across v0.0.1

Welcome to Across

This tool is currently experimental, DO NOT USE UNDER PRODUCTION ENVIRONMENTS JUST YET.

This small tool was created as a way of processing large data sets in parallel by using Node's worker threads.

With it, most complexity related to multi-threading is abstracted, so you don't need to worry.

Most of this work is more of a proof-of-concept by now.

Basic Example

// Import our library
const { Distributed } = require('across')

// Let's create a sorta large array to process
const numberArray = Array.from(Array(1000000).keys())

// Our "processing" function will calculate the square-root of the numbers and return it as a string.
const iterator = (number) => `sqrt(${ number }) = ${ Math.sqrt(number).toFixed(2) }`

// Now let's process it!
Distributed.map(numberArray, iterator)
  // Since our map returns a Promise, we can wait for its result by using then()
  .then(results => {
    // Do something with them, in this case print them
    console.log('Results:', results)

    // Exit
    process.exit(0)
  })

By running the above snippet your function will be executed across multiple threads, by default, matching your CPU's core count.

Sure, with simple loads there's a big chance Node's built-in Array.map will be faster, since there will be no overhead regarding threads.

But, if you're doing something really CPU intensive and without any non-blocking options available, then you should expect your code to run much faster.

More Examples

Under the examples directory you should find some other examples for this. One of them is a small speed test using the bcrypt NPM module as reference.

This test will show you how long the processing takes by using the module's provided async version and the sync version, both with single-threaded and multi-threaded variations, plus a basic Array.map of it too.

To-do

• Add new functions such as filter and reduce
• Allow functions to be chained
• Reduce Worker creation overhead
• Work with Worker Pools