How to profile Swift code

Posted 23 May 2023
Tagged with swift

Stack Overflow coughed up a snippet which I’ve adapted:

let start = DispatchTime.now()
var elapsed = start

func printElapsed(_ label: String) -> Void {
  let now = DispatchTime.now()

  let totalInterval = Double(now.uptimeNanoseconds - start.uptimeNanoseconds) / 1_000_000_000
  let elapsedInterval = Double(now.uptimeNanoseconds - elapsed.uptimeNanoseconds) / 1_000_000_000

  elapsed = DispatchTime.now()

  print("Time to \(label):\n  \(elapsedInterval) seconds (\(totalInterval) total)")
}

e.g. in this profiling:

$ time swift actions/run_action.swift 934BD809-D6D9-4147-AE7E-E1701C82AADD/L0/001 toggle-rejected
Time to waking up:
  1.3583e-05 seconds
Time to get photo:
  0.110328541 seconds
Time to enter switch:
  0.1103705 seconds
Time to toggle needs action:
  0.159904833 seconds
Time to all done:
  0.211530583 seconds

________________________________________________________
Executed in  523.46 millis    fish           external
   usr time  136.72 millis   60.00 micros  136.66 millis
   sys time   52.85 millis  683.00 micros   52.17 millis

it’s pretty obvious the “time to get photo” step is the slow one

The Stack Overflow answer suggests using ContinuousClock; that was in a newer version of Swift than I had available when I wrote this snippet – for now I’m sticking with what I’ve got and which I know works.