dheia / phpench

Realtime benchmarks for PHP code

PHPench

PHPench creates a graphical output for a PHP benchmark. Plot the runtime of any function in realtime with GnuPlot and create an image out of the result.

Why is it useful?

Algorithms are beautiful

Sometimes the difference between two algorithms is hard to explain but easy to show.
For instance, take two sorting algorithms which both have a best-case runtime of O(n*log n). Depending on the input, one can be much faster than the other. This tools helps you see what's going on.

Death to premature-optimizations

Whenever people tell you that using single quotes instead of double quotes around strings is a performance improvement, it's time to debunk some myths. Most of the time such programmer folklore turns out to be misguided and can actually be pretty harmful.
"Premature emphasis on efficiency is a big mistake which may well be the source of most programming complexity and grief." (Donald Knuth)
Let's be professionals. Let's measure.

Example

Using PHPench feels a bit like writing a visual unit test. Check it out:

<?php

require_once __DIR__.'/../vendor/autoload.php';

/*
 * You can use an closure or a class that implements TestInterface.
 *
 * Data that will be processed by the tested function can be executed
 * without including its execution time. This will provide more accurate data.
 */

abstract class AbstractBenchmark implements \mre\PHPench\BenchmarkInterface
{
    protected $test;

    function setUp($arrSize)
    {
        $this->test = array();
        for ($i=1; $i<$arrSize; $i++) {
            $this->test[$i]= $arrSize % $i;
        }

        return $this->test;
    }
}

class BenchmarkArrayFlip extends AbstractBenchmark
{
    public function execute() {
        $test = array_flip(array_flip($this->test));
    }
}

class BenchmarkArrayUnique extends AbstractBenchmark
{
    public function execute() {
        $test = array_unique($this->test);
    }
}

// Create a new benchmark instance
$phpench = new \mre\PHPench(new \mre\PHPench\Aggregator\MedianAggregator);

// Use GnuPlot for output
$oOutput = new \mre\PHPench\Output\GnuPlotOutput('test2.png', 1024, 768);

// Alternatively, print the values to the terminal
//$oOutput = new \mre\PHPench\Output\CliOutput();

$oOutput->setTitle('Compare array_flip and array_unique');
$phpench->setOutput($oOutput);

// Add your test to the instance
$phpench->addBenchmark(new BenchmarkArrayFlip, 'array_flip');
$phpench->addBenchmark(new BenchmarkArrayUnique, 'array_unique');

// Run the benchmark and plot the results in realtime.
// With the second parameter you can specify
// the start, end and step for each call
$phpench->setInput(range(1,pow(2,16), 1024));
$phpench->setRepetitions(4);
$phpench->run();

Installation

1.) Add this package to your composer.json

{
    "require": {
      "mre/phpench": "*@dev"
    }
}

2.) Install gnuplot (Version 4.6)

For Mac OS X you can install gnuplot via homebrew. For live generated charts you also need to install XQuartz.

Without X11 support:
$ brew install homebrew/versions/gnuplot4

With X11 supprt (recommended!):
$ brew install homebrew/versions/gnuplot4 --with-x11

For Linux use your package manager.

apt-get install gnuplot

3.) Look at the examples for usage

Maintainers

Matthias Endler (@matthiasendler)
Markus Poerschke (@markuspoerschke)

License

Apache License Version 2.0