News for hackers. Stuff that matters.
Re:Monotonic and wall clock time in the Go time package by jsw on Friday July 25, 2025 @02:53PM (#44686855)
For simple high-frequency time deltas, I've found this function wins on the perf front:

  //go:linkname nanotime runtime.nanotime
  func nanotime() int64
[ Parent | on HN ]
Re:Monotonic and wall clock time in the Go time package by deepsun on Friday July 25, 2025 @02:41PM (#44686703)
And not a word about fasttime.UnixTimestamp()? I thought that's the recommended way, it literally says:

    // It is faster than time.Now().Unix()
https://pkg.go.dev/github.com/VictoriaMetrics/VictoriaMetric...
[ Parent | on HN ]
Re:Monotonic and wall clock time in the Go time package by kiitos on Friday July 25, 2025 @03:47PM (#44687537)
This package spawns a goroutine that updates an atomic value every second based on a time.Time from a time.Ticker, and its functions load that atomic value. This is not really comparable to anything in package time in any useful way...
[ Parent | on HN ]
Re:Monotonic and wall clock time in the Go time package by deepsun on Friday July 25, 2025 @02:45PM (#44686762)
Wait, I see it's been removed actually [1]. Why, isn't it faster anymore?

[1] https://github.com/VictoriaMetrics/VictoriaMetrics/commit/c2...

[ Parent | on HN ]
Re:Monotonic and wall clock time in the Go time package by noselasd on Friday July 25, 2025 @02:52PM (#44686845)
It's not removed, it's moved to a separate file as it's problematic to use in tests, commit 06c2631 talks about why.
[ Parent | on HN ]
Re:Monotonic and wall clock time in the Go time package by noselasd on Friday July 25, 2025 @02:48PM (#44686784)
It's a bit specialized though, it's just a cached value of a time.NewTicker updating once per second and the article talks about this in the "Schedule Based on Monotonic Time" section
[ Parent | on HN ]