• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1#!/usr/local/bin/perl
2# * © 2016 and later: Unicode, Inc. and others.
3# * License & terms of use: http://www.unicode.org/copyright.html
4# *******************************************************************************
5# * Copyright (C) 2002-2007 International Business Machines Corporation and     *
6# * others. All Rights Reserved.                                                *
7# *******************************************************************************
8
9use strict;
10
11# Assume we are running within the icu4j root directory
12use lib 'src/com/ibm/icu/dev/test/perf';
13use Dataset;
14
15#---------------------------------------------------------------------
16# Test class
17my $TESTCLASS = 'com.ibm.icu.dev.test.perf.DateFormatPerformanceTest';
18
19my $CLASSES = './out/bin:../tools/misc/out/bin/:../icu4j.jar';
20
21# Methods to be tested.  Each pair represents a test method and
22# a baseline method which is used for comparison.
23my @METHODS  = (
24                 ['TestJDKConstruction',     'TestICUConstruction'],
25                 ['TestJDKParse',            'TestICUParse'],
26                 ['TestJDKFormat',           'TestICUFormat']
27               );
28# Patterns which define the set of characters used for testing.
29my @OPTIONS = (
30#                 locale    pattern              date string
31                [ "en_US",  "dddd MMM yyyy",     "15 Jan 2007"],
32                [ "sw_KE",  "dddd MMM yyyy",     "15 Jan 2007"],
33                [ "en_US",  "HH:mm",             "13:13"],
34                [ "en_US",  "HH:mm zzzz",        "13:13 Pacific Standard Time"],
35                [ "en_US",  "HH:mm z",           "13:13 PST"],
36                [ "en_US",  "HH:mm Z",           "13:13 -0800"],
37              );
38
39my $THREADS;        # number of threads (input from command-line args)
40my $CALIBRATE = 2;  # duration in seconds for initial calibration
41my $DURATION  = 10; # duration in seconds for each pass
42my $NUMPASSES = 4;  # number of passes.  If > 1 then the first pass
43                    # is discarded as a JIT warm-up pass.
44
45my $TABLEATTR = 'BORDER="1" CELLPADDING="4" CELLSPACING="0"';
46
47my $PLUS_MINUS = "±";
48
49if ($NUMPASSES < 3) {
50    die "Need at least 3 passes.  One is discarded (JIT warmup) and need two to have 1 degree of freedom (t distribution).";
51}
52
53my $OUT; # see out()
54
55# run all tests with the specified number of threads from command-line input
56# (if there is no arguments, use $THREADS = 1)
57foreach my $arg ($#ARGV >= 0 ? @ARGV : "1") {
58  $THREADS = $arg;
59  main();
60}
61
62
63#---------------------------------------------------------------------
64# ...
65sub main {
66    my $date = localtime;
67    my $threads = ($THREADS > 1) ? "($THREADS threads)" : "";
68    my $title = "ICU4J Performance Test $threads $date";
69
70    my $html = $date;
71    $html =~ s/://g; # ':' illegal
72    $html =~ s/\s*\d+$//; # delete year
73    $html =~ s/^\w+\s*//; # delete dow
74    $html = "perf $html.html";
75
76    open(HTML,">$html") or die "Can't write to $html: $!";
77
78    print HTML <<EOF;
79<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
80   "http://www.w3.org/TR/html4/strict.dtd">
81<HTML>
82   <HEAD>
83      <TITLE>$title</TITLE>
84   </HEAD>
85   <BODY>
86EOF
87    print HTML "<H1>$title</H1>\n";
88
89    print HTML "<H2>$TESTCLASS</H2>\n";
90
91    my $raw = "";
92
93    for my $methodPair (@METHODS) {
94
95        my $testMethod = $methodPair->[0];
96        my $baselineMethod = $methodPair->[1];
97
98        print HTML "<P><TABLE $TABLEATTR><TR><TD>\n";
99        print HTML "<P><B>$testMethod vs. $baselineMethod</B></P>\n";
100
101        print HTML "<P><TABLE $TABLEATTR BGCOLOR=\"#CCFFFF\">\n";
102        print HTML "<TR><TD>Options</TD><TD>$testMethod</TD>";
103        print HTML "<TD>$baselineMethod</TD><TD>Ratio</TD></TR>\n";
104
105        $OUT = '';
106
107        for my $pat (@OPTIONS) {
108            print HTML "<TR><TD>@$pat[0], \"@$pat[1]\", \"@$pat[2]\"</TD>\n";
109
110            out("<P><TABLE $TABLEATTR WIDTH=\"100%\">");
111
112            # measure the test method
113            out("<TR><TD>");
114            print "\n$testMethod [@$pat]\n";
115            my $t = measure2($testMethod, $pat, -$DURATION);
116            out("</TD></TR>");
117            print HTML "<TD>", formatSeconds(4, $t->getMean(), $t->getError);
118            print HTML "/event</TD>\n";
119
120            # measure baseline method
121            out("<TR><TD>");
122            print "\n$baselineMethod [@$pat]\n";
123            my $b = measure2($baselineMethod, $pat, -$DURATION);
124            out("</TD></TR>");
125            print HTML "<TD>", formatSeconds(4, $b->getMean(), $t->getError);
126            print HTML "/event</TD>\n";
127
128            out("</TABLE></P>");
129
130            # output ratio
131            my $r = $t->divide($b);
132            my $mean = $r->getMean() - 1;
133            my $color = $mean < 0 ? "RED" : "BLACK";
134            print HTML "<TD><B><FONT COLOR=\"$color\">", formatPercent(3, $mean, $r->getError);
135            print HTML "</FONT></B></TD></TR>\n";
136        }
137
138        print HTML "</TABLE></P>\n";
139
140        print HTML "<P>Raw data:</P>\n";
141        print HTML $OUT;
142        print HTML "</TABLE></P>\n";
143    }
144
145    print HTML <<EOF;
146   </BODY>
147</HTML>
148EOF
149    close(HTML) or die "Can't close $html: $!";
150}
151
152#---------------------------------------------------------------------
153# Append text to the global variable $OUT
154sub out {
155    $OUT .= join('', @_);
156}
157
158#---------------------------------------------------------------------
159# Append text to the global variable $OUT
160sub outln {
161    $OUT .= join('', @_) . "\n";
162}
163
164#---------------------------------------------------------------------
165# Measure a given test method with a give test pattern using the
166# global run parameters.
167#
168# @param the method to run
169# @param the pattern defining characters to test
170# @param if >0 then the number of iterations per pass.  If <0 then
171#        (negative of) the number of seconds per pass.
172#
173# @return a Dataset object, scaled by iterations per pass and
174#         events per iteration, to give time per event
175#
176sub measure2 {
177    my @data = measure1(@_);
178    my $iterPerPass = shift(@data);
179    my $eventPerIter = shift(@data);
180
181    shift(@data) if (@data > 1); # discard first run
182
183    my $ds = Dataset->new(@data);
184    $ds->setScale(1.0e-3 / ($iterPerPass * $eventPerIter));
185    $ds;
186}
187
188#---------------------------------------------------------------------
189# Measure a given test method with a give test pattern using the
190# global run parameters.
191#
192# @param the method to run
193# @param the pattern defining characters to test
194# @param if >0 then the number of iterations per pass.  If <0 then
195#        (negative of) the number of seconds per pass.
196#
197# @return array of:
198#         [0] iterations per pass
199#         [1] events per iteration
200#         [2..] ms reported for each pass, in order
201#
202sub measure1 {
203    my $method = shift;
204    my $pat = shift;
205    my $iterCount = shift; # actually might be -seconds/pass
206
207    out("<P>Measuring $method for input file @$pat[0] for encoding @$pat[2] , ");
208    if ($iterCount > 0) {
209        out("$iterCount iterations/pass, $NUMPASSES passes</P>\n");
210    } else {
211        out(-$iterCount, " seconds/pass, $NUMPASSES passes</P>\n");
212    }
213
214    # is $iterCount actually -seconds/pass?
215    if ($iterCount < 0) {
216
217        # calibrate: estimate ms/iteration
218        print "Calibrating...";
219        my @t = callJava($method, $pat, -$CALIBRATE, 1);
220        print "done.\n";
221
222        my @data = split(/\s+/, $t[0]->[2]);
223        $data[0] *= 1.0e+3;
224
225        my $timePerIter = 1.0e-3 * $data[0] / $data[1];
226
227        # determine iterations/pass
228        $iterCount = int(-$iterCount / $timePerIter + 0.5);
229
230        out("<P>Calibration pass ($CALIBRATE sec): ");
231        out("$data[0] ms, ");
232        out("$data[1] iterations = ");
233        out(formatSeconds(4, $timePerIter), "/iteration<BR>\n");
234    }
235
236    # run passes
237    print "Measuring $iterCount iterations x $NUMPASSES passes...";
238    my @t = callJava($method, $pat, $iterCount, $NUMPASSES);
239    print "done.\n";
240    my @ms = ();
241    my @b; # scratch
242    for my $a (@t) {
243        # $a->[0]: method name, corresponds to $method
244        # $a->[1]: 'begin' data, == $iterCount
245        # $a->[2]: 'end' data, of the form <ms> <loops> <eventsPerIter>
246        # $a->[3...]: gc messages from JVM during pass
247        @b = split(/\s+/, $a->[2]);
248        push(@ms, $b[0] * 1.0e+3);
249    }
250    my $eventsPerIter = $b[2];
251
252    out("Iterations per pass: $iterCount<BR>\n");
253    out("Events per iteration: $eventsPerIter<BR>\n");
254
255    my @ms_str = @ms;
256    $ms_str[0] .= " (discarded)" if (@ms_str > 1);
257    out("Raw times (ms/pass): ", join(", ", @ms_str), "<BR>\n");
258
259    ($iterCount, $eventsPerIter, @ms);
260}
261
262#---------------------------------------------------------------------
263# Invoke java to run $TESTCLASS, passing it the given parameters.
264#
265# @param the method to run
266# @param the number of iterations, or if negative, the duration
267#        in seconds.  If more than on pass is desired, pass in
268#        a string, e.g., "100 100 100".
269# @param the pattern defining characters to test
270#
271# @return an array of results.  Each result is an array REF
272#         describing one pass.  The array REF contains:
273#         ->[0]: The method name as reported
274#         ->[1]: The params on the '= <meth> begin ...' line
275#         ->[2]: The params on the '= <meth> end ...' line
276#         ->[3..]: GC messages from the JVM, if any
277#
278sub callJava {
279    my $method = shift;
280    my $pat = shift;
281    my $n = shift;
282    my $passes = shift;
283
284    my $n = ($n < 0) ? "-t ".(-$n) : "-i ".$n;
285
286    my $cmd = "java -classpath $CLASSES $TESTCLASS $method $n -p $passes -L @$pat[0] \"@$pat[1]\" \"@$pat[2]\" -r $THREADS";
287    print "[$cmd]\n"; # for debugging
288    open(PIPE, "$cmd|") or die "Can't run \"$cmd\"";
289    my @out;
290    while (<PIPE>) {
291        push(@out, $_);
292    }
293    close(PIPE) or die "Java failed: \"$cmd\"";
294
295    @out = grep(!/^\#/, @out);  # filter out comments
296
297    #print "[", join("\n", @out), "]\n";
298
299    my @results;
300    my $method = '';
301    my $data = [];
302    foreach (@out) {
303        next unless (/\S/);
304
305        if (/^=\s*(\w+)\s*(\w+)\s*(.*)/) {
306            my ($m, $state, $d) = ($1, $2, $3);
307            #print "$_ => [[$m $state $data]]\n";
308            if ($state eq 'begin') {
309                die "$method was begun but not finished" if ($method);
310                $method = $m;
311                push(@$data, $d);
312                push(@$data, ''); # placeholder for end data
313            } elsif ($state eq 'end') {
314                if ($m ne $method) {
315                    die "$method end does not match: $_";
316                }
317                $data->[1] = $d; # insert end data at [1]
318                #print "#$method:", join(";",@$data), "\n";
319                unshift(@$data, $method); # add method to start
320
321                push(@results, $data);
322                $method = '';
323                $data = [];
324            } else {
325                die "Can't parse: $_";
326            }
327        }
328
329        elsif (/^\[/) {
330            if ($method) {
331                push(@$data, $_);
332            } else {
333                # ignore extraneous GC notices
334            }
335        }
336
337        else {
338            die "Can't parse: $_";
339        }
340    }
341
342    die "$method was begun but not finished" if ($method);
343
344    @results;
345}
346
347#|#---------------------------------------------------------------------
348#|# Format a confidence interval, as given by a Dataset.  Output is as
349#|# as follows:
350#|#   241.23 - 241.98 => 241.5 +/- 0.3
351#|#   241.2 - 243.8 => 242 +/- 1
352#|#   211.0 - 241.0 => 226 +/- 15 or? 230 +/- 20
353#|#   220.3 - 234.3 => 227 +/- 7
354#|#   220.3 - 300.3 => 260 +/- 40
355#|#   220.3 - 1000 => 610 +/- 390 or? 600 +/- 400
356#|#   0.022 - 0.024 => 0.023 +/- 0.001
357#|#   0.022 - 0.032 => 0.027 +/- 0.005
358#|#   0.022 - 1.000 => 0.5 +/- 0.5
359#|# In other words, take one significant digit of the error value and
360#|# display the mean to the same precision.
361#|sub formatDataset {
362#|    my $ds = shift;
363#|    my $lower = $ds->getMean() - $ds->getError();
364#|    my $upper = $ds->getMean() + $ds->getError();
365#|    my $scale = 0;
366#|    # Find how many initial digits are the same
367#|    while ($lower < 1 ||
368#|           int($lower) == int($upper)) {
369#|        $lower *= 10;
370#|        $upper *= 10;
371#|        $scale++;
372#|    }
373#|    while ($lower >= 10 &&
374#|           int($lower) == int($upper)) {
375#|        $lower /= 10;
376#|        $upper /= 10;
377#|        $scale--;
378#|    }
379#|}
380
381#---------------------------------------------------------------------
382# Format a number, optionally with a +/- delta, to n significant
383# digits.
384#
385# @param significant digit, a value >= 1
386# @param multiplier
387# @param time in seconds to be formatted
388# @optional delta in seconds
389#
390# @return string of the form "23" or "23 +/- 10".
391#
392sub formatNumber {
393    my $sigdig = shift;
394    my $mult = shift;
395    my $a = shift;
396    my $delta = shift; # may be undef
397
398    my $result = formatSigDig($sigdig, $a*$mult);
399    if (defined($delta)) {
400        my $d = formatSigDig($sigdig, $delta*$mult);
401        # restrict PRECISION of delta to that of main number
402        if ($result =~ /\.(\d+)/) {
403            # TODO make this work for values with all significant
404            # digits to the left of the decimal, e.g., 1234000.
405
406            # TODO the other thing wrong with this is that it
407            # isn't rounding the $delta properly.  Have to put
408            # this logic into formatSigDig().
409            my $x = length($1);
410            $d =~ s/\.(\d{$x})\d+/.$1/;
411        }
412        $result .= " $PLUS_MINUS " . $d;
413    }
414    $result;
415}
416
417#---------------------------------------------------------------------
418# Format a time, optionally with a +/- delta, to n significant
419# digits.
420#
421# @param significant digit, a value >= 1
422# @param time in seconds to be formatted
423# @optional delta in seconds
424#
425# @return string of the form "23 ms" or "23 +/- 10 ms".
426#
427sub formatSeconds {
428    my $sigdig = shift;
429    my $a = shift;
430    my $delta = shift; # may be undef
431
432    my @MULT = (1   , 1e3,  1e6,  1e9);
433    my @SUFF = ('s' , 'ms', 'us', 'ns');
434
435    # Determine our scale
436    my $i = 0;
437    ++$i while ($a*$MULT[$i] < 1 && $i < @MULT);
438
439    formatNumber($sigdig, $MULT[$i], $a, $delta) . ' ' . $SUFF[$i];
440}
441
442#---------------------------------------------------------------------
443# Format a percentage, optionally with a +/- delta, to n significant
444# digits.
445#
446# @param significant digit, a value >= 1
447# @param value to be formatted, as a fraction, e.g. 0.5 for 50%
448# @optional delta, as a fraction
449#
450# @return string of the form "23 %" or "23 +/- 10 %".
451#
452sub formatPercent {
453    my $sigdig = shift;
454    my $a = shift;
455    my $delta = shift; # may be undef
456
457    formatNumber($sigdig, 100, $a, $delta) . ' %';
458}
459
460#---------------------------------------------------------------------
461# Format a number to n significant digits without using exponential
462# notation.
463#
464# @param significant digit, a value >= 1
465# @param number to be formatted
466#
467# @return string of the form "1234" "12.34" or "0.001234".  If
468#         number was negative, prefixed by '-'.
469#
470sub formatSigDig {
471    my $n = shift() - 1;
472    my $a = shift;
473
474    local $_ = sprintf("%.${n}e", $a);
475    my $sign = (s/^-//) ? '-' : '';
476
477    my $a_e;
478    my $result;
479    if (/^(\d)\.(\d+)e([-+]\d+)$/) {
480        my ($d, $dn, $e) = ($1, $2, $3);
481        $a_e = $e;
482        $d .= $dn;
483        $e++;
484        $d .= '0' while ($e > length($d));
485        while ($e < 1) {
486            $e++;
487            $d = '0' . $d;
488        }
489        if ($e == length($d)) {
490            $result = $sign . $d;
491        } else {
492            $result = $sign . substr($d, 0, $e) . '.' . substr($d, $e);
493        }
494    } else {
495        die "Can't parse $_";
496    }
497    $result;
498}
499
500#eof
501