Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:1862
Mailing-List: contact internals-help@lists.php.net; run by ezmlm
To: "Brian J. France" <list@firehawksystems.com>
Cc: internals@lists.php.net
In-Reply-To: <20030521093640446007.GyazMail.list@firehawksystems.com>
References: <20030521093640446007.GyazMail.list@firehawksystems.com>
Content-Type: text/plain
Organization: 
Message-ID: <1053530841.2177.51.camel@hasele>
Mime-Version: 1.0
Date: 21 May 2003 11:27:21 -0400
Content-Transfer-Encoding: 7bit
Subject: Re: [PHP-DEV] smart_str performance
From: sterling@bumblebury.com (Sterling Hughes)

Switched it to avoid a copy on convert_to_string().  Performance in PHP5
now beats the previous version.

-Sterling

On Wed, 2003-05-21 at 12:36, Brian J. France wrote:
> This was a email I sent to Rasmus and another co-worker a while ago and forgot 
> to post the question to the list.  I just saw a commit converting zlib to 
> smart_str and figured I would send it now.
> 
> Comments?
> 
> Brian
> 
> ----
> I think I found what you were looking for yesterday.  implode has change 
> between 4.3 and 4.2.  4.2 does the double pass while 4.3 has changed to use a 
> new smart_str struct with smart_str_appendl function (a struct and function 
> calls around what I was doing in my extension).  I decided to run some 
> quick tests and here is what I have found:
> 
> for ( $i=0; $i < $count; $i++ ) {
>   $arr[] = rand_string_function(); // 12 characters
> }
> 
> $start = gettimeofday();
> $str = implode( "", $arr );
> $end = gettimeofday();
> printf( "%8.8d: %8.8f\n", $count, time_diff( $start, $end ) );
> 
> give this output: 
> 
> 4.3:
>        1:        0.00005100
>        2:        0.00002800
>        5:        0.00003700
>       10:        0.00006000
>      100:        0.00045600
>     1000:        0.00478600
>    10000:        0.06507500
>   100000:        1.71620700
> 
> 4.2:
>        1:        0.00004700
>        2:        0.00002500
>        5:        0.00002800
>       10:        0.00004000
>      100:        0.00022400
>     1000:        0.00240700
>    10000:        0.03677200
>   100000:        0.37068200
> 
> While below 10 items I am sure it is hard to tell the difference, but it is 
> slower in all cases.  Any clue why one item is slower than two or even five?
> 
> Brian
-- 
"First they ignore you, then they laugh at you,  
 then they fight you, then you win."  
    - Gandhi