Hi!
Steffen Kamper wrote:
> ok, here are your numbers:
>
> $s = microtime(1);
> for ($i = 0; $i < 100000; $i++) { if($GLOBALS['not_existing']) {$k++;} }
> $e = microtime(1);
> echo 'Time needed without isset: ' . ($e-$s) . "\n";
>
> $s = microtime(1);
> for ($i = 0; $i < 100000; $i++) { if(isset($GLOBALS['not_existing']))
> {$k++;} }
> $e = microtime(1);
> echo 'Time needed with isset: ' . ($e-$s) . "\n";
>
> Output:
> Time needed without isset: 1.60987997055
> Time needed with isset: 0.0281138420105
>
> You see that performance increased with factor 80
>
> impressed?
Not at all. You did not reproduce the case we are talking about but made a test over $GLOBALS. This is not correct testing. My example:
======================
<?php
$t = new stdClass();
$t->config['config'] = array();
$k = 0;
$t1 = microtime(true);
for ($i = 0; $i < 100000; $i++) {
if ($t->config['config']['debug']) {
$k = $t->config['config']['debug'] + 1;
}
}
$t2 = microtime(true);
$t3 = microtime(true);
for ($i = 0; $i < 100000; $i++) {
if (isset($t->config['config']['debug'])) {
$k = $t->config['config']['debug'] + 1;
}
}
$t4 = microtime(true);
echo sprintf("%.5f %.5f\n", $t2 - $t1, $t4 - $t3);
?>
======================
Gives:
0.16026 0.06486
Difference is 2.66 times on 100000 iterations or 0.000026 times for a single use.
I went further and ran a profiler. 69.72 of time was spent in the first part and 30.23 in the second, which gives 2.33 times.
isset() is faster but not significantly faster.
--
Dmitry Dulepov
TYPO3 Core team
Latest article: http://typo3bloke.net/post-details/do_not_ask_more_than_you_need/
More about TYPO3: http://typo3bloke.net/
Subscribe: http://typo3bloke.net/rss.xml