array_map('intval',...) vs arrayToInt()

arrayToInt.php

<?php

$integers = range(100, 1000);

foreach ($integers as &$int) {
    $int = (string)$int;
}

function arrayToInt(array $arr)
{
    foreach ($arr as &$a) {
        $a = (int)$a;
    }
    return $arr;
}

$start = microtime(true);
$sum   = 0;
for ($i = 0; $i < 100000; $i++) {
    $converted = array_map('intval', $integers);
    $sum += array_sum($converted);
}
var_dump($sum);
printf("\n%.6f msec\n", microtime(true) - $start);
flush();

$start = microtime(true);
$sum   = 0;
for ($i = 0; $i < 100000; $i++) {
    $converted = arrayToInt($integers);
    $sum += array_sum($converted);
}
var_dump($sum);
printf("\n%.6f msec\n", microtime(true) - $start);

result.txt

$ php -f arrayToInt.php

31.362757 msec
23.988653 msec

You labeled the output 'msec' when it's really seconds. 7ms difference is ~80 pico seconds per element which is laughable.

Even so, I find it highly suspect that array_map is a real bottleneck here. I re-ran your gist with a second version of arrayToInt that was identical except used function call cast rather than operator (to get better comparison with array map which naturally has to call a function on each element).

<?php
// ...
function arrayToInt2(array $arr)
{
    foreach ($arr as &$a) {
        $a = intval($a);
    }
    return $arr;
}
// ...

array_map       35.559198 sec
arrayToInt      21.657530 sec
arrayToInt2     36.301321 sec

At any rate worrying about micro-optimisation like this seems misguided in all but the most extreme cases in PHP to be honest. If a few micro seconds here and there are so critical to your app then you probably shouldn't be writing it in PHP at all :).

My 2 cents: use array_map() if it is useful and makes your code cleaner, write C if PHP is too slow ;).

@banks A+

You said everything I was thinking as I read the original post.

You even wrote your function the way I would've.

Amen, my friend.

array_map(fn($num) => (int)$num, ['1', '2']);