Comparing for-loop vs Array.p.map

map.js

'use strict';

/*

Checked CL out using...
> git fetch https://chromium.googlesource.com/v8/v8 refs/changes/39/548639/17 && git checkout FETCH_HEAD


> d8 --turbo_inline_array_builtins --allow-natives-syntax map.js
doArrayMap 2194 ms
doForLoop 879 ms

> d8 --allow-natives-syntax map.js
doArrayMap 2188 ms
doForLoop 863 ms

*/

function doForLoop(arr) {
  const result = [];
  for(let i=0; i<arr.length; ++i) {
    const c = arr[i];
    result.push(c + c);
  }
  return result;
}

function doArrayMap(arr) { return arr.map(c => c + c); }

const ARRAYS = [
  [0,1,2,3,4,5],
  [5,0,1,2,3,4],
  [4,5,0,1,2,3],
  // ['0','1','2','3','4','5'],
  // [{a: 0},{a: 1},{a: 2},{a: 3},{a: 4},{a: 5}]
];

function bench(func, max){
  const start = Date.now();
  for (let i = 0; i < max; ++i) {
    func(ARRAYS[i % ARRAYS.length]);
  }
  const end = Date.now();
  return `${func.name} ${end - start} ms`;
}

// Speculation: Prevent one function getting an advantage (inlining?) on the other.
%NeverOptimizeFunction(bench);

// Warmup
bench(doArrayMap, 5);
bench(doForLoop, 5);

%OptimizeFunctionOnNextCall(doArrayMap);
%OptimizeFunctionOnNextCall(doForLoop);

bench(doArrayMap, 5);
bench(doForLoop, 5);

// Bench
print(bench(doArrayMap, 1e7));
print(bench(doForLoop, 1e7));