Benchmarks

[coolgames]

Well, this isn't really a BUG but i read:

**generic --> benchmarks resulted in a dramatic performance increase once the "while (list-each)" construct is replaced by i.e. "foreach" or "for". So I changed it in the entire core code (might have missed some). This should speed the game up a few milliseconds on a heavy trafficked server.

in the clangelog.

something seems to be in-correct
using (list-each) is known to be faster than foreach.

there are many companies that have tested it too   

[Nightborn]

wrong.

=)

http://benchmark.nophia.de/benchmarks-arrays-k-4-for-foreach-while-fuer-kleine-arrays-b-7.html

30ms for a foreach, 150ms for a while(list()) ...

what are your sources?

if you like, run the following scripts in rawsql->php:

Code:

$array=array();
for ($i=0;$i<10000;$i++) {
  $array[]="Test-$i";
}
//filled, now test.
$start=microtime();
while (list($key,$val)=each($array)) {
//do something
}
$end=microtime();
output("While-Each lasted: %s ms",$end-$start);

$start=microtime();
foreach($array as $key=>$val) {
//do something
}
$end=microtime();
output("Foreach with key-val lasted: %s ms",$end-$start);

Note: foreach is even faster when you do not care about the key, but about the value only.

[coolgames]

Note: foreach is even faster when you do not care about the key, but about the value only.

yep, hehe i think i misunderstood the results in http://www.php.lt/benchmark/phpbench.php

you're right 
sorry for wrong postification. 

[Nightborn]

+  415 %   1: foreach($aHash as $val);   Total time:    6[ms]
+  204 %   2: while(list(,$val) = each($aHash));   Total time:    3[ms]
+  853 %   3: foreach($aHash as $key=>$val);   Total time:   13[ms]
+  926 %   4: while(list($key,$val)= each($aHash));

if you have only one val, while -list *MIGHT* be faster, but it would be good to test that.

Here are my tests:

hirako:~/test# php test.php 100000
Test foreach took 0.045385ms.
Test foreach-ass took 0.065821ms.
Test for took 0.073692ms.
Test while took 0.069795ms.
Test while-list-each(ass) took 0.240931ms.
Test while-list-each(val) took 0.21557ms.

this is under PHP5, Zend, No accelerator, uncached.

The php source code was:

Code:

hirako:~/test# cat test.php
<?php
$tries=(int)$argv[1];
if ($tries==0) $tries=100000;
$ass=array();
for ($i=0;$i<$tries;$i++) {
        $ass["Ad$i"]="Test$i";
}
$num=array();
for ($i=0;$i<$tries;$i++) {
        $num[]="Test$i";
}
$test= array(
        "foreach"=>'reset($ass);foreach ($ass as $bla) {$b=0;}',
        "foreach-ass"=>'reset($ass);foreach ($ass as $key=>$val) {$b=0;}',
        "for"=>'for ($i=0;$i<$tries;$i++) {$num[$i]="x";}',
        "while"=>'$i=0; while ($i<$tries) {$num[$i]="y";$i++;}',
        "while-list-each(ass)"=>'reset($ass);while (list($key,$val)=each($ass)) {$b=0;}',
        "while-list-each(val)"=>'reset($ass);while (list(,$val)=each($ass)) {$b=0;}',
        );


foreach ($test as $name=>$code) {
        $start=microtime();
        eval($code);
        $end=microtime();
        echo "Test ".$name." took ".($end-$start)."ms.\n";
}

[coolgames]

if you have only one val, while -list *MIGHT* be faster, but it would be good to test that.

[color=Lime Green]Sadly, very few of the LoTGD core uses 1 Val..[/color]
i'll switch to foreach.