davidcai1111 / es6-benchmark

:speak_no_evil:A benchmark for some popular ES6 features in Node.js ( V8 / Chakra )

ES6-benchmark

A benchmark for some popular ES6 features in Node.js 5.10.0 and Node-chakracore 6.0.0-pre5.

Run Benchmark

npm run benchmark

Env

• CPU: Intel Core(TM) i5-2410M 2.30GHz
• Memory: 8GB 1600 MHz DDR3
• Node.js: 5.10.0 / Node-chakracore 6.0.0-pre5

Contributor

• Brooooooklyn
• CarterLi
• jucrouzet

Benchmark

concat-strings.js

V8:
template string vs use +
  1,524,602 op/s » template string
  1,391,110 op/s » use +

Chakra:
template string vs use +
  2,316,501 op/s » template string
  2,277,366 op/s » use +

for-of-for-loop.js

V8:
for...of vs forEach
  260,761 op/s    » for...of
  210,823 op/s    » Array#forEach

Chakra:
for...of vs forEach
  276,133,193 op/s  » for...of
  695,715,320 op/s  » Array#forEach

merge-objects.js

V8:
merge objects
    314,921 op/s  » Object.assign
  3,854,182 op/s  » for...in loop and assign

Chakra:
merge objects
  2,263,889 op/s  » Object.assign
  1,647,837 op/s  » for...in loop and assign

declear-class.js

V8:
declear a class
  47,864 op/s  » class statement
  42,662 op/s  » use function and prototype

Chakra:
declear a class
  191,705 op/s » class statement
  201,991 op/s » use function and prototype

repeat-string.js

V8:
String#repeat() vs polyfill
  7,774,842 op/s  » String#repeat
    222,137 op/s  » use Array#join trick

Chakra:
String#repeat() vs polyfill
  1,674,259 op/s  » String#repeat
   169,631 op/s   » use Array#join trick

spread-operator

V8:
spread operator
   1,042,842 op/s » noop(...[1,2,3])
  12,241,137 op/s » noop.apply(null, [1,2,3])

Chakra:
spread operator
  3,409,259 op/s  » noop(...[1,2,3])
  5,406,631 op/s  » noop.apply(null, [1,2,3])

array-like-to-array.js

V8:
array like object to array
  1,453,649 op/s » Array.from
    562,458 op/s » Array.prototype.slice.call

Chakra:
array like object to array
  1,994,649 op/s » Array.from
  2,897,458 op/s » Array.prototype.slice.call

promise-bluebird.js

promise vs bluebird
V8:
    322,534 op/s » promise
  1,763,186 op/s » bluebird

Chakra:
   93,534 op/s » promise
  205,186 op/s » bluebird

var-let-const.js

V8:
var let const
  134,028,614 op/s » let
  132,193,000 op/s » const
  438,460,321 op/s » var

Chakra:
var let const
  168,028,614 op/s » let
  171,193,000 op/s » const
  153,460,321 op/s » var

string-start-with.js

V8:
string starts with
  10,774,987 op/s » string.startsWith(value)
  15,127,611 op/s » use regexp

Chakra:
string starts with
  30,774,987 op/s » string.startsWith(value)
  12,127,611 op/s » use regexp

define-a-funciton-with-this.js

V8:
define a function with inherited this
  28,661,143 op/s » function statement with self = this
     479,220 op/s » function statement with bind
  26,874,220 op/s » arrow function () =>

Chakra:
define a function with inherited this
  62,661,143 op/s » function statement with self = this
   6,874,220 op/s » function statement with bind
  61,237,253 op/s » arrow function () =>

parse-int.js

V8:
global.parseInt() vs Number.parseInt()
  53,940,634 op/s  » Number.parseInt()
  81,509,873 op/s  » global.parseInt()

Chakra:
global.parseInt() vs Number.parseInt()
  23,940,634 op/s  » Number.parseInt()
  24,509,873 op/s  » global.parseInt()

es6-benchmark's Issues

We should not only bench them on node, but also on real browsers

As mentioned before, they may have different performance.
The best way is using jsperf

Several cases done by me:

1. http://jsperf.com/merging-objects-vs-for-in
2. http://jsperf.com/string-startswith-vs-regexp-test ES6 wins on FF & Safari
3. http://jsperf.com/native-promise-vs-bluebird
4. http://jsperf.com/template-string-vs-use-add-v2 ES6 wins on all 3 browsers
5. http://jsperf.com/declare-a-class ES6 wins on FF
6. http://jsperf.com/spread-operator-vs-apply

FYI, very diff. metrics on let/const/var

I have some very different metrics on let/var/const :

$ node -v 
v4.2.2
$ grep "model name" /proc/cpuinfo
model name  : Intel(R) Core(TM) i5-5300U CPU @ 2.30GHz
model name  : Intel(R) Core(TM) i5-5300U CPU @ 2.30GHz

[...]
                      var let const
      45,260,944 op/s » var
      46,583,073 op/s » let
      46,108,278 op/s » const