Gist to compare two ways of creating the string for flake (https://npmjs.org/package/flake) IDs. I suspect the construction of the array far outweighs the construction of the string.

speed-comparison.js

var t1, t2, t3;
var i;
const TOP = 1000000000;
var tsHex = '013c798f75ad';
var counterHex = '0000';
var pidHex = '1154';
var macHex = '984be1b8b104';

t1 = Date.now();

for (i=0; i< TOP; i++) {
    tsHex + '-' + counterHex + '-' + pidHex + '-' + macHex;
}

t2 = Date.now();

for (i=0; i< TOP; i++) {
    [tsHex, counterHex, pidHex, macHex].join('-');
}

t3 = Date.now();

// about 85ms
console.log('1=' + (t2-t1) + 'ms');

// about 53000ms (53s)!!
console.log('2=' + (t3-t2) + 'ms');