JS performance notes

---

NOTE: All of the notes below will use these timing functions:

function time(fn) {
  const start = new Date()
  fn()
  return new Date() - start
}

function timeAsync(fn) {
  const start = new Date()
  await fn()
  return new Date() - start
}

Loops

When working with arrays, always prefer for loops to forEach, map, filter, etc. Plain ol' for loops can be orders of magnitude faster! For example:

function loopWithPlainOldForLoop(x, fn) {
  for (let i = 0; i < x.length; i++) {
    fn(x[i])
  }
}

function loopWithArrayForEach(x, fn) {
  x.forEach(v => fn(v))
}

const x = []

for (let i = 0; i < 100000000; i++) {
  x.push(Math.random())
}

const dubbleFn = v => v * 2
let meanPlainOldForLoopTime = 0
let meanArrayForEachTime = 0

for (let i = 0; i < 100; i++) {
  const t = time(() => loopWithPlainOldForLoop(x, dubbleFn))
  meanPlainOldForLoopTime += t / 100
}

for (let i = 0; i < 100; i++) {
  const t = time(() => loopWithArrayForEach(x, dubbleFn))
  meanArrayForEachTime += t / 100
}

console.log("mean `for` loop time:", meanPlainOldForLoopTime.toFixed(2), "ms")
// mean `for` loop time: 42.43 ms
console.log("mean `Array.forEach` loop time:", meanArrayForEachTime.toFixed(2), "ms")
// mean `Array.forEach` loop time: 1245.44 ms

ASIDE: This makes me wonder if I should try monkey-patching the Array.prototype methods. Hm... It's probably a horrible idea, but it's tempting nonetheless!