Question

I was writing some simple file manipulation and a thought came up if it wouldn’t be faster if I saved the string size in variable. It showed up that it is 10 times faster.
Using this code:

include "../classes/Timer.class.php";
$t = new Timer();             //Timer class I've written for this purpose [link below]
$multiplyer = 3000000;        //Times to try the operation
$string = str_repeat("ggggggggggg",2);  //I first tried 2000 here, but for 2 there are same results
$t("calling");     //Saving time
for($i=0; $i<$multiplyer; $i++) {
  $size =  strlen($string);
  $size2 = strlen($string);
  $size3 = strlen($string);
}
$t("clover");
$t("caching");     //Saving time
for($i=0; $i<$multiplyer; $i++) {
  $size =  strlen($string);
  $size2 = $size;
  $size3 = $size;
}
$t("chover");
$total = $t["calling-clover"]+$t["caching-chover"];  //percents are usefull :)
echo "Calling: {$t["calling-clover"]} (".round(($t["calling-clover"]/$total)*100)."%)<br>\n";
echo "Caching in variables: {$t["caching-chover"]} (".round(($t["caching-chover"]/$total)*100)."%)<br>\n";

Results:

Calling: 1.988455057 (67%)
Caching in variables: 0.984993458 (33%)

What is even more interesting is the fact, that it does not matter what number I put in the str_repeat call, so the strlen obviously does not compute anything – the size must be saved somewhere and strlen is just function that returns value.
This implies:
Are really function calls so slow?
If not, is this strlen specific?

---

Timer.class.php

Answer 1

There is much more to a function call than a variable retrieval. Every time you execute a function:

1. A new stack is created and arguments to the function are stored on the stack.
2. Memory is allocated on the stack for the return value of the function
3. The previously allocated memory address is stored on the stack
4. The address of the function is called
5. The function reads the arguments from the stack
6. Return value is stored in the stack
7. Execution is restored to the caller and the stack is cleared