Is there a performance difference between 'let' and 'var' in JavaScript

Question

The difference between these two keywords in terms of scoping has already been thoroughly discussed here, but I was wondering if there is any kind of performance difference

Answer 1

var: Declare a variable, value initialization optional. Let is faster in outside scope.

let: Declare a local variable with block scope. Let is a little bit slow in inside loop.

Ex:

var a;
a = 1;
a = 2; //re-intilize possibe
var a = 3; //re-declare
console.log(a); //3

let b;
b = 5;
b = 6; //re-intilize possibe
// let b = 7; //re-declare not possible
console.log(b);

Answer 2

After testing this on http://jsperf.com, I got the following results: jsperf has been down for a while; see the replacing code below.

To check this, I'll use the following performance test based on this answer, which led me to write this function:

/**
 * Finds the performance for a given function
 * function fn the function to be executed
 * int n the amount of times to repeat
 * return array [time for n iterations, average execution frequency (executions per second)]
 */
function getPerf(fn, n) {
  var t0, t1;
  t0 = performance.now();
  for (var i = 0; i < n; i++) {
    fn(i)
  }
  t1 = performance.now();
  return [parseFloat((t1 - t0).toFixed(3)), parseFloat((repeat * 1000 / (t1 - t0)).toFixed(3))];
}

var repeat = 100000000;
var msg = '';

//-------inside a scope------------
var letperf1 = getPerf(function(i) {
  if (true) {
    let a = i;
  }
}, repeat);
msg += '<code>let</code> inside an if() takes ' + letperf1[0] + ' ms for ' + repeat + ' iterations (' + letperf1[1] + ' per sec).<br>'

var varperf1 = getPerf(function(i) {
  if (true) {
    var a = i;
  }
}, repeat);
msg += '<code>var</code> inside an if() takes ' + varperf1[0] + ' ms for ' + repeat + ' iterations (' + varperf1[1] + ' per sec).<br>'

//-------outside a scope-----------

var letperf2 = getPerf(function(i) {
  if (true) {}
  let a = i;
}, repeat);
msg += '<code>let</code> outside an if() takes ' + letperf2[0] + ' ms for ' + repeat + ' iterations (' + letperf2[1] + ' per sec).<br>'

var varperf2 = getPerf(function(i) {
  if (true) {}
  var a = i;
}, repeat);
msg += '<code>var</code> outside an if() takes ' + varperf1[0] + ' ms for ' + repeat + ' iterations (' + varperf1[1] + ' per sec).<br>'

document.getElementById('out').innerHTML = msg

<output id="out" style="font-family: monospace;white-space: pre-wrap;"></output>

After testing this in Chrome and Firefox, this shows that let is faster than var, but only when inside a different scope than the main scope of a function. In the main scope, var and let are roughly identical in performance. In IE11 and MS Edge, let and var are roughly equal in performance in both cases.

Press the big blue button to see for yourself in your favourite browser.

Currently let has support from only newer browsers, but older browsers are still being used relatively much, which would be a reason to generally not use it yet. If you want to use it somewhere where older browsers wouldn't function otherwise, there should be no problem with it.

Edit: revamped answer since jsperf is not working (see revision history for old version).

Answer 3

Inside loops let is significantly slower see: https://jsperf.com/let-vs-var-loop

838,602 ±0.77% 61% slower

(function() {

  "use strict";
  var a=0;
  for(let i=0;i<100;i++) {
    a+=i;
  }
})();

vs.

2,136,387 ±1.09% fastest

(function() {

  "use strict";
  var a=0;
  for(var i=0;i<100;i++) {
    a+=i;
  }
})();

This is because when using let, for every loop iteration the variable is scoped. example:

for (let i = 0; i < 10 ; i++) {
    setTimeout(function() { console.log(i); }, 100 * i);
}

yields to

0,1,2,3,4,5,6,7,8,9

using var yields to

10,10,10,10,10,10,10,10,10,10

If you want to have the same result, but using var you have to use an IIFE:

for (var i = 0; i < 10; i++) {
  // capture the current state of 'i'
  // by invoking a function with its current value
  (function(i) {
    setTimeout(function() { console.log(i); }, 100 * i);
  })(i);
}

which on the other hand is significantly slower than using let.

Answer 4

$ node --version
v6.0.0
$ node
> timeit = (times, func) => {
     let start = (new Date()).getTime();
     for (let i = 0; i < times; i++) {
       func();
     };
     return (new Date()).getTime() - start;
   };
[Function]
> timeit(1000000, () => {
     let sum = 0;  // <-- here's LET
     for (let i = 0; i < 1000; i++) {
       sum += i;
       if (sum > 1000000) { sum = 0; }
     }
     return sum;
   })
12144
> timeit(1000000, () => {
     var sum = 0;  // <-- here's VAR
     for (let i = 0; i < 1000; i++) {
       sum += i;
       if (sum > 1000000) { sum = 0; }
     }
     return sum;
   })
2459

Same scope (function), same code, 5 times difference. Similar results in chrome 49.0.2623.75.

Answer 5

FYI; After Chrome v60, no further regressions have cropped up. var and let are neck and neck, with var only ever winning by less than 1%. Real world scenarios sometimes give var an advantage due to hoisting and re-use, but at that point you're comparing apples to oranges, as let is intended to allow you to avoid that behavior because the semantics are different.

Benchmark. Firefox, IE and Edge like let just fine.