Concern about native map and reduce speed

[stevensacks]

The native Javascript map and reduce array functions are ~134x slower than writing a loop, and ~2x as slow as Lodash (which is 67x times slower than writing a loop).

For example, doing a reduction where you double the value:

function reduce(arr, total = 0) {
    var i, len = arr.length;
    for (i = 0; i < len; i++)
    {
        total += arr[i] * 2;
    }
    return total;
}

An array with 100,000 integers (going from 0 to 99,999) gets reduced in 0.19ms using this function.

Native Array.reduce() takes 25.44ms.
arr.reduce((total, val) => total + (val * 2), 0)

Lodash takes 12.81ms
_(arr).reduce((total, val) => total + (val * 2), 0);

Lodash and native performance are abysmal by comparison. And while I know that arrays of 100,000 items are uncommon, it's worth noting.

I'm wondering if there's a way to use your reduce function to call my own insanely faster reduce instead of JavaScript's native reduce.

[aikeru]

Looks like this might do what you want...

(reduce, arr) => _.reduce(arr, reduce)

But native performance is expected to improve over time and may vary by browser and platform.

PS I'm not affiliated with 1-liners, though their work is very cool!