Question

I was considering ways to create arrays containing a default value using native methods and ended up with

function pushMap(length, fill){
    var a = [], b = [];
    a.length = length;
    b.push.apply(b,a);
    return b.map(function(){return fill;});
}

Expecting it to be 2 or 3 times slower than a while loop, as the native methods have to loop twice whereas while loops only once, so I compared it on jsperf against

function whileLengthNew(len, val) {
    var rv = new Array(len);
    while (--len >= 0) {
        rv[len] = val;
    }
    return rv;
}

and it is actually 18 to 27 times slower (tested with Google Chrome on Ubuntu, browsers/OSs welcome).

What is happening that causes such a big difference?

Answer 1

I would expect that this is due to two major factors:

1. Memory allocation — whileLengthNew creates an array of the correct size first, and then operates on it, pushMap creates the final array one element at a time with map. This may cause multiple allocations, especially if the source array is large. (The way that you create the initial a and b arrays is basically irrelevant, since map is building up a new array to return anyway — it doesn’t actually change anything in b)

2. Function call overhead — in your call to map, you are calling a function for every element of the array. This involves quite a lot of overhead; setting up activation records and scope chains, stack manipulation, and passing the return value back. — all of this to access a variable which is constant within the function. On top of that, you have set up a closure, so even accessing the fill variable is slower than it is in the whileLengthNew version.