sandra

sandra

Sandra is a simple, promise-only, low overhead benchmarking library for ES2015. It only contains benchmark and suite classes, both of which have a single method run.

The name refers to a popular Brazilian song by Sidney Magal.

API

As aforementioned, you have two classes at your disposal:

import {Benchmark, Suite} from 'sandra'

(async () => {
  // basic operation
  const bench1 = new Benchmark('test', async () => {
    await longRunningOp()
  })

  // you can pass arguments to the benchmarked function
  const bench2 = new Benchmark('universe', async answer => {
    await longRunningOp(answer)
  }, 42)

  // don't forget you can instead refer to the function. avoid extra calls!
  const bench3 = new Benchmark('another test', longRunningOp, 42)

  // run a benchmark directly specifying the timeout to run (in ms)
  const result = await bench1.run(2000)
  // result contains an array of elapsed times, you do the maths
  console.log(result)

  // you can create a suite to run multiple benchmarks
  const suite = new Suite('nice suite')
  suite.push(bench2)
  suite.push(bench3)

  // it's possible to pass benchmark arguments to .push(), std::forward-like
  suite.push('nice one', longRunningOp, 42)

  // hook a callback to the 'cycle' event to get the results:
  suite.on('cycle', event => {
    // event contains 'average', 'deviation' and 'runs' fields too
    console.log(event.toString())
  })
  // there's also 'start', 'error' and 'complete' events

  // to run, call .run() passing an option object (currently, only timeout)
  await suite.run()
})()

You are ready to rock!

Example

You can check node-argon2 benchmarks for a real example on how to use.

License

Work licensed under the MIT License.