icuSources/test/perf/collationperf/CollPerf.pl

   1 #!/usr/bin/perl
   2 #  ********************************************************************
   3 #  * Copyright (C) 2016 and later: Unicode, Inc. and others.
   4 #  * License & terms of use: http://www.unicode.org/copyright.html#License
   5 #  ********************************************************************
   6 #  * COPYRIGHT:
   7 #  * Copyright (c) 2002-2013, International Business Machines Corporation and
   8 #  * others. All Rights Reserved.
   9 #  ********************************************************************
  10
  11 require "../perldriver/Common.pl";
  12
  13 use lib '../perldriver';
  14
  15 my $p;
  16 if ($OnWindows) {
  17     $p = "cd ".$ICULatest."/bin && ".$ICUPathLatest . "/collationperf/$WindowsPlatform/Release/collationperf.exe";
  18 }
  19 else {
  20     $p = "LD_LIBRARY_PATH=".$ICULatest."/source/lib:".$ICULatest."/source/tools/ctestfw ".$ICUPathLatest . "/collationperf/collperf";
  21 }
  22
  23 my @locale = (
  24     "en_US",
  25     "da_DK",
  26     "de_DE",
  27     "fr_FR",
  28     "ja_JP",
  29     "ja_JP",
  30     "ja_JP",
  31     "ja_JP",
  32     "zh_CN",
  33     "zh_CN",
  34     "zh_CN",
  35     "zh_TW",
  36     "zh_TW",
  37     "ko_KR",
  38     "ko_KR",
  39     "ru_RU",
  40     "ru_RU",
  41     "th_TH",
  42     "th_TH"
  43 );
  44
  45 my $filePath   = $CollationDataPath . "/";
  46 my $filePrefix = "TestNames_";
  47 my @data       = (
  48     $filePrefix."Latin.txt",
  49     $filePrefix."Latin.txt",
  50     $filePrefix."Latin.txt",
  51     $filePrefix."Latin.txt",
  52     $filePrefix."Latin.txt",
  53     $filePrefix."Japanese_h.txt",
  54     $filePrefix."Japanese_k.txt",
  55     $filePrefix."Asian.txt",
  56     $filePrefix."Latin.txt",
  57     $filePrefix."Chinese.txt",
  58     $filePrefix."Simplified_Chinese.txt",
  59     $filePrefix."Latin.txt",
  60     $filePrefix."Chinese.txt",
  61     $filePrefix."Latin.txt",
  62     $filePrefix."Korean.txt",
  63     $filePrefix."Latin.txt",
  64     $filePrefix."Russian.txt",
  65     $filePrefix."Latin.txt",
  66     $filePrefix."Thai.txt"
  67 );
  68
  69 my @resultPER;
  70 my @resultFIN;
  71
  72 for ( $n = 0 ; $n < @data ; $n++ ) {
  73     my $resultICU;
  74     my $resultNIX;
  75     $resultICU = @locale[$n].",".@data[$n].",";
  76     $resultNIX = @locale[$n].",".@data[$n].",";
  77     @resultFIN[$n] = @locale[$n].",".@data[$n].",";
  78
  79     #quicksort
  80     my @icu = `$p -locale @locale[$n] -loop 1000 -file $filePath@data[$n] -qsort`;
  81     my @nix = `$p -locale @locale[$n] -unix -loop 1000 -file $filePath@data[$n] -qsort`;
  82
  83     my @icua = split( ' = ', $icu[2] );
  84     my @icub = split( ' ',   $icua[1] );
  85     my @nixa = split( ' = ', $nix[2] );
  86     my @nixb = split( ' ',   $nixa[1] );
  87
  88     $resultICU = $resultICU.$icub[0].",";
  89     $resultNIX = $resultNIX.$nixb[0].",";
  90
  91     #keygen time
  92     @icu = `$p -locale @locale[$n] -loop 1000 -file $filePath@data[$n] -keygen`;
  93     @nix = `$p -locale @locale[$n] -unix -loop 1000 -file $filePath@data[$n] -keygen`;
  94
  95     @icua = split( ' = ', $icu[2] );
  96     @icub = split( ' ',   $icua[1] );
  97     @nixa = split( ' = ', $nix[2] );
  98     @nixb = split( ' ',   $nixa[1] );
  99
 100     $resultICU = $resultICU.$icub[0].",";
 101     $resultNIX = $resultNIX.$nixb[0].",";
 102
 103     #keygen len
 104     @icua = split( ' = ', $icu[3] );
 105     @nixa = split( ' = ', $nix[3] );
 106
 107     chomp( @icua[1] );
 108     chomp( @nixa[1] );
 109
 110     $resultICU = $resultICU.$icua[1].",";
 111     $resultNIX = $resultNIX.$nixa[1].",";
 112
 113     my @resultSplitICU;
 114     my @resultSplitNIX;
 115
 116     #percent
 117     for ( $i = 0 ; $i < 3 ; $i++ ) {
 118         my $percent = 0;
 119         @resultSplitICU = split( ',', $resultICU );
 120         @resultSplitNIX = split( ',', $resultNIX );
 121         if ( @resultSplitICU[ 2 + $i ] > 0 ) {
 122             $percent = substr((((
 123                 @resultSplitNIX[ 2 + $i ] - @resultSplitICU[ 2 + $i ]) / @resultSplitICU[ 2 + $i ]) * 100),
 124                 0, 7);
 125         }
 126         @resultPER[$n] = @resultPER[$n].$percent."%,";
 127     }
 128
 129     #store ICU result
 130     for ( $j = 0 ; $j < 3 ; $j++ ) {
 131         @resultFIN[$n] = @resultFIN[$n].@resultSplitICU[ 2 + $j ].",";
 132     }
 133
 134     #store Unix result
 135     for ( $j = 0 ; $j < 3 ; $j++ ) {
 136         @resultFIN[$n] = @resultFIN[$n].@resultSplitNIX[ 2 + $j ].",";
 137     }
 138
 139     #store Percent result
 140     @resultFIN[$n] = @resultFIN[$n].@resultPER[$n];
 141 }
 142
 143 # Print the results in a HTML page
 144 printOutput();
 145
 146 exit(0);
 147
 148 # This subroutine creates the web page and prints out the results in a table
 149 sub printOutput {
 150     my $title = "Collation:  ICU " . $ICULatestVersion . " vs GLIBC";
 151     my $html  = localtime;
 152     $html =~ s/://g;         # ':' illegal
 153     $html =~ s/\s*\d+$//;    # delete year
 154     $html =~ s/^\w+\s*//;    # delete dow
 155     $html = "CollationPerformance $html.html";
 156     $html = "../results/" . $html;
 157     $html =~ s/ /_/g;
 158     open( HTML, ">$html" ) or die "Can't write to $html: $!";
 159     print HTML <<EOF;
 160 <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
 161 "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
 162 <html xmlns="http://www.w3.org/1999/xhtml">
 163 <head>
 164 <meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
 165 <title>Collation: ICU4C vs. glibc</title>
 166 <link rel="stylesheet" href="../icu.css" type="text/css" />
 167 </head>
 168 <body>
 169 <!--#include virtual="../ssi/header.html" -->
 170 EOF
 171
 172     print HTML "<h2>Collation: ICU4C ".$ICULatestVersion." vs. GLIBC</h2>\n";
 173
 174     print HTML <<EOF;
 175 <p>The performance test takes a locale and creates a RuleBasedCollator with
 176 default options. A large list of names is used as data in each test, where the
 177 names vary according to language. Each Collation operation over the whole list
 178 is repeated 1000 times. The percentage values in the final column are the most
 179 useful. They measure differences, where positive is better for ICU4C, and
 180 negative is better for the compared implementation.</p>
 181 <h3>Key</h3>
 182 <table border="1" cellspacing="0" cellpadding="4">
 183 <tr>
 184 <th align="left">Operation</th>
 185 <th align="left">Units</th>
 186 <th align="left">Description</th>
 187 </tr>
 188 <tr>
 189 <td>strcoll</td>
 190 <td>nanosecs</td>
 191 <td>Timing for string collation, an incremental compare of strings.</td>
 192 </tr>
 193 <tr>
 194 <td>keygen</td>
 195 <td>nanosecs</td>
 196 <td>Timing for generation of sort keys, used to 'precompile' information so
 197 that subsequent operations can use binary comparison.</td>
 198 </tr>
 199 <tr>
 200 <td>keylen</td>
 201 <td>bytes/char</td>
 202 <td>The average length of the generated sort keys, in bytes per character
 203 (Unicode/ISO 10646 code point). Generally this is the important field for sort
 204 key performance, since it directly impacts the time necessary for binary
 205 comparison, and the overhead of memory usage and retrieval time for sort
 206 keys.</td>
 207 </tr>
 208 </table>
 209 EOF
 210     printData();
 211
 212     print HTML <<EOF;
 213 <h3><i>Notes</i></h3>
 214 <ol>
 215 <li>As with all performance measurements, the results will vary according to
 216 the hardware and compiler. The strcoll operation is particularly sensitive; we
 217 have found that even slight changes in code alignment can produce 10%
 218 differences.</li>
 219 <li>For more information on incremental vs. sort key comparison, the importance
 220 of multi-level sorting, and other features of collation, see <a href=
 221 "http://www.unicode.org/reports/tr10/">Unicode Collation (UCA)</a>.</li>
 222 <li>For general information on ICU collation see <a href=
 223 "/userguide/Collate_Intro.html">User Guide</a>.</li>
 224 <li>For information on APIs, see <a href="/apiref/icu4c/ucol_8h.html">C</a>,
 225 <a href="/apiref/icu4c/classCollator.html">C++</a>, or <a href=
 226 "/apiref/icu4j/com/ibm/icu/text/Collator.html">Java</a>.</li>
 227 </ol>
 228 <!--#include virtual="../ssi/footer.html" -->
 229 </body>
 230 </html>
 231
 232 EOF
 233
 234     close(HTML) or die "Can't close $html: $!";
 235 }
 236
 237 # This subroutine formats and prints the table.
 238 sub printData() {
 239     print HTML <<EOF;
 240 <h3>Data</h3>
 241 <table border="1" cellspacing="0" cellpadding="4">
 242 <tr>
 243 <td align="left"><b>Locale</b></td>
 244 <td align="left"><b>Data file</b></td>
 245 <td align="left"><b>strcoll</b> <i>(ICU)</i></td>
 246 <td align="left"><b>keygen</b> <i>(ICU)</i></td>
 247 <td align="left"><b>keylen</b> <i>(ICU)</i></td>
 248 <td align="left"><b>strcoll</b> <i>(GLIBC)</i></td>
 249 <td align="left"><b>keygen</b> <i>(GLIBC)</i></td>
 250 <td align="left"><b>keylen</b> <i>(GLIBC)</i></td>
 251 <td align="left"><b>strcoll</b> <i>(GLIBC-ICU)/ICU)</i></td>
 252 <td align="left"><b>keygen</b> <i>(GLIBC-ICU)/ICU)</i></td>
 253 <td align="left"><b>keylen</b> <i>(GLIBC-ICU)/ICU)</i></td>
 254 </tr>
 255 EOF
 256
 257     for ( $n = 0 ; $n < @resultFIN ; $n++ ) {
 258         print HTML "<tr>";
 259         my @parsed = split( ',', @resultFIN[$n] );
 260         for ( $i = 0 ; $i < @parsed ; $i++ ) {
 261             my $value = @parsed[$i];
 262             print HTML "<td align=\"center\">";
 263
 264             if ( $value =~ m/^[-]/ ) {
 265                 print HTML "<font color=\"red\">$value</font>";
 266             }
 267             else {
 268                 print HTML "$value";
 269             }
 270
 271             print HTML "</td>";
 272
 273         }
 274         print HTML "</tr>\n";
 275     }
 276
 277     print HTML<<EOF;
 278 </table>
 279 EOF
 280 }