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 ** $Global ');
62 =head2 Standard Exports
64 The following routines will be exported into your namespace
65 if you use the Benchmark module:
69 =item timeit(COUNT, CODE)
71 Arguments: COUNT is the number of times to run the loop, and CODE is
72 the code to run. CODE may be either a code reference or a string to
73 be eval'd; either way it will be run in the caller's package.
75 Returns: a Benchmark object.
77 =item timethis ( COUNT, CODE, [ TITLE, [ STYLE ]] )
79 Time COUNT iterations of CODE. CODE may be a string to eval or a
80 code reference; either way the CODE will run in the caller's package.
81 Results will be printed to STDOUT as TITLE followed by the times.
82 TITLE defaults to "timethis COUNT" if none is provided. STYLE
83 determines the format of the output, as described for timestr() below.
85 The COUNT can be zero or negative: this means the I<minimum number of
86 CPU seconds> to run. A zero signifies the default of 3 seconds. For
87 example to run at least for 10 seconds:
91 or to run two pieces of code tests for at least 3 seconds:
93 timethese(0, { test1 => '...', test2 => '...'})
95 CPU seconds is, in UNIX terms, the user time plus the system time of
96 the process itself, as opposed to the real (wallclock) time and the
97 time spent by the child processes. Less than 0.1 seconds is not
98 accepted (-0.01 as the count, for example, will cause a fatal runtime
101 Note that the CPU seconds is the B<minimum> time: CPU scheduling and
102 other operating system factors may complicate the attempt so that a
103 little bit more time is spent. The benchmark output will, however,
104 also tell the number of C<$code> runs/second, which should be a more
105 interesting number than the actually spent seconds.
107 Returns a Benchmark object.
109 =item timethese ( COUNT, CODEHASHREF, [ STYLE ] )
111 The CODEHASHREF is a reference to a hash containing names as keys
112 and either a string to eval or a code reference for each value.
113 For each (KEY, VALUE) pair in the CODEHASHREF, this routine will
116 timethis(COUNT, VALUE, KEY, STYLE)
118 The Count can be zero or negative, see timethis().
120 =item timediff ( T1, T2 )
122 Returns the difference between two Benchmark times as a Benchmark
123 object suitable for passing to timestr().
125 =item timestr ( TIMEDIFF, [ STYLE, [ FORMAT ] ] )
127 Returns a string that formats the times in the TIMEDIFF object in
128 the requested STYLE. TIMEDIFF is expected to be a Benchmark object
129 similar to that returned by timediff().
131 STYLE can be any of 'all', 'noc', 'nop' or 'auto'. 'all' shows each
132 of the 5 times available ('wallclock' time, user time, system time,
133 user time of children, and system time of children). 'noc' shows all
134 except the two children times. 'nop' shows only wallclock and the
135 two children times. 'auto' (the default) will act as 'all' unless
136 the children times are both zero, in which case it acts as 'noc'.
138 FORMAT is the L<printf(3)>-style format specifier (without the
139 leading '%') to use to print the times. It defaults to '5.2f'.
143 =head2 Optional Exports
145 The following routines will be exported into your namespace
146 if you specifically ask that they be imported:
150 =item clearcache ( COUNT )
152 Clear the cached time for COUNT rounds of the null loop.
154 =item clearallcache ( )
156 Clear all cached times.
158 =item disablecache ( )
160 Disable caching of timings for the null loop. This will force Benchmark
161 to recalculate these timings for each new piece of code timed.
163 =item enablecache ( )
165 Enable caching of timings for the null loop. The time taken for COUNT
166 rounds of the null loop will be calculated only once for each
167 different COUNT used.
173 The data is stored as a list of values from the time and times
176 ($real, $user, $system, $children_user, $children_system)
178 in seconds for the whole loop (not divided by the number of rounds).
180 The timing is done using time(3) and times(3).
182 Code is executed in the caller's package.
184 The time of the null loop (a loop with the same
185 number of rounds but empty loop body) is subtracted
186 from the time of the real loop.
188 The null loop times are cached, the key being the
189 number of rounds. The caching can be controlled using
200 Benchmark inherits from no other class, except of course
205 Comparing eval'd strings with code references will give you
206 inaccurate results: a code reference will show a slower
207 execution time than the equivalent eval'd string.
209 The real time timing is done using time(2) and
210 the granularity is therefore only one second.
212 Short tests may produce negative figures because perl
213 can appear to take longer to execute the empty loop
214 than a short test; try:
218 The system time of the null loop might be slightly
219 more than the system time of the loop with the actual
220 code and therefore the difference might end up being E<lt> 0.
224 Jarkko Hietaniemi <F<jhi@iki.fi>>, Tim Bunce <F<Tim.Bunce@ig.co.uk>>
226 =head1 MODIFICATION HISTORY
228 September 8th, 1994; by Tim Bunce.
230 March 28th, 1997; by Hugo van der Sanden: added support for code
231 references and the already documented 'debug' method; revamped
234 April 04-07th, 1997: by Jarkko Hietaniemi, added the run-for-some-time
242 @EXPORT=qw(timeit timethis timethese timediff timestr);
243 @EXPORT_OK=qw(clearcache clearallcache disablecache enablecache);
251 $defaultfmt = '5.2f';
252 $defaultstyle = 'auto';
253 # The cache can cause a slight loss of sys time accuracy. If a
254 # user does many tests (>10) with *very* large counts (>10000)
255 # or works on a very slow machine the cache may be useful.
260 sub debug { $debug = ($_[1] != 0); }
262 sub clearcache { delete $cache{$_[0]}; }
263 sub clearallcache { %cache = (); }
264 sub enablecache { $cache = 1; }
265 sub disablecache { $cache = 0; }
267 # --- Functions to process the 'time' data type
269 sub new { my @t = (time, times, @_ == 2 ? $_[1] : 0);
270 print "new=@t\n" if $debug;
273 sub cpu_p { my($r,$pu,$ps,$cu,$cs) = @{$_[0]}; $pu+$ps ; }
274 sub cpu_c { my($r,$pu,$ps,$cu,$cs) = @{$_[0]}; $cu+$cs ; }
275 sub cpu_a { my($r,$pu,$ps,$cu,$cs) = @{$_[0]}; $pu+$ps+$cu+$cs ; }
276 sub real { my($r,$pu,$ps,$cu,$cs) = @{$_[0]}; $r ; }
281 for ($i=0; $i < @$a; ++$i) {
282 push(@r, $a->[$i] - $b->[$i]);
288 my($tr, $style, $f) = @_;
290 warn "bad time value (@t)" unless @t==6;
291 my($r, $pu, $ps, $cu, $cs, $n) = @t;
292 my($pt, $ct, $t) = ($tr->cpu_p, $tr->cpu_c, $tr->cpu_a);
293 $f = $defaultfmt unless defined $f;
294 # format a time in the required style, other formats may be added here
295 $style ||= $defaultstyle;
296 $style = ($ct>0) ? 'all' : 'noc' if $style eq 'auto';
297 my $s = "@t $style"; # default for unknown style
298 $s=sprintf("%2d wallclock secs (%$f usr %$f sys + %$f cusr %$f csys = %$f CPU secs)",
299 @t,$t) if $style eq 'all';
300 $s=sprintf("%$f CPU secs (%$f usr + %$f sys)",
301 $pt,$pu,$ps) if $style eq 'noc';
302 $s=sprintf("%$f CPU secs (%$f cusr %$f csys)",
303 $ct,$cu,$cs) if $style eq 'nop';
304 $s .= sprintf(" @ %$f/s (n=$n)", $n / ( $pu + $ps )) if $n;
310 print STDERR "$msg",timestr($t),"\n" if $debug;
313 # --- Functions implementing low-level support for timing loops
318 $n+=0; # force numeric now, so garbage won't creep into the eval
319 croak "negative loopcount $n" if $n<0;
320 confess "Usage: runloop(number, [string | coderef])" unless defined $c;
321 my($t0, $t1, $td); # before, after, difference
323 # find package of caller so we can execute code there
324 my($curpack) = caller(0);
326 while (($pack) = caller(++$i)) {
327 last if $pack ne $curpack;
330 my $subcode = (ref $c eq 'CODE')
331 ? "sub { package $pack; my(\$_i)=$n; while (\$_i--){&\$c;} }"
332 : "sub { package $pack; my(\$_i)=$n; while (\$_i--){$c;} }";
333 my $subref = eval $subcode;
334 croak "runloop unable to compile '$c': $@\ncode: $subcode\n" if $@;
335 print STDERR "runloop $n '$subcode'\n" if $debug;
337 $t0 = Benchmark->new(0);
339 $t1 = Benchmark->new($n);
340 $td = &timediff($t1, $t0);
342 timedebug("runloop:",$td);
351 printf STDERR "timeit $n $code\n" if $debug;
353 if ($cache && exists $cache{$n}) {
356 $wn = &runloop($n, '');
360 $wc = &runloop($n, $code);
362 $wd = timediff($wc, $wn);
364 timedebug("timeit: ",$wc);
365 timedebug(" - ",$wn);
366 timedebug(" = ",$wd);
376 my ($code, $tmax) = @_;
378 if ( not defined $tmax or $tmax == 0 ) {
379 $tmax = $default_for;
380 } elsif ( $tmax < 0 ) {
384 die "runfor(..., $tmax): timelimit cannot be less than $min_for.\n"
387 my ($n, $td, $tc, $ntot, $rtot, $utot, $stot, $cutot, $cstot );
389 # First find the minimum $n that gives a non-zero timing.
393 for ($n = 1, $tc = 0; $tc <= 0; $n *= 2 ) {
394 $td = timeit($n, $code);
395 $tc = $td->[1] + $td->[2];
401 my $tpra = 0.05 * $tmax; # Target/time practice.
403 # Double $n until we have think we have practiced enough.
404 for ( $n = 1; $ttot < $tpra; $n *= 2 ) {
405 $td = timeit($n, $code);
411 $ttot = $utot + $stot;
418 # Then iterate towards the $tmax.
419 while ( $ttot < $tmax ) {
420 $r = $tmax / $ttot - 1; # Linear approximation.
422 $n = $nmin if $n < $nmin;
423 $td = timeit($n, $code);
428 $ttot = $utot + $stot;
433 return bless [ $rtot, $utot, $stot, $cutot, $cstot, $ntot ];
436 # --- Functions implementing high-level time-then-print utilities
440 return $n == 0 ? $default_for : $n < 0 ? -$n : undef;
444 my($n, $code, $title, $style) = @_;
448 croak "non-integer loopcount $n, stopped" if int($n)<$n;
449 $t = timeit($n, $code);
450 $title = "timethis $n" unless defined $title;
452 $fort = n_to_for( $n );
453 $t = runfor($code, $fort);
454 $title = "timethis for $fort" unless defined $title;
458 $style = "" unless defined $style;
459 printf("%10s: ", $title);
460 print timestr($t, $style, $defaultfmt),"\n";
462 $n = $forn if defined $forn;
464 # A conservative warning to spot very silly tests.
465 # Don't assume that your benchmark is ok simply because
466 # you don't get this warning!
467 print " (warning: too few iterations for a reliable count)\n"
469 || ($t->real < 1 && $n < 1000)
470 || $t->cpu_a < $min_cpu;
475 my($n, $alt, $style) = @_;
476 die "usage: timethese(count, { 'Name1'=>'code1', ... }\n"
477 unless ref $alt eq HASH;
478 my @names = sort keys %$alt;
479 $style = "" unless defined $style;
482 croak "non-integer loopcount $n, stopped" if int($n)<$n;
483 print "timing $n iterations of";
487 print " ", join(', ',@names);
489 my $for = n_to_for( $n );
490 print ", each for at least $for CPU seconds";
494 # we could save the results in an array and produce a summary here
495 # sum, min, max, avg etc etc
496 foreach my $name (@names) {
497 timethis ($n, $alt -> {$name}, $name, $style);