Multicore

A simple Javascript library for handling Web Workers.

Currently being developed. Only works on the web.

Progress so far:

• Faster for arithmetic operations than ParallelJS. Test: Operating on a list of 100k elements, multiplying and reducing them to one value. Result: Multicore: ~200ms - ParallelJS ~20s
• Now supports TypedArrays (and any kind of transferable) for futher increase in speed and memory use.

Installation

Use yarn or npm:

yarn add multicore --save
npm install --save multicore

This assumes that you’re using npm package manager with a module bundler like Webpack or Browserify to consume CommonJS modules.

Documentation

Usage

TLDR;

• Use .data or, if you know size/type of array, a TypedArray function like .uInt8.
• Then manipulate data by chaining API-methods: Multicore.data([1,2,3]).reduce((acc,val) => acc+val);
• Retrieve the data from the instance promise: [...].reduce().map().then(result => { doSomething(result); })

Import Multicore

Using either:

const Multicore = require('multicore').default;
import Multicore from 'multicore';

Create an instance

Start an instance for data you wish you manipulate (can either be used as a constructor or with these static shorthand-methods):

Constructor

Create an instance with constructor and save it to a variable. The Constructor can also be subclassed as per ES6 classes.

const data = new Multicore([1,2,3]); // Constructor

Static converter-functions

Use the generic .data method to get the same behavior as with constructor.

Use any of the TypedArray functions to pass in an array and use parallelized conversion to convert it to a typed array internally, and thereby being able to use buffers internally. Recommended for large arrays containing integers or floats, or anytime you know your array will only contain numbers.

TypedArray functions for the different types: .int8, .int16, .int32, .uInt8, .uInt16, .uInt32, .float32, .float16.

const data = Multicore.data([1,2,3]); // Use for generic data.
const data = Multicore.uInt8([1,2,3); // Will convert array to typed-array and be able to utilize buffers internally

Start manipulating

// Manipulate saved instance
data.map(val => val*2)
  .map(val => val*3)
  .reduce((acc, val) => {
    return acc + val;
  })
  .then(result => {
    console.log(`The result is ${result}`);
  });

// The result is 37

// Direct manipulation and conversion
Multicore.uInt8([1,2,3])
    .map(val => val*2)
    .then(console.log); 

// [2,4,6]
  

Methods

A method receives only one argument, the current data being manipulated.

.spawn(fn)

Apply given function to current data. Usage:

const data = new Multicore(['a','b','c']);
data.spawn(data => {
    return data.join(',').toUpperCase();
});

// "A,B,C"

.map(fn)

Map over elements of array applying given function to every element. Similar to Javascript Array.map Usage:

const data = new Multicore([1,2,3]);
data.map(val => val*2);

// [2,4,6]

.reduce(fn, [accumulator = 0])

Applies a function against an accumulator and each element in the array in parallel to reduce it to a single value. Important: This function is optimized to run in parallel and needs your reduction operators to be associative, if your operators are not associative then use foldr() instead. Usage:

const data = new Multicore([1,2,3]);
data.reduce((acc, val) => {
    return acc + val;
  });

// 6

.foldr(fn, [accumulator = 0])

Applies a function against an accumulator and each element in the array in from left-to-right to reduce it to a single value. Similar to Javascript Array.reduce. Usage:

const data = new Multicore([1,2,3]);
data.foldr((acc, val) => {
    return acc - val;
  }, 0);

// -4

.filter(fn)

The filter method filters the data and keeps only elements that pass the test implemented by the provided function. Similar to Javascript Array.filter Usage:

const data = new Multicore([1,2,3]);
data.filter((val) => {
    return val > 2;
  });

// 3

More methods under development.

Work In Progress

Work that is currently underway and lives in feature-branches.

• Adding full test-coverage.
• Switching to babel-implementation and building minified-js-files for CDN.
• Adding support for new operators: .parallel and .merge.

License

MIT