Home | History | Annotate | Download | only in perf-tests 
      1 #!/usr/local/bin/perl
      2 # *******************************************************************************
      3 # * Copyright (C) 2002-2007 International Business Machines Corporation and     *
      4 # * others. All Rights Reserved.                                                *
      5 # *******************************************************************************
      6 
      7 use strict;
      8 
      9 # Assume we are running within the icu4j root directory
     10 use lib 'src/com/ibm/icu/dev/test/perf';
     11 use Dataset;
     12 
     13 #---------------------------------------------------------------------
     14 # Test class
     15 my $TESTCLASS = 'com.ibm.icu.dev.test.perf.DateFormatPerformanceTest';
     16 
     17 # Methods to be tested.  Each pair represents a test method and
     18 # a baseline method which is used for comparison.
     19 my @METHODS  = (
     20                  ['TestJDKConstruction',     'TestICUConstruction'],
     21                  ['TestJDKParse',            'TestICUParse'],
     22                  ['TestJDKFormat',           'TestICUFormat']
     23                );
     24 # Patterns which define the set of characters used for testing.
     25 my @OPTIONS = (
     26 #                 locale    pattern              date string
     27                 [ "en_US",  "dddd MMM yyyy",     "15 Jan 2007"],
     28                 [ "sw_KE",  "dddd MMM yyyy",     "15 Jan 2007"],
     29                 [ "en_US",  "HH:mm",             "13:13"],
     30                 [ "en_US",  "HH:mm zzzz",        "13:13 Pacific Standard Time"],
     31                 [ "en_US",  "HH:mm z",           "13:13 PST"],
     32                 [ "en_US",  "HH:mm Z",           "13:13 -0800"],
     33               );
     34 
     35 my $THREADS;        # number of threads (input from command-line args)
     36 my $CALIBRATE = 2;  # duration in seconds for initial calibration
     37 my $DURATION  = 10; # duration in seconds for each pass
     38 my $NUMPASSES = 4;  # number of passes.  If > 1 then the first pass
     39                     # is discarded as a JIT warm-up pass.
     40 
     41 my $TABLEATTR = 'BORDER="1" CELLPADDING="4" CELLSPACING="0"';
     42 
     43 my $PLUS_MINUS = "±";
     44 
     45 if ($NUMPASSES < 3) {
     46     die "Need at least 3 passes.  One is discarded (JIT warmup) and need two to have 1 degree of freedom (t distribution).";
     47 }
     48 
     49 my $OUT; # see out()
     50 
     51 # run all tests with the specified number of threads from command-line input
     52 # (if there is no arguments, use $THREADS = 1)
     53 foreach my $arg ($#ARGV >= 0 ? @ARGV : "1") {
     54   $THREADS = $arg;
     55   main();
     56 }
     57 
     58 
     59 #---------------------------------------------------------------------
     60 # ...
     61 sub main {
     62     my $date = localtime;
     63     my $threads = ($THREADS > 1) ? "($THREADS threads)" : "";
     64     my $title = "ICU4J Performance Test $threads $date";
     65 
     66     my $html = $date;
     67     $html =~ s/://g; # ':' illegal
     68     $html =~ s/\s*\d+$//; # delete year
     69     $html =~ s/^\w+\s*//; # delete dow
     70     $html = "perf $html.html";
     71 
     72     open(HTML,">$html") or die "Can't write to $html: $!";
     73 
     74     print HTML <<EOF;
     75 <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
     76    "http://www.w3.org/TR/html4/strict.dtd">
     77 <HTML>
     78    <HEAD>
     79       <TITLE>$title</TITLE>
     80    </HEAD>
     81    <BODY>
     82 EOF
     83     print HTML "<H1>$title</H1>\n";
     84 
     85     print HTML "<H2>$TESTCLASS</H2>\n";
     86 
     87     my $raw = "";
     88 
     89     for my $methodPair (@METHODS) {
     90 
     91         my $testMethod = $methodPair->[0];
     92         my $baselineMethod = $methodPair->[1];
     93 
     94         print HTML "<P><TABLE $TABLEATTR><TR><TD>\n";
     95         print HTML "<P><B>$testMethod vs. $baselineMethod</B></P>\n";
     96         
     97         print HTML "<P><TABLE $TABLEATTR BGCOLOR=\"#CCFFFF\">\n";
     98         print HTML "<TR><TD>Options</TD><TD>$testMethod</TD>";
     99         print HTML "<TD>$baselineMethod</TD><TD>Ratio</TD></TR>\n";
    100 
    101         $OUT = '';
    102 
    103         for my $pat (@OPTIONS) {
    104             print HTML "<TR><TD>@$pat[0], \"@$pat[1]\", \"@$pat[2]\"</TD>\n";
    105 
    106             out("<P><TABLE $TABLEATTR WIDTH=\"100%\">");
    107 
    108             # measure the test method
    109             out("<TR><TD>");
    110             print "\n$testMethod [@$pat]\n";
    111             my $t = measure2($testMethod, $pat, -$DURATION);
    112             out("</TD></TR>");
    113             print HTML "<TD>", formatSeconds(4, $t->getMean(), $t->getError);
    114             print HTML "/event</TD>\n";
    115 
    116             # measure baseline method
    117             out("<TR><TD>");
    118             print "\n$baselineMethod [@$pat]\n";
    119             my $b = measure2($baselineMethod, $pat, -$DURATION);
    120             out("</TD></TR>");
    121             print HTML "<TD>", formatSeconds(4, $b->getMean(), $t->getError);
    122             print HTML "/event</TD>\n";
    123 
    124             out("</TABLE></P>");
    125 
    126             # output ratio
    127             my $r = $t->divide($b);
    128             my $mean = $r->getMean() - 1;
    129             my $color = $mean < 0 ? "RED" : "BLACK";
    130             print HTML "<TD><B><FONT COLOR=\"$color\">", formatPercent(3, $mean, $r->getError);
    131             print HTML "</FONT></B></TD></TR>\n";
    132         }
    133 
    134         print HTML "</TABLE></P>\n";
    135 
    136         print HTML "<P>Raw data:</P>\n";
    137         print HTML $OUT;
    138         print HTML "</TABLE></P>\n";
    139     }
    140 
    141     print HTML <<EOF;
    142    </BODY>
    143 </HTML>
    144 EOF
    145     close(HTML) or die "Can't close $html: $!";
    146 }
    147 
    148 #---------------------------------------------------------------------
    149 # Append text to the global variable $OUT
    150 sub out {
    151     $OUT .= join('', @_);
    152 }
    153 
    154 #---------------------------------------------------------------------
    155 # Append text to the global variable $OUT
    156 sub outln {
    157     $OUT .= join('', @_) . "\n";
    158 }
    159 
    160 #---------------------------------------------------------------------
    161 # Measure a given test method with a give test pattern using the
    162 # global run parameters.
    163 #
    164 # @param the method to run
    165 # @param the pattern defining characters to test
    166 # @param if >0 then the number of iterations per pass.  If <0 then
    167 #        (negative of) the number of seconds per pass.
    168 #
    169 # @return a Dataset object, scaled by iterations per pass and
    170 #         events per iteration, to give time per event
    171 #
    172 sub measure2 {
    173     my @data = measure1(@_);
    174     my $iterPerPass = shift(@data);
    175     my $eventPerIter = shift(@data);
    176 
    177     shift(@data) if (@data > 1); # discard first run
    178 
    179     my $ds = Dataset->new(@data);
    180     $ds->setScale(1.0e-3 / ($iterPerPass * $eventPerIter));
    181     $ds;
    182 }
    183 
    184 #---------------------------------------------------------------------
    185 # Measure a given test method with a give test pattern using the
    186 # global run parameters.
    187 #
    188 # @param the method to run
    189 # @param the pattern defining characters to test
    190 # @param if >0 then the number of iterations per pass.  If <0 then
    191 #        (negative of) the number of seconds per pass.
    192 #
    193 # @return array of:
    194 #         [0] iterations per pass
    195 #         [1] events per iteration
    196 #         [2..] ms reported for each pass, in order
    197 #
    198 sub measure1 {
    199     my $method = shift;
    200     my $pat = shift;
    201     my $iterCount = shift; # actually might be -seconds/pass
    202 
    203     out("<P>Measuring $method for input file @$pat[0] for encoding @$pat[2] , ");
    204     if ($iterCount > 0) {
    205         out("$iterCount iterations/pass, $NUMPASSES passes</P>\n");
    206     } else {
    207         out(-$iterCount, " seconds/pass, $NUMPASSES passes</P>\n");
    208     }
    209 
    210     # is $iterCount actually -seconds/pass?
    211     if ($iterCount < 0) {
    212 
    213         # calibrate: estimate ms/iteration
    214         print "Calibrating...";
    215         my @t = callJava($method, $pat, -$CALIBRATE, 1);
    216         print "done.\n";
    217 
    218         my @data = split(/\s+/, $t[0]->[2]);
    219         $data[0] *= 1.0e+3;
    220 
    221         my $timePerIter = 1.0e-3 * $data[0] / $data[1];
    222         
    223         # determine iterations/pass
    224         $iterCount = int(-$iterCount / $timePerIter + 0.5);
    225         
    226         out("<P>Calibration pass ($CALIBRATE sec): ");
    227         out("$data[0] ms, ");
    228         out("$data[1] iterations = ");
    229         out(formatSeconds(4, $timePerIter), "/iteration<BR>\n");
    230     }
    231     
    232     # run passes
    233     print "Measuring $iterCount iterations x $NUMPASSES passes...";
    234     my @t = callJava($method, $pat, $iterCount, $NUMPASSES);
    235     print "done.\n";
    236     my @ms = ();
    237     my @b; # scratch
    238     for my $a (@t) {
    239         # $a->[0]: method name, corresponds to $method
    240         # $a->[1]: 'begin' data, == $iterCount
    241         # $a->[2]: 'end' data, of the form <ms> <loops> <eventsPerIter>
    242         # $a->[3...]: gc messages from JVM during pass
    243         @b = split(/\s+/, $a->[2]);
    244         push(@ms, $b[0] * 1.0e+3);
    245     }
    246     my $eventsPerIter = $b[2];
    247 
    248     out("Iterations per pass: $iterCount<BR>\n");
    249     out("Events per iteration: $eventsPerIter<BR>\n");
    250 
    251     my @ms_str = @ms;
    252     $ms_str[0] .= " (discarded)" if (@ms_str > 1);
    253     out("Raw times (ms/pass): ", join(", ", @ms_str), "<BR>\n");
    254 
    255     ($iterCount, $eventsPerIter, @ms);
    256 }
    257 
    258 #---------------------------------------------------------------------
    259 # Invoke java to run $TESTCLASS, passing it the given parameters.
    260 #
    261 # @param the method to run
    262 # @param the number of iterations, or if negative, the duration
    263 #        in seconds.  If more than on pass is desired, pass in
    264 #        a string, e.g., "100 100 100".
    265 # @param the pattern defining characters to test
    266 #
    267 # @return an array of results.  Each result is an array REF
    268 #         describing one pass.  The array REF contains:
    269 #         ->[0]: The method name as reported
    270 #         ->[1]: The params on the '= <meth> begin ...' line
    271 #         ->[2]: The params on the '= <meth> end ...' line
    272 #         ->[3..]: GC messages from the JVM, if any
    273 #
    274 sub callJava {
    275     my $method = shift;
    276     my $pat = shift;
    277     my $n = shift;
    278     my $passes = shift;
    279     
    280     my $n = ($n < 0) ? "-t ".(-$n) : "-i ".$n;
    281     
    282     my $cmd = "java -classpath classes $TESTCLASS $method $n -p $passes -L @$pat[0] \"@$pat[1]\" \"@$pat[2]\" -r $THREADS";
    283     print "[$cmd]\n"; # for debugging
    284     open(PIPE, "$cmd|") or die "Can't run \"$cmd\"";
    285     my @out;
    286     while (<PIPE>) {
    287         push(@out, $_);
    288     }
    289     close(PIPE) or die "Java failed: \"$cmd\"";
    290 
    291     @out = grep(!/^\#/, @out);  # filter out comments
    292 
    293     #print "[", join("\n", @out), "]\n";
    294 
    295     my @results;
    296     my $method = '';
    297     my $data = [];
    298     foreach (@out) {
    299         next unless (/\S/);
    300 
    301         if (/^=\s*(\w+)\s*(\w+)\s*(.*)/) {
    302             my ($m, $state, $d) = ($1, $2, $3);
    303             #print "$_ => [[$m $state $data]]\n";
    304             if ($state eq 'begin') {
    305                 die "$method was begun but not finished" if ($method);
    306                 $method = $m;
    307                 push(@$data, $d);
    308                 push(@$data, ''); # placeholder for end data
    309             } elsif ($state eq 'end') {
    310                 if ($m ne $method) {
    311                     die "$method end does not match: $_";
    312                 }
    313                 $data->[1] = $d; # insert end data at [1]
    314                 #print "#$method:", join(";",@$data), "\n";
    315                 unshift(@$data, $method); # add method to start
    316 
    317                 push(@results, $data);
    318                 $method = '';
    319                 $data = [];
    320             } else {
    321                 die "Can't parse: $_";
    322             }
    323         }
    324 
    325         elsif (/^\[/) {
    326             if ($method) {
    327                 push(@$data, $_);
    328             } else {
    329                 # ignore extraneous GC notices
    330             }
    331         }
    332 
    333         else {
    334             die "Can't parse: $_";
    335         }
    336     }
    337 
    338     die "$method was begun but not finished" if ($method);
    339 
    340     @results;
    341 }
    342 
    343 #|#---------------------------------------------------------------------
    344 #|# Format a confidence interval, as given by a Dataset.  Output is as
    345 #|# as follows:
    346 #|#   241.23 - 241.98 => 241.5 +/- 0.3
    347 #|#   241.2 - 243.8 => 242 +/- 1
    348 #|#   211.0 - 241.0 => 226 +/- 15 or? 230 +/- 20
    349 #|#   220.3 - 234.3 => 227 +/- 7
    350 #|#   220.3 - 300.3 => 260 +/- 40
    351 #|#   220.3 - 1000 => 610 +/- 390 or? 600 +/- 400
    352 #|#   0.022 - 0.024 => 0.023 +/- 0.001
    353 #|#   0.022 - 0.032 => 0.027 +/- 0.005
    354 #|#   0.022 - 1.000 => 0.5 +/- 0.5
    355 #|# In other words, take one significant digit of the error value and
    356 #|# display the mean to the same precision.
    357 #|sub formatDataset {
    358 #|    my $ds = shift;
    359 #|    my $lower = $ds->getMean() - $ds->getError();
    360 #|    my $upper = $ds->getMean() + $ds->getError();
    361 #|    my $scale = 0;
    362 #|    # Find how many initial digits are the same
    363 #|    while ($lower < 1 ||
    364 #|           int($lower) == int($upper)) {
    365 #|        $lower *= 10;
    366 #|        $upper *= 10;
    367 #|        $scale++;
    368 #|    }
    369 #|    while ($lower >= 10 &&
    370 #|           int($lower) == int($upper)) {
    371 #|        $lower /= 10;
    372 #|        $upper /= 10;
    373 #|        $scale--;
    374 #|    }
    375 #|}
    376 
    377 #---------------------------------------------------------------------
    378 # Format a number, optionally with a +/- delta, to n significant
    379 # digits.
    380 #
    381 # @param significant digit, a value >= 1
    382 # @param multiplier
    383 # @param time in seconds to be formatted
    384 # @optional delta in seconds
    385 #
    386 # @return string of the form "23" or "23 +/- 10".
    387 #
    388 sub formatNumber {
    389     my $sigdig = shift;
    390     my $mult = shift;
    391     my $a = shift;
    392     my $delta = shift; # may be undef
    393     
    394     my $result = formatSigDig($sigdig, $a*$mult);
    395     if (defined($delta)) {
    396         my $d = formatSigDig($sigdig, $delta*$mult);
    397         # restrict PRECISION of delta to that of main number
    398         if ($result =~ /\.(\d+)/) {
    399             # TODO make this work for values with all significant
    400             # digits to the left of the decimal, e.g., 1234000.
    401 
    402             # TODO the other thing wrong with this is that it
    403             # isn't rounding the $delta properly.  Have to put
    404             # this logic into formatSigDig().
    405             my $x = length($1);
    406             $d =~ s/\.(\d{$x})\d+/.$1/;
    407         }
    408         $result .= " $PLUS_MINUS " . $d;
    409     }
    410     $result;
    411 }
    412 
    413 #---------------------------------------------------------------------
    414 # Format a time, optionally with a +/- delta, to n significant
    415 # digits.
    416 #
    417 # @param significant digit, a value >= 1
    418 # @param time in seconds to be formatted
    419 # @optional delta in seconds
    420 #
    421 # @return string of the form "23 ms" or "23 +/- 10 ms".
    422 #
    423 sub formatSeconds {
    424     my $sigdig = shift;
    425     my $a = shift;
    426     my $delta = shift; # may be undef
    427 
    428     my @MULT = (1   , 1e3,  1e6,  1e9);
    429     my @SUFF = ('s' , 'ms', 'us', 'ns');
    430 
    431     # Determine our scale
    432     my $i = 0;
    433     ++$i while ($a*$MULT[$i] < 1 && $i < @MULT);
    434     
    435     formatNumber($sigdig, $MULT[$i], $a, $delta) . ' ' . $SUFF[$i];
    436 }
    437 
    438 #---------------------------------------------------------------------
    439 # Format a percentage, optionally with a +/- delta, to n significant
    440 # digits.
    441 #
    442 # @param significant digit, a value >= 1
    443 # @param value to be formatted, as a fraction, e.g. 0.5 for 50%
    444 # @optional delta, as a fraction
    445 #
    446 # @return string of the form "23 %" or "23 +/- 10 %".
    447 #
    448 sub formatPercent {
    449     my $sigdig = shift;
    450     my $a = shift;
    451     my $delta = shift; # may be undef
    452     
    453     formatNumber($sigdig, 100, $a, $delta) . ' %';
    454 }
    455 
    456 #---------------------------------------------------------------------
    457 # Format a number to n significant digits without using exponential
    458 # notation.
    459 #
    460 # @param significant digit, a value >= 1
    461 # @param number to be formatted
    462 #
    463 # @return string of the form "1234" "12.34" or "0.001234".  If
    464 #         number was negative, prefixed by '-'.
    465 #
    466 sub formatSigDig {
    467     my $n = shift() - 1;
    468     my $a = shift;
    469 
    470     local $_ = sprintf("%.${n}e", $a);
    471     my $sign = (s/^-//) ? '-' : '';
    472 
    473     my $a_e;
    474     my $result;
    475     if (/^(\d)\.(\d+)e([-+]\d+)$/) {
    476         my ($d, $dn, $e) = ($1, $2, $3);
    477         $a_e = $e;
    478         $d .= $dn;
    479         $e++;
    480         $d .= '0' while ($e > length($d));
    481         while ($e < 1) {
    482             $e++;
    483             $d = '0' . $d;
    484         }
    485         if ($e == length($d)) {
    486             $result = $sign . $d;
    487         } else {
    488             $result = $sign . substr($d, 0, $e) . '.' . substr($d, $e);
    489         }
    490     } else {
    491         die "Can't parse $_";
    492     }
    493     $result;
    494 }
    495 
    496 #eof
    497