github 地址：https://github.com/alsotang/fast-js 大家如果有其他的 benchmark 要贡献的话，欢迎提 issue 或者在此留言给我。
fast-js
:heart_eyes: Writing Fast JavaScript
env
- platform: OS X 10.10.4
- cpu: 1.4 GHz Intel Core i5
- iojs: v2.3.0
- v8: 4.2.77.20
benchmark
arguments_to_array
[].slice.apply ................................. 3,846,459 op/s
[].slice.call .................................. 3,718,698 op/s
Array.prototype.slice.apply .................... 3,240,713 op/s
Array.prototype.slice.call ..................... 4,390,122 op/s
lodash.toArray ................................. 13,271,549 op/s
clone_object
JSON.parse(JSON.stringify) ..................... 192,176 op/s
_.cloneDeep .................................... 151,387 op/s
_.clone. this is shadow clone .................. 918,130 op/s
for_loop
normal for loop. i < arr.length ................ 7,143 op/s
normal for loop. cache arr.length .............. 7,070 op/s
native forEach ................................. 145 op/s
lodash.forEach ................................. 656 op/s
hidden_class
withoutHiddenClass ............................. 72,448,120 op/s
withHiddenClass ................................ 130,504,857 op/s
inner_function
inner .......................................... 23,634,031 op/s
outter ......................................... 94,173,853 op/s
iterate_object
for .. in .. ................................... 19,144,156 op/s
Object.keys .................................... 1,858,462 op/s
lodash.forEach ................................. 863,574 op/s
map_loop
normal loop. use push .......................... 796 op/s
normal loop. use index ......................... 1,223 op/s
new Array(arr.length) .......................... 3,539 op/s
native map ..................................... 132 op/s
lodash.forEach ................................. 611 op/s
new_array
new Array() .................................... 58,084,959 op/s
[] ............................................. 162,278,430 op/s
next_tick
process.nextTick ............................... 2,315 op/s
setTimeout(0) .................................. 1,938 op/s
setImmediate ................................... 1,349 op/s
start_with
regex /^ab/ .................................... 6,753,906 op/s
indexOf === 0 .................................. 12,967,224 op/s
lodash.startsWith .............................. 14,404,678 op/s
str_concat
+ .............................................. 609,026,862 op/s
+= ............................................. 121,310,295 op/s
arr.join("") ................................... 3,004,308 op/s
str.concat ..................................... 31,938,318 op/s
str_to_int_number
+str ........................................... 18,200,137 op/s
Number(str) .................................... 35,489,770 op/s
parseInt(str) .................................. 42,522,640 op/s
parseInt(str, 10) .............................. 63,504,924 op/s
~~str .......................................... 17,663,670 op/s
str | 0 ........................................ 19,359,115 op/s
str >> 0 ....................................... 20,045,162 op/s
str >>> 0 ...................................... 18,191,777 op/s
str * 1 ........................................ 20,565,845 op/s
contribute
- add your test to
benchmarkdir - run
$ make buildand it will update README.md including new test
@alsotang https://github.com/airbnb/javascript/tree/master/es5#strings 我是看到这里这么说的，不过他说针对ES5
When programmatically building up a string, use Array#join instead of string concatenation. Mostly for IE: jsPerf.
难道是考虑到IE的原因 =。=
str_concat.js
var c = '';
c += longstra;
c += longstrb;
//改成下面试试呢
var c = longstra;
c += longstrb;
我以为你要帖的是这个： https://github.com/codemix/fast.js
光 Benchmark ，最好还要补上说明为什么快，不然人云亦云，流出一些莫名其妙的性能优化方法，害处不小，比如我就听人说过 while 比 for循环快。Benchmark 方法也最好严谨些，两个不一样的东西不便放在一起比较，+str 、 str | 0 、 str * 1 这三者一样吗？
@JexCheng 现在只保留这几种了
+str ........................................... 23,294,152 op/s
~~str .......................................... 24,858,884 op/s
Number(str) .................................... 46,948,016 op/s
parseInt(str) .................................. 41,145,945 op/s
parseInt(str, 10) .............................. 61,879,685 op/s
@alsotang
+str 、 str *1 倒是一样的，都是 ToNumber，支持16进制，支持小数、当然也支持 +“Infinity”（所以parseInt当然不支持这些值，应该拿parseFloat来比，但parseFloat又不支持16进制） 。 str|0 不但只支持整数，而且是只支持ToInt32，大了会overflow的。同样 ~~str 这些位操作都是只支持 Int32，~~"2147483648" 就溢出了
@JexCheng 了解的，前不久刚优化过一个涉及 js 中 int32 算 hash 的库。https://github.com/node-modules/node-murmurhash/pull/10/files
但确实这都是常见用法，比较一下的意义还是有的。就是注释太少哈哈
