generic-library/boost
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity
main
boost/doc/html/intrusive/performance.html

2023 lines
77 KiB
HTML
Raw Permalink Normal View History

[DEV] add v1.76.0
2021-10-05 21:37:46 +02:00
<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
<title>Performance</title>
<link rel="stylesheet" href="../../../doc/src/boostbook.css" type="text/css">
<meta name="generator" content="DocBook XSL Stylesheets V1.79.1">
<link rel="home" href="../index.html" title="The Boost C++ Libraries BoostBook Documentation Subset">
<link rel="up" href="../intrusive.html" title="Chapter 19. Boost.Intrusive">
<link rel="prev" href="design_notes.html" title="Design Notes">
<link rel="next" href="release_notes.html" title="Release Notes">
</head>
<body bgcolor="white" text="black" link="#0000FF" vlink="#840084" alink="#0000FF">
<table cellpadding="2" width="100%"><tr>
<td valign="top"><img alt="Boost C++ Libraries" width="277" height="86" src="../../../boost.png"></td>
<td align="center"><a href="../../../index.html">Home</a></td>
<td align="center"><a href="../../../libs/libraries.htm">Libraries</a></td>
<td align="center"><a href="http://www.boost.org/users/people.html">People</a></td>
<td align="center"><a href="http://www.boost.org/users/faq.html">FAQ</a></td>
<td align="center"><a href="../../../more/index.htm">More</a></td>
</tr></table>
<hr>
<div class="spirit-nav">
<a accesskey="p" href="design_notes.html"><img src="../../../doc/src/images/prev.png" alt="Prev"></a><a accesskey="u" href="../intrusive.html"><img src="../../../doc/src/images/up.png" alt="Up"></a><a accesskey="h" href="../index.html"><img src="../../../doc/src/images/home.png" alt="Home"></a><a accesskey="n" href="release_notes.html"><img src="../../../doc/src/images/next.png" alt="Next"></a>
</div>
<div class="section">
<div class="titlepage"><div><div><h2 class="title" style="clear: both">
<a name="intrusive.performance"></a><a class="link" href="performance.html" title="Performance">Performance</a>
</h2></div></div></div>
<div class="toc"><dl class="toc">
<dt><span class="section"><a href="performance.html#intrusive.performance.performance_results_push_back">Back
insertion and destruction</a></span></dt>
<dt><span class="section"><a href="performance.html#intrusive.performance.performance_results_reversing">Reversing</a></span></dt>
<dt><span class="section"><a href="performance.html#intrusive.performance.performance_results_sorting">Sorting</a></span></dt>
<dt><span class="section"><a href="performance.html#intrusive.performance.performance_results_write_access">Write
access</a></span></dt>
<dt><span class="section"><a href="performance.html#intrusive.performance.performance_results_conclusions">Conclusions</a></span></dt>
</dl></div>
<p>
<span class="bold"><strong>Boost.Intrusive</strong></span> containers offer speed improvements
compared to non-intrusive containers primarily because:
</p>
<div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; ">
<li class="listitem">
They minimize memory allocation/deallocation calls.
</li>
<li class="listitem">
They obtain better memory locality.
</li>
</ul></div>
<p>
This section will show performance tests comparing some operations on <code class="computeroutput"><span class="identifier">boost</span><span class="special">::</span><span class="identifier">intrusive</span><span class="special">::</span><span class="identifier">list</span></code> and
<code class="computeroutput"><span class="identifier">std</span><span class="special">::</span><span class="identifier">list</span></code>:
</p>
<div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; ">
<li class="listitem">
Insertions using <code class="computeroutput"><span class="identifier">push_back</span></code>
and container destruction will show the overhead associated with memory
allocation/deallocation.
</li>
<li class="listitem">
The <code class="computeroutput"><span class="identifier">reverse</span></code> member function
will show the advantages of the compact memory representation that can
be achieved with intrusive containers.
</li>
<li class="listitem">
The <code class="computeroutput"><span class="identifier">sort</span></code> and <code class="computeroutput"><span class="identifier">write</span> <span class="identifier">access</span></code>
tests will show the advantage of intrusive containers minimizing memory
accesses compared to containers of pointers.
</li>
</ul></div>
<p>
Given an object of type <code class="computeroutput"><span class="identifier">T</span></code>,
<code class="computeroutput"><a class="link" href="../boost/intrusive/list.html" title="Class template list">boost::intrusive::list&lt;T&gt;</a></code>
can replace <code class="computeroutput"><span class="identifier">std</span><span class="special">::</span><span class="identifier">list</span><span class="special">&lt;</span><span class="identifier">T</span><span class="special">&gt;</span></code> to
avoid memory allocation overhead, or it can replace <code class="computeroutput"><span class="identifier">std</span><span class="special">::</span><span class="identifier">list</span><span class="special">&lt;</span><span class="identifier">T</span><span class="special">*&gt;</span></code> when
the user wants containers with polymorphic values or wants to share values
between several containers. Because of this versatility, the performance tests
will be executed for 6 different list types:
</p>
<div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; ">
<li class="listitem">
3 intrusive lists holding a class named <code class="computeroutput"><span class="identifier">itest_class</span></code>,
each one with a different linking policy (<code class="computeroutput"><span class="identifier">normal_link</span></code>,
<code class="computeroutput"><span class="identifier">safe_link</span></code>, <code class="computeroutput"><span class="identifier">auto_unlink</span></code>). The <code class="computeroutput"><span class="identifier">itest_class</span></code>
objects will be tightly packed in a <code class="computeroutput"><span class="identifier">std</span><span class="special">::</span><span class="identifier">vector</span><span class="special">&lt;</span><span class="identifier">itest_class</span><span class="special">&gt;</span></code> object.
</li>
<li class="listitem">
<code class="computeroutput"><span class="identifier">std</span><span class="special">::</span><span class="identifier">list</span><span class="special">&lt;</span><span class="identifier">test_class</span><span class="special">&gt;</span></code>,
where <code class="computeroutput"><span class="identifier">test_class</span></code> is exactly
the same as <code class="computeroutput"><span class="identifier">itest_class</span></code>,
but without the intrusive hook.
</li>
<li class="listitem">
<code class="computeroutput"><span class="identifier">std</span><span class="special">::</span><span class="identifier">list</span><span class="special">&lt;</span><span class="identifier">test_class</span><span class="special">*&gt;</span></code>.
The list will contain pointers to <code class="computeroutput"><span class="identifier">test_class</span></code>
objects tightly packed in a <code class="computeroutput"><span class="identifier">std</span><span class="special">::</span><span class="identifier">vector</span><span class="special">&lt;</span><span class="identifier">test_class</span><span class="special">&gt;</span></code> object. We'll call this configuration
<span class="emphasis"><em>compact pointer list</em></span>
</li>
<li class="listitem">
<code class="computeroutput"><span class="identifier">std</span><span class="special">::</span><span class="identifier">list</span><span class="special">&lt;</span><span class="identifier">test_class</span><span class="special">*&gt;</span></code>.
The list will contain pointers to <code class="computeroutput"><span class="identifier">test_class</span></code>
objects owned by a <code class="computeroutput"><span class="identifier">std</span><span class="special">::</span><span class="identifier">list</span><span class="special">&lt;</span><span class="identifier">test_class</span><span class="special">&gt;</span></code> object. We'll call this configuration
<span class="emphasis"><em>disperse pointer list</em></span>.
</li>
</ul></div>
<p>
Both <code class="computeroutput"><span class="identifier">test_class</span></code> and <code class="computeroutput"><span class="identifier">itest_class</span></code> are templatized classes that
can be configured with a boolean to increase the size of the object. This way,
the tests can be executed with small and big objects. Here is the first part
of the testing code, which shows the definitions of <code class="computeroutput"><span class="identifier">test_class</span></code>
and <code class="computeroutput"><span class="identifier">itest_class</span></code> classes, and
some other utilities:
</p>
<pre class="programlisting"><span class="comment">//Iteration and element count defines</span>
<span class="keyword">const</span> <span class="keyword">int</span> <span class="identifier">NumIter</span> <span class="special">=</span> <span class="number">4</span><span class="special">;</span>
<span class="keyword">const</span> <span class="keyword">int</span> <span class="identifier">NumElements</span> <span class="special">=</span> <span class="number">50000</span><span class="special">;</span>
<span class="keyword">using</span> <span class="keyword">namespace</span> <span class="identifier">boost</span><span class="special">::</span><span class="identifier">intrusive</span><span class="special">;</span>
<span class="keyword">template</span><span class="special">&lt;</span><span class="keyword">bool</span> <span class="identifier">BigSize</span><span class="special">&gt;</span> <span class="keyword">struct</span> <span class="identifier">filler</span> <span class="special">{</span> <span class="keyword">int</span> <span class="identifier">dummy</span><span class="special">[</span><span class="number">10</span><span class="special">];</span> <span class="special">};</span>
<span class="keyword">template</span> <span class="special">&lt;&gt;</span> <span class="keyword">struct</span> <span class="identifier">filler</span><span class="special">&lt;</span><span class="keyword">false</span><span class="special">&gt;</span> <span class="special">{};</span>
<span class="keyword">template</span><span class="special">&lt;</span><span class="keyword">bool</span> <span class="identifier">BigSize</span><span class="special">&gt;</span> <span class="comment">//The object for non-intrusive containers</span>
<span class="keyword">struct</span> <span class="identifier">test_class</span> <span class="special">:</span> <span class="keyword">private</span> <span class="identifier">filler</span><span class="special">&lt;</span><span class="identifier">BigSize</span><span class="special">&gt;</span>
<span class="special">{</span>
<span class="keyword">int</span> <span class="identifier">i_</span><span class="special">;</span>
<span class="identifier">test_class</span><span class="special">()</span> <span class="special">{}</span>
<span class="identifier">test_class</span><span class="special">(</span><span class="keyword">int</span> <span class="identifier">i</span><span class="special">)</span> <span class="special">:</span> <span class="identifier">i_</span><span class="special">(</span><span class="identifier">i</span><span class="special">)</span> <span class="special">{}</span>
<span class="keyword">friend</span> <span class="keyword">bool</span> <span class="keyword">operator</span> <span class="special">&lt;(</span><span class="keyword">const</span> <span class="identifier">test_class</span> <span class="special">&amp;</span><span class="identifier">l</span><span class="special">,</span> <span class="keyword">const</span> <span class="identifier">test_class</span> <span class="special">&amp;</span><span class="identifier">r</span><span class="special">)</span> <span class="special">{</span> <span class="keyword">return</span> <span class="identifier">l</span><span class="special">.</span><span class="identifier">i_</span> <span class="special">&lt;</span> <span class="identifier">r</span><span class="special">.</span><span class="identifier">i_</span><span class="special">;</span> <span class="special">}</span>
<span class="keyword">friend</span> <span class="keyword">bool</span> <span class="keyword">operator</span> <span class="special">&gt;(</span><span class="keyword">const</span> <span class="identifier">test_class</span> <span class="special">&amp;</span><span class="identifier">l</span><span class="special">,</span> <span class="keyword">const</span> <span class="identifier">test_class</span> <span class="special">&amp;</span><span class="identifier">r</span><span class="special">)</span> <span class="special">{</span> <span class="keyword">return</span> <span class="identifier">l</span><span class="special">.</span><span class="identifier">i_</span> <span class="special">&gt;</span> <span class="identifier">r</span><span class="special">.</span><span class="identifier">i_</span><span class="special">;</span> <span class="special">}</span>
<span class="special">};</span>
<span class="keyword">template</span> <span class="special">&lt;</span><span class="keyword">bool</span> <span class="identifier">BigSize</span><span class="special">,</span> <span class="identifier">link_mode_type</span> <span class="identifier">LinkMode</span><span class="special">&gt;</span>
<span class="keyword">struct</span> <span class="identifier">itest_class</span> <span class="comment">//The object for intrusive containers</span>
<span class="special">:</span> <span class="keyword">public</span> <span class="identifier">list_base_hook</span><span class="special">&lt;</span><span class="identifier">link_mode</span><span class="special">&lt;</span><span class="identifier">LinkMode</span><span class="special">&gt;</span> <span class="special">&gt;,</span> <span class="keyword">public</span> <span class="identifier">test_class</span><span class="special">&lt;</span><span class="identifier">BigSize</span><span class="special">&gt;</span>
<span class="special">{</span>
<span class="identifier">itest_class</span><span class="special">()</span> <span class="special">{}</span>
<span class="identifier">itest_class</span><span class="special">(</span><span class="keyword">int</span> <span class="identifier">i</span><span class="special">)</span> <span class="special">:</span> <span class="identifier">test_class</span><span class="special">&lt;</span><span class="identifier">BigSize</span><span class="special">&gt;(</span><span class="identifier">i</span><span class="special">)</span> <span class="special">{}</span>
<span class="special">};</span>
<span class="keyword">template</span><span class="special">&lt;</span><span class="keyword">class</span> <span class="identifier">FuncObj</span><span class="special">&gt;</span> <span class="comment">//Adapts functors taking values to functors taking pointers</span>
<span class="keyword">struct</span> <span class="identifier">func_ptr_adaptor</span> <span class="special">:</span> <span class="keyword">public</span> <span class="identifier">FuncObj</span>
<span class="special">{</span>
<span class="keyword">typedef</span> <span class="keyword">typename</span> <span class="identifier">FuncObj</span><span class="special">::</span><span class="identifier">first_argument_type</span><span class="special">*</span> <span class="identifier">first_argument_type</span><span class="special">;</span>
<span class="keyword">typedef</span> <span class="keyword">typename</span> <span class="identifier">FuncObj</span><span class="special">::</span><span class="identifier">second_argument_type</span><span class="special">*</span> <span class="identifier">second_argument_type</span><span class="special">;</span>
<span class="keyword">typedef</span> <span class="keyword">typename</span> <span class="identifier">FuncObj</span><span class="special">::</span><span class="identifier">result_type</span> <span class="identifier">result_type</span><span class="special">;</span>
<span class="identifier">result_type</span> <span class="keyword">operator</span><span class="special">()(</span><span class="identifier">first_argument_type</span> <span class="identifier">a</span><span class="special">,</span> <span class="identifier">second_argument_type</span> <span class="identifier">b</span><span class="special">)</span> <span class="keyword">const</span>
<span class="special">{</span> <span class="keyword">return</span> <span class="identifier">FuncObj</span><span class="special">::</span><span class="keyword">operator</span><span class="special">()(*</span><span class="identifier">a</span><span class="special">,</span> <span class="special">*</span><span class="identifier">b</span><span class="special">);</span> <span class="special">}</span>
<span class="special">};</span>
</pre>
<p>
As we can see, <code class="computeroutput"><span class="identifier">test_class</span></code> is
a very simple class holding an <code class="computeroutput"><span class="keyword">int</span></code>.
<code class="computeroutput"><span class="identifier">itest_class</span></code> is just a class
that has a base hook (<code class="computeroutput"><a class="link" href="../boost/intrusive/list_base_hook.html" title="Class template list_base_hook">list_base_hook</a></code>)
and also derives from <code class="computeroutput"><span class="identifier">test_class</span></code>.
</p>
<p>
<code class="computeroutput"><span class="identifier">func_ptr_adaptor</span></code> is just a
functor adaptor to convert function objects taking <code class="computeroutput"><span class="identifier">test_list</span></code>
objects to function objects taking pointers to them.
</p>
<p>
You can find the full test code in the <a href="../../../libs/intrusive/perf/perf_list.cpp" target="_top">perf_list.cpp</a>
source file.
</p>
<div class="section">
<div class="titlepage"><div><div><h3 class="title">
<a name="intrusive.performance.performance_results_push_back"></a><a class="link" href="performance.html#intrusive.performance.performance_results_push_back" title="Back insertion and destruction">Back
insertion and destruction</a>
</h3></div></div></div>
<p>
The first test will measure the benefits we can obtain with intrusive containers
avoiding memory allocations and deallocations. All the objects to be inserted
in intrusive containers are allocated in a single allocation call, whereas
<code class="computeroutput"><span class="identifier">std</span><span class="special">::</span><span class="identifier">list</span></code> will need to allocate memory for each
object and deallocate it for every erasure (or container destruction).
</p>
<p>
Let's compare the code to be executed for each container type for different
insertion tests:
</p>
<pre class="programlisting"><span class="identifier">std</span><span class="special">::</span><span class="identifier">vector</span><span class="special">&lt;</span><span class="keyword">typename</span> <span class="identifier">ilist</span><span class="special">::</span><span class="identifier">value_type</span><span class="special">&gt;</span> <span class="identifier">objects</span><span class="special">(</span><span class="identifier">NumElements</span><span class="special">);</span>
<span class="identifier">ilist</span> <span class="identifier">l</span><span class="special">;</span>
<span class="keyword">for</span><span class="special">(</span><span class="keyword">int</span> <span class="identifier">i</span> <span class="special">=</span> <span class="number">0</span><span class="special">;</span> <span class="identifier">i</span> <span class="special">&lt;</span> <span class="identifier">NumElements</span><span class="special">;</span> <span class="special">++</span><span class="identifier">i</span><span class="special">)</span>
<span class="identifier">l</span><span class="special">.</span><span class="identifier">push_back</span><span class="special">(</span><span class="identifier">objects</span><span class="special">[</span><span class="identifier">i</span><span class="special">]);</span>
<span class="comment">//Elements are unlinked in ilist's destructor</span>
<span class="comment">//Elements are destroyed in vector's destructor</span>
</pre>
<p>
For intrusive containers, all the values are created in a vector and after
that inserted in the intrusive list.
</p>
<pre class="programlisting"><span class="identifier">stdlist</span> <span class="identifier">l</span><span class="special">;</span>
<span class="keyword">for</span><span class="special">(</span><span class="keyword">int</span> <span class="identifier">i</span> <span class="special">=</span> <span class="number">0</span><span class="special">;</span> <span class="identifier">i</span> <span class="special">&lt;</span> <span class="identifier">NumElements</span><span class="special">;</span> <span class="special">++</span><span class="identifier">i</span><span class="special">)</span>
<span class="identifier">l</span><span class="special">.</span><span class="identifier">push_back</span><span class="special">(</span><span class="keyword">typename</span> <span class="identifier">stdlist</span><span class="special">::</span><span class="identifier">value_type</span><span class="special">(</span><span class="identifier">i</span><span class="special">));</span>
<span class="comment">//Elements unlinked and destroyed in stdlist's destructor</span>
</pre>
<p>
For a standard list, elements are pushed back using push_back().
</p>
<pre class="programlisting"><span class="identifier">std</span><span class="special">::</span><span class="identifier">vector</span><span class="special">&lt;</span><span class="keyword">typename</span> <span class="identifier">stdlist</span><span class="special">::</span><span class="identifier">value_type</span><span class="special">&gt;</span> <span class="identifier">objects</span><span class="special">(</span><span class="identifier">NumElements</span><span class="special">);</span>
<span class="identifier">stdptrlist</span> <span class="identifier">l</span><span class="special">;</span>
<span class="keyword">for</span><span class="special">(</span><span class="keyword">int</span> <span class="identifier">i</span> <span class="special">=</span> <span class="number">0</span><span class="special">;</span> <span class="identifier">i</span> <span class="special">&lt;</span> <span class="identifier">NumElements</span><span class="special">;</span> <span class="special">++</span><span class="identifier">i</span><span class="special">)</span>
<span class="identifier">l</span><span class="special">.</span><span class="identifier">push_back</span><span class="special">(&amp;</span><span class="identifier">objects</span><span class="special">[</span><span class="identifier">i</span><span class="special">]);</span>
<span class="comment">//Pointers to elements unlinked and destroyed in stdptrlist's destructor</span>
<span class="comment">//Elements destroyed in vector's destructor</span>
</pre>
<p>
For a standard compact pointer list, elements are created in a vector and
pushed back in the pointer list using push_back().
</p>
<pre class="programlisting"><span class="identifier">stdlist</span> <span class="identifier">objects</span><span class="special">;</span> <span class="identifier">stdptrlist</span> <span class="identifier">l</span><span class="special">;</span>
<span class="keyword">for</span><span class="special">(</span><span class="keyword">int</span> <span class="identifier">i</span> <span class="special">=</span> <span class="number">0</span><span class="special">;</span> <span class="identifier">i</span> <span class="special">&lt;</span> <span class="identifier">NumElements</span><span class="special">;</span> <span class="special">++</span><span class="identifier">i</span><span class="special">){</span>
<span class="identifier">objects</span><span class="special">.</span><span class="identifier">push_back</span><span class="special">(</span><span class="keyword">typename</span> <span class="identifier">stdlist</span><span class="special">::</span><span class="identifier">value_type</span><span class="special">(</span><span class="identifier">i</span><span class="special">));</span>
<span class="identifier">l</span><span class="special">.</span><span class="identifier">push_back</span><span class="special">(&amp;</span><span class="identifier">objects</span><span class="special">.</span><span class="identifier">back</span><span class="special">());</span>
<span class="special">}</span>
<span class="comment">//Pointers to elements unlinked and destroyed in stdptrlist's destructor</span>
<span class="comment">//Elements unlinked and destroyed in stdlist's destructor</span>
</pre>
<p>
For a <span class="emphasis"><em>disperse pointer list</em></span>, elements are created in
a list and pushed back in the pointer list using push_back().
</p>
<p>
These are the times in microseconds for each case, and the normalized time:
</p>
<div class="table">
<a name="intrusive.performance.performance_results_push_back.back_insertion_destruction_times"></a><p class="title"><b>Table 19.2. Back insertion + destruction times for Visual C++ 7.1 / Windows XP</b></p>
<div class="table-contents"><table class="table" summary="Back insertion + destruction times for Visual C++ 7.1 / Windows XP">
<colgroup>
<col>
<col>
<col>
</colgroup>
<thead><tr>
<th>
<p>
Container
</p>
</th>
<th>
<p>
Time in us/iteration (small object / big object)
</p>
</th>
<th>
<p>
Normalized time (small object / big object)
</p>
</th>
</tr></thead>
<tbody>
<tr>
<td>
<p>
<code class="computeroutput"><span class="identifier">normal_link</span></code> intrusive
list
</p>
</td>
<td>
<p>
5000 / 22500
</p>
</td>
<td>
<p>
1 / 1
</p>
</td>
</tr>
<tr>
<td>
<p>
<code class="computeroutput"><span class="identifier">safe_link</span></code> intrusive
list
</p>
</td>
<td>
<p>
7812 / 32187
</p>
</td>
<td>
<p>
1.56 / 1.43
</p>
</td>
</tr>
<tr>
<td>
<p>
<code class="computeroutput"><span class="identifier">auto_unlink</span></code> intrusive
list
</p>
</td>
<td>
<p>
10156 / 41562
</p>
</td>
<td>
<p>
2.03 / 1.84
</p>
</td>
</tr>
<tr>
<td>
<p>
Standard list
</p>
</td>
<td>
<p>
26875 / 97500
</p>
</td>
<td>
<p>
5.37 / 4.33
</p>
</td>
</tr>
<tr>
<td>
<p>
Standard compact pointer list
</p>
</td>
<td>
<p>
76406 / 86718
</p>
</td>
<td>
<p>
15.28 / 3.85
</p>
</td>
</tr>
<tr>
<td>
<p>
Standard disperse pointer list
</p>
</td>
<td>
<p>
146562 / 175625
</p>
</td>
<td>
<p>
29.31 / 7.80
</p>
</td>
</tr>
</tbody>
</table></div>
</div>
<br class="table-break"><div class="table">
<a name="intrusive.performance.performance_results_push_back.back_insertion_destruction_time0"></a><p class="title"><b>Table 19.3. Back insertion + destruction times for GCC 4.1.1 / MinGW over Windows
XP</b></p>
<div class="table-contents"><table class="table" summary="Back insertion + destruction times for GCC 4.1.1 / MinGW over Windows
XP">
<colgroup>
<col>
<col>
<col>
</colgroup>
<thead><tr>
<th>
<p>
Container
</p>
</th>
<th>
<p>
Time in us/iteration (small object / big object)
</p>
</th>
<th>
<p>
Normalized time (small object / big object)
</p>
</th>
</tr></thead>
<tbody>
<tr>
<td>
<p>
<code class="computeroutput"><span class="identifier">normal_link</span></code> intrusive
list
</p>
</td>
<td>
<p>
4375 / 22187
</p>
</td>
<td>
<p>
1 / 1
</p>
</td>
</tr>
<tr>
<td>
<p>
<code class="computeroutput"><span class="identifier">safe_link</span></code> intrusive
list
</p>
</td>
<td>
<p>
7812 / 32812
</p>
</td>
<td>
<p>
1.78 / 1.47
</p>
</td>
</tr>
<tr>
<td>
<p>
<code class="computeroutput"><span class="identifier">auto_unlink</span></code> intrusive
list
</p>
</td>
<td>
<p>
10468 / 42031
</p>
</td>
<td>
<p>
2.39 / 1.89
</p>
</td>
</tr>
<tr>
<td>
<p>
Standard list
</p>
</td>
<td>
<p>
81250 / 98125
</p>
</td>
<td>
<p>
18.57 / 4.42
</p>
</td>
</tr>
<tr>
<td>
<p>
Standard compact pointer list
</p>
</td>
<td>
<p>
83750 / 94218
</p>
</td>
<td>
<p>
19.14 / 4.24
</p>
</td>
</tr>
<tr>
<td>
<p>
Standard disperse pointer list
</p>
</td>
<td>
<p>
155625 / 175625
</p>
</td>
<td>
<p>
35.57 / 7.91
</p>
</td>
</tr>
</tbody>
</table></div>
</div>
<br class="table-break"><div class="table">
<a name="intrusive.performance.performance_results_push_back.back_insertion_destruction_time1"></a><p class="title"><b>Table 19.4. Back insertion + destruction times for GCC 4.1.2 / Linux Kernel 2.6.18
(OpenSuse 10.2)</b></p>
<div class="table-contents"><table class="table" summary="Back insertion + destruction times for GCC 4.1.2 / Linux Kernel 2.6.18
(OpenSuse 10.2)">
<colgroup>
<col>
<col>
<col>
</colgroup>
<thead><tr>
<th>
<p>
Container
</p>
</th>
<th>
<p>
Time in us/iteration (small object / big object)
</p>
</th>
<th>
<p>
Normalized time (small object / big object)
</p>
</th>
</tr></thead>
<tbody>
<tr>
<td>
<p>
<code class="computeroutput"><span class="identifier">normal_link</span></code> intrusive
list
</p>
</td>
<td>
<p>
4792 / 20495
</p>
</td>
<td>
<p>
1 / 1
</p>
</td>
</tr>
<tr>
<td>
<p>
<code class="computeroutput"><span class="identifier">safe_link</span></code> intrusive
list
</p>
</td>
<td>
<p>
7709 / 30803
</p>
</td>
<td>
<p>
1.60 / 1.5
</p>
</td>
</tr>
<tr>
<td>
<p>
<code class="computeroutput"><span class="identifier">auto_unlink</span></code> intrusive
list
</p>
</td>
<td>
<p>
10180 / 41183
</p>
</td>
<td>
<p>
2.12 / 2.0
</p>
</td>
</tr>
<tr>
<td>
<p>
Standard list
</p>
</td>
<td>
<p>
17031 / 32586
</p>
</td>
<td>
<p>
3.55 / 1.58
</p>
</td>
</tr>
<tr>
<td>
<p>
Standard compact pointer list
</p>
</td>
<td>
<p>
27221 / 34823
</p>
</td>
<td>
<p>
5.68 / 1.69
</p>
</td>
</tr>
<tr>
<td>
<p>
Standard disperse pointer list
</p>
</td>
<td>
<p>
102272 / 60056
</p>
</td>
<td>
<p>
21.34 / 2.93
</p>
</td>
</tr>
</tbody>
</table></div>
</div>
<br class="table-break"><p>
The results are logical: intrusive lists just need one allocation. The destruction
time of the <code class="computeroutput"><span class="identifier">normal_link</span></code> intrusive
container is trivial (complexity: <code class="computeroutput"><span class="identifier">O</span><span class="special">(</span><span class="number">1</span><span class="special">)</span></code>),
whereas <code class="computeroutput"><span class="identifier">safe_link</span></code> and <code class="computeroutput"><span class="identifier">auto_unlink</span></code> intrusive containers need to
put the hooks of erased values in the default state (complexity: <code class="computeroutput"><span class="identifier">O</span><span class="special">(</span><span class="identifier">NumElements</span><span class="special">)</span></code>). That's why <code class="computeroutput"><span class="identifier">normal_link</span></code>
intrusive list shines in this test.
</p>
<p>
Non-intrusive containers need to make many more allocations and that's why
they lag behind. The <code class="computeroutput"><span class="identifier">disperse</span> <span class="identifier">pointer</span> <span class="identifier">list</span></code>
needs to make <code class="computeroutput"><span class="identifier">NumElements</span><span class="special">*</span><span class="number">2</span></code> allocations,
so the result is not surprising.
</p>
<p>
The Linux test shows that standard containers perform very well against intrusive
containers with big objects. Nearly the same GCC version in MinGW performs
worse, so maybe a good memory allocator is the reason for these excellent
results.
</p>
</div>
<div class="section">
<div class="titlepage"><div><div><h3 class="title">
<a name="intrusive.performance.performance_results_reversing"></a><a class="link" href="performance.html#intrusive.performance.performance_results_reversing" title="Reversing">Reversing</a>
</h3></div></div></div>
<p>
The next test measures the time needed to complete calls to the member function
<code class="computeroutput"><span class="identifier">reverse</span><span class="special">()</span></code>.
Values (<code class="computeroutput"><span class="identifier">test_class</span></code> and <code class="computeroutput"><span class="identifier">itest_class</span></code>) and lists are created as explained
in the previous section.
</p>
<p>
Note that for pointer lists, <code class="computeroutput"><span class="identifier">reverse</span></code>
<span class="bold"><strong>does not need to access <code class="computeroutput"><span class="identifier">test_class</span></code>
values stored in another list or vector</strong></span>, since this function just
needs to adjust internal pointers, so in theory all tested lists need to
perform the same operations.
</p>
<p>
These are the results:
</p>
<div class="table">
<a name="intrusive.performance.performance_results_reversing.reverse_times_for_visual_c_7_1_w"></a><p class="title"><b>Table 19.5. Reverse times for Visual C++ 7.1 / Windows XP</b></p>
<div class="table-contents"><table class="table" summary="Reverse times for Visual C++ 7.1 / Windows XP">
<colgroup>
<col>
<col>
<col>
</colgroup>
<thead><tr>
<th>
<p>
Container
</p>
</th>
<th>
<p>
Time in us/iteration (small object / big object)
</p>
</th>
<th>
<p>
Normalized time (small object / big object)
</p>
</th>
</tr></thead>
<tbody>
<tr>
<td>
<p>
<code class="computeroutput"><span class="identifier">normal_link</span></code> intrusive
list
</p>
</td>
<td>
<p>
2656 / 10625
</p>
</td>
<td>
<p>
1 / 1.83
</p>
</td>
</tr>
<tr>
<td>
<p>
<code class="computeroutput"><span class="identifier">safe_link</span></code> intrusive
list
</p>
</td>
<td>
<p>
2812 / 10937
</p>
</td>
<td>
<p>
1.05 / 1.89
</p>
</td>
</tr>
<tr>
<td>
<p>
<code class="computeroutput"><span class="identifier">auto_unlink</span></code> intrusive
list
</p>
</td>
<td>
<p>
2710 / 10781
</p>
</td>
<td>
<p>
1.02 / 1.86
</p>
</td>
</tr>
<tr>
<td>
<p>
Standard list
</p>
</td>
<td>
<p>
5781 / 14531
</p>
</td>
<td>
<p>
2.17 / 2.51
</p>
</td>
</tr>
<tr>
<td>
<p>
Standard compact pointer list
</p>
</td>
<td>
<p>
5781 / 5781
</p>
</td>
<td>
<p>
2.17 / 1
</p>
</td>
</tr>
<tr>
<td>
<p>
Standard disperse pointer list
</p>
</td>
<td>
<p>
10781 / 15781
</p>
</td>
<td>
<p>
4.05 / 2.72
</p>
</td>
</tr>
</tbody>
</table></div>
</div>
<br class="table-break"><div class="table">
<a name="intrusive.performance.performance_results_reversing.reverse_times_for_gcc_4_1_1_ming"></a><p class="title"><b>Table 19.6. Reverse times for GCC 4.1.1 / MinGW over Windows XP</b></p>
<div class="table-contents"><table class="table" summary="Reverse times for GCC 4.1.1 / MinGW over Windows XP">
<colgroup>
<col>
<col>
<col>
</colgroup>
<thead><tr>
<th>
<p>
Container
</p>
</th>
<th>
<p>
Time in us/iteration (small object / big object)
</p>
</th>
<th>
<p>
Normalized time (small object / big object)
</p>
</th>
</tr></thead>
<tbody>
<tr>
<td>
<p>
<code class="computeroutput"><span class="identifier">normal_link</span></code> intrusive
list
</p>
</td>
<td>
<p>
2656 / 10781
</p>
</td>
<td>
<p>
1 / 2.22
</p>
</td>
</tr>
<tr>
<td>
<p>
<code class="computeroutput"><span class="identifier">safe_link</span></code> intrusive
list
</p>
</td>
<td>
<p>
2656 / 10781
</p>
</td>
<td>
<p>
1 / 2.22
</p>
</td>
</tr>
<tr>
<td>
<p>
<code class="computeroutput"><span class="identifier">auto_unlink</span></code> intrusive
list
</p>
</td>
<td>
<p>
2812 / 10781
</p>
</td>
<td>
<p>
1.02 / 2.22
</p>
</td>
</tr>
<tr>
<td>
<p>
Standard list
</p>
</td>
<td>
<p>
4843 / 12500
</p>
</td>
<td>
<p>
1.82 / 2.58
</p>
</td>
</tr>
<tr>
<td>
<p>
Standard compact pointer list
</p>
</td>
<td>
<p>
4843 / 4843
</p>
</td>
<td>
<p>
1.82 / 1
</p>
</td>
</tr>
<tr>
<td>
<p>
Standard disperse pointer list
</p>
</td>
<td>
<p>
9218 / 12968
</p>
</td>
<td>
<p>
3.47 / 2.67
</p>
</td>
</tr>
</tbody>
</table></div>
</div>
<br class="table-break"><div class="table">
<a name="intrusive.performance.performance_results_reversing.reverse_times_for_gcc_4_1_2_linu"></a><p class="title"><b>Table 19.7. Reverse times for GCC 4.1.2 / Linux Kernel 2.6.18 (OpenSuse 10.2)</b></p>
<div class="table-contents"><table class="table" summary="Reverse times for GCC 4.1.2 / Linux Kernel 2.6.18 (OpenSuse 10.2)">
<colgroup>
<col>
<col>
<col>
</colgroup>
<thead><tr>
<th>
<p>
Container
</p>
</th>
<th>
<p>
Time in us/iteration (small object / big object)
</p>
</th>
<th>
<p>
Normalized time (small object / big object)
</p>
</th>
</tr></thead>
<tbody>
<tr>
<td>
<p>
<code class="computeroutput"><span class="identifier">normal_link</span></code> intrusive
list
</p>
</td>
<td>
<p>
2742 / 10847
</p>
</td>
<td>
<p>
1 / 3.41
</p>
</td>
</tr>
<tr>
<td>
<p>
<code class="computeroutput"><span class="identifier">safe_link</span></code> intrusive
list
</p>
</td>
<td>
<p>
2742 / 10847
</p>
</td>
<td>
<p>
1 / 3.41
</p>
</td>
</tr>
<tr>
<td>
<p>
<code class="computeroutput"><span class="identifier">auto_unlink</span></code> intrusive
list
</p>
</td>
<td>
<p>
2742 / 11027
</p>
</td>
<td>
<p>
1 / 3.47
</p>
</td>
</tr>
<tr>
<td>
<p>
Standard list
</p>
</td>
<td>
<p>
3184 / 10942
</p>
</td>
<td>
<p>
1.16 / 3.44
</p>
</td>
</tr>
<tr>
<td>
<p>
Standard compact pointer list
</p>
</td>
<td>
<p>
3207 / 3176
</p>
</td>
<td>
<p>
1.16 / 1
</p>
</td>
</tr>
<tr>
<td>
<p>
Standard disperse pointer list
</p>
</td>
<td>
<p>
5814 / 13381
</p>
</td>
<td>
<p>
2.12 / 4.21
</p>
</td>
</tr>
</tbody>
</table></div>
</div>
<br class="table-break"><p>
For small objects the results show that the compact storage of values in
intrusive containers improve locality and reversing is faster than with standard
containers, whose values might be dispersed in memory because each value
is independently allocated. Note that the dispersed pointer list (a list
of pointers to values allocated in another list) suffers more because nodes
of the pointer list might be more dispersed, since allocations from both
lists are interleaved in the code:
</p>
<pre class="programlisting"><span class="comment">//Object list (holding `test_class`)</span>
<span class="identifier">stdlist</span> <span class="identifier">objects</span><span class="special">;</span>
<span class="comment">//Pointer list (holding `test_class` pointers)</span>
<span class="identifier">stdptrlist</span> <span class="identifier">l</span><span class="special">;</span>
<span class="keyword">for</span><span class="special">(</span><span class="keyword">int</span> <span class="identifier">i</span> <span class="special">=</span> <span class="number">0</span><span class="special">;</span> <span class="identifier">i</span> <span class="special">&lt;</span> <span class="identifier">NumElements</span><span class="special">;</span> <span class="special">++</span><span class="identifier">i</span><span class="special">){</span>
<span class="comment">//Allocation from the object list</span>
<span class="identifier">objects</span><span class="special">.</span><span class="identifier">push_back</span><span class="special">(</span><span class="identifier">stdlist</span><span class="special">::</span><span class="identifier">value_type</span><span class="special">(</span><span class="identifier">i</span><span class="special">));</span>
<span class="comment">//Allocation from the pointer list</span>
<span class="identifier">l</span><span class="special">.</span><span class="identifier">push_back</span><span class="special">(&amp;</span><span class="identifier">objects</span><span class="special">.</span><span class="identifier">back</span><span class="special">());</span>
<span class="special">}</span>
</pre>
<p>
For big objects the compact pointer list wins because the reversal test doesn't
need access to values stored in another container. Since all the allocations
for nodes of this pointer list are likely to be close (since there is no
other allocation in the process until the pointer list is created) locality
is better than with intrusive containers. The dispersed pointer list, as
with small values, has poor locality.
</p>
</div>
<div class="section">
<div class="titlepage"><div><div><h3 class="title">
<a name="intrusive.performance.performance_results_sorting"></a><a class="link" href="performance.html#intrusive.performance.performance_results_sorting" title="Sorting">Sorting</a>
</h3></div></div></div>
<p>
The next test measures the time needed to complete calls to the member function
<code class="computeroutput"><span class="identifier">sort</span><span class="special">(</span><span class="identifier">Pred</span> <span class="identifier">pred</span><span class="special">)</span></code>. Values (<code class="computeroutput"><span class="identifier">test_class</span></code>
and <code class="computeroutput"><span class="identifier">itest_class</span></code>) and lists
are created as explained in the first section. The values will be sorted
in ascending and descending order each iteration. For example, if <span class="emphasis"><em>l</em></span>
is a list:
</p>
<pre class="programlisting"><span class="keyword">for</span><span class="special">(</span><span class="keyword">int</span> <span class="identifier">i</span> <span class="special">=</span> <span class="number">0</span><span class="special">;</span> <span class="identifier">i</span> <span class="special">&lt;</span> <span class="identifier">NumIter</span><span class="special">;</span> <span class="special">++</span><span class="identifier">i</span><span class="special">){</span>
<span class="keyword">if</span><span class="special">(!(</span><span class="identifier">i</span> <span class="special">%</span> <span class="number">2</span><span class="special">))</span>
<span class="identifier">l</span><span class="special">.</span><span class="identifier">sort</span><span class="special">(</span><span class="identifier">std</span><span class="special">::</span><span class="identifier">greater</span><span class="special">&lt;</span><span class="identifier">stdlist</span><span class="special">::</span><span class="identifier">value_type</span><span class="special">&gt;());</span>
<span class="keyword">else</span>
<span class="identifier">l</span><span class="special">.</span><span class="identifier">sort</span><span class="special">(</span><span class="identifier">std</span><span class="special">::</span><span class="identifier">less</span><span class="special">&lt;</span><span class="identifier">stdlist</span><span class="special">::</span><span class="identifier">value_type</span><span class="special">&gt;());</span>
<span class="special">}</span>
</pre>
<p>
For a pointer list, the function object will be adapted using <code class="computeroutput"><span class="identifier">func_ptr_adaptor</span></code>:
</p>
<pre class="programlisting"><span class="keyword">for</span><span class="special">(</span><span class="keyword">int</span> <span class="identifier">i</span> <span class="special">=</span> <span class="number">0</span><span class="special">;</span> <span class="identifier">i</span> <span class="special">&lt;</span> <span class="identifier">NumIter</span><span class="special">;</span> <span class="special">++</span><span class="identifier">i</span><span class="special">){</span>
<span class="keyword">if</span><span class="special">(!(</span><span class="identifier">i</span> <span class="special">%</span> <span class="number">2</span><span class="special">))</span>
<span class="identifier">l</span><span class="special">.</span><span class="identifier">sort</span><span class="special">(</span><span class="identifier">func_ptr_adaptor</span><span class="special">&lt;</span><span class="identifier">std</span><span class="special">::</span><span class="identifier">greater</span><span class="special">&lt;</span><span class="identifier">stdlist</span><span class="special">::</span><span class="identifier">value_type</span><span class="special">&gt;</span> <span class="special">&gt;());</span>
<span class="keyword">else</span>
<span class="identifier">l</span><span class="special">.</span><span class="identifier">sort</span><span class="special">(</span><span class="identifier">func_ptr_adaptor</span><span class="special">&lt;</span><span class="identifier">std</span><span class="special">::</span><span class="identifier">less</span><span class="special">&lt;</span><span class="identifier">stdlist</span><span class="special">::</span><span class="identifier">value_type</span><span class="special">&gt;</span> <span class="special">&gt;());</span>
<span class="special">}</span>
</pre>
<p>
Note that for pointer lists, <code class="computeroutput"><span class="identifier">sort</span></code>
will take a function object that <span class="bold"><strong>will access <code class="computeroutput"><span class="identifier">test_class</span></code> values stored in another list
or vector</strong></span>, so pointer lists will suffer an extra indirection:
they will need to access the <code class="computeroutput"><span class="identifier">test_class</span></code>
values stored in another container to compare two elements.
</p>
<p>
These are the results:
</p>
<div class="table">
<a name="intrusive.performance.performance_results_sorting.sort_times_for_visual_c_7_1_wind"></a><p class="title"><b>Table 19.8. Sort times for Visual C++ 7.1 / Windows XP</b></p>
<div class="table-contents"><table class="table" summary="Sort times for Visual C++ 7.1 / Windows XP">
<colgroup>
<col>
<col>
<col>
</colgroup>
<thead><tr>
<th>
<p>
Container
</p>
</th>
<th>
<p>
Time in us/iteration (small object / big object)
</p>
</th>
<th>
<p>
Normalized time (small object / big object)
</p>
</th>
</tr></thead>
<tbody>
<tr>
<td>
<p>
<code class="computeroutput"><span class="identifier">normal_link</span></code> intrusive
list
</p>
</td>
<td>
<p>
16093 / 38906
</p>
</td>
<td>
<p>
1 / 1
</p>
</td>
</tr>
<tr>
<td>
<p>
<code class="computeroutput"><span class="identifier">safe_link</span></code> intrusive
list
</p>
</td>
<td>
<p>
16093 / 39062
</p>
</td>
<td>
<p>
1 / 1
</p>
</td>
</tr>
<tr>
<td>
<p>
<code class="computeroutput"><span class="identifier">auto_unlink</span></code> intrusive
list
</p>
</td>
<td>
<p>
16093 / 38906
</p>
</td>
<td>
<p>
1 / 1
</p>
</td>
</tr>
<tr>
<td>
<p>
Standard list
</p>
</td>
<td>
<p>
32343 / 56406
</p>
</td>
<td>
<p>
2.0 / 1.44
</p>
</td>
</tr>
<tr>
<td>
<p>
Standard compact pointer list
</p>
</td>
<td>
<p>
33593 / 46093
</p>
</td>
<td>
<p>
2.08 / 1.18
</p>
</td>
</tr>
<tr>
<td>
<p>
Standard disperse pointer list
</p>
</td>
<td>
<p>
46875 / 68593
</p>
</td>
<td>
<p>
2.91 / 1.76
</p>
</td>
</tr>
</tbody>
</table></div>
</div>
<br class="table-break"><div class="table">
<a name="intrusive.performance.performance_results_sorting.sort_times_for_gcc_4_1_1_mingw_o"></a><p class="title"><b>Table 19.9. Sort times for GCC 4.1.1 / MinGW over Windows XP</b></p>
<div class="table-contents"><table class="table" summary="Sort times for GCC 4.1.1 / MinGW over Windows XP">
<colgroup>
<col>
<col>
<col>
</colgroup>
<thead><tr>
<th>
<p>
Container
</p>
</th>
<th>
<p>
Time in us/iteration (small object / big object)
</p>
</th>
<th>
<p>
Normalized time (small object / big object)
</p>
</th>
</tr></thead>
<tbody>
<tr>
<td>
<p>
<code class="computeroutput"><span class="identifier">normal_link</span></code> intrusive
list
</p>
</td>
<td>
<p>
15000 / 39218
</p>
</td>
<td>
<p>
1 / 1
</p>
</td>
</tr>
<tr>
<td>
<p>
<code class="computeroutput"><span class="identifier">safe_link</span></code> intrusive
list
</p>
</td>
<td>
<p>
15156 / 39531
</p>
</td>
<td>
<p>
1.01 / 1.01
</p>
</td>
</tr>
<tr>
<td>
<p>
<code class="computeroutput"><span class="identifier">auto_unlink</span></code> intrusive
list
</p>
</td>
<td>
<p>
15156 / 39531
</p>
</td>
<td>
<p>
1.01 / 1.01
</p>
</td>
</tr>
<tr>
<td>
<p>
Standard list
</p>
</td>
<td>
<p>
34218 / 56875
</p>
</td>
<td>
<p>
2.28 / 1.45
</p>
</td>
</tr>
<tr>
<td>
<p>
Standard compact pointer list
</p>
</td>
<td>
<p>
35468 / 49218
</p>
</td>
<td>
<p>
2.36 / 1.25
</p>
</td>
</tr>
<tr>
<td>
<p>
Standard disperse pointer list
</p>
</td>
<td>
<p>
47656 / 70156
</p>
</td>
<td>
<p>
3.17 / 1.78
</p>
</td>
</tr>
</tbody>
</table></div>
</div>
<br class="table-break"><div class="table">
<a name="intrusive.performance.performance_results_sorting.sort_times_for_gcc_4_1_2_linux_k"></a><p class="title"><b>Table 19.10. Sort times for GCC 4.1.2 / Linux Kernel 2.6.18 (OpenSuse 10.2)</b></p>
<div class="table-contents"><table class="table" summary="Sort times for GCC 4.1.2 / Linux Kernel 2.6.18 (OpenSuse 10.2)">
<colgroup>
<col>
<col>
<col>
</colgroup>
<thead><tr>
<th>
<p>
Container
</p>
</th>
<th>
<p>
Time in us/iteration (small object / big object)
</p>
</th>
<th>
<p>
Normalized time (small object / big object)
</p>
</th>
</tr></thead>
<tbody>
<tr>
<td>
<p>
<code class="computeroutput"><span class="identifier">normal_link</span></code> intrusive
list
</p>
</td>
<td>
<p>
18003 / 40795
</p>
</td>
<td>
<p>
1 / 1
</p>
</td>
</tr>
<tr>
<td>
<p>
<code class="computeroutput"><span class="identifier">safe_link</span></code> intrusive
list
</p>
</td>
<td>
<p>
18003 / 41017
</p>
</td>
<td>
<p>
1 / 1
</p>
</td>
</tr>
<tr>
<td>
<p>
<code class="computeroutput"><span class="identifier">auto_unlink</span></code> intrusive
list
</p>
</td>
<td>
<p>
18230 / 40941
</p>
</td>
<td>
<p>
1.01 / 1
</p>
</td>
</tr>
<tr>
<td>
<p>
Standard list
</p>
</td>
<td>
<p>
26273 / 49643
</p>
</td>
<td>
<p>
1.45 / 1.21
</p>
</td>
</tr>
<tr>
<td>
<p>
Standard compact pointer list
</p>
</td>
<td>
<p>
28540 / 43172
</p>
</td>
<td>
<p>
1.58 / 1.05
</p>
</td>
</tr>
<tr>
<td>
<p>
Standard disperse pointer list
</p>
</td>
<td>
<p>
35077 / 57638
</p>
</td>
<td>
<p>
1.94 / 1.41
</p>
</td>
</tr>
</tbody>
</table></div>
</div>
<br class="table-break"><p>
The results show that intrusive containers are faster than standard containers.
We can see that the pointer list holding pointers to values stored in a vector
is quite fast, so the extra indirection that is needed to access the value
is minimized because all the values are tightly stored, improving caching.
The disperse list, on the other hand, is slower because the indirection to
access values stored in the object list is more expensive than accessing
values stored in a vector.
</p>
</div>
<div class="section">
<div class="titlepage"><div><div><h3 class="title">
<a name="intrusive.performance.performance_results_write_access"></a><a class="link" href="performance.html#intrusive.performance.performance_results_write_access" title="Write access">Write
access</a>
</h3></div></div></div>
<p>
The next test measures the time needed to iterate through all the elements
of a list, and increment the value of the internal <code class="computeroutput"><span class="identifier">i_</span></code>
member:
</p>
<pre class="programlisting"><span class="identifier">stdlist</span><span class="special">::</span><span class="identifier">iterator</span> <span class="identifier">it</span><span class="special">(</span><span class="identifier">l</span><span class="special">.</span><span class="identifier">begin</span><span class="special">()),</span> <span class="identifier">end</span><span class="special">(</span><span class="identifier">l</span><span class="special">.</span><span class="identifier">end</span><span class="special">());</span>
<span class="keyword">for</span><span class="special">(;</span> <span class="identifier">it</span> <span class="special">!=</span> <span class="identifier">end</span><span class="special">;</span> <span class="special">++</span><span class="identifier">it</span><span class="special">)</span>
<span class="special">++(</span><span class="identifier">it</span><span class="special">-&gt;</span><span class="identifier">i_</span><span class="special">);</span>
</pre>
<p>
Values (<code class="computeroutput"><span class="identifier">test_class</span></code> and <code class="computeroutput"><span class="identifier">itest_class</span></code>) and lists are created as explained
in the first section. Note that for pointer lists, the iteration will suffer
an extra indirection: they will need to access the <code class="computeroutput"><span class="identifier">test_class</span></code>
values stored in another container:
</p>
<pre class="programlisting"><span class="identifier">stdptrlist</span><span class="special">::</span><span class="identifier">iterator</span> <span class="identifier">it</span><span class="special">(</span><span class="identifier">l</span><span class="special">.</span><span class="identifier">begin</span><span class="special">()),</span> <span class="identifier">end</span><span class="special">(</span><span class="identifier">l</span><span class="special">.</span><span class="identifier">end</span><span class="special">());</span>
<span class="keyword">for</span><span class="special">(;</span> <span class="identifier">it</span> <span class="special">!=</span> <span class="identifier">end</span><span class="special">;</span> <span class="special">++</span><span class="identifier">it</span><span class="special">)</span>
<span class="special">++((*</span><span class="identifier">it</span><span class="special">)-&gt;</span><span class="identifier">i_</span><span class="special">);</span>
</pre>
<p>
These are the results:
</p>
<div class="table">
<a name="intrusive.performance.performance_results_write_access.write_access_times_for_visual_c_"></a><p class="title"><b>Table 19.11. Write access times for Visual C++ 7.1 / Windows XP</b></p>
<div class="table-contents"><table class="table" summary="Write access times for Visual C++ 7.1 / Windows XP">
<colgroup>
<col>
<col>
<col>
</colgroup>
<thead><tr>
<th>
<p>
Container
</p>
</th>
<th>
<p>
Time in us/iteration (small object / big object)
</p>
</th>
<th>
<p>
Normalized time (small object / big object)
</p>
</th>
</tr></thead>
<tbody>
<tr>
<td>
<p>
<code class="computeroutput"><span class="identifier">normal_link</span></code> intrusive
list
</p>
</td>
<td>
<p>
2031 / 8125
</p>
</td>
<td>
<p>
1 / 1
</p>
</td>
</tr>
<tr>
<td>
<p>
<code class="computeroutput"><span class="identifier">safe_link</span></code> intrusive
list
</p>
</td>
<td>
<p>
2031 / 8281
</p>
</td>
<td>
<p>
1 / 1.01
</p>
</td>
</tr>
<tr>
<td>
<p>
<code class="computeroutput"><span class="identifier">auto_unlink</span></code> intrusive
list
</p>
</td>
<td>
<p>
2031 / 8281
</p>
</td>
<td>
<p>
1 / 1.01
</p>
</td>
</tr>
<tr>
<td>
<p>
Standard list
</p>
</td>
<td>
<p>
4218 / 10000
</p>
</td>
<td>
<p>
2.07 / 1.23
</p>
</td>
</tr>
<tr>
<td>
<p>
Standard compact pointer list
</p>
</td>
<td>
<p>
4062 / 8437
</p>
</td>
<td>
<p>
2.0 / 1.03
</p>
</td>
</tr>
<tr>
<td>
<p>
Standard disperse pointer list
</p>
</td>
<td>
<p>
8593 / 13125
</p>
</td>
<td>
<p>
4.23 / 1.61
</p>
</td>
</tr>
</tbody>
</table></div>
</div>
<br class="table-break"><div class="table">
<a name="intrusive.performance.performance_results_write_access.write_access_times_for_gcc_4_1_1"></a><p class="title"><b>Table 19.12. Write access times for GCC 4.1.1 / MinGW over Windows XP</b></p>
<div class="table-contents"><table class="table" summary="Write access times for GCC 4.1.1 / MinGW over Windows XP">
<colgroup>
<col>
<col>
<col>
</colgroup>
<thead><tr>
<th>
<p>
Container
</p>
</th>
<th>
<p>
Time in us/iteration (small object / big object)
</p>
</th>
<th>
<p>
Normalized time (small object / big object)
</p>
</th>
</tr></thead>
<tbody>
<tr>
<td>
<p>
<code class="computeroutput"><span class="identifier">normal_link</span></code> intrusive
list
</p>
</td>
<td>
<p>
2343 / 8281
</p>
</td>
<td>
<p>
1 / 1
</p>
</td>
</tr>
<tr>
<td>
<p>
<code class="computeroutput"><span class="identifier">safe_link</span></code> intrusive
list
</p>
</td>
<td>
<p>
2500 / 8281
</p>
</td>
<td>
<p>
1.06 / 1
</p>
</td>
</tr>
<tr>
<td>
<p>
<code class="computeroutput"><span class="identifier">auto_unlink</span></code> intrusive
list
</p>
</td>
<td>
<p>
2500 / 8281
</p>
</td>
<td>
<p>
1.06 / 1
</p>
</td>
</tr>
<tr>
<td>
<p>
Standard list
</p>
</td>
<td>
<p>
4218 / 10781
</p>
</td>
<td>
<p>
1.8 / 1.3
</p>
</td>
</tr>
<tr>
<td>
<p>
Standard compact pointer list
</p>
</td>
<td>
<p>
3906 / 8281
</p>
</td>
<td>
<p>
1.66 / 1
</p>
</td>
</tr>
<tr>
<td>
<p>
Standard disperse pointer list
</p>
</td>
<td>
<p>
8281 / 13750
</p>
</td>
<td>
<p>
3.53 / 1.66
</p>
</td>
</tr>
</tbody>
</table></div>
</div>
<br class="table-break"><div class="table">
<a name="intrusive.performance.performance_results_write_access.write_access_times_for_gcc_4_1_2"></a><p class="title"><b>Table 19.13. Write access times for GCC 4.1.2 / Linux Kernel 2.6.18 (OpenSuse 10.2)</b></p>
<div class="table-contents"><table class="table" summary="Write access times for GCC 4.1.2 / Linux Kernel 2.6.18 (OpenSuse 10.2)">
<colgroup>
<col>
<col>
<col>
</colgroup>
<thead><tr>
<th>
<p>
Container
</p>
</th>
<th>
<p>
Time in us/iteration (small object / big object)
</p>
</th>
<th>
<p>
Normalized time (small object / big object)
</p>
</th>
</tr></thead>
<tbody>
<tr>
<td>
<p>
<code class="computeroutput"><span class="identifier">normal_link</span></code> intrusive
list
</p>
</td>
<td>
<p>
2286 / 8468
</p>
</td>
<td>
<p>
1 / 1.1
</p>
</td>
</tr>
<tr>
<td>
<p>
<code class="computeroutput"><span class="identifier">safe_link</span></code> intrusive
list
</p>
</td>
<td>
<p>
2381 / 8412
</p>
</td>
<td>
<p>
1.04 / 1.09
</p>
</td>
</tr>
<tr>
<td>
<p>
<code class="computeroutput"><span class="identifier">auto_unlink</span></code> intrusive
list
</p>
</td>
<td>
<p>
2301 / 8437
</p>
</td>
<td>
<p>
1.01 / 1.1
</p>
</td>
</tr>
<tr>
<td>
<p>
Standard list
</p>
</td>
<td>
<p>
3044 / 9061
</p>
</td>
<td>
<p>
1.33 / 1.18
</p>
</td>
</tr>
<tr>
<td>
<p>
Standard compact pointer list
</p>
</td>
<td>
<p>
2755 / 7660
</p>
</td>
<td>
<p>
1.20 / 1
</p>
</td>
</tr>
<tr>
<td>
<p>
Standard disperse pointer list
</p>
</td>
<td>
<p>
6118 / 12453
</p>
</td>
<td>
<p>
2.67 / 1.62
</p>
</td>
</tr>
</tbody>
</table></div>
</div>
<br class="table-break"><p>
As with the read access test, the results show that intrusive containers
outperform all other containers if the values are tightly packed in a vector.
The disperse list is again the slowest.
</p>
</div>
<div class="section">
<div class="titlepage"><div><div><h3 class="title">
<a name="intrusive.performance.performance_results_conclusions"></a><a class="link" href="performance.html#intrusive.performance.performance_results_conclusions" title="Conclusions">Conclusions</a>
</h3></div></div></div>
<p>
Intrusive containers can offer performance benefits that cannot be achieved
with equivalent non-intrusive containers. Memory locality improvements are
noticeable when the objects to be inserted are small. Minimizing memory allocation/deallocation
calls is also an important factor and intrusive containers make this simple
if all objects to be inserted in intrusive containers are allocated using
<code class="computeroutput"><span class="identifier">std</span><span class="special">::</span><span class="identifier">vector</span></code> or <code class="computeroutput"><span class="identifier">std</span><span class="special">::</span><span class="identifier">deque</span></code>.
</p>
</div>
</div>
<table xmlns:rev="http://www.cs.rpi.edu/~gregod/boost/tools/doc/revision" width="100%"><tr>
<td align="left"></td>
<td align="right"><div class="copyright-footer">Copyright © 2005 Olaf Krzikalla<br>Copyright © 2006-2015 Ion Gaztanaga<p>
Distributed under the Boost Software License, Version 1.0. (See accompanying
file LICENSE_1_0.txt or copy at <a href="http://www.boost.org/LICENSE_1_0.txt" target="_top">http://www.boost.org/LICENSE_1_0.txt</a>)
</p>
</div></td>
</tr></table>
<hr>
<div class="spirit-nav">
<a accesskey="p" href="design_notes.html"><img src="../../../doc/src/images/prev.png" alt="Prev"></a><a accesskey="u" href="../intrusive.html"><img src="../../../doc/src/images/up.png" alt="Up"></a><a accesskey="h" href="../index.html"><img src="../../../doc/src/images/home.png" alt="Home"></a><a accesskey="n" href="release_notes.html"><img src="../../../doc/src/images/next.png" alt="Next"></a>
</div>
</body>
</html>
Reference in New Issue Copy Permalink