GitHub - jaaamesey/parra: a `run-with-worker` wrapper for easy parallel computation in JavaScript

2 min read Original article ↗

parra: parallel computations made easy

To install:

npm i parra

Note: This library is still in a very experimental stage of development. Feel free to report any issues here.

This library wraps run-with-worker with Array functions that enable easy, safe, and platform-agnostic parallel computations in JavaScript Web Workers.

It currently supports:

  1. a parallel Array.map equivalent
  2. a parallel Array.reduce equivalent

As with the core run-with-worker package, the amount of overhead Web Workers introduce makes parallelization only useful for computations that would take more than a few milliseconds.

Examples

test("parallelMap: powers of 2", async () => {
  const res = await parallelMap(
    // number of threads (Workers) to split work between
    4,
    // items to process
    [1, 2, 3, 4, 5, 6, 7, 8, 9, 10],
    // computation to perform on each item
    (item, [base]) => base ** item,
    // any extra dependencies to send to the workers
    [2],
  );
  expect(res).toEqual([2, 4, 8, 16, 32, 64, 128, 256, 512, 1024]);
});

test("parallelReduce: sum of numbers", async () => {
  const res = await parallelReduce(
    4,
    Array.from({ length: 100 }).map((_, i) => i + 1),
    (acc, item) => acc + item,
    0,
    [],
  );
  expect(res).toEqual(5050);
});

test("parallelReduce: product of numbers", async () => {
  const res = await parallelReduce(
    4,
    Array.from({ length: 10 }).map((_, i) => i + 1),
    (acc, item) => acc * item,
    1,
    [],
  );
  expect(res).toEqual(3628800);
});