JS Benchmarker (NodeJS)

Benchmarker is a basic JS function benchmarking utility. This version is built to run with NodeJS apps and not compatible for use with browsers. For the browser-friendly version, check the version/vanilla-js branch. It uses Node's built-in perf_hook library to measure a function's completion time. To get readings that are more accurate, the number of times a function runs can be specified using the cycles argument.

Dependencies

• perf_hooks (JS built-in)
• mathjs

Installation (npm)

Run npm install @darylt/benchmarker in your npm project folder.

Installation (Github Source Code)

Run npm install to install both dependencies.

Usage

The benchmarker takes in an array of callback functions and runs them n number of times. It returns an array of objects containing each tested function's name and average completion times. Anonymous functions will be named AnonymousFuncN, where N is the index position of that function in the callback array.

The benchmark(functionsToRun, cycles) function takes in two arguments.

1. functionsToRun - an array containing all the functions to benchmark.
2. cycles - an integer value that determines how many times to run each function.

To use this in the browser, use const { benchmark } = require('path_to/benchmark.js') near the start of the script, then call benchmark() as described above.

Caution

The current iteration does not support asynchronous functions.

The benchmarked function will run normally, benchmark() does not prevent any functional side-effects. If your function's side effects depends on external factors, runtimes may differ.

For example, consider the following function:

let status = true;

// ...

const exampleFunc = function() {
  if (status) {
    let i = 0;
    while (i < 1000000000) {
      i++;
    }
    status = false;
  }
};

results = benchmark([exampleFunc], 100);

The first time exampleFunc runs, it will execute the while loop. However, any subsequent runs will not because status has been set to false. Therefore the first reported completion time will be much higher than all subsequent runs of this function. To check the logged completion times, use results.exampleFunc.perfHistory or results[0].perfHistory. Use the latter if you passed in anonymous functions, or use its automatically-assigned name AnonymousFuncN, where N is the index of that function in the callback function array.