Benchmark One - High

0.52939612 - 2000 cycles

Benchmark Two - High

0.5329661 - 2000 cycles

Benchmark One - Low

0.58751399000001 - 2000 cycles

Benchmark Two - Low

0.69158023000001 - 2000 cycles

Current Benchmark

<?php 

$template 
'
<html><head><title>{$title}</title></head><body><?php if(true) { ?>TRUE!<?php } else { ?>FALSE!<?php } ?></body></html>
'
;
$array = array (
    
'<?php' => '";',
    
'?>' => 'echo "',
);
$title 'Page Title';
eval(
'echo "'.strtr($template$array).'";');

 
?>
vs.
<?php 

$template 
'
<html><head><title>{$title}</title></head><body><?php if(true) { ?>TRUE!<?php } else { ?>FALSE!<?php } ?></body></html>
'
;
$array = array (
    
'<?php' => '";',
    
'?>' => 'echo "',
);
$title 'Page Title';
eval(
'echo "'.strtr($template$array).'";');

 
?>

Conclusion

This test shows that this type of benchmarking is technically inaccurate. I have repeated each test multiple times to, hopefully, average out all the differences.
However, with the low-load tests, (marked as Low), the more times I repeat the test, the more inaccurate the results become.
With the high-load tests, things are far more accurate, with a 0.02 second margin of error, compared with low-load tests having a 0.15 second margin of error.

Conclusion results are determined by an average of all results.
For any indecisive (are the same, when taking into account the margins of accuracy), I will continue the test until a visible difference occurs.

Eventually, I may decide to restart the test if results appear to be consistantly inconclusive.


Copyright Syridium DV