Calculate the appropriate Argon2id parameters for the local hardware environment.

v0.2.2 2019-07-10 18:49 UTC

This package is auto-updated.

Last update: 2020-07-11 05:27:12 UTC


Build Status Latest Stable Version Latest Unstable Version License Downloads

Easily and effectively benchmark the real time to perform Argon2id password hashes on your machine.

Warning: This might take many seconds or minutes to complete.

Installation Instructions

Use Composer.

composer require paragonie/argon2-refiner

Alternatively, you can install this with Git.

git clone
cd argon2-refiner
composer install

Usage Instructions

Command Line

Run the bundled benchmark script like so:

# Installed via Composer:
vendor/bin/benchmark [milliseconds=500] [tolerance=250]

# Installed via Git:
composer run-benchmarks [milliseconds=500] [tolerance=250]

The expected output will look something like this:

$ vendor/bin/benchmark 125
 Recommended Argon2id parameters:
 	       Memory cost (sodium): 79691776
 	Memory cost (password_hash): 77824
 	                  Time cost: 3
 Real time: 124ms

This means that if you set your Argon2id mem_cost to 79691776 bytes (or 77824 KiB, which is what password_hash() expects) and the time_cost to 3, you will get the closest parameters that take about 125 milliseconds to process (in this example, it took 124).

Object-Oriented API

You can fine-tune your min/max costs to search within from the object by invoking the appropriate methods.

use ParagonIE\Argon2Refiner\ParameterRecommender;

$refiner = (new ParameterRecommender(125))
    ->setMinMemory(1 << 20)
    ->setMaxMemory(1 << 31)

$results = $refiner->runBenchmarks();

The runBenchmarks() method returns a two-dimensional array of arrays. Each child array consists of the following data:

  • mem_cost (int) -- Candidate parameter
  • time_cost (int) -- Candidate parameter
  • bench_time (int) -- Milliseconds elapsed in Argon2id calculation

From this data, you can devise your own strategy for selecting which parameters set is most suitable for your environment.