Comparison of delete vs Reflect.deleteProperty

benchmark.js

/**
 * Benchmark process:
 *  - generate an array with 100.000 random elements (0 - 999.999)
 *  - in each sampler function, remove first element from array
 * Removing first element might turn out with better performance
 * compared to real world usage.
 *
 *
 * Tested in io.js v3.0.0, babel-node v5.8.21
 * before using, run: npm install benchmark harmony-reflect
 */

'use strict';
var Benchmark = require('benchmark');
var Reflect = require('harmony-reflect');

var array = Array.apply(null, {length: 1e5}).map(function () {
  return Math.floor(Math.random() * 1e6);
});
var arr = [];

new Benchmark.Suite()
  .on('cycle start', function () {
    arr = array.slice(0);
  })
  .add('delete [operator]', function () {
    delete arr[0];
  })
  .add('Reflect#deleteProperty', function () {
    Reflect.deleteProperty(arr, 0);
  })
  .add('Array#splice', function () {
    arr.splice(0, 1);
  })
  .on('start', function () {
    console.log('begin benchmark');
  })
  .on('cycle', function (event) {
    console.log(String(event.target));
  })
  .run({async: true});

results.md

node (io.js) v3.0.0

delete [operator] x 10,035,966 ops/sec ±0.82% (98 runs sampled)
Reflect#deleteProperty x 5,588,987 ops/sec ±0.78% (93 runs sampled)
Array#splice x 9,494,060 ops/sec ±0.76% (91 runs sampled)

node v0.12.7

delete [operator] x 10,184,585 ops/sec ±0.84% (96 runs sampled)
Reflect#deleteProperty x 4,802,106 ops/sec ±0.66% (99 runs sampled)
Array#splice x 7,278,378 ops/sec ±0.77% (94 runs sampled)

babel-node v5.8.21

delete [operator] x 9,879,017 ops/sec ±0.95% (78 runs sampled)
Reflect#deleteProperty x 5,537,061 ops/sec ±0.82% (93 runs sampled)
Array#splice x 9,126,094 ops/sec ±0.61% (91 runs sampled)