/tinybench

🔎 A simple, tiny and lightweight benchmarking library!

Primary LanguageTypeScriptMIT LicenseMIT

I'm transitioning to a full-time open source career. Your support would be greatly appreciated 🙌 Subscription Tiers on Polar

Tinybench 🔎

CI NPM version

Benchmark your code easily with Tinybench, a simple, tiny and light-weight 7KB (2KB minified and gzipped) benchmarking library! You can run your benchmarks in multiple JavaScript runtimes, Tinybench is completely based on the Web APIs with proper timing using process.hrtime or performance.now.

  • Accurate and precise timing based on the environment
  • Event and EventTarget compatible events
  • Statistically analyzed values
  • Calculated Percentiles
  • Fully detailed results
  • No dependencies

In case you need more tiny libraries like tinypool or tinyspy, please consider submitting an RFC

Installing

$ npm install -D tinybench

Usage

You can start benchmarking by instantiating the Bench class and adding benchmark tasks to it.

import { Bench } from 'tinybench';

const bench = new Bench({ time: 100 });

bench
  .add('faster task', () => {
    console.log('I am faster')
  })
  .add('slower task', async () => {
    await new Promise(r => setTimeout(r, 1)) // we wait 1ms :)
    console.log('I am slower')
  })
  .todo('unimplemented bench')

await bench.warmup(); // make results more reliable, ref: https://github.com/tinylibs/tinybench/pull/50
await bench.run();

console.table(bench.table());

// Output:
// ┌─────────┬───────────────┬──────────┬────────────────────┬───────────┬─────────┐
// │ (index) │   Task Name   │ ops/sec  │ Average Time (ns)  │  Margin   │ Samples │
// ├─────────┼───────────────┼──────────┼────────────────────┼───────────┼─────────┤
// │    0    │ 'faster task' │ '41,621' │ 24025.791819761525 │ '±20.50%' │  4257   │
// │    1    │ 'slower task' │  '828'   │ 1207382.7838323202 │ '±7.07%'  │   83    │
// └─────────┴───────────────┴──────────┴────────────────────┴───────────┴─────────┘

console.table(
  bench.table((task) => ({'Task name': task.name}))
);

// Output:
// ┌─────────┬───────────────────────┐
// │ (index) │       Task name       │
// ├─────────┼───────────────────────┤
// │    0    │ 'unimplemented bench' │
// └─────────┴───────────────────────┘

The add method accepts a task name and a task function, so it can benchmark it! This method returns a reference to the Bench instance, so it's possible to use it to create an another task for that instance.

Note that the task name should always be unique in an instance, because Tinybench stores the tasks based on their names in a Map.

Also note that tinybench does not log any result by default. You can extract the relevant stats from bench.tasks or any other API after running the benchmark, and process them however you want.

Docs

Bench

The Benchmark instance for keeping track of the benchmark tasks and controlling them.

Options:

export type Options = {
  /**
   * time needed for running a benchmark task (milliseconds) @default 500
   */
  time?: number;

  /**
   * number of times that a task should run if even the time option is finished @default 10
   */
  iterations?: number;

  /**
   * function to get the current timestamp in milliseconds
   */
  now?: () => number;

  /**
   * An AbortSignal for aborting the benchmark
   */
  signal?: AbortSignal;

  /**
   * Throw if a task fails (events will not work if true)
   */
  throws?: boolean;

  /**
   * warmup time (milliseconds) @default 100ms
   */
  warmupTime?: number;

  /**
   * warmup iterations @default 5
   */
  warmupIterations?: number;

  /**
   * setup function to run before each benchmark task (cycle)
   */
  setup?: Hook;

  /**
   * teardown function to run after each benchmark task (cycle)
   */
  teardown?: Hook;
};

export type Hook = (task: Task, mode: "warmup" | "run") => void | Promise<void>;
  • async run(): run the added tasks that were registered using the add method
  • async runConcurrently(threshold: number = Infinity, mode: "bench" | "task" = "bench"): similar to the run method but runs concurrently rather than sequentially. See the Concurrency section.
  • async warmup(): warm up the benchmark tasks
  • async warmupConcurrently(threshold: number = Infinity, mode: "bench" | "task" = "bench"): warm up the benchmark tasks concurrently
  • reset(): reset each task and remove its result
  • add(name: string, fn: Fn, opts?: FnOpts): add a benchmark task to the task map
    • Fn: () => any | Promise<any>
    • FnOpts: {}: a set of optional functions run during the benchmark lifecycle that can be used to set up or tear down test data or fixtures without affecting the timing of each task
      • beforeAll?: () => any | Promise<any>: invoked once before iterations of fn begin
      • beforeEach?: () => any | Promise<any>: invoked before each time fn is executed
      • afterEach?: () => any | Promise<any>: invoked after each time fn is executed
      • afterAll?: () => any | Promise<any>: invoked once after all iterations of fn have finished
  • remove(name: string): remove a benchmark task from the task map
  • table(): table of the tasks results
  • get results(): (TaskResult | undefined)[]: (getter) tasks results as an array
  • get tasks(): Task[]: (getter) tasks as an array
  • getTask(name: string): Task | undefined: get a task based on the name
  • todo(name: string, fn?: Fn, opts: FnOptions): add a benchmark todo to the todo map
  • get todos(): Task[]: (getter) tasks todos as an array

Task

A class that represents each benchmark task in Tinybench. It keeps track of the results, name, Bench instance, the task function and the number of times the task function has been executed.

  • constructor(bench: Bench, name: string, fn: Fn, opts: FnOptions = {})
  • bench: Bench
  • name: string: task name
  • fn: Fn: the task function
  • opts: FnOptions: Task options
  • runs: number: the number of times the task function has been executed
  • result?: TaskResult: the result object
  • async run(): run the current task and write the results in Task.result object
  • async warmup(): warm up the current task
  • setResult(result: Partial<TaskResult>): change the result object values
  • reset(): reset the task to make the Task.runs a zero-value and remove the Task.result object
export interface FnOptions {
  /**
   * An optional function that is run before iterations of this task begin
   */
  beforeAll?: (this: Task) => void | Promise<void>;

  /**
   * An optional function that is run before each iteration of this task
   */
  beforeEach?: (this: Task) => void | Promise<void>;

  /**
   * An optional function that is run after each iteration of this task
   */
  afterEach?: (this: Task) => void | Promise<void>;

  /**
   * An optional function that is run after all iterations of this task end
   */
  afterAll?: (this: Task) => void | Promise<void>;
}

TaskResult

the benchmark task result object.

export type TaskResult = {

  /*
   * the last error that was thrown while running the task
   */
  error?: unknown;

  /**
   * The amount of time in milliseconds to run the benchmark task (cycle).
   */
  totalTime: number;

  /**
   * the minimum value in the samples
   */
  min: number;
  /**
   * the maximum value in the samples
   */
  max: number;

  /**
   * the number of operations per second
   */
  hz: number;

  /**
   * how long each operation takes (ms)
   */
  period: number;

  /**
   * task samples of each task iteration time (ms)
   */
  samples: number[];

  /**
   * samples mean/average (estimate of the population mean)
   */
  mean: number;

  /**
   * samples variance (estimate of the population variance)
   */
  variance: number;

  /**
   * samples standard deviation (estimate of the population standard deviation)
   */
  sd: number;

  /**
   * standard error of the mean (a.k.a. the standard deviation of the sampling distribution of the sample mean)
   */
  sem: number;

  /**
   * degrees of freedom
   */
  df: number;

  /**
   * critical value of the samples
   */
  critical: number;

  /**
   * margin of error
   */
  moe: number;

  /**
   * relative margin of error
   */
  rme: number;

  /**
   * p75 percentile
   */
  p75: number;

  /**
   * p99 percentile
   */
  p99: number;

  /**
   * p995 percentile
   */
  p995: number;

  /**
   * p999 percentile
   */
  p999: number;
};

Events

Both the Task and Bench objects extend the EventTarget object, so you can attach listeners to different types of events in each class instance using the universal addEventListener and removeEventListener.

/**
 * Bench events
 */
export type BenchEvents =
  | "abort" // when a signal aborts
  | "complete" // when running a benchmark finishes
  | "error" // when the benchmark task throws
  | "reset" // when the reset function gets called
  | "start" // when running the benchmarks gets started
  | "warmup" // when the benchmarks start getting warmed up (before start)
  | "cycle" // when running each benchmark task gets done (cycle)
  | "add" // when a Task gets added to the Bench
  | "remove" // when a Task gets removed of the Bench
  | "todo"; // when a todo Task gets added to the Bench

/**
 * task events
 */
export type TaskEvents =
  | "abort"
  | "complete"
  | "error"
  | "reset"
  | "start"
  | "warmup"
  | "cycle";

For instance:

// runs on each benchmark task's cycle
bench.addEventListener("cycle", (e) => {
  const task = e.task!;
});

// runs only on this benchmark task's cycle
task.addEventListener("cycle", (e) => {
  const task = e.task!;
});

BenchEvent

export type BenchEvent = Event & {
  task: Task | null;
};

process.hrtime

if you want more accurate results for nodejs with process.hrtime, then import the hrtimeNow function from the library and pass it to the Bench options.

import { hrtimeNow } from 'tinybench';

It may make your benchmarks slower, check #42.

Concurrency

  • When mode is set to null (default), concurrency is disabled.
  • When mode is set to 'task', each task's iterations (calls of a task function) run concurrently.
  • When mode is set to 'bench', different tasks within the bench run concurrently. Concurrent cycles.
// options way (recommended)
bench.threshold = 10 // The maximum number of concurrent tasks to run. Defaults to Infinity.
bench.concurrency = "task" // The concurrency mode to determine how tasks are run.  
// await bench.warmup()
await bench.run()

// standalone method way
// await bench.warmupConcurrently(10, "task")
await bench.runConcurrently(10, "task") // with runConcurrently, mode is set to 'bench' by default

Prior art

Authors


Mohammad Bagher

Credits


Uzlopak

poyoho

Contributing

Feel free to create issues/discussions and then PRs for the project!

Sponsors

Your sponsorship can make a huge difference in continuing our work in open source!