1#!/usr/local/bin/perl
2# * © 2016 and later: Unicode, Inc. and others.
3# * License & terms of use: http://www.unicode.org/copyright.html#License
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.ConverterPerformanceTest';
18
19# Methods to be tested.  Each pair represents a test method and
20# a baseline method which is used for comparison.
21my @METHODS  = (
22##               ['TestByteToCharConverter', 'TestByteToCharConverterICU'],
23##               ['TestCharToByteConverter', 'TestCharToByteConverterICU'],
24                 ['TestCharsetDecoder',      'TestCharsetDecoderICU'],
25                 ['TestCharsetEncoder',      'TestCharsetEncoderICU']
26               );
27
28# Patterns which define the set of characters used for testing.
29
30my $SOURCEDIR ="src/com/ibm/icu/dev/test/perf/data/conversion/";
31
32my @OPTIONS = (
33#                                 src text          src encoding    test encoding
34                                [ "arabic.txt",     "UTF-8",        "csisolatinarabic"],
35                                [ "french.txt",     "UTF-8",        "csisolatin1"],
36                                [ "greek.txt",      "UTF-8",        "csisolatingreek"],
37                                [ "hebrew.txt",     "UTF-8",        "csisolatinhebrew"],
38#                               [ "hindi.txt" ,     "UTF-8",        "iscii"],
39                                [ "japanese.txt",   "UTF-8",        "EUC-JP"],
40#                               [ "japanese.txt",   "UTF-8",        "csiso2022jp"],
41                                [ "japanese.txt",   "UTF-8",        "shift_jis"],
42#                               [ "korean.txt",     "UTF-8",        "csiso2022kr"],
43                                [ "korean.txt",     "UTF-8",        "EUC-KR"],
44                                [ "s-chinese.txt",  "UTF-8",        "EUC_CN"],
45                                [ "arabic.txt",     "UTF-8",        "UTF-8"],
46                                [ "french.txt",     "UTF-8",        "UTF-8"],
47                                [ "greek.txt",      "UTF-8",        "UTF-8"],
48                                [ "hebrew.txt",     "UTF-8",        "UTF-8"],
49                                [ "hindi.txt" ,     "UTF-8",        "UTF-8"],
50                                [ "japanese.txt",   "UTF-8",        "UTF-8"],
51                                [ "korean.txt",     "UTF-8",        "UTF-8"],
52                                [ "s-chinese.txt",  "UTF-8",        "UTF-8"],
53                                [ "french.txt",     "UTF-8",        "UTF-16BE"],
54                                [ "french.txt",     "UTF-8",        "UTF-16LE"],
55                                [ "english.txt",    "UTF-8",        "US-ASCII"],
56                          );
57
58my $CALIBRATE = 2;  # duration in seconds for initial calibration
59my $DURATION  = 10; # duration in seconds for each pass
60my $NUMPASSES = 4;  # number of passes.  If > 1 then the first pass
61                    # is discarded as a JIT warm-up pass.
62
63my $TABLEATTR = 'BORDER="1" CELLPADDING="4" CELLSPACING="0"';
64
65my $PLUS_MINUS = "±";
66
67if ($NUMPASSES < 3) {
68    die "Need at least 3 passes.  One is discarded (JIT warmup) and need two to have 1 degree of freedom (t distribution).";
69}
70
71my $OUT; # see out()
72
73main();
74
75#---------------------------------------------------------------------
76# ...
77sub main {
78    my $date = localtime;
79    my $title = "ICU4J Performance Test $date";
80
81    my $html = $date;
82    $html =~ s/://g; # ':' illegal
83    $html =~ s/\s*\d+$//; # delete year
84    $html =~ s/^\w+\s*//; # delete dow
85    $html = "perf $html.html";
86
87    open(HTML,">$html") or die "Can't write to $html: $!";
88
89    print HTML <<EOF;
90<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
91   "http://www.w3.org/TR/html4/strict.dtd">
92<HTML>
93   <HEAD>
94      <TITLE>$title</TITLE>
95   </HEAD>
96   <BODY>
97EOF
98    print HTML "<H1>$title</H1>\n";
99
100    print HTML "<H2>$TESTCLASS</H2>\n";
101
102    my $raw = "";
103
104    for my $methodPair (@METHODS) {
105
106        my $testMethod = $methodPair->[0];
107        my $baselineMethod = $methodPair->[1];
108
109        print HTML "<P><TABLE $TABLEATTR><TR><TD>\n";
110        print HTML "<P><B>$testMethod vs. $baselineMethod</B></P>\n";
111
112        print HTML "<P><TABLE $TABLEATTR BGCOLOR=\"#CCFFFF\">\n";
113        print HTML "<TR><TD>Options</TD><TD>$testMethod</TD>";
114        print HTML "<TD>$baselineMethod</TD><TD>Ratio</TD></TR>\n";
115
116        $OUT = '';
117
118        for my $pat (@OPTIONS) {
119            print HTML "<TR><TD>@$pat[0], @$pat[2]</TD>\n";
120
121            out("<P><TABLE $TABLEATTR WIDTH=\"100%\">");
122
123            # measure the test method
124            out("<TR><TD>");
125            print "\n$testMethod [@$pat]\n";
126            my $t = measure2($testMethod, $pat, -$DURATION);
127            out("</TD></TR>");
128            print HTML "<TD>", formatSeconds(4, $t->getMean(), $t->getError);
129            print HTML "/event</TD>\n";
130
131            # measure baseline method
132            out("<TR><TD>");
133            print "\n$baselineMethod [@$pat]\n";
134            my $b = measure2($baselineMethod, $pat, -$DURATION);
135            out("</TD></TR>");
136            print HTML "<TD>", formatSeconds(4, $b->getMean(), $t->getError);
137            print HTML "/event</TD>\n";
138
139            out("</TABLE></P>");
140
141            # output ratio
142            my $r = $t->divide($b);
143            my $mean = $r->getMean() - 1;
144            my $color = $mean < 0 ? "RED" : "BLACK";
145            print HTML "<TD><B><FONT COLOR=\"$color\">", formatPercent(3, $mean, $r->getError);
146            print HTML "</FONT></B></TD></TR>\n";
147        }
148
149        print HTML "</TABLE></P>\n";
150
151        print HTML "<P>Raw data:</P>\n";
152        print HTML $OUT;
153        print HTML "</TABLE></P>\n";
154    }
155
156    print HTML <<EOF;
157   </BODY>
158</HTML>
159EOF
160    close(HTML) or die "Can't close $html: $!";
161}
162
163#---------------------------------------------------------------------
164# Append text to the global variable $OUT
165sub out {
166    $OUT .= join('', @_);
167}
168
169#---------------------------------------------------------------------
170# Append text to the global variable $OUT
171sub outln {
172    $OUT .= join('', @_) . "\n";
173}
174
175#---------------------------------------------------------------------
176# Measure a given test method with a give test pattern using the
177# global run parameters.
178#
179# @param the method to run
180# @param the pattern defining characters to test
181# @param if >0 then the number of iterations per pass.  If <0 then
182#        (negative of) the number of seconds per pass.
183#
184# @return a Dataset object, scaled by iterations per pass and
185#         events per iteration, to give time per event
186#
187sub measure2 {
188    my @data = measure1(@_);
189    my $iterPerPass = shift(@data);
190    my $eventPerIter = shift(@data);
191
192    shift(@data) if (@data > 1); # discard first run
193
194    my $ds = Dataset->new(@data);
195    $ds->setScale(1.0e-3 / ($iterPerPass * $eventPerIter));
196    $ds;
197}
198
199#---------------------------------------------------------------------
200# Measure a given test method with a give test pattern using the
201# global run parameters.
202#
203# @param the method to run
204# @param the pattern defining characters to test
205# @param if >0 then the number of iterations per pass.  If <0 then
206#        (negative of) the number of seconds per pass.
207#
208# @return array of:
209#         [0] iterations per pass
210#         [1] events per iteration
211#         [2..] ms reported for each pass, in order
212#
213sub measure1 {
214    my $method = shift;
215    my $pat = shift;
216    my $iterCount = shift; # actually might be -seconds/pass
217
218    out("<P>Measuring $method for input file @$pat[0] for encoding @$pat[2] , ");
219    if ($iterCount > 0) {
220        out("$iterCount iterations/pass, $NUMPASSES passes</P>\n");
221    } else {
222        out(-$iterCount, " seconds/pass, $NUMPASSES passes</P>\n");
223    }
224
225    # is $iterCount actually -seconds/pass?
226    if ($iterCount < 0) {
227
228        # calibrate: estimate ms/iteration
229        print "Calibrating...";
230        my @t = callJava($method, $pat, -$CALIBRATE, 1);
231        print "done.\n";
232
233        my @data = split(/\s+/, $t[0]->[2]);
234        $data[0] *= 1.0e+3;
235
236        my $timePerIter = 1.0e-3 * $data[0] / $data[1];
237
238        # determine iterations/pass
239        $iterCount = int(-$iterCount / $timePerIter + 0.5);
240
241        out("<P>Calibration pass ($CALIBRATE sec): ");
242        out("$data[0] ms, ");
243        out("$data[1] iterations = ");
244        out(formatSeconds(4, $timePerIter), "/iteration<BR>\n");
245    }
246
247    # run passes
248    print "Measuring $iterCount iterations x $NUMPASSES passes...";
249    my @t = callJava($method, $pat, $iterCount, $NUMPASSES);
250    print "done.\n";
251    my @ms = ();
252    my @b; # scratch
253    for my $a (@t) {
254        # $a->[0]: method name, corresponds to $method
255        # $a->[1]: 'begin' data, == $iterCount
256        # $a->[2]: 'end' data, of the form <ms> <loops> <eventsPerIter>
257        # $a->[3...]: gc messages from JVM during pass
258        @b = split(/\s+/, $a->[2]);
259        push(@ms, $b[0] * 1.0e+3);
260    }
261    my $eventsPerIter = $b[2];
262
263    out("Iterations per pass: $iterCount<BR>\n");
264    out("Events per iteration: $eventsPerIter<BR>\n");
265
266    my @ms_str = @ms;
267    $ms_str[0] .= " (discarded)" if (@ms_str > 1);
268    out("Raw times (ms/pass): ", join(", ", @ms_str), "<BR>\n");
269
270    ($iterCount, $eventsPerIter, @ms);
271}
272
273#---------------------------------------------------------------------
274# Invoke java to run $TESTCLASS, passing it the given parameters.
275#
276# @param the method to run
277# @param the number of iterations, or if negative, the duration
278#        in seconds.  If more than on pass is desired, pass in
279#        a string, e.g., "100 100 100".
280# @param the pattern defining characters to test
281#
282# @return an array of results.  Each result is an array REF
283#         describing one pass.  The array REF contains:
284#         ->[0]: The method name as reported
285#         ->[1]: The params on the '= <meth> begin ...' line
286#         ->[2]: The params on the '= <meth> end ...' line
287#         ->[3..]: GC messages from the JVM, if any
288#
289sub callJava {
290    my $method = shift;
291    my $pat = shift;
292    my $n = shift;
293    my $passes = shift;
294
295    my $fileName = $SOURCEDIR.@$pat[0] ;
296    my $n = ($n < 0) ? "-t ".(-$n) : "-i ".$n;
297
298    my $cmd = "java -classpath classes $TESTCLASS $method $n -p $passes -f $fileName -e @$pat[1] -T @$pat[2]";
299    print "[$cmd]\n"; # for debugging
300    open(PIPE, "$cmd|") or die "Can't run \"$cmd\"";
301    my @out;
302    while (<PIPE>) {
303        push(@out, $_);
304    }
305    close(PIPE) or die "Java failed: \"$cmd\"";
306
307    @out = grep(!/^\#/, @out);  # filter out comments
308
309    #print "[", join("\n", @out), "]\n";
310
311    my @results;
312    my $method = '';
313    my $data = [];
314    foreach (@out) {
315        next unless (/\S/);
316
317        if (/^=\s*(\w+)\s*(\w+)\s*(.*)/) {
318            my ($m, $state, $d) = ($1, $2, $3);
319            #print "$_ => [[$m $state $data]]\n";
320            if ($state eq 'begin') {
321                die "$method was begun but not finished" if ($method);
322                $method = $m;
323                push(@$data, $d);
324                push(@$data, ''); # placeholder for end data
325            } elsif ($state eq 'end') {
326                if ($m ne $method) {
327                    die "$method end does not match: $_";
328                }
329                $data->[1] = $d; # insert end data at [1]
330                #print "#$method:", join(";",@$data), "\n";
331                unshift(@$data, $method); # add method to start
332
333                push(@results, $data);
334                $method = '';
335                $data = [];
336            } else {
337                die "Can't parse: $_";
338            }
339        }
340
341        elsif (/^\[/) {
342            if ($method) {
343                push(@$data, $_);
344            } else {
345                # ignore extraneous GC notices
346            }
347        }
348
349        else {
350            die "Can't parse: $_";
351        }
352    }
353
354    die "$method was begun but not finished" if ($method);
355
356    @results;
357}
358
359#|#---------------------------------------------------------------------
360#|# Format a confidence interval, as given by a Dataset.  Output is as
361#|# as follows:
362#|#   241.23 - 241.98 => 241.5 +/- 0.3
363#|#   241.2 - 243.8 => 242 +/- 1
364#|#   211.0 - 241.0 => 226 +/- 15 or? 230 +/- 20
365#|#   220.3 - 234.3 => 227 +/- 7
366#|#   220.3 - 300.3 => 260 +/- 40
367#|#   220.3 - 1000 => 610 +/- 390 or? 600 +/- 400
368#|#   0.022 - 0.024 => 0.023 +/- 0.001
369#|#   0.022 - 0.032 => 0.027 +/- 0.005
370#|#   0.022 - 1.000 => 0.5 +/- 0.5
371#|# In other words, take one significant digit of the error value and
372#|# display the mean to the same precision.
373#|sub formatDataset {
374#|    my $ds = shift;
375#|    my $lower = $ds->getMean() - $ds->getError();
376#|    my $upper = $ds->getMean() + $ds->getError();
377#|    my $scale = 0;
378#|    # Find how many initial digits are the same
379#|    while ($lower < 1 ||
380#|           int($lower) == int($upper)) {
381#|        $lower *= 10;
382#|        $upper *= 10;
383#|        $scale++;
384#|    }
385#|    while ($lower >= 10 &&
386#|           int($lower) == int($upper)) {
387#|        $lower /= 10;
388#|        $upper /= 10;
389#|        $scale--;
390#|    }
391#|}
392
393#---------------------------------------------------------------------
394# Format a number, optionally with a +/- delta, to n significant
395# digits.
396#
397# @param significant digit, a value >= 1
398# @param multiplier
399# @param time in seconds to be formatted
400# @optional delta in seconds
401#
402# @return string of the form "23" or "23 +/- 10".
403#
404sub formatNumber {
405    my $sigdig = shift;
406    my $mult = shift;
407    my $a = shift;
408    my $delta = shift; # may be undef
409
410    my $result = formatSigDig($sigdig, $a*$mult);
411    if (defined($delta)) {
412        my $d = formatSigDig($sigdig, $delta*$mult);
413        # restrict PRECISION of delta to that of main number
414        if ($result =~ /\.(\d+)/) {
415            # TODO make this work for values with all significant
416            # digits to the left of the decimal, e.g., 1234000.
417
418            # TODO the other thing wrong with this is that it
419            # isn't rounding the $delta properly.  Have to put
420            # this logic into formatSigDig().
421            my $x = length($1);
422            $d =~ s/\.(\d{$x})\d+/.$1/;
423        }
424        $result .= " $PLUS_MINUS " . $d;
425    }
426    $result;
427}
428
429#---------------------------------------------------------------------
430# Format a time, optionally with a +/- delta, to n significant
431# digits.
432#
433# @param significant digit, a value >= 1
434# @param time in seconds to be formatted
435# @optional delta in seconds
436#
437# @return string of the form "23 ms" or "23 +/- 10 ms".
438#
439sub formatSeconds {
440    my $sigdig = shift;
441    my $a = shift;
442    my $delta = shift; # may be undef
443
444    my @MULT = (1   , 1e3,  1e6,  1e9);
445    my @SUFF = ('s' , 'ms', 'us', 'ns');
446
447    # Determine our scale
448    my $i = 0;
449    ++$i while ($a*$MULT[$i] < 1 && $i < @MULT);
450
451    formatNumber($sigdig, $MULT[$i], $a, $delta) . ' ' . $SUFF[$i];
452}
453
454#---------------------------------------------------------------------
455# Format a percentage, optionally with a +/- delta, to n significant
456# digits.
457#
458# @param significant digit, a value >= 1
459# @param value to be formatted, as a fraction, e.g. 0.5 for 50%
460# @optional delta, as a fraction
461#
462# @return string of the form "23 %" or "23 +/- 10 %".
463#
464sub formatPercent {
465    my $sigdig = shift;
466    my $a = shift;
467    my $delta = shift; # may be undef
468
469    formatNumber($sigdig, 100, $a, $delta) . ' %';
470}
471
472#---------------------------------------------------------------------
473# Format a number to n significant digits without using exponential
474# notation.
475#
476# @param significant digit, a value >= 1
477# @param number to be formatted
478#
479# @return string of the form "1234" "12.34" or "0.001234".  If
480#         number was negative, prefixed by '-'.
481#
482sub formatSigDig {
483    my $n = shift() - 1;
484    my $a = shift;
485
486    local $_ = sprintf("%.${n}e", $a);
487    my $sign = (s/^-//) ? '-' : '';
488
489    my $a_e;
490    my $result;
491    if (/^(\d)\.(\d+)e([-+]\d+)$/) {
492        my ($d, $dn, $e) = ($1, $2, $3);
493        $a_e = $e;
494        $d .= $dn;
495        $e++;
496        $d .= '0' while ($e > length($d));
497        while ($e < 1) {
498            $e++;
499            $d = '0' . $d;
500        }
501        if ($e == length($d)) {
502            $result = $sign . $d;
503        } else {
504            $result = $sign . substr($d, 0, $e) . '.' . substr($d, $e);
505        }
506    } else {
507        die "Can't parse $_";
508    }
509    $result;
510}
511
512#eof
513