5 Benchmark - benchmark running times of code
7 timethis - run a chunk of code several times
9 timethese - run several chunks of code several times
11 timeit - run a chunk of code and see how long it goes
15 timethis ($count, "code");
18 'Name1' => '...code1...',
19 'Name2' => sub { ...code2... },
22 $t = timeit($count, '...other code...')
23 print "$count loops of other code took:",timestr($t),"\n";
27 The Benchmark module encapsulates a number of routines to help you
28 figure out how long it takes to execute some code.
36 Returns the current time. Example:
40 # ... your code here ...
42 $td = timediff($t1, $t0);
43 print "the code took:",timestr($td),"\n";
47 Enables or disable debugging by setting the C<$Benchmark::Debug> flag:
50 $t = timeit(10, ' 5 ** $Global ');
55 =head2 Standard Exports
57 The following routines will be exported into your namespace
58 if you use the Benchmark module:
62 =item timeit(COUNT, CODE)
64 Arguments: COUNT is the number of times to run the loop, and CODE is
65 the code to run. CODE may be either a code reference or a string to
66 be eval'd; either way it will be run in the caller's package.
68 Returns: a Benchmark object.
70 =item timethis ( COUNT, CODE, [ TITLE, [ STYLE ]] )
72 Time COUNT iterations of CODE. CODE may be a string to eval or a
73 code reference; either way the CODE will run in the caller's package.
74 Results will be printed to STDOUT as TITLE followed by the times.
75 TITLE defaults to "timethis COUNT" if none is provided. STYLE
76 determines the format of the output, as described for timestr() below.
78 =item timethese ( COUNT, CODEHASHREF, [ STYLE ] )
80 The CODEHASHREF is a reference to a hash containing names as keys
81 and either a string to eval or a code reference for each value.
82 For each (KEY, VALUE) pair in the CODEHASHREF, this routine will
85 timethis(COUNT, VALUE, KEY, STYLE)
87 =item timediff ( T1, T2 )
89 Returns the difference between two Benchmark times as a Benchmark
90 object suitable for passing to timestr().
92 =item timestr ( TIMEDIFF, [ STYLE, [ FORMAT ]] )
94 Returns a string that formats the times in the TIMEDIFF object in
95 the requested STYLE. TIMEDIFF is expected to be a Benchmark object
96 similar to that returned by timediff().
98 STYLE can be any of 'all', 'noc', 'nop' or 'auto'. 'all' shows each
99 of the 5 times available ('wallclock' time, user time, system time,
100 user time of children, and system time of children). 'noc' shows all
101 except the two children times. 'nop' shows only wallclock and the
102 two children times. 'auto' (the default) will act as 'all' unless
103 the children times are both zero, in which case it acts as 'noc'.
105 FORMAT is the L<printf(3)>-style format specifier (without the
106 leading '%') to use to print the times. It defaults to '5.2f'.
110 =head2 Optional Exports
112 The following routines will be exported into your namespace
113 if you specifically ask that they be imported:
117 =item clearcache ( COUNT )
119 Clear the cached time for COUNT rounds of the null loop.
121 =item clearallcache ( )
123 Clear all cached times.
125 =item disablecache ( )
127 Disable caching of timings for the null loop. This will force Benchmark
128 to recalculate these timings for each new piece of code timed.
130 =item enablecache ( )
132 Enable caching of timings for the null loop. The time taken for COUNT
133 rounds of the null loop will be calculated only once for each
134 different COUNT used.
140 The data is stored as a list of values from the time and times
143 ($real, $user, $system, $children_user, $children_system)
145 in seconds for the whole loop (not divided by the number of rounds).
147 The timing is done using time(3) and times(3).
149 Code is executed in the caller's package.
151 The time of the null loop (a loop with the same
152 number of rounds but empty loop body) is subtracted
153 from the time of the real loop.
155 The null loop times are cached, the key being the
156 number of rounds. The caching can be controlled using
167 Benchmark inherits from no other class, except of course
172 The real time timing is done using time(2) and
173 the granularity is therefore only one second.
175 Short tests may produce negative figures because perl
176 can appear to take longer to execute the empty loop
177 than a short test; try:
181 The system time of the null loop might be slightly
182 more than the system time of the loop with the actual
183 code and therefore the difference might end up being E<lt> 0.
187 Jarkko Hietaniemi E<lt>F<Jarkko.Hietaniemi@hut.fi>E<gt>,
188 Tim Bunce E<lt>F<Tim.Bunce@ig.co.uk>E<gt>
190 =head1 MODIFICATION HISTORY
192 September 8th, 1994; by Tim Bunce.
194 March 28th, 1997; by Hugo van der Sanden: added support for code
195 references and the already documented 'debug' method; revamped
203 @EXPORT=qw(timeit timethis timethese timediff timestr);
204 @EXPORT_OK=qw(clearcache clearallcache disablecache enablecache);
212 $defaultfmt = '5.2f';
213 $defaultstyle = 'auto';
214 # The cache can cause a slight loss of sys time accuracy. If a
215 # user does many tests (>10) with *very* large counts (>10000)
216 # or works on a very slow machine the cache may be useful.
221 sub debug { $debug = ($_[1] != 0); }
223 sub clearcache { delete $cache{$_[0]}; }
224 sub clearallcache { %cache = (); }
225 sub enablecache { $cache = 1; }
226 sub disablecache { $cache = 0; }
228 # --- Functions to process the 'time' data type
230 sub new { my @t = (time, times); print "new=@t\n" if $debug; bless \@t; }
232 sub cpu_p { my($r,$pu,$ps,$cu,$cs) = @{$_[0]}; $pu+$ps ; }
233 sub cpu_c { my($r,$pu,$ps,$cu,$cs) = @{$_[0]}; $cu+$cs ; }
234 sub cpu_a { my($r,$pu,$ps,$cu,$cs) = @{$_[0]}; $pu+$ps+$cu+$cs ; }
235 sub real { my($r,$pu,$ps,$cu,$cs) = @{$_[0]}; $r ; }
240 for ($i=0; $i < @$a; ++$i) {
241 push(@r, $a->[$i] - $b->[$i]);
247 my($tr, $style, $f) = @_;
249 warn "bad time value" unless @t==5;
250 my($r, $pu, $ps, $cu, $cs) = @t;
251 my($pt, $ct, $t) = ($tr->cpu_p, $tr->cpu_c, $tr->cpu_a);
252 $f = $defaultfmt unless defined $f;
253 # format a time in the required style, other formats may be added here
254 $style = $defaultstyle unless defined $style;
255 $style = ($ct>0) ? 'all' : 'noc' if $style eq 'auto';
256 my $s = "@t $style"; # default for unknown style
257 $s=sprintf("%2d secs (%$f usr %$f sys + %$f cusr %$f csys = %$f cpu)",
258 @t,$t) if $style eq 'all';
259 $s=sprintf("%2d secs (%$f usr %$f sys = %$f cpu)",
260 $r,$pu,$ps,$pt) if $style eq 'noc';
261 $s=sprintf("%2d secs (%$f cusr %$f csys = %$f cpu)",
262 $r,$cu,$cs,$ct) if $style eq 'nop';
268 print STDERR "$msg",timestr($t),"\n" if $debug;
271 # --- Functions implementing low-level support for timing loops
276 $n+=0; # force numeric now, so garbage won't creep into the eval
277 croak "negative loopcount $n" if $n<0;
278 confess "Usage: runloop(number, [string | coderef])" unless defined $c;
279 my($t0, $t1, $td); # before, after, difference
281 # find package of caller so we can execute code there
282 my($curpack) = caller(0);
284 while (($pack) = caller(++$i)) {
285 last if $pack ne $curpack;
288 my $subcode = (ref $c eq 'CODE')
289 ? "sub { package $pack; my(\$_i)=$n; while (\$_i--){&\$c;} }"
290 : "sub { package $pack; my(\$_i)=$n; while (\$_i--){$c;} }";
291 my $subref = eval $subcode;
292 croak "runloop unable to compile '$c': $@\ncode: $subcode\n" if $@;
293 print STDERR "runloop $n '$subcode'\n" if $debug;
298 $td = &timediff($t1, $t0);
300 timedebug("runloop:",$td);
309 printf STDERR "timeit $n $code\n" if $debug;
311 if ($cache && exists $cache{$n}) {
314 $wn = &runloop($n, '');
318 $wc = &runloop($n, $code);
320 $wd = timediff($wc, $wn);
322 timedebug("timeit: ",$wc);
323 timedebug(" - ",$wn);
324 timedebug(" = ",$wd);
329 # --- Functions implementing high-level time-then-print utilities
332 my($n, $code, $title, $style) = @_;
333 my $t = timeit($n, $code);
335 $title = "timethis $n" unless defined $title;
336 $style = "" unless defined $style;
337 printf("%10s: ", $title);
338 print timestr($t, $style),"\n";
340 # A conservative warning to spot very silly tests.
341 # Don't assume that your benchmark is ok simply because
342 # you don't get this warning!
343 print " (warning: too few iterations for a reliable count)\n"
345 || ($t->real < 1 && $n < 1000)
346 || $t->cpu_a < $min_cpu;
351 my($n, $alt, $style) = @_;
352 die "usage: timethese(count, { 'Name1'=>'code1', ... }\n"
353 unless ref $alt eq HASH;
354 my @names = sort keys %$alt;
355 $style = "" unless defined $style;
356 print "Benchmark: timing $n iterations of ",join(', ',@names),"...\n";
358 # we could save the results in an array and produce a summary here
359 # sum, min, max, avg etc etc
360 map timethis($n, $alt->{$_}, $_, $style), @names;