php.net |  support |  documentation |  report a bug |  advanced search |  search howto |  statistics |  random bug |  login
Bug #34540 dramatic count and strlen slow down on variable reference
Submitted: 2005-09-18 08:18 UTC Modified: 2005-09-18 22:54 UTC
From: php at vicaya dot com Assigned:
Status: Not a bug Package: Performance problem
PHP Version: 4.4.0 OS: Linux 2.6.12
Private report: No CVE-ID: None
Welcome back! If you're the original bug submitter, here's where you can edit the bug or add additional notes.
If this is not your bug, you can add a comment by following this link.
If this is your bug, but you forgot your password, you can retrieve your password here.
Password:
Status:
Package:
Bug Type:
Summary:
From: php at vicaya dot com
New email:
PHP Version: OS:

 

 [2005-09-18 08:18 UTC] php at vicaya dot com
Description:
------------
There is a dramatic (talking about orders of magnitude here) slow down when use count or strlen on referenced variables (including pass by reference and using the global keyword). strace shows excessive brk and mremap activities in such cases.

If you use the foreach loop (commented out in the attached code) to count the elements, the version using reference is faster. See attached code and benchmark for details.

# pass by value and count()
$ /usr/bin/time php passva.php
2.43user 0.30system 0:02.77elapsed 98%CPU (0avgtext+0avgdata 0maxresident)k
0inputs+0outputs (0major+47193minor)pagefaults 0swaps

# pass by reference and count()
/usr/bin/time php passra.php
70.32user 14.52system 1:27.73elapsed 96%CPU (0avgtext+0avgdata 0maxresident)k
0inputs+0outputs (0major+2982532minor)pagefaults 0swaps

# pass by value and foreach loop
$ /usr/bin/time php passva.php
127.99user 1.72system 2:13.97elapsed 96%CPU (0avgtext+0avgdata 0maxresident)k
0inputs+0outputs (0major+276203minor)pagefaults 0swaps

# pass by reference and foreach loop
$ /usr/bin/time php passra.php
47.73user 0.37system 0:49.36elapsed 97%CPU (0avgtext+0avgdata 0maxresident)k
0inputs+0outputs (0major+47194minor)pagefaults 0swaps


Reproduce code:
---------------
<?
# passva.php
for ($i = 0; $i < 1000000; ++$i) { $a[] = $i; }
for($i = 0 ; $i < 100 ; ++$i ) doit($a);
#function doit($arg) { $s = 0; foreach($arg as $i) { ++$s; } }
function doit($arg) { count($arg); }
?>

<?
# passra.php
for ($i = 0; $i < 1000000; ++$i) { $a[] = $i; }
for($i = 0 ; $i < 100 ; ++$i ) doit($a);
#function doit(&$arg) { $s = 0; foreach($arg as $i) { ++$s; } }
function doit(&$arg) { count($arg); }
?>



Patches

Add a Patch

Pull Requests

Add a Pull Request

History

AllCommentsChangesGit/SVN commitsRelated reports
 [2005-09-18 20:13 UTC] derick@php.net
Thank you for taking the time to write to us, but this is not
a bug. Please double-check the documentation available at
http://www.php.net/manual/ and the instructions on how to report
a bug at http://bugs.php.net/how-to-report.php

In case you pass it by reference to the function, and then use the same variable as parameter to a function that doesn\'t accept by reference, then PHP needs to make a copy of the array. In case you do not pass by reference, in this case PHP doesn\'t have to copy the array because it\'s internal refcounting mechanism is smart enough.
If you are just passing by reference to ensure PHP doesn\'t create a copy, then you\'re doing things wrong. You should only pass by reference when you are actually modifying the array inside the function.
 [2005-09-18 22:54 UTC] php at vicaya dot com
Fair enough, as strlen and count is documented to accept a mixed var not a reference. But the builtin functions like count and strlen should be able to handle reference efficiently. Why it'd be so hard to do so? There're a lot of cases you want to use count and strlen on a reference to variable. eg. you want to modify some huge array or string but you want to use count and strlen first to find out the length...

Maybe it's not a language bug per se, But a limitation/problem of the interface of the builtin functions like strlen and count.
 [2013-05-12 16:25 UTC] sergio dot nalin at gmail dot com
May I ask why does PHP need to make a copy of the variable? 
Shouldn't copy-on-write suffice to handle this?

strlen(), count(), etc aren't going to write anything to the variable and 
therefore should not cause a copy (at least for performance's sake).
 
PHP Copyright © 2001-2024 The PHP Group
All rights reserved.
Last updated: Fri Mar 29 15:01:28 2024 UTC