Array<Array>, Array<{}> or Array<Object>? tests made with node 6.3.0 on a mac

array.js

// Execution time: 1048
// { rss: 1028186112, heapTotal: 1015271424, heapUsed: 969132048 }

var start = new Date().getTime();

var arr = [];
for (i = 0; i < 10000000; ++i) {
  arr.push([Math.random(),  Math.random()])
}

var end = new Date().getTime();
var time = end - start;
console.log('Execution time: ' + time);
console.log(process.memoryUsage())

object.js

// this example performs the worse in term of speed!
// probably because of the manual allocation of x and y
//Execution time: 3109
// { rss: 740888576, heapTotal: 722391040, heapUsed: 693790992 }

var start = new Date().getTime();

class Obj {
  constructor(x,y){
    this.x = x;
    this.y = y;
  }
}
var arr = [];
for (i = 0; i < 10000000; ++i) {
  arr.push(new Obj(Math.random(), Math.random()))
}

var end = new Date().getTime();
var time = end - start;
console.log('Execution time: ' + time);
console.log(process.memoryUsage())

plain.js

// objects have overhead, but I suspect that v8 optimizes
// when it sees that the size of the object, ie. number of stored references (keys) never change
// Execution time: 987
// { rss: 788381696, heapTotal: 771977216, heapUsed: 729104864 }

var start = new Date().getTime();

var arr = [];
for (i = 0; i < 10000000; ++i) {
  arr.push({x: Math.random(), y: Math.random()})
}

var end = new Date().getTime();
var time = end - start;
console.log('Execution time: ' + time);
console.log(process.memoryUsage())