal, but at least they wouldn't have to completely write off older or lower-powered devices.

View on HN

Distributing threads and processes across available cores is the job of the OS, not the application.

In Rust, (right now) you can choose between a 1:1 or a N:M mapping between OS threads and Rust tasks. With N:M threading, the runtime necessarily does (some of) that internally.

Incidentally, Servo uses both: we use 1:1 for the parallel layout worker threads and M:N for the script and layout tasks.

Reply on news.ycombinator.com