Share your experience with the alternative timer

[ADD-SP]

We recently added an alternative timer implementation in Tokio for better multicore scalability, and we’d love to know what you think! If you’ve tried out the new timer, please share your story.

tokio::runtime::Builder::enable_alt_timer

let runtime = Builder::new_multi_thread()
    .enable_alt_timer() // <-- enable the alternative timer
    .build()
    .unwrap();

What can you talk about? Pretty much anything, including:

• Where and how you tried the alternative timer
• What feels different compared to the traditional timer
• Any speed ups, slow downs, or bugs you’ve spotted
• Cool benchmarks, numbers, or logs (or just vibes!)
• Anything that surprised you, confused you, or didn’t work as expected
• Stuff you wish the timer did better or ideas you have

Seriously, any details or thoughts are super helpful. Let us know so we can make Tokio even better for you and everyone else. Thanks for helping out!

---

Reference:

• Root Issue: Reduce contention in timer driver
• RFC: RFC: Delayed cancellation of timer
• Design Document: Desgin document of the alternative timer
• Pull Request: time: add alternative timer for better multicore scalability

[alex]

This is a wonderful Thanksgiving present :-) I'll make a note to figure out how I can test on a larger project next week, but in the immediate term I just want to say thanks for working on performance issues that show up at large scale, its very much appreciated!