| 1 |
<?xml version="1.0" encoding="ISO-8859-1" standalone="yes"?>
|
| 2 |
<page name="bench">
|
| 3 |
|
| 4 |
<title>Benchmarks</title>
|
| 5 |
|
| 6 |
<left>
|
| 7 |
<p>This page briefly presents preliminary benchmarks on the CDuce
|
| 8 |
prototype.
|
| 9 |
</p>
|
| 10 |
<boxes-toc/>
|
| 11 |
<p>See also:</p>
|
| 12 |
<local-links href="index,proto"/>
|
| 13 |
</left>
|
| 14 |
|
| 15 |
<box title="Methodology" link="meth">
|
| 16 |
<p>
|
| 17 |
The benchmarks were performed on a Linux box, Athlon 750, 128 Mb RAM.
|
| 18 |
The results are given in seconds. They were measured with
|
| 19 |
the Unix <tt>time</tt> command. The user time is reported unless
|
| 20 |
otherwise specified.
|
| 21 |
</p>
|
| 22 |
|
| 23 |
<p>
|
| 24 |
These are only <b>preliminary benchmarks</b>, and results
|
| 25 |
should be taken with much care.
|
| 26 |
</p>
|
| 27 |
</box>
|
| 28 |
|
| 29 |
<box title="CDuce vs. XDuce" link="xduce">
|
| 30 |
|
| 31 |
<section title="Addrbook">
|
| 32 |
|
| 33 |
<p>
|
| 34 |
This example was taken from XDuce distribution.
|
| 35 |
The transformation is simple;
|
| 36 |
the example is intended to evaluate run-time efficiency.
|
| 37 |
</p>
|
| 38 |
|
| 39 |
<p>
|
| 40 |
<a href="benchmarks/addrbook.cd">CDuce source</a> -
|
| 41 |
<a href="benchmarks/addrbook.q">XDuce source</a> -
|
| 42 |
<a href="benchmarks/addrbook.ml">OCaml file</a> (to generate random documents).
|
| 43 |
</p>
|
| 44 |
|
| 45 |
|
| 46 |
<table border="1">
|
| 47 |
<tr> <td>Addrbook</td>
|
| 48 |
<th>XDuce 0.2.4</th>
|
| 49 |
<th>XDuce 0.2.4 + patopt</th>
|
| 50 |
<th>XDuce 0.4.0 + patopt</th>
|
| 51 |
<th>CDuce</th> </tr>
|
| 52 |
|
| 53 |
<tr align="right"> <th>Type-checking</th>
|
| 54 |
<td>0.04</td> <td>0.05</td> <td>0.15</td> <td>0.04</td> </tr>
|
| 55 |
<tr align="right"> <th>Small file (60 Ko)</th>
|
| 56 |
<td>1.56</td> <td>0.14</td> <td>9.6</td> <td>0.09</td> </tr>
|
| 57 |
<tr align="right"> <th>Medium file (600 Ko)</th>
|
| 58 |
<td>260</td> <td>1.16</td> <td>*</td> <td>0.46</td> </tr>
|
| 59 |
<tr align ="right"> <th>Large file (6 Mb)</th>
|
| 60 |
<td>*</td> <td>31.05</td> <td>*</td> <td>6.37</td> </tr>
|
| 61 |
|
| 62 |
</table>
|
| 63 |
|
| 64 |
<p>
|
| 65 |
The * corresponds to checks stopped after ten minutes.
|
| 66 |
</p>
|
| 67 |
</section>
|
| 68 |
|
| 69 |
<section title="Html2latex">
|
| 70 |
|
| 71 |
<p>
|
| 72 |
This example was taken from XDuce distribution and simplified
|
| 73 |
(removed support for tables). The transformation is more complex;
|
| 74 |
the example is intended to evaluate type-checking time.
|
| 75 |
</p>
|
| 76 |
|
| 77 |
<p>
|
| 78 |
<a href="benchmarks/html2latex.cd">CDuce source</a> -
|
| 79 |
<a href="benchmarks/xhtml-trans.cd">CDuce source (XHTML Transitional DTD)</a> -
|
| 80 |
<a href="benchmarks/html2latex.q">XDuce source</a> -
|
| 81 |
<a href="benchmarks/tst_html2latex.xml">Sample input XML file</a>.
|
| 82 |
</p>
|
| 83 |
|
| 84 |
<table border="1">
|
| 85 |
<tr> <td>Html2latex</td>
|
| 86 |
<th>XDuce 0.2.4</th>
|
| 87 |
<th>XDuce 0.2.4 + patopt</th>
|
| 88 |
<th>XDuce 0.4.0 + patopt</th>
|
| 89 |
<th>CDuce</th> </tr>
|
| 90 |
|
| 91 |
<tr align="right"> <th>Type-checking</th>
|
| 92 |
<td>1.5</td> <td>21.4</td> <td>9.5</td> <td>0.4</td> </tr>
|
| 93 |
<tr align="right"> <th>Total</th>
|
| 94 |
<td>2.3</td> <td>26.5</td> <td>10.0</td> <td>0.8</td> </tr>
|
| 95 |
</table>
|
| 96 |
|
| 97 |
</section>
|
| 98 |
|
| 99 |
</box>
|
| 100 |
|
| 101 |
<box title="CDuce vs. XSLT" link="xslt">
|
| 102 |
|
| 103 |
<p>
|
| 104 |
We also performed benchmarks between CDuce and a fast XSLT processor (the
|
| 105 |
<tt>xsltproc</tt> program from <a href="http://xmlsoft.org/XSLT/">
|
| 106 |
Gnome libxslt library</a>).
|
| 107 |
</p>
|
| 108 |
|
| 109 |
<p>
|
| 110 |
The example is intended to evaluate run-time efficiency. Even though
|
| 111 |
the transformation is quite complex, type-checking time for CDuce
|
| 112 |
is negligible. A few remarks:
|
| 113 |
</p>
|
| 114 |
<ul>
|
| 115 |
<li>type-based optimization
|
| 116 |
brings almost no gain (only tag checks can be avoided);</li>
|
| 117 |
<li>the CDuce program must validate its input document (the XSLT
|
| 118 |
stylesheet can operate on any document);</li>
|
| 119 |
<li>the XSLT processor is written in C; the CDuce interpreter
|
| 120 |
is written in Objective Caml.</li>
|
| 121 |
</ul>
|
| 122 |
|
| 123 |
<p>
|
| 124 |
Taking these remarks into account, the results below are more than
|
| 125 |
encouraging.
|
| 126 |
</p>
|
| 127 |
|
| 128 |
<p>
|
| 129 |
<a href="benchmarks/split.cd">CDuce source</a> -
|
| 130 |
<a href="benchmarks/split.xslt1.xml">XSLT source</a> -
|
| 131 |
<a href="benchmarks/split.xslt2.xml">XSLT source (another version)</a> -
|
| 132 |
<a href="benchmarks/split.ml">OCaml file</a> (to generate random documents).
|
| 133 |
</p>
|
| 134 |
|
| 135 |
|
| 136 |
<table border="1">
|
| 137 |
<tr> <td>split</td>
|
| 138 |
<th>XSLT (1)</th>
|
| 139 |
<th>XSLT (2)</th>
|
| 140 |
<th>CDuce</th> </tr>
|
| 141 |
|
| 142 |
<tr align="right"> <th>5 Mb</th>
|
| 143 |
<td>8.90</td> <td>8.05</td> <td>5.30</td> </tr>
|
| 144 |
<tr align="right"> <th>27 Mb</th>
|
| 145 |
<td>47.10</td> <td>68.40</td> <td>33.85</td> </tr>
|
| 146 |
</table>
|
| 147 |
|
| 148 |
<p>
|
| 149 |
The table above displays CPU user time (parsing + transformation) .
|
| 150 |
The table below, instead, displays total wall-clock time
|
| 151 |
(including system time). The results indicate that CDuce
|
| 152 |
behave much better when the XML document
|
| 153 |
does'nt fit in memory (and swapping occurs).
|
| 154 |
</p>
|
| 155 |
|
| 156 |
<table border="1">
|
| 157 |
<tr> <td>split</td>
|
| 158 |
<th>XSLT (1)</th>
|
| 159 |
<th>XSLT (2)</th>
|
| 160 |
<th>CDuce</th> </tr>
|
| 161 |
|
| 162 |
<tr align="right"> <th>5 Mb</th>
|
| 163 |
<td>9.05</td> <td>8.26</td> <td>5.40</td> </tr>
|
| 164 |
<tr align="right"> <th>27 Mb</th>
|
| 165 |
<td>320.80</td> <td>324.50</td> <td>83.65</td> </tr>
|
| 166 |
</table>
|
| 167 |
|
| 168 |
</box>
|
| 169 |
|
| 170 |
</page>
|
Parent Directory
| 
Revision Log