Commit | Line | Data |
---|---|---|
a0d0e21e LW |
1 | package Benchmark; |
2 | ||
f06db76b AD |
3 | =head1 NAME |
4 | ||
8a4f6ac2 | 5 | Benchmark - benchmark running times of Perl code |
431d98c2 | 6 | |
f06db76b AD |
7 | =head1 SYNOPSIS |
8 | ||
f36484b0 BS |
9 | use Benchmark qw(:all) ; |
10 | ||
f06db76b AD |
11 | timethis ($count, "code"); |
12 | ||
523cc92b | 13 | # Use Perl code in strings... |
f06db76b AD |
14 | timethese($count, { |
15 | 'Name1' => '...code1...', | |
16 | 'Name2' => '...code2...', | |
17 | }); | |
18 | ||
523cc92b CS |
19 | # ... or use subroutine references. |
20 | timethese($count, { | |
21 | 'Name1' => sub { ...code1... }, | |
22 | 'Name2' => sub { ...code2... }, | |
23 | }); | |
24 | ||
431d98c2 BS |
25 | # cmpthese can be used both ways as well |
26 | cmpthese($count, { | |
27 | 'Name1' => '...code1...', | |
28 | 'Name2' => '...code2...', | |
29 | }); | |
30 | ||
31 | cmpthese($count, { | |
32 | 'Name1' => sub { ...code1... }, | |
33 | 'Name2' => sub { ...code2... }, | |
34 | }); | |
35 | ||
36 | # ...or in two stages | |
37 | $results = timethese($count, | |
38 | { | |
39 | 'Name1' => sub { ...code1... }, | |
40 | 'Name2' => sub { ...code2... }, | |
41 | }, | |
42 | 'none' | |
43 | ); | |
44 | cmpthese( $results ) ; | |
45 | ||
f06db76b AD |
46 | $t = timeit($count, '...other code...') |
47 | print "$count loops of other code took:",timestr($t),"\n"; | |
48 | ||
431d98c2 BS |
49 | $t = countit($time, '...other code...') |
50 | $count = $t->iters ; | |
51 | print "$count loops of other code took:",timestr($t),"\n"; | |
52 | ||
e3d6de9a JH |
53 | # enable hires wallclock timing if possible |
54 | use Benchmark ':hireswallclock'; | |
55 | ||
f06db76b AD |
56 | =head1 DESCRIPTION |
57 | ||
58 | The Benchmark module encapsulates a number of routines to help you | |
59 | figure out how long it takes to execute some code. | |
60 | ||
8a4f6ac2 GS |
61 | timethis - run a chunk of code several times |
62 | ||
63 | timethese - run several chunks of code several times | |
64 | ||
65 | cmpthese - print results of timethese as a comparison chart | |
66 | ||
67 | timeit - run a chunk of code and see how long it goes | |
68 | ||
69 | countit - see how many times a chunk of code runs in a given time | |
70 | ||
71 | ||
f06db76b AD |
72 | =head2 Methods |
73 | ||
74 | =over 10 | |
75 | ||
76 | =item new | |
77 | ||
78 | Returns the current time. Example: | |
79 | ||
80 | use Benchmark; | |
81 | $t0 = new Benchmark; | |
82 | # ... your code here ... | |
83 | $t1 = new Benchmark; | |
84 | $td = timediff($t1, $t0); | |
a24a9dfe | 85 | print "the code took:",timestr($td),"\n"; |
f06db76b AD |
86 | |
87 | =item debug | |
88 | ||
89 | Enables or disable debugging by setting the C<$Benchmark::Debug> flag: | |
90 | ||
523cc92b | 91 | debug Benchmark 1; |
f06db76b | 92 | $t = timeit(10, ' 5 ** $Global '); |
523cc92b | 93 | debug Benchmark 0; |
f06db76b | 94 | |
431d98c2 BS |
95 | =item iters |
96 | ||
97 | Returns the number of iterations. | |
98 | ||
f06db76b AD |
99 | =back |
100 | ||
101 | =head2 Standard Exports | |
102 | ||
523cc92b | 103 | The following routines will be exported into your namespace |
f06db76b AD |
104 | if you use the Benchmark module: |
105 | ||
106 | =over 10 | |
107 | ||
108 | =item timeit(COUNT, CODE) | |
109 | ||
523cc92b CS |
110 | Arguments: COUNT is the number of times to run the loop, and CODE is |
111 | the code to run. CODE may be either a code reference or a string to | |
112 | be eval'd; either way it will be run in the caller's package. | |
113 | ||
114 | Returns: a Benchmark object. | |
115 | ||
116 | =item timethis ( COUNT, CODE, [ TITLE, [ STYLE ]] ) | |
117 | ||
118 | Time COUNT iterations of CODE. CODE may be a string to eval or a | |
119 | code reference; either way the CODE will run in the caller's package. | |
120 | Results will be printed to STDOUT as TITLE followed by the times. | |
121 | TITLE defaults to "timethis COUNT" if none is provided. STYLE | |
122 | determines the format of the output, as described for timestr() below. | |
123 | ||
6ee623d5 GS |
124 | The COUNT can be zero or negative: this means the I<minimum number of |
125 | CPU seconds> to run. A zero signifies the default of 3 seconds. For | |
126 | example to run at least for 10 seconds: | |
127 | ||
128 | timethis(-10, $code) | |
129 | ||
130 | or to run two pieces of code tests for at least 3 seconds: | |
131 | ||
132 | timethese(0, { test1 => '...', test2 => '...'}) | |
133 | ||
134 | CPU seconds is, in UNIX terms, the user time plus the system time of | |
135 | the process itself, as opposed to the real (wallclock) time and the | |
136 | time spent by the child processes. Less than 0.1 seconds is not | |
137 | accepted (-0.01 as the count, for example, will cause a fatal runtime | |
138 | exception). | |
139 | ||
140 | Note that the CPU seconds is the B<minimum> time: CPU scheduling and | |
141 | other operating system factors may complicate the attempt so that a | |
142 | little bit more time is spent. The benchmark output will, however, | |
143 | also tell the number of C<$code> runs/second, which should be a more | |
144 | interesting number than the actually spent seconds. | |
145 | ||
146 | Returns a Benchmark object. | |
147 | ||
523cc92b | 148 | =item timethese ( COUNT, CODEHASHREF, [ STYLE ] ) |
f06db76b | 149 | |
523cc92b CS |
150 | The CODEHASHREF is a reference to a hash containing names as keys |
151 | and either a string to eval or a code reference for each value. | |
152 | For each (KEY, VALUE) pair in the CODEHASHREF, this routine will | |
153 | call | |
f06db76b | 154 | |
523cc92b | 155 | timethis(COUNT, VALUE, KEY, STYLE) |
f06db76b | 156 | |
1d2dff63 GS |
157 | The routines are called in string comparison order of KEY. |
158 | ||
159 | The COUNT can be zero or negative, see timethis(). | |
6ee623d5 | 160 | |
3c6312e9 BS |
161 | Returns a hash of Benchmark objects, keyed by name. |
162 | ||
523cc92b | 163 | =item timediff ( T1, T2 ) |
f06db76b | 164 | |
523cc92b CS |
165 | Returns the difference between two Benchmark times as a Benchmark |
166 | object suitable for passing to timestr(). | |
f06db76b | 167 | |
6ee623d5 | 168 | =item timestr ( TIMEDIFF, [ STYLE, [ FORMAT ] ] ) |
f06db76b | 169 | |
523cc92b CS |
170 | Returns a string that formats the times in the TIMEDIFF object in |
171 | the requested STYLE. TIMEDIFF is expected to be a Benchmark object | |
172 | similar to that returned by timediff(). | |
173 | ||
3c6312e9 BS |
174 | STYLE can be any of 'all', 'none', 'noc', 'nop' or 'auto'. 'all' shows |
175 | each of the 5 times available ('wallclock' time, user time, system time, | |
523cc92b CS |
176 | user time of children, and system time of children). 'noc' shows all |
177 | except the two children times. 'nop' shows only wallclock and the | |
178 | two children times. 'auto' (the default) will act as 'all' unless | |
179 | the children times are both zero, in which case it acts as 'noc'. | |
3c6312e9 | 180 | 'none' prevents output. |
523cc92b CS |
181 | |
182 | FORMAT is the L<printf(3)>-style format specifier (without the | |
183 | leading '%') to use to print the times. It defaults to '5.2f'. | |
f06db76b AD |
184 | |
185 | =back | |
186 | ||
187 | =head2 Optional Exports | |
188 | ||
189 | The following routines will be exported into your namespace | |
190 | if you specifically ask that they be imported: | |
191 | ||
192 | =over 10 | |
193 | ||
523cc92b CS |
194 | =item clearcache ( COUNT ) |
195 | ||
196 | Clear the cached time for COUNT rounds of the null loop. | |
197 | ||
198 | =item clearallcache ( ) | |
f06db76b | 199 | |
523cc92b | 200 | Clear all cached times. |
f06db76b | 201 | |
8962dfd6 | 202 | =item cmpthese ( COUNT, CODEHASHREF, [ STYLE ] ) |
ac8eabc1 | 203 | |
d1083c7a | 204 | =item cmpthese ( RESULTSHASHREF, [ STYLE ] ) |
ac8eabc1 | 205 | |
d1083c7a | 206 | Optionally calls timethese(), then outputs comparison chart. This: |
ac8eabc1 | 207 | |
d1083c7a BS |
208 | cmpthese( -1, { a => "++\$i", b => "\$i *= 2" } ) ; |
209 | ||
210 | outputs a chart like: | |
211 | ||
212 | Rate b a | |
213 | b 2831802/s -- -61% | |
214 | a 7208959/s 155% -- | |
215 | ||
216 | This chart is sorted from slowest to fastest, and shows the percent speed | |
217 | difference between each pair of tests. | |
218 | ||
219 | c<cmpthese> can also be passed the data structure that timethese() returns: | |
220 | ||
221 | $results = timethese( -1, { a => "++\$i", b => "\$i *= 2" } ) ; | |
ac8eabc1 JH |
222 | cmpthese( $results ); |
223 | ||
d1083c7a BS |
224 | in case you want to see both sets of results. |
225 | ||
226 | Returns a reference to an ARRAY of rows, each row is an ARRAY of cells from the | |
227 | above chart, including labels. This: | |
228 | ||
229 | my $rows = cmpthese( -1, { a => '++$i', b => '$i *= 2' }, "none" ); | |
230 | ||
231 | returns a data structure like: | |
232 | ||
233 | [ | |
234 | [ '', 'Rate', 'b', 'a' ], | |
235 | [ 'b', '2885232/s', '--', '-59%' ], | |
236 | [ 'a', '7099126/s', '146%', '--' ], | |
237 | ] | |
238 | ||
239 | B<NOTE>: This result value differs from previous versions, which returned | |
240 | the C<timethese()> result structure. If you want that, just use the two | |
241 | statement C<timethese>...C<cmpthese> idiom shown above. | |
242 | ||
243 | Incidently, note the variance in the result values between the two examples; | |
244 | this is typical of benchmarking. If this were a real benchmark, you would | |
245 | probably want to run a lot more iterations. | |
ac8eabc1 JH |
246 | |
247 | =item countit(TIME, CODE) | |
248 | ||
249 | Arguments: TIME is the minimum length of time to run CODE for, and CODE is | |
250 | the code to run. CODE may be either a code reference or a string to | |
251 | be eval'd; either way it will be run in the caller's package. | |
252 | ||
253 | TIME is I<not> negative. countit() will run the loop many times to | |
254 | calculate the speed of CODE before running it for TIME. The actual | |
255 | time run for will usually be greater than TIME due to system clock | |
256 | resolution, so it's best to look at the number of iterations divided | |
257 | by the times that you are concerned with, not just the iterations. | |
258 | ||
259 | Returns: a Benchmark object. | |
260 | ||
523cc92b | 261 | =item disablecache ( ) |
f06db76b | 262 | |
523cc92b CS |
263 | Disable caching of timings for the null loop. This will force Benchmark |
264 | to recalculate these timings for each new piece of code timed. | |
265 | ||
266 | =item enablecache ( ) | |
267 | ||
268 | Enable caching of timings for the null loop. The time taken for COUNT | |
269 | rounds of the null loop will be calculated only once for each | |
270 | different COUNT used. | |
f06db76b | 271 | |
ac8eabc1 JH |
272 | =item timesum ( T1, T2 ) |
273 | ||
274 | Returns the sum of two Benchmark times as a Benchmark object suitable | |
275 | for passing to timestr(). | |
276 | ||
f06db76b AD |
277 | =back |
278 | ||
e3d6de9a JH |
279 | =head2 :hireswallclock |
280 | ||
281 | If the Time::HiRes module has been installed, you can specify the | |
282 | special tag C<:hireswallclock> for Benchmark (if Time::HiRes is not | |
283 | available, the tag will be silently ignored). This tag will cause the | |
284 | wallclock time to be measured in microseconds, instead of integer | |
702fa71c HS |
285 | seconds. Note though that the speed computations are still conducted |
286 | in CPU time, not wallclock time. | |
e3d6de9a | 287 | |
f06db76b AD |
288 | =head1 NOTES |
289 | ||
290 | The data is stored as a list of values from the time and times | |
523cc92b | 291 | functions: |
f06db76b | 292 | |
431d98c2 | 293 | ($real, $user, $system, $children_user, $children_system, $iters) |
f06db76b AD |
294 | |
295 | in seconds for the whole loop (not divided by the number of rounds). | |
296 | ||
297 | The timing is done using time(3) and times(3). | |
298 | ||
299 | Code is executed in the caller's package. | |
300 | ||
f06db76b AD |
301 | The time of the null loop (a loop with the same |
302 | number of rounds but empty loop body) is subtracted | |
303 | from the time of the real loop. | |
304 | ||
3c6312e9 | 305 | The null loop times can be cached, the key being the |
f06db76b AD |
306 | number of rounds. The caching can be controlled using |
307 | calls like these: | |
308 | ||
523cc92b | 309 | clearcache($key); |
f06db76b AD |
310 | clearallcache(); |
311 | ||
523cc92b | 312 | disablecache(); |
f06db76b AD |
313 | enablecache(); |
314 | ||
3c6312e9 BS |
315 | Caching is off by default, as it can (usually slightly) decrease |
316 | accuracy and does not usually noticably affect runtimes. | |
317 | ||
54e82ce5 GS |
318 | =head1 EXAMPLES |
319 | ||
320 | For example, | |
321 | ||
14393033 BS |
322 | use Benchmark qw( cmpthese ) ; |
323 | $x = 3; | |
324 | cmpthese( -5, { | |
325 | a => sub{$x*$x}, | |
326 | b => sub{$x**2}, | |
327 | } ); | |
54e82ce5 GS |
328 | |
329 | outputs something like this: | |
330 | ||
331 | Benchmark: running a, b, each for at least 5 CPU seconds... | |
14393033 BS |
332 | Rate b a |
333 | b 1559428/s -- -62% | |
334 | a 4152037/s 166% -- | |
335 | ||
54e82ce5 GS |
336 | |
337 | while | |
338 | ||
14393033 BS |
339 | use Benchmark qw( timethese cmpthese ) ; |
340 | $x = 3; | |
341 | $r = timethese( -5, { | |
342 | a => sub{$x*$x}, | |
343 | b => sub{$x**2}, | |
344 | } ); | |
345 | cmpthese $r; | |
54e82ce5 GS |
346 | |
347 | outputs something like this: | |
348 | ||
14393033 BS |
349 | Benchmark: running a, b, each for at least 5 CPU seconds... |
350 | a: 10 wallclock secs ( 5.14 usr + 0.13 sys = 5.27 CPU) @ 3835055.60/s (n=20210743) | |
351 | b: 5 wallclock secs ( 5.41 usr + 0.00 sys = 5.41 CPU) @ 1574944.92/s (n=8520452) | |
352 | Rate b a | |
353 | b 1574945/s -- -59% | |
354 | a 3835056/s 144% -- | |
54e82ce5 GS |
355 | |
356 | ||
f06db76b AD |
357 | =head1 INHERITANCE |
358 | ||
359 | Benchmark inherits from no other class, except of course | |
360 | for Exporter. | |
361 | ||
362 | =head1 CAVEATS | |
363 | ||
80eab818 | 364 | Comparing eval'd strings with code references will give you |
431d98c2 | 365 | inaccurate results: a code reference will show a slightly slower |
80eab818 CS |
366 | execution time than the equivalent eval'd string. |
367 | ||
f06db76b AD |
368 | The real time timing is done using time(2) and |
369 | the granularity is therefore only one second. | |
370 | ||
371 | Short tests may produce negative figures because perl | |
523cc92b CS |
372 | can appear to take longer to execute the empty loop |
373 | than a short test; try: | |
f06db76b AD |
374 | |
375 | timethis(100,'1'); | |
376 | ||
377 | The system time of the null loop might be slightly | |
378 | more than the system time of the loop with the actual | |
a24a9dfe | 379 | code and therefore the difference might end up being E<lt> 0. |
f06db76b | 380 | |
8a4f6ac2 GS |
381 | =head1 SEE ALSO |
382 | ||
383 | L<Devel::DProf> - a Perl code profiler | |
384 | ||
f06db76b AD |
385 | =head1 AUTHORS |
386 | ||
5aabfad6 | 387 | Jarkko Hietaniemi <F<jhi@iki.fi>>, Tim Bunce <F<Tim.Bunce@ig.co.uk>> |
f06db76b AD |
388 | |
389 | =head1 MODIFICATION HISTORY | |
390 | ||
391 | September 8th, 1994; by Tim Bunce. | |
392 | ||
523cc92b CS |
393 | March 28th, 1997; by Hugo van der Sanden: added support for code |
394 | references and the already documented 'debug' method; revamped | |
395 | documentation. | |
f06db76b | 396 | |
6ee623d5 GS |
397 | April 04-07th, 1997: by Jarkko Hietaniemi, added the run-for-some-time |
398 | functionality. | |
399 | ||
3c6312e9 BS |
400 | September, 1999; by Barrie Slaymaker: math fixes and accuracy and |
401 | efficiency tweaks. Added cmpthese(). A result is now returned from | |
431d98c2 | 402 | timethese(). Exposed countit() (was runfor()). |
3c6312e9 | 403 | |
0e74ff8e JH |
404 | December, 2001; by Nicholas Clark: make timestr() recognise the style 'none' |
405 | and return an empty string. If cmpthese is calling timethese, make it pass the | |
406 | style in. (so that 'none' will suppress output). Make sub new dump its | |
407 | debugging output to STDERR, to be consistent with everything else. | |
408 | All bugs found while writing a regression test. | |
409 | ||
e3d6de9a JH |
410 | September, 2002; by Jarkko Hietaniemi: add ':hireswallclock' special tag. |
411 | ||
523cc92b | 412 | =cut |
a0d0e21e | 413 | |
3f943bd9 GS |
414 | # evaluate something in a clean lexical environment |
415 | sub _doeval { eval shift } | |
416 | ||
417 | # | |
418 | # put any lexicals at file scope AFTER here | |
419 | # | |
420 | ||
4aa0a1f7 | 421 | use Carp; |
a0d0e21e LW |
422 | use Exporter; |
423 | @ISA=(Exporter); | |
ac8eabc1 JH |
424 | @EXPORT=qw(timeit timethis timethese timediff timestr); |
425 | @EXPORT_OK=qw(timesum cmpthese countit | |
426 | clearcache clearallcache disablecache enablecache); | |
f36484b0 | 427 | %EXPORT_TAGS=( all => [ @EXPORT, @EXPORT_OK ] ) ; |
a0d0e21e | 428 | |
e3d6de9a JH |
429 | $VERSION = 1.0501; |
430 | ||
431 | # --- ':hireswallclock' special handling | |
432 | ||
433 | my $hirestime; | |
434 | ||
435 | sub mytime () { time } | |
8a4f6ac2 | 436 | |
a0d0e21e LW |
437 | &init; |
438 | ||
e3d6de9a JH |
439 | sub BEGIN { |
440 | if (eval 'require Time::HiRes') { | |
441 | import Time::HiRes qw(time); | |
442 | $hirestime = \&Time::HiRes::time; | |
443 | } | |
444 | } | |
445 | ||
446 | sub import { | |
447 | my $class = shift; | |
448 | if (grep { $_ eq ":hireswallclock" } @_) { | |
449 | @_ = grep { $_ ne ":hireswallclock" } @_; | |
450 | *mytime = $hirestime if defined $hirestime; | |
451 | } | |
452 | Benchmark->export_to_level(1, $class, @_); | |
453 | } | |
454 | ||
a0d0e21e LW |
455 | sub init { |
456 | $debug = 0; | |
457 | $min_count = 4; | |
458 | $min_cpu = 0.4; | |
459 | $defaultfmt = '5.2f'; | |
460 | $defaultstyle = 'auto'; | |
461 | # The cache can cause a slight loss of sys time accuracy. If a | |
462 | # user does many tests (>10) with *very* large counts (>10000) | |
463 | # or works on a very slow machine the cache may be useful. | |
464 | &disablecache; | |
465 | &clearallcache; | |
466 | } | |
467 | ||
523cc92b CS |
468 | sub debug { $debug = ($_[1] != 0); } |
469 | ||
bba8fca5 BS |
470 | # The cache needs two branches: 's' for strings and 'c' for code. The |
471 | # emtpy loop is different in these two cases. | |
472 | sub clearcache { delete $cache{"$_[0]c"}; delete $cache{"$_[0]s"}; } | |
a0d0e21e LW |
473 | sub clearallcache { %cache = (); } |
474 | sub enablecache { $cache = 1; } | |
475 | sub disablecache { $cache = 0; } | |
476 | ||
a0d0e21e LW |
477 | # --- Functions to process the 'time' data type |
478 | ||
e3d6de9a | 479 | sub new { my @t = (mytime, times, @_ == 2 ? $_[1] : 0); |
0e74ff8e | 480 | print STDERR "new=@t\n" if $debug; |
6ee623d5 | 481 | bless \@t; } |
a0d0e21e LW |
482 | |
483 | sub cpu_p { my($r,$pu,$ps,$cu,$cs) = @{$_[0]}; $pu+$ps ; } | |
484 | sub cpu_c { my($r,$pu,$ps,$cu,$cs) = @{$_[0]}; $cu+$cs ; } | |
485 | sub cpu_a { my($r,$pu,$ps,$cu,$cs) = @{$_[0]}; $pu+$ps+$cu+$cs ; } | |
486 | sub real { my($r,$pu,$ps,$cu,$cs) = @{$_[0]}; $r ; } | |
431d98c2 | 487 | sub iters { $_[0]->[5] ; } |
a0d0e21e | 488 | |
523cc92b | 489 | sub timediff { |
a0d0e21e | 490 | my($a, $b) = @_; |
523cc92b | 491 | my @r; |
3f943bd9 | 492 | for (my $i=0; $i < @$a; ++$i) { |
a0d0e21e LW |
493 | push(@r, $a->[$i] - $b->[$i]); |
494 | } | |
495 | bless \@r; | |
496 | } | |
497 | ||
705cc255 TB |
498 | sub timesum { |
499 | my($a, $b) = @_; | |
500 | my @r; | |
501 | for (my $i=0; $i < @$a; ++$i) { | |
502 | push(@r, $a->[$i] + $b->[$i]); | |
503 | } | |
504 | bless \@r; | |
505 | } | |
506 | ||
523cc92b | 507 | sub timestr { |
a0d0e21e | 508 | my($tr, $style, $f) = @_; |
523cc92b | 509 | my @t = @$tr; |
6ee623d5 GS |
510 | warn "bad time value (@t)" unless @t==6; |
511 | my($r, $pu, $ps, $cu, $cs, $n) = @t; | |
ce9550df | 512 | my($pt, $ct, $tt) = ($tr->cpu_p, $tr->cpu_c, $tr->cpu_a); |
523cc92b | 513 | $f = $defaultfmt unless defined $f; |
a0d0e21e | 514 | # format a time in the required style, other formats may be added here |
80eab818 | 515 | $style ||= $defaultstyle; |
0e74ff8e | 516 | return '' if $style eq 'none'; |
523cc92b CS |
517 | $style = ($ct>0) ? 'all' : 'noc' if $style eq 'auto'; |
518 | my $s = "@t $style"; # default for unknown style | |
e3d6de9a JH |
519 | my $w = $hirestime ? "%2g" : "%2d"; |
520 | $s=sprintf("$w wallclock secs (%$f usr %$f sys + %$f cusr %$f csys = %$f CPU)", | |
ce9550df | 521 | $r,$pu,$ps,$cu,$cs,$tt) if $style eq 'all'; |
e3d6de9a | 522 | $s=sprintf("$w wallclock secs (%$f usr + %$f sys = %$f CPU)", |
7be077a2 | 523 | $r,$pu,$ps,$pt) if $style eq 'noc'; |
e3d6de9a | 524 | $s=sprintf("$w wallclock secs (%$f cusr + %$f csys = %$f CPU)", |
7be077a2 | 525 | $r,$cu,$cs,$ct) if $style eq 'nop'; |
cc31225e | 526 | $s .= sprintf(" @ %$f/s (n=$n)", $n / ( $pu + $ps )) if $n && $pu+$ps; |
a0d0e21e LW |
527 | $s; |
528 | } | |
523cc92b CS |
529 | |
530 | sub timedebug { | |
a0d0e21e | 531 | my($msg, $t) = @_; |
523cc92b | 532 | print STDERR "$msg",timestr($t),"\n" if $debug; |
a0d0e21e LW |
533 | } |
534 | ||
a0d0e21e LW |
535 | # --- Functions implementing low-level support for timing loops |
536 | ||
537 | sub runloop { | |
538 | my($n, $c) = @_; | |
4aa0a1f7 AD |
539 | |
540 | $n+=0; # force numeric now, so garbage won't creep into the eval | |
523cc92b CS |
541 | croak "negative loopcount $n" if $n<0; |
542 | confess "Usage: runloop(number, [string | coderef])" unless defined $c; | |
a0d0e21e LW |
543 | my($t0, $t1, $td); # before, after, difference |
544 | ||
545 | # find package of caller so we can execute code there | |
523cc92b CS |
546 | my($curpack) = caller(0); |
547 | my($i, $pack)= 0; | |
a0d0e21e LW |
548 | while (($pack) = caller(++$i)) { |
549 | last if $pack ne $curpack; | |
550 | } | |
551 | ||
3f943bd9 GS |
552 | my ($subcode, $subref); |
553 | if (ref $c eq 'CODE') { | |
554 | $subcode = "sub { for (1 .. $n) { local \$_; package $pack; &\$c; } }"; | |
555 | $subref = eval $subcode; | |
556 | } | |
557 | else { | |
558 | $subcode = "sub { for (1 .. $n) { local \$_; package $pack; $c;} }"; | |
559 | $subref = _doeval($subcode); | |
560 | } | |
4aa0a1f7 | 561 | croak "runloop unable to compile '$c': $@\ncode: $subcode\n" if $@; |
523cc92b | 562 | print STDERR "runloop $n '$subcode'\n" if $debug; |
a0d0e21e | 563 | |
3c6312e9 BS |
564 | # Wait for the user timer to tick. This makes the error range more like |
565 | # -0.01, +0. If we don't wait, then it's more like -0.01, +0.01. This | |
566 | # may not seem important, but it significantly reduces the chances of | |
567 | # getting a too low initial $n in the initial, 'find the minimum' loop | |
431d98c2 | 568 | # in &countit. This, in turn, can reduce the number of calls to |
bba8fca5 BS |
569 | # &runloop a lot, and thus reduce additive errors. |
570 | my $tbase = Benchmark->new(0)->[1]; | |
277427cf | 571 | while ( ( $t0 = Benchmark->new(0) )->[1] == $tbase ) {} ; |
a0d0e21e | 572 | &$subref; |
6ee623d5 | 573 | $t1 = Benchmark->new($n); |
a0d0e21e | 574 | $td = &timediff($t1, $t0); |
a0d0e21e LW |
575 | timedebug("runloop:",$td); |
576 | $td; | |
577 | } | |
578 | ||
579 | ||
580 | sub timeit { | |
581 | my($n, $code) = @_; | |
582 | my($wn, $wc, $wd); | |
583 | ||
584 | printf STDERR "timeit $n $code\n" if $debug; | |
3c6312e9 | 585 | my $cache_key = $n . ( ref( $code ) ? 'c' : 's' ); |
bba8fca5 BS |
586 | if ($cache && exists $cache{$cache_key} ) { |
587 | $wn = $cache{$cache_key}; | |
523cc92b | 588 | } else { |
6bf773bc | 589 | $wn = &runloop($n, ref( $code ) ? sub { } : '' ); |
3c6312e9 BS |
590 | # Can't let our baseline have any iterations, or they get subtracted |
591 | # out of the result. | |
592 | $wn->[5] = 0; | |
bba8fca5 | 593 | $cache{$cache_key} = $wn; |
a0d0e21e LW |
594 | } |
595 | ||
596 | $wc = &runloop($n, $code); | |
597 | ||
598 | $wd = timediff($wc, $wn); | |
a0d0e21e LW |
599 | timedebug("timeit: ",$wc); |
600 | timedebug(" - ",$wn); | |
601 | timedebug(" = ",$wd); | |
602 | ||
603 | $wd; | |
604 | } | |
605 | ||
6ee623d5 GS |
606 | |
607 | my $default_for = 3; | |
608 | my $min_for = 0.1; | |
609 | ||
3c6312e9 | 610 | |
431d98c2 BS |
611 | sub countit { |
612 | my ( $tmax, $code ) = @_; | |
6ee623d5 GS |
613 | |
614 | if ( not defined $tmax or $tmax == 0 ) { | |
615 | $tmax = $default_for; | |
616 | } elsif ( $tmax < 0 ) { | |
617 | $tmax = -$tmax; | |
618 | } | |
619 | ||
431d98c2 | 620 | die "countit($tmax, ...): timelimit cannot be less than $min_for.\n" |
6ee623d5 GS |
621 | if $tmax < $min_for; |
622 | ||
3c6312e9 | 623 | my ($n, $tc); |
6ee623d5 | 624 | |
bba8fca5 | 625 | # First find the minimum $n that gives a significant timing. |
3c6312e9 BS |
626 | for ($n = 1; ; $n *= 2 ) { |
627 | my $td = timeit($n, $code); | |
628 | $tc = $td->[1] + $td->[2]; | |
629 | last if $tc > 0.1; | |
630 | } | |
6ee623d5 | 631 | |
3c6312e9 BS |
632 | my $nmin = $n; |
633 | ||
634 | # Get $n high enough that we can guess the final $n with some accuracy. | |
635 | my $tpra = 0.1 * $tmax; # Target/time practice. | |
636 | while ( $tc < $tpra ) { | |
637 | # The 5% fudge is to keep us from iterating again all | |
638 | # that often (this speeds overall responsiveness when $tmax is big | |
639 | # and we guess a little low). This does not noticably affect | |
640 | # accuracy since we're not couting these times. | |
641 | $n = int( $tpra * 1.05 * $n / $tc ); # Linear approximation. | |
642 | my $td = timeit($n, $code); | |
c5d57293 A |
643 | my $new_tc = $td->[1] + $td->[2]; |
644 | # Make sure we are making progress. | |
645 | $tc = $new_tc > 1.2 * $tc ? $new_tc : 1.2 * $tc; | |
6ee623d5 GS |
646 | } |
647 | ||
3c6312e9 BS |
648 | # Now, do the 'for real' timing(s), repeating until we exceed |
649 | # the max. | |
650 | my $ntot = 0; | |
651 | my $rtot = 0; | |
652 | my $utot = 0.0; | |
653 | my $stot = 0.0; | |
654 | my $cutot = 0.0; | |
655 | my $cstot = 0.0; | |
656 | my $ttot = 0.0; | |
657 | ||
658 | # The 5% fudge is because $n is often a few % low even for routines | |
659 | # with stable times and avoiding extra timeit()s is nice for | |
660 | # accuracy's sake. | |
661 | $n = int( $n * ( 1.05 * $tmax / $tc ) ); | |
662 | ||
663 | while () { | |
664 | my $td = timeit($n, $code); | |
665 | $ntot += $n; | |
666 | $rtot += $td->[0]; | |
667 | $utot += $td->[1]; | |
668 | $stot += $td->[2]; | |
6ee623d5 GS |
669 | $cutot += $td->[3]; |
670 | $cstot += $td->[4]; | |
3c6312e9 BS |
671 | $ttot = $utot + $stot; |
672 | last if $ttot >= $tmax; | |
6ee623d5 | 673 | |
c5d57293 | 674 | $ttot = 0.01 if $ttot < 0.01; |
3c6312e9 | 675 | my $r = $tmax / $ttot - 1; # Linear approximation. |
bba8fca5 | 676 | $n = int( $r * $ntot ); |
6ee623d5 | 677 | $n = $nmin if $n < $nmin; |
6ee623d5 GS |
678 | } |
679 | ||
680 | return bless [ $rtot, $utot, $stot, $cutot, $cstot, $ntot ]; | |
681 | } | |
682 | ||
a0d0e21e LW |
683 | # --- Functions implementing high-level time-then-print utilities |
684 | ||
6ee623d5 GS |
685 | sub n_to_for { |
686 | my $n = shift; | |
687 | return $n == 0 ? $default_for : $n < 0 ? -$n : undef; | |
688 | } | |
689 | ||
a0d0e21e LW |
690 | sub timethis{ |
691 | my($n, $code, $title, $style) = @_; | |
6ee623d5 GS |
692 | my($t, $for, $forn); |
693 | ||
694 | if ( $n > 0 ) { | |
695 | croak "non-integer loopcount $n, stopped" if int($n)<$n; | |
696 | $t = timeit($n, $code); | |
697 | $title = "timethis $n" unless defined $title; | |
698 | } else { | |
699 | $fort = n_to_for( $n ); | |
431d98c2 | 700 | $t = countit( $fort, $code ); |
6ee623d5 GS |
701 | $title = "timethis for $fort" unless defined $title; |
702 | $forn = $t->[-1]; | |
703 | } | |
523cc92b | 704 | local $| = 1; |
523cc92b | 705 | $style = "" unless defined $style; |
3c6312e9 BS |
706 | printf("%10s: ", $title) unless $style eq 'none'; |
707 | print timestr($t, $style, $defaultfmt),"\n" unless $style eq 'none'; | |
6ee623d5 GS |
708 | |
709 | $n = $forn if defined $forn; | |
523cc92b | 710 | |
a0d0e21e LW |
711 | # A conservative warning to spot very silly tests. |
712 | # Don't assume that your benchmark is ok simply because | |
713 | # you don't get this warning! | |
714 | print " (warning: too few iterations for a reliable count)\n" | |
523cc92b | 715 | if $n < $min_count |
a0d0e21e | 716 | || ($t->real < 1 && $n < 1000) |
523cc92b | 717 | || $t->cpu_a < $min_cpu; |
a0d0e21e LW |
718 | $t; |
719 | } | |
720 | ||
a0d0e21e LW |
721 | sub timethese{ |
722 | my($n, $alt, $style) = @_; | |
723 | die "usage: timethese(count, { 'Name1'=>'code1', ... }\n" | |
724 | unless ref $alt eq HASH; | |
523cc92b CS |
725 | my @names = sort keys %$alt; |
726 | $style = "" unless defined $style; | |
3c6312e9 | 727 | print "Benchmark: " unless $style eq 'none'; |
6ee623d5 GS |
728 | if ( $n > 0 ) { |
729 | croak "non-integer loopcount $n, stopped" if int($n)<$n; | |
3c6312e9 | 730 | print "timing $n iterations of" unless $style eq 'none'; |
6ee623d5 | 731 | } else { |
3c6312e9 | 732 | print "running" unless $style eq 'none'; |
6ee623d5 | 733 | } |
3c6312e9 | 734 | print " ", join(', ',@names) unless $style eq 'none'; |
6ee623d5 GS |
735 | unless ( $n > 0 ) { |
736 | my $for = n_to_for( $n ); | |
df7779cf T |
737 | print ", each" if $n > 1 && $style ne 'none'; |
738 | print " for at least $for CPU seconds" unless $style eq 'none'; | |
6ee623d5 | 739 | } |
3c6312e9 | 740 | print "...\n" unless $style eq 'none'; |
523cc92b CS |
741 | |
742 | # we could save the results in an array and produce a summary here | |
a0d0e21e | 743 | # sum, min, max, avg etc etc |
3c6312e9 | 744 | my %results; |
4dbb2df9 | 745 | foreach my $name (@names) { |
3c6312e9 | 746 | $results{$name} = timethis ($n, $alt -> {$name}, $name, $style); |
4dbb2df9 | 747 | } |
3c6312e9 BS |
748 | |
749 | return \%results; | |
a0d0e21e LW |
750 | } |
751 | ||
3c6312e9 | 752 | sub cmpthese{ |
8962dfd6 A |
753 | my ($results, $style) = |
754 | ref $_ [0] ? @_ | |
755 | : (timethese (@_ [0, 1], @_ > 2 ? $_ [2] : "none"), $_ [2]); | |
3c6312e9 | 756 | |
d1083c7a | 757 | $style = "" unless defined $style; |
3c6312e9 BS |
758 | |
759 | # Flatten in to an array of arrays with the name as the first field | |
760 | my @vals = map{ [ $_, @{$results->{$_}} ] } keys %$results; | |
761 | ||
762 | for (@vals) { | |
763 | # The epsilon fudge here is to prevent div by 0. Since clock | |
764 | # resolutions are much larger, it's below the noise floor. | |
765 | my $rate = $_->[6] / ( $_->[2] + $_->[3] + 0.000000000000001 ); | |
766 | $_->[7] = $rate; | |
767 | } | |
768 | ||
769 | # Sort by rate | |
770 | @vals = sort { $a->[7] <=> $b->[7] } @vals; | |
771 | ||
772 | # If more than half of the rates are greater than one... | |
773 | my $display_as_rate = $vals[$#vals>>1]->[7] > 1; | |
774 | ||
775 | my @rows; | |
776 | my @col_widths; | |
777 | ||
778 | my @top_row = ( | |
779 | '', | |
780 | $display_as_rate ? 'Rate' : 's/iter', | |
781 | map { $_->[0] } @vals | |
782 | ); | |
783 | ||
784 | push @rows, \@top_row; | |
785 | @col_widths = map { length( $_ ) } @top_row; | |
786 | ||
787 | # Build the data rows | |
788 | # We leave the last column in even though it never has any data. Perhaps | |
789 | # it should go away. Also, perhaps a style for a single column of | |
790 | # percentages might be nice. | |
791 | for my $row_val ( @vals ) { | |
792 | my @row; | |
793 | ||
794 | # Column 0 = test name | |
795 | push @row, $row_val->[0]; | |
796 | $col_widths[0] = length( $row_val->[0] ) | |
797 | if length( $row_val->[0] ) > $col_widths[0]; | |
798 | ||
799 | # Column 1 = performance | |
800 | my $row_rate = $row_val->[7]; | |
801 | ||
802 | # We assume that we'll never get a 0 rate. | |
803 | my $a = $display_as_rate ? $row_rate : 1 / $row_rate; | |
804 | ||
805 | # Only give a few decimal places before switching to sci. notation, | |
806 | # since the results aren't usually that accurate anyway. | |
807 | my $format = | |
808 | $a >= 100 ? | |
809 | "%0.0f" : | |
810 | $a >= 10 ? | |
811 | "%0.1f" : | |
812 | $a >= 1 ? | |
813 | "%0.2f" : | |
814 | $a >= 0.1 ? | |
815 | "%0.3f" : | |
816 | "%0.2e"; | |
817 | ||
818 | $format .= "/s" | |
819 | if $display_as_rate; | |
820 | # Using $b here due to optimizing bug in _58 through _61 | |
821 | my $b = sprintf( $format, $a ); | |
822 | push @row, $b; | |
823 | $col_widths[1] = length( $b ) | |
824 | if length( $b ) > $col_widths[1]; | |
825 | ||
826 | # Columns 2..N = performance ratios | |
827 | my $skip_rest = 0; | |
828 | for ( my $col_num = 0 ; $col_num < @vals ; ++$col_num ) { | |
829 | my $col_val = $vals[$col_num]; | |
830 | my $out; | |
831 | if ( $skip_rest ) { | |
832 | $out = ''; | |
833 | } | |
834 | elsif ( $col_val->[0] eq $row_val->[0] ) { | |
835 | $out = "--"; | |
836 | # $skip_rest = 1; | |
837 | } | |
838 | else { | |
839 | my $col_rate = $col_val->[7]; | |
840 | $out = sprintf( "%.0f%%", 100*$row_rate/$col_rate - 100 ); | |
841 | } | |
842 | push @row, $out; | |
843 | $col_widths[$col_num+2] = length( $out ) | |
844 | if length( $out ) > $col_widths[$col_num+2]; | |
845 | ||
846 | # A little wierdness to set the first column width properly | |
847 | $col_widths[$col_num+2] = length( $col_val->[0] ) | |
848 | if length( $col_val->[0] ) > $col_widths[$col_num+2]; | |
849 | } | |
850 | push @rows, \@row; | |
851 | } | |
852 | ||
d1083c7a BS |
853 | return \@rows if $style eq "none"; |
854 | ||
3c6312e9 BS |
855 | # Equalize column widths in the chart as much as possible without |
856 | # exceeding 80 characters. This does not use or affect cols 0 or 1. | |
857 | my @sorted_width_refs = | |
858 | sort { $$a <=> $$b } map { \$_ } @col_widths[2..$#col_widths]; | |
859 | my $max_width = ${$sorted_width_refs[-1]}; | |
860 | ||
277427cf | 861 | my $total = @col_widths - 1 ; |
3c6312e9 BS |
862 | for ( @col_widths ) { $total += $_ } |
863 | ||
864 | STRETCHER: | |
865 | while ( $total < 80 ) { | |
866 | my $min_width = ${$sorted_width_refs[0]}; | |
867 | last | |
868 | if $min_width == $max_width; | |
869 | for ( @sorted_width_refs ) { | |
870 | last | |
871 | if $$_ > $min_width; | |
872 | ++$$_; | |
873 | ++$total; | |
874 | last STRETCHER | |
875 | if $total >= 80; | |
876 | } | |
877 | } | |
878 | ||
879 | # Dump the output | |
880 | my $format = join( ' ', map { "%${_}s" } @col_widths ) . "\n"; | |
881 | substr( $format, 1, 0 ) = '-'; | |
882 | for ( @rows ) { | |
883 | printf $format, @$_; | |
884 | } | |
885 | ||
d1083c7a | 886 | return \@rows ; |
3c6312e9 BS |
887 | } |
888 | ||
889 | ||
a0d0e21e | 890 | 1; |