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");
17 # Use Perl code in strings...
19 'Name1' => '...code1...',
20 'Name2' => '...code2...',
23 # ... or use subroutine references.
25 'Name1' => sub { ...code1... },
26 'Name2' => sub { ...code2... },
29 $t = timeit($count, '...other code...')
30 print "$count loops of other code took:",timestr($t),"\n";
34 The Benchmark module encapsulates a number of routines to help you
35 figure out how long it takes to execute some code.
43 Returns the current time. Example:
47 # ... your code here ...
49 $td = timediff($t1, $t0);
50 print "the code took:",timestr($td),"\n";
54 Enables or disable debugging by setting the C<$Benchmark::Debug> flag:
57 $t = timeit(10, ' 5 ** $Global ');
62 =head2 Standard Exports
64 The following routines will be exported into your namespace
65 if you use the Benchmark module:
69 =item timeit(COUNT, CODE)
71 Arguments: COUNT is the number of times to run the loop, and CODE is
72 the code to run. CODE may be either a code reference or a string to
73 be eval'd; either way it will be run in the caller's package.
75 Returns: a Benchmark object.
77 =item timethis ( COUNT, CODE, [ TITLE, [ STYLE ]] )
79 Time COUNT iterations of CODE. CODE may be a string to eval or a
80 code reference; either way the CODE will run in the caller's package.
81 Results will be printed to STDOUT as TITLE followed by the times.
82 TITLE defaults to "timethis COUNT" if none is provided. STYLE
83 determines the format of the output, as described for timestr() below.
85 =item timethese ( COUNT, CODEHASHREF, [ STYLE ] )
87 The CODEHASHREF is a reference to a hash containing names as keys
88 and either a string to eval or a code reference for each value.
89 For each (KEY, VALUE) pair in the CODEHASHREF, this routine will
92 timethis(COUNT, VALUE, KEY, STYLE)
94 =item timediff ( T1, T2 )
96 Returns the difference between two Benchmark times as a Benchmark
97 object suitable for passing to timestr().
99 =item timestr ( TIMEDIFF, [ STYLE, [ FORMAT ]] )
101 Returns a string that formats the times in the TIMEDIFF object in
102 the requested STYLE. TIMEDIFF is expected to be a Benchmark object
103 similar to that returned by timediff().
105 STYLE can be any of 'all', 'noc', 'nop' or 'auto'. 'all' shows each
106 of the 5 times available ('wallclock' time, user time, system time,
107 user time of children, and system time of children). 'noc' shows all
108 except the two children times. 'nop' shows only wallclock and the
109 two children times. 'auto' (the default) will act as 'all' unless
110 the children times are both zero, in which case it acts as 'noc'.
112 FORMAT is the L<printf(3)>-style format specifier (without the
113 leading '%') to use to print the times. It defaults to '5.2f'.
117 =head2 Optional Exports
119 The following routines will be exported into your namespace
120 if you specifically ask that they be imported:
124 =item clearcache ( COUNT )
126 Clear the cached time for COUNT rounds of the null loop.
128 =item clearallcache ( )
130 Clear all cached times.
132 =item disablecache ( )
134 Disable caching of timings for the null loop. This will force Benchmark
135 to recalculate these timings for each new piece of code timed.
137 =item enablecache ( )
139 Enable caching of timings for the null loop. The time taken for COUNT
140 rounds of the null loop will be calculated only once for each
141 different COUNT used.
147 The data is stored as a list of values from the time and times
150 ($real, $user, $system, $children_user, $children_system)
152 in seconds for the whole loop (not divided by the number of rounds).
154 The timing is done using time(3) and times(3).
156 Code is executed in the caller's package.
158 The time of the null loop (a loop with the same
159 number of rounds but empty loop body) is subtracted
160 from the time of the real loop.
162 The null loop times are cached, the key being the
163 number of rounds. The caching can be controlled using
174 Benchmark inherits from no other class, except of course
179 The real time timing is done using time(2) and
180 the granularity is therefore only one second.
182 Short tests may produce negative figures because perl
183 can appear to take longer to execute the empty loop
184 than a short test; try:
188 The system time of the null loop might be slightly
189 more than the system time of the loop with the actual
190 code and therefore the difference might end up being E<lt> 0.
194 Jarkko Hietaniemi E<lt>F<Jarkko.Hietaniemi@hut.fi>E<gt>,
195 Tim Bunce E<lt>F<Tim.Bunce@ig.co.uk>E<gt>
197 =head1 MODIFICATION HISTORY
199 September 8th, 1994; by Tim Bunce.
201 March 28th, 1997; by Hugo van der Sanden: added support for code
202 references and the already documented 'debug' method; revamped
210 @EXPORT=qw(timeit timethis timethese timediff timestr);
211 @EXPORT_OK=qw(clearcache clearallcache disablecache enablecache);
219 $defaultfmt = '5.2f';
220 $defaultstyle = 'auto';
221 # The cache can cause a slight loss of sys time accuracy. If a
222 # user does many tests (>10) with *very* large counts (>10000)
223 # or works on a very slow machine the cache may be useful.
228 sub debug { $debug = ($_[1] != 0); }
230 sub clearcache { delete $cache{$_[0]}; }
231 sub clearallcache { %cache = (); }
232 sub enablecache { $cache = 1; }
233 sub disablecache { $cache = 0; }
235 # --- Functions to process the 'time' data type
237 sub new { my @t = (time, times); print "new=@t\n" if $debug; bless \@t; }
239 sub cpu_p { my($r,$pu,$ps,$cu,$cs) = @{$_[0]}; $pu+$ps ; }
240 sub cpu_c { my($r,$pu,$ps,$cu,$cs) = @{$_[0]}; $cu+$cs ; }
241 sub cpu_a { my($r,$pu,$ps,$cu,$cs) = @{$_[0]}; $pu+$ps+$cu+$cs ; }
242 sub real { my($r,$pu,$ps,$cu,$cs) = @{$_[0]}; $r ; }
247 for ($i=0; $i < @$a; ++$i) {
248 push(@r, $a->[$i] - $b->[$i]);
254 my($tr, $style, $f) = @_;
256 warn "bad time value" unless @t==5;
257 my($r, $pu, $ps, $cu, $cs) = @t;
258 my($pt, $ct, $t) = ($tr->cpu_p, $tr->cpu_c, $tr->cpu_a);
259 $f = $defaultfmt unless defined $f;
260 # format a time in the required style, other formats may be added here
261 $style = $defaultstyle unless defined $style;
262 $style = ($ct>0) ? 'all' : 'noc' if $style eq 'auto';
263 my $s = "@t $style"; # default for unknown style
264 $s=sprintf("%2d secs (%$f usr %$f sys + %$f cusr %$f csys = %$f cpu)",
265 @t,$t) if $style eq 'all';
266 $s=sprintf("%2d secs (%$f usr %$f sys = %$f cpu)",
267 $r,$pu,$ps,$pt) if $style eq 'noc';
268 $s=sprintf("%2d secs (%$f cusr %$f csys = %$f cpu)",
269 $r,$cu,$cs,$ct) if $style eq 'nop';
275 print STDERR "$msg",timestr($t),"\n" if $debug;
278 # --- Functions implementing low-level support for timing loops
283 $n+=0; # force numeric now, so garbage won't creep into the eval
284 croak "negative loopcount $n" if $n<0;
285 confess "Usage: runloop(number, [string | coderef])" unless defined $c;
286 my($t0, $t1, $td); # before, after, difference
288 # find package of caller so we can execute code there
289 my($curpack) = caller(0);
291 while (($pack) = caller(++$i)) {
292 last if $pack ne $curpack;
295 my $subcode = (ref $c eq 'CODE')
296 ? "sub { package $pack; my(\$_i)=$n; while (\$_i--){&\$c;} }"
297 : "sub { package $pack; my(\$_i)=$n; while (\$_i--){$c;} }";
298 my $subref = eval $subcode;
299 croak "runloop unable to compile '$c': $@\ncode: $subcode\n" if $@;
300 print STDERR "runloop $n '$subcode'\n" if $debug;
305 $td = &timediff($t1, $t0);
307 timedebug("runloop:",$td);
316 printf STDERR "timeit $n $code\n" if $debug;
318 if ($cache && exists $cache{$n}) {
321 $wn = &runloop($n, '');
325 $wc = &runloop($n, $code);
327 $wd = timediff($wc, $wn);
329 timedebug("timeit: ",$wc);
330 timedebug(" - ",$wn);
331 timedebug(" = ",$wd);
336 # --- Functions implementing high-level time-then-print utilities
339 my($n, $code, $title, $style) = @_;
340 my $t = timeit($n, $code);
342 $title = "timethis $n" unless defined $title;
343 $style = "" unless defined $style;
344 printf("%10s: ", $title);
345 print timestr($t, $style),"\n";
347 # A conservative warning to spot very silly tests.
348 # Don't assume that your benchmark is ok simply because
349 # you don't get this warning!
350 print " (warning: too few iterations for a reliable count)\n"
352 || ($t->real < 1 && $n < 1000)
353 || $t->cpu_a < $min_cpu;
358 my($n, $alt, $style) = @_;
359 die "usage: timethese(count, { 'Name1'=>'code1', ... }\n"
360 unless ref $alt eq HASH;
361 my @names = sort keys %$alt;
362 $style = "" unless defined $style;
363 print "Benchmark: timing $n iterations of ",join(', ',@names),"...\n";
365 # we could save the results in an array and produce a summary here
366 # sum, min, max, avg etc etc
367 map timethis($n, $alt->{$_}, $_, $style), @names;