The purpose of this test is to see if casting an object to a string by concatenating it was actually slower than using normal type casting. Here is my source:

<?php

// Setup
$tests = 1000000;
class Foo {
	public function __toString() {
		return "bar";
	}
}

// Test 1
$x = new Foo();
$timer = microtime();

for ($i = 0; $i < $tests; $i++) {
	$y = "" . $x;
}

$timer -= microtime();
echo "Test 1 complete. Total time: {$timer}\n" . PHP_EOL;

// Test 2
$x = new Foo();
$timer = microtime();

for ($i = 0; $i < $tests; $i++) {
	$y = (string)$x;
}

$timer -= microtime();
echo "Test 2 complete. Total time: {$timer}\n" . PHP_EOL;

// Done.
exit();

This is what I got in my terminal:

$php -f test-typecast.php
Test 1 complete. Total time: -0.203729
Test 2 complete. Total time: -0.16766

Explanation

Test 1 tells PHP to 1) create a string, 2) concatenate it to an object, 3) PHP can’t do that so it tries to call the $x->__toString() method, 4) now concatenate "" . $x->__toString();.

Test 2 in this case will 1) check the type of $x, and 2) return the value of $x->__toString().

Naturally the process with the least steps would be faster.