Graphics Code Sites Notes

There have been several comments in the notes of the uniqid() manual page about uniqid() being very slow. I set off to see for myself if this were true, using several combinations of input. I tried different length prefixes, as well as toggling the function's more_entropy parameter. For those interested, the code looks something like this:

<?php	
    function microtime_float() {	
        list($usec, $sec) = explode(" ", microtime());	
        return ((float)$usec + (float)$sec);	
    }	
    	
    $s = microtime_float();	
    for($i=0;$i<1000;$i++) {	
        uniqid("longprefix");	
    }	
    echo "Long prefix: ".(microtime_float()-$s)."\n";	
    	
    $s = microtime_float();	
    for($i=0;$i<1000;$i++) {	
        uniqid("s");	
    }	
    echo "Short prefix: ".(microtime_float()-$s)."\n";	
    	
    $s = microtime_float();	
    for($i=0;$i<1000;$i++) {	
        uniqid("");	
    }	
    echo "No prefix: ".(microtime_float()-$s)."\n";	
    	
    	
    $s = microtime_float();	
    for($i=0;$i<1000;$i++) {	
        uniqid("longprefix",true);	
    }	
    echo "Long prefix (More Entropy): ".(microtime_float()-$s)."\n";	
    	
    $s = microtime_float();	
    for($i=0;$i<1000;$i++) {	
        uniqid("s",true);	
    }	
    echo "Short prefix (More Entropy): ".(microtime_float()-$s)."\n";	
    	
    $s = microtime_float();	
    for($i=0;$i<1000;$i++) {	
        uniqid("",true);	
    }	
    echo "No prefix (More Entropy): ".(microtime_float()-$s)."\n";	
?>

And after about 20 trials, the results look something like this:



The y axis being the number of seconds to do 1000 iterations of the function, and the x axis being trial number (funky decimals due to the fact that they're evenly spaced.)

As you can see, simply setting more_entropy makes a world of difference. Prefix appears to not have much effect, but the 3 sets with more_entropy set to true were always faster than the three without it. more_entropy has the added bonus of producing a "more unique" result, according to the manual.