Favorite Loop Benchmark

Thanks to a suggestion from Johnathan Snook, here is a summary of the results of the benchmarking I did on my favorite loop:

Method 1 uses for (var i=0; i<test.length; i++)
Method 2 uses for (var i=0, l=test.length; i<l; i++)
Method 3 uses for (var i=0, o; o=test[i]; i++)

1000 Iterations
Method	Firefox 1.5 PC	IE 6.0 PC	Safari
1	0ms	16ms	11ms
2	0ms	0ms	12ms
3	0ms	0ms	10ms

10000 Iterations
Method	Firefox 1.5 PC	IE 6.0 PC	Safari
1	31ms	32ms	125ms
2	16ms	15ms	117ms
3	16ms	16ms	96ms

100000 Iterations
Method	Firefox 1.5 PC	IE 6.0 PC	Safari
1	297ms	266ms	1937ms
2	296ms	250ms	1924ms
3	250ms	203ms	1878ms

Here’s a link to the script I used for benchmarking.

Both comments and pings are currently closed.

2 Responses to “Favorite Loop Benchmark”

Mike Haugland said on April 15th, 2006 at 6:39 pm
Just adding the results of the test for Opera 9.0 weekly build 8367.

1,000 Iterations
———————
1. 0ms
2. 0ms
3. 0ms

10,000 Iterations
———————
1. 40ms
2. 30ms
3. 30ms

100,000 Iterations
———————
1. 340ms
2. 341ms
3. 280ms

Doing the test multiple times resulted in different results, but method 3 was consistently the fastest.
Danilo said on April 16th, 2006 at 6:34 am
Unfortunately the loops as shown are not equilivant folks.

Try running the 3rd one with 0, “”, false, or null as any one of the array values instead of the “1 to iterations” integers within the test array, and you’ll find that your looping will not loop around all of the elements in the array, but rather stop at the first ocurance of 0, “”, or null.

I suppose that if you can guarantee that you’ll only ever have non-0, non-”", non-false, non-null data in your array you might be able to get away with using this type of construct, but even then, you be giving up guaranteed working for just a few milliseconds. And even at 100,000 iterations (when’s the last time anyone really needed an array that large?), in the worst performing browser, the difference is only 46 milliseconds, which isn’t going to be noticable.