PHP, MySQL, Advanced JS | Tutorials, Tips & Tricks: November 2012

Wednesday, November 7, 2012

PHP Accessing array performance test / benchmark

Hi, All
Today we are going to benchmark array access, the key stored in variable, double quoted and single. Let we go, as usual, we go short and simple! As usual we run 1 million times.

Test Stand:
Hardware:
Intel i7 860 @ 2.8Ghz
Asus P7H55-M PRO
8GB DDR @ 1600Mhz Timings 8-8-8-24
SSD 80GB Intel SA2M080

Software
Windows 7 x64
Apache 2.2.17 (x86)
PHP 5.3.5

Let's begin.

Our test script looks such simple as:

set_time_limit(0);
$time_start = microtime(true);

$arr = array('name'=>'vasya', 'last_name'=>'patrik');
$tmp = NULL;
$i = 0;
while($i < 1000000){
    $tmp = NULL;
    $tmp = $arr["name"]; // different access method
    ++$i;
}
echo number_format(microtime(true)-$time_start, 6, '.', '').' seconds';

First we will run with double quotes. We will use $tmp = $arr["name"].
The results will be quiet interesting, and, as usual we calculate avg time for 10 runs.
First Case: $tmp = $arr["name"]
Result: 0.1790sec.

Second Case: $tmp = $arr['name']
Result: 0.1763sec.

Third Case: $tmp = $arr[$key]
Result: 0.1832sec.

Fourth Case: $tmp = $arr[0](after array refined to use int keys)
Result: 0.1659sec.

PHP Accessing array alternatives benchmark / performance test conclusion:
By the test results you can see that accessing array almost doesn't make sens from performance point of view, as we can see for 1 million assignments of $arr[key] to $tmp the coded runes very quick, 0.16-0.18s. Although makes sens, the fastest way of accessing is by integer key, as not surprisingly accessing array with a key stored in variable the slowest way. Just remember to use int keys whenever is possible, as this way will speed up the execution time of the code.

Sincerely,
Ruskevych Valentin

PHP echo vs print vs printf Performance test / Benchmark / Alternatives

Hi, All
Today we going to test the performance of php's functions: echo versus print versus printf. The test stand as usual, my workstation:)
The test will run 1 million of outputs.

Test Stand:
Hardware:
Intel i7 860 @ 2.8Ghz
Asus P7H55-M PRO
8GB DDR @ 1600Mhz Timings 8-8-8-24
SSD 80GB Intel SA2M080

Software
Windows 7 x64
Apache 2.2.17 (x86)
PHP 5.3.5

Let's begin.

Our test script looks such simple as:

set_time_limit(0);
$time_start = microtime(true);

$i=0;
while($i < 1000000){
    // initial command, whether echo, print of print_r
    ++$i;
}

echo number_format(microtime(true)-$time_start, 6, '.', ''),' seconds \n';

PHP Echo vs Print vs Printf Benchmark results:

* Note: Remember, tests timings are calculated as average based on 10 runs.
Output of string '1' single quoted and double quoted:
echo : Single quotes: 0.111sec. (Double quotes: 0.112sec. )
print : Single quotes: 0.120sec. (Double quotes: 0.120sec. )
printf : Single quotes: 0.468sec. (Double quotes: 0.468sec. )
In the case of sting w/o concatenation, variables, etc. the echo is fastest, while printf is slowest. Also as we can notice the double quotes makes a little difference in execution times. Lets proceed to a bit more complicated benchmarks.

This time we will output 1 character stored in variable for 1 million times, let's see whether we will have the difference.
echo : 0.117sec.
print : 0.128sec.
printf : 0.400sec.
The execution time are growing up, as expected, but printf shown increased performance towards the competitors. Anyways print_r stays slowest option of three.

Case: 'string'.$static_var.'string'.
*Note: static var i assume is 'z', not $i that changes every loop as in next test case.
echo : Single quoted: 0.282sec (Double quoted: 0.261sec )
print : Single quoted: 0.298sec (Double quoted: 0.274sec )
printf : Single quoted: 0.605sec (Double quoted: 0.547sec )
Seems very interesting:) No matter which scenario, the echo is fastest yet.

Next test case will perform string concatenation of three string in a way 'string'.var.'string' (using dots), the var is dynamic variable $i.
In case of printf used %s.
echo : single quoted: 0.481sec. (Double quoted: 0.483sec)
print : single quoted: 0.492sec. (Double quoted: 0.492sec)
printf : single quoted: 0.879sec (Double quoted: 0.879sec)
Conclusion: The echo is fastest yet, while printf is slowest. let's proceed testing other methods.

Next test case is output with commas. ',' rather than concatenating.
Static variable, assumes not changes every loop. Always the same. echo : Single quoted: 0.216sec (Double quoted: 0.243sec)
print : DOESN'T SUPPORT
printf : Single quoted: 0.534sec (Double quoted: 0.572sec)

PHP echo vs. print vs. printf performance/benchmark conclusions:
The conclusion is anyhow the echo is always faster, so preferred way to output data, also you may notice that using single quotes makes the execution a little, but faster. So take it to your coding standard and go with echo and single quotes.

Sincerely,
Valentin Ruskevych

Sunday, November 4, 2012

PHP Switch vs In_Array vs IsSet, Performance test / Alternatives / Benchmark

Hi, All :)
Today I just made a benchmark of PHP's Alternatives of looking up in array with ~30 keys and is around 60 keys. Not that big, but, that is the one i was needed to test for the work ;-)

Test Stand:
Hardware:
Intel i7 860 @ 2.8Ghz
Asus P7H55-M PRO
8GB DDR @ 1600Mhz Timings 8-8-8-24
SSD 80GB Intel SA2M080

Software
Windows 7 x64
Apache 2.2.17 (x86)
PHP 5.3.5

Let's begin.

The goal: Find language and locale by country in predefined values.
Starting point: The switch is used there by one of previous developers.

Test case php's Switch:

set_time_limit(0);
$time_start = microtime(true);
$country='ve';


$iterations = 1000000;
$i=0;
while($i<$iterations){
    $property ='';
    $lang='';
    $locale='';
    switch ($country) {
            case 'ar': $property .= 'ar';  $lang = 'es';  $locale = 'ar';  break; 
            case 'au': $property .= 'au';  $lang = 'en';  $locale = 'au';  break; 
            case 'br': $property .= 'br';  $lang = 'pt-BR'; $locale = 'br';  break; 
            case 'ca': $property .= 'ca';  $lang = 'en';  $locale = 'ca';  break; 
            case 'co': $property .= 'co';  $lang = 'es';  $locale = 'co';  break; 
            case 'fr': $property .= 'fr';  $lang = 'fr';  $locale = 'fr';  break; 
            case 'de': $property .= 'de';  $lang = 'de';  $locale = 'de';  break; 
            case 'hu': $property .= 'hu';  $lang = 'hu';  $locale = 'hu';  break; 
            case 'il': $property .= 'il';  $lang = 'he';  $locale = 'il';  break; 
            case 'it': $property .= 'it';  $lang = 'it';  $locale = 'it';  break; 
            case 'mx': $property .= 'mx';  $lang = 'es';  $locale = 'mx';  break; 
            case 'nl': $property .= 'nl';  $lang = 'nl';  $locale = 'nl';  break; 
            case 'no': $property .= 'no';  $lang = 'no';  $locale = 'no';  break; 
            case 'pt': $property .= 'pt';  $lang = 'pt-PT'; $locale = 'pt';  break; 
            case 'sa': $property .= 'sa';  $lang = 'ar';  $locale = 'sa';  break; 
            case 'za': $property .= 'za';  $lang = null;  $locale = 'za';  break; 
            case 'es': $property .= 'es';  $lang = 'es';  $locale = 'es';  break; 
            case 'se': $property .= 'se';  $lang = 'sv';  $locale = 'se';  break; 
            case 'ch': $property .= 'ch';  $lang = null;  $locale = 'ch';  break; 
            case 'th': $property .= 'th';  $lang = 'th';  $locale = 'th';  break; 
            case 'tr': $property .= 'tr';  $lang = 'tr';  $locale = 'tr';  break; 
            case 'ae': $property .= 'ae';  $lang = 'ar';  $locale = 'ae';  break; 
            case 'gb':
            case 'uk': $property .= 'uk';     $locale = 'uk';  break; 
            case 'us': $property .= 'us';        break; 
            case 've': $property .= 've';  $lang = 'es';  $locale = 've';  break; 
            default: $property .= 'row';     $locale = $country; break; 
    }
    ++$i;
}
$time_end = microtime(true);
echo number_format($time_end-$time_start, 6, '.', '')." seconds - switch 26 options\n";

Test case for php's In_array:

set_time_limit(0);
$time_start = microtime(true);
$country='ve';

$arr = array(
    'ar'=>array('property'=>'ar','lang'=>'es','locale'=>'ar')
    ,'au'=>array('property'=>'au','lang'=>'en','locale'=>'au')
    ,'br'=>array('property'=>'br','lang'=>'pt-BR','locale'=>'br')
    ,'ca'=>array('property'=>'ca','lang'=>'en','locale'=>'ca')
    ,'co'=>array('property'=>'co','lang'=>'es','locale'=>'co')
    ,'fr'=>array('property'=>'fr','lang'=>'fr','locale'=>'fr')
    ,'de'=>array('property'=>'de','lang'=>'de','locale'=>'de')
    ,'hu'=>array('property'=>'hu','lang'=>'hu','locale'=>'hu')
    ,'il'=>array('property'=>'il','lang'=>'he','locale'=>'il')
    ,'it'=>array('property'=>'it','lang'=>'it','locale'=>'it')
    ,'mx'=>array('property'=>'mx','lang'=>'es','locale'=>'mx')
    ,'nl'=>array('property'=>'nl','lang'=>'nl','locale'=>'nl')
    ,'no'=>array('property'=>'no','lang'=>'no','locale'=>'no')
    ,'pt'=>array('property'=>'pt','lang'=>'pt-PT','locale'=>'pt')
    ,'sa'=>array('property'=>'sa','lang'=>'ar','locale'=>'sa')
    ,'za'=>array('property'=>'za','lang'=>NULL,'locale'=>'za')
    ,'es'=>array('property'=>'es','lang'=>'es','locale'=>'es')
    ,'se'=>array('property'=>'se','lang'=>'sv','locale'=>'se')
    ,'ch'=>array('property'=>'ch','lang'=>NULL,'locale'=>'ch')
    ,'th'=>array('property'=>'th','lang'=>'th','locale'=>'th')
    ,'tr'=>array('property'=>'tr','lang'=>'tr','locale'=>'tr')
    ,'ae'=>array('property'=>'ae','lang'=>'ar','locale'=>'ae')
    ,'gb'=>array('property'=>'gb','lang'=>'en','locale'=>'uk')
    ,'uk'=>array('property'=>'uk','lang'=>'en','locale'=>'uk')
    ,'us'=>array('property'=>'us','lang'=>'en','locale'=>'us')
    ,'ve'=>array('property'=>'ve','lang'=>'es','locale'=>'ve')
);

$iterations = 1000000;
$i=0;
while($i<$iterations){
    $property ='';
    $lang='';
    $locale='';
    if(in_array($country, $arr) !== FALSE){
        $property = $arr[$country]['property'];
        $lang = $arr[$country]['lang'];
        $locale = $arr[$country]['locale'];
    }else{
        $property = 'row';
        $lang = 'en';
        $locale = $country;
    }
    ++$i;
}
$time_end = microtime(true);
echo number_format($time_end-$time_start, 6, '.', '')." seconds - in_array 26 options \n";

Test case for PHP's IsSet:

set_time_limit(0);
$time_start = microtime(true);
$country='ve';

$arr = array(
    'ar'=>array('property'=>'ar','lang'=>'es','locale'=>'ar')
    ,'au'=>array('property'=>'au','lang'=>'en','locale'=>'au')
    ,'br'=>array('property'=>'br','lang'=>'pt-BR','locale'=>'br')
    ,'ca'=>array('property'=>'ca','lang'=>'en','locale'=>'ca')
    ,'co'=>array('property'=>'co','lang'=>'es','locale'=>'co')
    ,'fr'=>array('property'=>'fr','lang'=>'fr','locale'=>'fr')
    ,'de'=>array('property'=>'de','lang'=>'de','locale'=>'de')
    ,'hu'=>array('property'=>'hu','lang'=>'hu','locale'=>'hu')
    ,'il'=>array('property'=>'il','lang'=>'he','locale'=>'il')
    ,'it'=>array('property'=>'it','lang'=>'it','locale'=>'it')
    ,'mx'=>array('property'=>'mx','lang'=>'es','locale'=>'mx')
    ,'nl'=>array('property'=>'nl','lang'=>'nl','locale'=>'nl')
    ,'no'=>array('property'=>'no','lang'=>'no','locale'=>'no')
    ,'pt'=>array('property'=>'pt','lang'=>'pt-PT','locale'=>'pt')
    ,'sa'=>array('property'=>'sa','lang'=>'ar','locale'=>'sa')
    ,'za'=>array('property'=>'za','lang'=>NULL,'locale'=>'za')
    ,'es'=>array('property'=>'es','lang'=>'es','locale'=>'es')
    ,'se'=>array('property'=>'se','lang'=>'sv','locale'=>'se')
    ,'ch'=>array('property'=>'ch','lang'=>NULL,'locale'=>'ch')
    ,'th'=>array('property'=>'th','lang'=>'th','locale'=>'th')
    ,'tr'=>array('property'=>'tr','lang'=>'tr','locale'=>'tr')
    ,'ae'=>array('property'=>'ae','lang'=>'ar','locale'=>'ae')
    ,'gb'=>array('property'=>'gb','lang'=>'en','locale'=>'uk')
    ,'uk'=>array('property'=>'uk','lang'=>'en','locale'=>'uk')
    ,'us'=>array('property'=>'us','lang'=>'en','locale'=>'us')
    ,'ve'=>array('property'=>'ve','lang'=>'es','locale'=>'ve')
);

$iterations = 1000000;
$i=0;
while($i<$iterations){
    $property ='';
    $lang='';
    $locale='';
    if(isset($arr[$country])){
        $property = $arr[$country]['property'];
        $lang = $arr[$country]['lang'];
        $locale = $arr[$country]['locale'];
    }else{
        $property = 'row';
        $lang = 'en';
        $locale = $country;
    }
    ++$i;
}
$time_end = microtime(true);
echo number_format($time_end-$time_start, 6, '.', '')." seconds - isset 26 options array \n";

To tell you the truth i was waited for In_Array to be slower than switch :), but let's take a look at the benchmark results.
PHP's Switch vs In_Array vs IsSet Benchmark Results:
*Notes:
The runs are (loop) 1 Million times
The time is counted as average for 10 runs * 1 million loops.
The time surrounded by closures is the time for the same run, but with *2 amount of keys in arrays.

The times
The key is set to last key in array
----------------------
For the Switch took: 2.094s. (3.726s)
For the In_Array took: 1.617s. (2.414s)
For the IsSet took: 0.848s. (0.899s.)
----------------------

The match is set to first key in array case:
----------------------
For the Switch: 0.569s. (0.629s) (great speed increase).
For the In_Array: 1.617s. (2.425s) (the execution still unchanged).
For the IsSet: 0.860s. (0.897s) (same with isset).
----------------------

No Match Case:
----------------------
For the Switch: 2.032s.(3.509s) (faster than last match. LOL)
For the In_Array: 1.657s. (2.372s) (Time almost not changing here)
For the IsSet: 0.540s. (0.507s) (Great increase!)
----------------------

Conclusion:
The best way to find whether the key is set in array is to use IsSet php's function.
As we see by the tests, Switch can bring performance only in case the key one of the first, the in_array in any case scanning array to the end and isset is as fast as lightning and becomes even faster when no match is found !
Also please note, as the amount of cases and array keys will become bigger, the switch and in_array will become slower, while isset doesn't have such issue.

The code isset($arr[$country]) brought to us best results.
Have fun playing around.

Sincerely,
Ruskevych Valentin