SHISHUA for Node.js: The fastest PRNG

The official page for SHISHUA is here.
This npm package provides bindings for Node.js.

const shishua = require('shishua');
const buffer = Buffer.alloc(1 << 30);  // 1 GiB
shishua().fill(buffer);                // ~100ms on my laptop.

The package is API-compatible with seedrandom, so that you can use it with the random library:

const random = require('random');
random.use(shishua('seed'));

// Build a Poisson sampling.
const poissonSample = random.poisson();
new Array(12).fill(0).map(poissonSample)

API

shishua(seed, options, callback)

Returns a function that produces random floating-point values between 0 (included) and 1 (excluded).

The seed determines the sequence of numbers that this function returns:

new Array(3).fill(0).map(shishua('seed').int32)
// [ 1534767448, 1726267546, 2477584171 ]
new Array(3).fill(0).map(shishua('seed').int32)
// [ 1534767448, 1726267546, 2477584171 ]
new Array(3).fill(0).map(shishua('different seed, different values').int32)
// [ 69339860, 2922872123, 3883034659 ]

The seed can be either:

• A 32-byte Buffer, to map exactly to the functionality provided by the underlying SHISHUA algorithm,
• A String, which is hashed into the 32-byte buffer,
• If not provided (or undefined), a random seed is generated from the system's CSPRNG.

The other parameters, options and callback, are only there for compatibility with seedrandom.

Associated methods

The object returned by calling shishua() also has the following methods, which all tap into the stream generated by the seed:

• .fill(buffer), which fills a provided buffer with random bytes,
• .int32(), which generates a uniformly random 32-bit number,
• .double(), which produces a random floating-point value between 0 (included) and 1 (excluded),
• .quick(), which also produces a random floating-point value between 0 (included) and 1 (excluded), but with only 32 bits of entropy.