• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1<html>
2<head>
3<title>V8 Benchmark Suite Revisions</title>
4<link type="text/css" rel="stylesheet" href="style.css" />
5</head>
6<body>
7<div>
8  <div class="title"><h1>V8 Benchmark Suite Revisions</h1></div>
9  <table>
10    <tr>
11      <td class="contents">
12
13<p>
14
15The V8 benchmark suite is changed from time to time as we fix bugs or
16expand the scope of the benchmarks.  Here is a list of revisions, with
17a description of the changes made.  Note that benchmark results are
18not comparable unless both results are run with the same revision of
19the benchmark suite.
20
21</p>
22
23<div class="subtitle"><h3>Version 5 (<a href="http://v8.googlecode.com/svn/data/benchmarks/v5/run.html">link</a>)</h3></div>
24
25<p>Removed duplicate line in random seed code, and changed the name of
26the Object.prototype.inherits function in the DeltaBlue benchmark to
27inheritsFrom to avoid name clashes when running in Chromium with
28extensions enabled.
29</p>
30
31<div class="subtitle"><h3>Version 4 (<a href="http://v8.googlecode.com/svn/data/benchmarks/v4/run.html">link</a>)</h3></div>
32
33<p>The <i>Splay</i> benchmark is a newcomer in version 4.  It
34manipulates a splay tree by adding and removing data nodes, thus
35exercising the memory management subsystem of the JavaScript engine.
36</p>
37
38<p>
39Furthermore, all the unused parts of the Prototype library were
40removed from the RayTrace benchmark. This does not affect the running
41of the benchmark.
42</p>
43
44
45<div class="subtitle"><h3>Version 3 (<a href="http://v8.googlecode.com/svn/data/benchmarks/v3/run.html">link</a>)</h3></div>
46
47<p>Version 3 adds a new benchmark, <i>RegExp</i>.  The RegExp
48benchmark is generated by loading 50 of the most popular pages on the
49web and logging all regexp operations performed.  Each operation is
50given a weight that is calculated from an estimate of the popularity
51of the pages where it occurs and the number of times it is executed
52while loading each page.  Finally the literal letters in the data are
53encoded using ROT13 in a way that does not affect how the regexps
54match their input.
55</p>
56
57
58<div class="subtitle"><h3>Version 2 (<a href="http://v8.googlecode.com/svn/data/benchmarks/v2/run.html">link</a>)</h3></div>
59
60<p>For version 2 the Crypto benchmark was fixed.  Previously, the
61decryption stage was given plaintext as input, which resulted in an
62error.  Now, the decryption stage is given the output of the
63encryption stage as input.  The result is checked against the original
64plaintext.  For this to give the correct results the crypto objects
65are reset for each iteration of the benchmark.  In addition, the size
66of the plain text has been increased a little and the use of
67Math.random() and new Date() to build an RNG pool has been
68removed. </p>
69
70<p>Other benchmarks were fixed to do elementary verification of the
71results of their calculations.  This is to avoid accidentally
72obtaining scores that are the result of an incorrect JavaScript engine
73optimization.</p>
74
75
76<div class="subtitle"><h3>Version 1 (<a href="http://v8.googlecode.com/svn/data/benchmarks/v1/run.html">link</a>)</h3></div>
77
78<p>Initial release.</p>
79
80</td><td style="text-align: center">
81</td></tr></table>
82
83</div>
84
85</body>
86</html>
87