PHPBench array_push vs array_merge

PushVsMerge.php

<?php

declare(strict_types = 1);

use PhpBench\Benchmark\Metadata\Annotations\BeforeMethods;
use PhpBench\Benchmark\Metadata\Annotations\Revs;

require 'vendor/autoload.php';

/**
 * @BeforeMethods({"prepare"})
 */
class PushVsMergeBench
{
    private $a = [];
    private $b = [];

    public function prepare()
    {
        $this->a = [];
        $this->b = [];
        for ($i = 0; $i < 100000; $i++) {
            $this->a[] = new \stdClass();
            $this->b[] = new \stdClass();
        }
    }

    /**
     * @Revs(100)
     */
    public function benchArrayMerge()
    {
        $this->a = array_merge($this->a, $this->b);
    }

    /**
     * @Revs(100)
     */
    public function benchArrayPush()
    {
        array_push($this->a, ...$this->b);
    }
}

Generation of data is included in benchmarking which might dillute actual results. Chock http://phpbench.readthedocs.io/en/latest/writing-benchmarks.html#subject-runtime-state-before-and-after

Updated to have @BeforeMethods({"prepare"}). Now the results are pretty different:

+------------------+-----------------+--------+--------+------+------+----------------+---------------+--------------+----------------+
| benchmark        | subject         | groups | params | revs | iter | mem_peak       | time_rev      | comp_z_value | comp_deviation |
+------------------+-----------------+--------+--------+------+------+----------------+---------------+--------------+----------------+
| PushVsMergeBench | benchArrayMerge |        | []     | 100  | 0    | 1,088,799,600b | 259,224.550μs | 0.00σ        | 0.00%          |
| PushVsMergeBench | benchArrayPush  |        | []     | 100  | 0    | 553,759,544b   | 7,336.010μs   | 0.00σ        | 0.00%          |
+------------------+-----------------+--------+--------+------+------+----------------+---------------+--------------+----------------+

It was php 7.0 right?

I told you :P