Commit | Line | Data |
---|---|---|
a0d0e21e LW |
1 | package Benchmark; |
2 | ||
53aa2791 MS |
3 | use strict; |
4 | ||
5 | ||
f06db76b AD |
6 | =head1 NAME |
7 | ||
8a4f6ac2 | 8 | Benchmark - benchmark running times of Perl code |
431d98c2 | 9 | |
f06db76b AD |
10 | =head1 SYNOPSIS |
11 | ||
f36484b0 BS |
12 | use Benchmark qw(:all) ; |
13 | ||
f06db76b AD |
14 | timethis ($count, "code"); |
15 | ||
523cc92b | 16 | # Use Perl code in strings... |
f06db76b AD |
17 | timethese($count, { |
18 | 'Name1' => '...code1...', | |
19 | 'Name2' => '...code2...', | |
20 | }); | |
21 | ||
523cc92b CS |
22 | # ... or use subroutine references. |
23 | timethese($count, { | |
24 | 'Name1' => sub { ...code1... }, | |
25 | 'Name2' => sub { ...code2... }, | |
26 | }); | |
27 | ||
431d98c2 BS |
28 | # cmpthese can be used both ways as well |
29 | cmpthese($count, { | |
30 | 'Name1' => '...code1...', | |
31 | 'Name2' => '...code2...', | |
32 | }); | |
33 | ||
34 | cmpthese($count, { | |
35 | 'Name1' => sub { ...code1... }, | |
36 | 'Name2' => sub { ...code2... }, | |
37 | }); | |
38 | ||
39 | # ...or in two stages | |
40 | $results = timethese($count, | |
41 | { | |
42 | 'Name1' => sub { ...code1... }, | |
43 | 'Name2' => sub { ...code2... }, | |
44 | }, | |
45 | 'none' | |
46 | ); | |
47 | cmpthese( $results ) ; | |
48 | ||
f06db76b AD |
49 | $t = timeit($count, '...other code...') |
50 | print "$count loops of other code took:",timestr($t),"\n"; | |
51 | ||
431d98c2 BS |
52 | $t = countit($time, '...other code...') |
53 | $count = $t->iters ; | |
54 | print "$count loops of other code took:",timestr($t),"\n"; | |
55 | ||
e3d6de9a JH |
56 | # enable hires wallclock timing if possible |
57 | use Benchmark ':hireswallclock'; | |
58 | ||
f06db76b AD |
59 | =head1 DESCRIPTION |
60 | ||
61 | The Benchmark module encapsulates a number of routines to help you | |
62 | figure out how long it takes to execute some code. | |
63 | ||
8a4f6ac2 GS |
64 | timethis - run a chunk of code several times |
65 | ||
66 | timethese - run several chunks of code several times | |
67 | ||
68 | cmpthese - print results of timethese as a comparison chart | |
69 | ||
70 | timeit - run a chunk of code and see how long it goes | |
71 | ||
72 | countit - see how many times a chunk of code runs in a given time | |
73 | ||
74 | ||
f06db76b AD |
75 | =head2 Methods |
76 | ||
77 | =over 10 | |
78 | ||
79 | =item new | |
80 | ||
81 | Returns the current time. Example: | |
82 | ||
83 | use Benchmark; | |
84 | $t0 = new Benchmark; | |
85 | # ... your code here ... | |
86 | $t1 = new Benchmark; | |
87 | $td = timediff($t1, $t0); | |
a24a9dfe | 88 | print "the code took:",timestr($td),"\n"; |
f06db76b AD |
89 | |
90 | =item debug | |
91 | ||
92 | Enables or disable debugging by setting the C<$Benchmark::Debug> flag: | |
93 | ||
523cc92b | 94 | debug Benchmark 1; |
f06db76b | 95 | $t = timeit(10, ' 5 ** $Global '); |
523cc92b | 96 | debug Benchmark 0; |
f06db76b | 97 | |
431d98c2 BS |
98 | =item iters |
99 | ||
100 | Returns the number of iterations. | |
101 | ||
f06db76b AD |
102 | =back |
103 | ||
104 | =head2 Standard Exports | |
105 | ||
523cc92b | 106 | The following routines will be exported into your namespace |
f06db76b AD |
107 | if you use the Benchmark module: |
108 | ||
109 | =over 10 | |
110 | ||
111 | =item timeit(COUNT, CODE) | |
112 | ||
523cc92b CS |
113 | Arguments: COUNT is the number of times to run the loop, and CODE is |
114 | the code to run. CODE may be either a code reference or a string to | |
115 | be eval'd; either way it will be run in the caller's package. | |
116 | ||
117 | Returns: a Benchmark object. | |
118 | ||
119 | =item timethis ( COUNT, CODE, [ TITLE, [ STYLE ]] ) | |
120 | ||
121 | Time COUNT iterations of CODE. CODE may be a string to eval or a | |
122 | code reference; either way the CODE will run in the caller's package. | |
123 | Results will be printed to STDOUT as TITLE followed by the times. | |
124 | TITLE defaults to "timethis COUNT" if none is provided. STYLE | |
125 | determines the format of the output, as described for timestr() below. | |
126 | ||
6ee623d5 GS |
127 | The COUNT can be zero or negative: this means the I<minimum number of |
128 | CPU seconds> to run. A zero signifies the default of 3 seconds. For | |
129 | example to run at least for 10 seconds: | |
130 | ||
131 | timethis(-10, $code) | |
132 | ||
133 | or to run two pieces of code tests for at least 3 seconds: | |
134 | ||
135 | timethese(0, { test1 => '...', test2 => '...'}) | |
136 | ||
137 | CPU seconds is, in UNIX terms, the user time plus the system time of | |
138 | the process itself, as opposed to the real (wallclock) time and the | |
139 | time spent by the child processes. Less than 0.1 seconds is not | |
140 | accepted (-0.01 as the count, for example, will cause a fatal runtime | |
141 | exception). | |
142 | ||
143 | Note that the CPU seconds is the B<minimum> time: CPU scheduling and | |
144 | other operating system factors may complicate the attempt so that a | |
145 | little bit more time is spent. The benchmark output will, however, | |
146 | also tell the number of C<$code> runs/second, which should be a more | |
147 | interesting number than the actually spent seconds. | |
148 | ||
149 | Returns a Benchmark object. | |
150 | ||
523cc92b | 151 | =item timethese ( COUNT, CODEHASHREF, [ STYLE ] ) |
f06db76b | 152 | |
523cc92b CS |
153 | The CODEHASHREF is a reference to a hash containing names as keys |
154 | and either a string to eval or a code reference for each value. | |
155 | For each (KEY, VALUE) pair in the CODEHASHREF, this routine will | |
156 | call | |
f06db76b | 157 | |
523cc92b | 158 | timethis(COUNT, VALUE, KEY, STYLE) |
f06db76b | 159 | |
1d2dff63 GS |
160 | The routines are called in string comparison order of KEY. |
161 | ||
162 | The COUNT can be zero or negative, see timethis(). | |
6ee623d5 | 163 | |
3c6312e9 BS |
164 | Returns a hash of Benchmark objects, keyed by name. |
165 | ||
523cc92b | 166 | =item timediff ( T1, T2 ) |
f06db76b | 167 | |
523cc92b CS |
168 | Returns the difference between two Benchmark times as a Benchmark |
169 | object suitable for passing to timestr(). | |
f06db76b | 170 | |
6ee623d5 | 171 | =item timestr ( TIMEDIFF, [ STYLE, [ FORMAT ] ] ) |
f06db76b | 172 | |
523cc92b CS |
173 | Returns a string that formats the times in the TIMEDIFF object in |
174 | the requested STYLE. TIMEDIFF is expected to be a Benchmark object | |
175 | similar to that returned by timediff(). | |
176 | ||
3c6312e9 BS |
177 | STYLE can be any of 'all', 'none', 'noc', 'nop' or 'auto'. 'all' shows |
178 | each of the 5 times available ('wallclock' time, user time, system time, | |
523cc92b CS |
179 | user time of children, and system time of children). 'noc' shows all |
180 | except the two children times. 'nop' shows only wallclock and the | |
181 | two children times. 'auto' (the default) will act as 'all' unless | |
182 | the children times are both zero, in which case it acts as 'noc'. | |
3c6312e9 | 183 | 'none' prevents output. |
523cc92b CS |
184 | |
185 | FORMAT is the L<printf(3)>-style format specifier (without the | |
186 | leading '%') to use to print the times. It defaults to '5.2f'. | |
f06db76b AD |
187 | |
188 | =back | |
189 | ||
190 | =head2 Optional Exports | |
191 | ||
192 | The following routines will be exported into your namespace | |
193 | if you specifically ask that they be imported: | |
194 | ||
195 | =over 10 | |
196 | ||
523cc92b CS |
197 | =item clearcache ( COUNT ) |
198 | ||
199 | Clear the cached time for COUNT rounds of the null loop. | |
200 | ||
201 | =item clearallcache ( ) | |
f06db76b | 202 | |
523cc92b | 203 | Clear all cached times. |
f06db76b | 204 | |
8962dfd6 | 205 | =item cmpthese ( COUNT, CODEHASHREF, [ STYLE ] ) |
ac8eabc1 | 206 | |
d1083c7a | 207 | =item cmpthese ( RESULTSHASHREF, [ STYLE ] ) |
ac8eabc1 | 208 | |
d1083c7a | 209 | Optionally calls timethese(), then outputs comparison chart. This: |
ac8eabc1 | 210 | |
d1083c7a BS |
211 | cmpthese( -1, { a => "++\$i", b => "\$i *= 2" } ) ; |
212 | ||
213 | outputs a chart like: | |
214 | ||
215 | Rate b a | |
216 | b 2831802/s -- -61% | |
217 | a 7208959/s 155% -- | |
218 | ||
219 | This chart is sorted from slowest to fastest, and shows the percent speed | |
220 | difference between each pair of tests. | |
221 | ||
222 | c<cmpthese> can also be passed the data structure that timethese() returns: | |
223 | ||
224 | $results = timethese( -1, { a => "++\$i", b => "\$i *= 2" } ) ; | |
ac8eabc1 JH |
225 | cmpthese( $results ); |
226 | ||
d1083c7a BS |
227 | in case you want to see both sets of results. |
228 | ||
229 | Returns a reference to an ARRAY of rows, each row is an ARRAY of cells from the | |
230 | above chart, including labels. This: | |
231 | ||
232 | my $rows = cmpthese( -1, { a => '++$i', b => '$i *= 2' }, "none" ); | |
233 | ||
234 | returns a data structure like: | |
235 | ||
236 | [ | |
237 | [ '', 'Rate', 'b', 'a' ], | |
238 | [ 'b', '2885232/s', '--', '-59%' ], | |
239 | [ 'a', '7099126/s', '146%', '--' ], | |
240 | ] | |
241 | ||
242 | B<NOTE>: This result value differs from previous versions, which returned | |
243 | the C<timethese()> result structure. If you want that, just use the two | |
244 | statement C<timethese>...C<cmpthese> idiom shown above. | |
245 | ||
246 | Incidently, note the variance in the result values between the two examples; | |
247 | this is typical of benchmarking. If this were a real benchmark, you would | |
248 | probably want to run a lot more iterations. | |
ac8eabc1 JH |
249 | |
250 | =item countit(TIME, CODE) | |
251 | ||
252 | Arguments: TIME is the minimum length of time to run CODE for, and CODE is | |
253 | the code to run. CODE may be either a code reference or a string to | |
254 | be eval'd; either way it will be run in the caller's package. | |
255 | ||
256 | TIME is I<not> negative. countit() will run the loop many times to | |
257 | calculate the speed of CODE before running it for TIME. The actual | |
258 | time run for will usually be greater than TIME due to system clock | |
259 | resolution, so it's best to look at the number of iterations divided | |
260 | by the times that you are concerned with, not just the iterations. | |
261 | ||
262 | Returns: a Benchmark object. | |
263 | ||
523cc92b | 264 | =item disablecache ( ) |
f06db76b | 265 | |
523cc92b CS |
266 | Disable caching of timings for the null loop. This will force Benchmark |
267 | to recalculate these timings for each new piece of code timed. | |
268 | ||
269 | =item enablecache ( ) | |
270 | ||
271 | Enable caching of timings for the null loop. The time taken for COUNT | |
272 | rounds of the null loop will be calculated only once for each | |
273 | different COUNT used. | |
f06db76b | 274 | |
ac8eabc1 JH |
275 | =item timesum ( T1, T2 ) |
276 | ||
277 | Returns the sum of two Benchmark times as a Benchmark object suitable | |
278 | for passing to timestr(). | |
279 | ||
f06db76b AD |
280 | =back |
281 | ||
e3d6de9a JH |
282 | =head2 :hireswallclock |
283 | ||
284 | If the Time::HiRes module has been installed, you can specify the | |
285 | special tag C<:hireswallclock> for Benchmark (if Time::HiRes is not | |
286 | available, the tag will be silently ignored). This tag will cause the | |
287 | wallclock time to be measured in microseconds, instead of integer | |
702fa71c HS |
288 | seconds. Note though that the speed computations are still conducted |
289 | in CPU time, not wallclock time. | |
e3d6de9a | 290 | |
f06db76b AD |
291 | =head1 NOTES |
292 | ||
293 | The data is stored as a list of values from the time and times | |
523cc92b | 294 | functions: |
f06db76b | 295 | |
431d98c2 | 296 | ($real, $user, $system, $children_user, $children_system, $iters) |
f06db76b AD |
297 | |
298 | in seconds for the whole loop (not divided by the number of rounds). | |
299 | ||
300 | The timing is done using time(3) and times(3). | |
301 | ||
302 | Code is executed in the caller's package. | |
303 | ||
f06db76b AD |
304 | The time of the null loop (a loop with the same |
305 | number of rounds but empty loop body) is subtracted | |
306 | from the time of the real loop. | |
307 | ||
3c6312e9 | 308 | The null loop times can be cached, the key being the |
f06db76b AD |
309 | number of rounds. The caching can be controlled using |
310 | calls like these: | |
311 | ||
523cc92b | 312 | clearcache($key); |
f06db76b AD |
313 | clearallcache(); |
314 | ||
523cc92b | 315 | disablecache(); |
f06db76b AD |
316 | enablecache(); |
317 | ||
3c6312e9 BS |
318 | Caching is off by default, as it can (usually slightly) decrease |
319 | accuracy and does not usually noticably affect runtimes. | |
320 | ||
54e82ce5 GS |
321 | =head1 EXAMPLES |
322 | ||
323 | For example, | |
324 | ||
14393033 BS |
325 | use Benchmark qw( cmpthese ) ; |
326 | $x = 3; | |
327 | cmpthese( -5, { | |
328 | a => sub{$x*$x}, | |
329 | b => sub{$x**2}, | |
330 | } ); | |
54e82ce5 GS |
331 | |
332 | outputs something like this: | |
333 | ||
334 | Benchmark: running a, b, each for at least 5 CPU seconds... | |
14393033 BS |
335 | Rate b a |
336 | b 1559428/s -- -62% | |
337 | a 4152037/s 166% -- | |
338 | ||
54e82ce5 GS |
339 | |
340 | while | |
341 | ||
14393033 BS |
342 | use Benchmark qw( timethese cmpthese ) ; |
343 | $x = 3; | |
344 | $r = timethese( -5, { | |
345 | a => sub{$x*$x}, | |
346 | b => sub{$x**2}, | |
347 | } ); | |
348 | cmpthese $r; | |
54e82ce5 GS |
349 | |
350 | outputs something like this: | |
351 | ||
14393033 BS |
352 | Benchmark: running a, b, each for at least 5 CPU seconds... |
353 | a: 10 wallclock secs ( 5.14 usr + 0.13 sys = 5.27 CPU) @ 3835055.60/s (n=20210743) | |
354 | b: 5 wallclock secs ( 5.41 usr + 0.00 sys = 5.41 CPU) @ 1574944.92/s (n=8520452) | |
355 | Rate b a | |
356 | b 1574945/s -- -59% | |
357 | a 3835056/s 144% -- | |
54e82ce5 GS |
358 | |
359 | ||
f06db76b AD |
360 | =head1 INHERITANCE |
361 | ||
362 | Benchmark inherits from no other class, except of course | |
363 | for Exporter. | |
364 | ||
365 | =head1 CAVEATS | |
366 | ||
80eab818 | 367 | Comparing eval'd strings with code references will give you |
431d98c2 | 368 | inaccurate results: a code reference will show a slightly slower |
80eab818 CS |
369 | execution time than the equivalent eval'd string. |
370 | ||
f06db76b AD |
371 | The real time timing is done using time(2) and |
372 | the granularity is therefore only one second. | |
373 | ||
374 | Short tests may produce negative figures because perl | |
523cc92b CS |
375 | can appear to take longer to execute the empty loop |
376 | than a short test; try: | |
f06db76b AD |
377 | |
378 | timethis(100,'1'); | |
379 | ||
380 | The system time of the null loop might be slightly | |
381 | more than the system time of the loop with the actual | |
a24a9dfe | 382 | code and therefore the difference might end up being E<lt> 0. |
f06db76b | 383 | |
8a4f6ac2 GS |
384 | =head1 SEE ALSO |
385 | ||
386 | L<Devel::DProf> - a Perl code profiler | |
387 | ||
f06db76b AD |
388 | =head1 AUTHORS |
389 | ||
5aabfad6 | 390 | Jarkko Hietaniemi <F<jhi@iki.fi>>, Tim Bunce <F<Tim.Bunce@ig.co.uk>> |
f06db76b AD |
391 | |
392 | =head1 MODIFICATION HISTORY | |
393 | ||
394 | September 8th, 1994; by Tim Bunce. | |
395 | ||
523cc92b CS |
396 | March 28th, 1997; by Hugo van der Sanden: added support for code |
397 | references and the already documented 'debug' method; revamped | |
398 | documentation. | |
f06db76b | 399 | |
6ee623d5 GS |
400 | April 04-07th, 1997: by Jarkko Hietaniemi, added the run-for-some-time |
401 | functionality. | |
402 | ||
3c6312e9 BS |
403 | September, 1999; by Barrie Slaymaker: math fixes and accuracy and |
404 | efficiency tweaks. Added cmpthese(). A result is now returned from | |
431d98c2 | 405 | timethese(). Exposed countit() (was runfor()). |
3c6312e9 | 406 | |
0e74ff8e JH |
407 | December, 2001; by Nicholas Clark: make timestr() recognise the style 'none' |
408 | and return an empty string. If cmpthese is calling timethese, make it pass the | |
409 | style in. (so that 'none' will suppress output). Make sub new dump its | |
410 | debugging output to STDERR, to be consistent with everything else. | |
411 | All bugs found while writing a regression test. | |
412 | ||
e3d6de9a JH |
413 | September, 2002; by Jarkko Hietaniemi: add ':hireswallclock' special tag. |
414 | ||
523cc92b | 415 | =cut |
a0d0e21e | 416 | |
3f943bd9 | 417 | # evaluate something in a clean lexical environment |
53aa2791 | 418 | sub _doeval { no strict; eval shift } |
3f943bd9 GS |
419 | |
420 | # | |
421 | # put any lexicals at file scope AFTER here | |
422 | # | |
423 | ||
4aa0a1f7 | 424 | use Carp; |
a0d0e21e | 425 | use Exporter; |
53aa2791 MS |
426 | |
427 | our(@ISA, @EXPORT, @EXPORT_OK, %EXPORT_TAGS, $VERSION); | |
428 | ||
429 | @ISA=qw(Exporter); | |
ac8eabc1 JH |
430 | @EXPORT=qw(timeit timethis timethese timediff timestr); |
431 | @EXPORT_OK=qw(timesum cmpthese countit | |
432 | clearcache clearallcache disablecache enablecache); | |
f36484b0 | 433 | %EXPORT_TAGS=( all => [ @EXPORT, @EXPORT_OK ] ) ; |
a0d0e21e | 434 | |
53aa2791 | 435 | $VERSION = 1.051; |
e3d6de9a JH |
436 | |
437 | # --- ':hireswallclock' special handling | |
438 | ||
439 | my $hirestime; | |
440 | ||
441 | sub mytime () { time } | |
8a4f6ac2 | 442 | |
359218de | 443 | init(); |
a0d0e21e | 444 | |
e3d6de9a JH |
445 | sub BEGIN { |
446 | if (eval 'require Time::HiRes') { | |
447 | import Time::HiRes qw(time); | |
448 | $hirestime = \&Time::HiRes::time; | |
449 | } | |
450 | } | |
451 | ||
452 | sub import { | |
453 | my $class = shift; | |
454 | if (grep { $_ eq ":hireswallclock" } @_) { | |
455 | @_ = grep { $_ ne ":hireswallclock" } @_; | |
456 | *mytime = $hirestime if defined $hirestime; | |
457 | } | |
458 | Benchmark->export_to_level(1, $class, @_); | |
459 | } | |
460 | ||
53aa2791 MS |
461 | our($Debug, $Min_Count, $Min_CPU, $Default_Format, $Default_Style, |
462 | %_Usage, %Cache, $Do_Cache); | |
463 | ||
a0d0e21e | 464 | sub init { |
53aa2791 MS |
465 | $Debug = 0; |
466 | $Min_Count = 4; | |
467 | $Min_CPU = 0.4; | |
468 | $Default_Format = '5.2f'; | |
469 | $Default_Style = 'auto'; | |
a0d0e21e LW |
470 | # The cache can cause a slight loss of sys time accuracy. If a |
471 | # user does many tests (>10) with *very* large counts (>10000) | |
472 | # or works on a very slow machine the cache may be useful. | |
359218de JH |
473 | disablecache(); |
474 | clearallcache(); | |
a0d0e21e LW |
475 | } |
476 | ||
53aa2791 MS |
477 | sub debug { $Debug = ($_[1] != 0); } |
478 | ||
479 | sub usage { | |
480 | my $calling_sub = (caller(1))[3]; | |
481 | $calling_sub =~ s/^Benchmark:://; | |
482 | return $_Usage{$calling_sub} || ''; | |
483 | } | |
484 | ||
bba8fca5 | 485 | # The cache needs two branches: 's' for strings and 'c' for code. The |
359218de | 486 | # empty loop is different in these two cases. |
53aa2791 | 487 | |
359218de | 488 | sub clearcache ($) { |
53aa2791 MS |
489 | delete $Cache{"$_[0]c"}; delete $Cache{"$_[0]s"}; |
490 | } | |
491 | ||
359218de | 492 | sub clearallcache () { |
53aa2791 MS |
493 | %Cache = (); |
494 | } | |
495 | ||
359218de | 496 | sub enablecache () { |
53aa2791 MS |
497 | $Do_Cache = 1; |
498 | } | |
499 | ||
359218de | 500 | sub disablecache () { |
53aa2791 MS |
501 | $Do_Cache = 0; |
502 | } | |
503 | ||
a0d0e21e | 504 | |
a0d0e21e LW |
505 | # --- Functions to process the 'time' data type |
506 | ||
e3d6de9a | 507 | sub new { my @t = (mytime, times, @_ == 2 ? $_[1] : 0); |
53aa2791 | 508 | print STDERR "new=@t\n" if $Debug; |
6ee623d5 | 509 | bless \@t; } |
a0d0e21e LW |
510 | |
511 | sub cpu_p { my($r,$pu,$ps,$cu,$cs) = @{$_[0]}; $pu+$ps ; } | |
512 | sub cpu_c { my($r,$pu,$ps,$cu,$cs) = @{$_[0]}; $cu+$cs ; } | |
513 | sub cpu_a { my($r,$pu,$ps,$cu,$cs) = @{$_[0]}; $pu+$ps+$cu+$cs ; } | |
514 | sub real { my($r,$pu,$ps,$cu,$cs) = @{$_[0]}; $r ; } | |
431d98c2 | 515 | sub iters { $_[0]->[5] ; } |
a0d0e21e | 516 | |
53aa2791 MS |
517 | |
518 | $_Usage{timediff} = <<'USAGE'; | |
519 | usage: $result_diff = timediff($result1, $result2); | |
520 | USAGE | |
521 | ||
523cc92b | 522 | sub timediff { |
a0d0e21e | 523 | my($a, $b) = @_; |
53aa2791 MS |
524 | |
525 | die usage unless ref $a and ref $b; | |
526 | ||
523cc92b | 527 | my @r; |
3f943bd9 | 528 | for (my $i=0; $i < @$a; ++$i) { |
a0d0e21e LW |
529 | push(@r, $a->[$i] - $b->[$i]); |
530 | } | |
531 | bless \@r; | |
532 | } | |
533 | ||
53aa2791 MS |
534 | $_Usage{timesum} = <<'USAGE'; |
535 | usage: $sum = timesum($result1, $result2); | |
536 | USAGE | |
537 | ||
705cc255 | 538 | sub timesum { |
53aa2791 MS |
539 | my($a, $b) = @_; |
540 | ||
541 | die usage unless ref $a and ref $b; | |
542 | ||
543 | my @r; | |
544 | for (my $i=0; $i < @$a; ++$i) { | |
705cc255 | 545 | push(@r, $a->[$i] + $b->[$i]); |
53aa2791 MS |
546 | } |
547 | bless \@r; | |
705cc255 TB |
548 | } |
549 | ||
53aa2791 MS |
550 | |
551 | $_Usage{timestr} = <<'USAGE'; | |
552 | usage: $formatted_result = timestr($result1); | |
553 | USAGE | |
554 | ||
523cc92b | 555 | sub timestr { |
a0d0e21e | 556 | my($tr, $style, $f) = @_; |
53aa2791 MS |
557 | |
558 | die usage unless ref $tr; | |
559 | ||
523cc92b | 560 | my @t = @$tr; |
6ee623d5 GS |
561 | warn "bad time value (@t)" unless @t==6; |
562 | my($r, $pu, $ps, $cu, $cs, $n) = @t; | |
ce9550df | 563 | my($pt, $ct, $tt) = ($tr->cpu_p, $tr->cpu_c, $tr->cpu_a); |
53aa2791 | 564 | $f = $Default_Format unless defined $f; |
a0d0e21e | 565 | # format a time in the required style, other formats may be added here |
53aa2791 | 566 | $style ||= $Default_Style; |
0e74ff8e | 567 | return '' if $style eq 'none'; |
523cc92b CS |
568 | $style = ($ct>0) ? 'all' : 'noc' if $style eq 'auto'; |
569 | my $s = "@t $style"; # default for unknown style | |
e3d6de9a JH |
570 | my $w = $hirestime ? "%2g" : "%2d"; |
571 | $s=sprintf("$w wallclock secs (%$f usr %$f sys + %$f cusr %$f csys = %$f CPU)", | |
ce9550df | 572 | $r,$pu,$ps,$cu,$cs,$tt) if $style eq 'all'; |
e3d6de9a | 573 | $s=sprintf("$w wallclock secs (%$f usr + %$f sys = %$f CPU)", |
7be077a2 | 574 | $r,$pu,$ps,$pt) if $style eq 'noc'; |
e3d6de9a | 575 | $s=sprintf("$w wallclock secs (%$f cusr + %$f csys = %$f CPU)", |
7be077a2 | 576 | $r,$cu,$cs,$ct) if $style eq 'nop'; |
cc31225e | 577 | $s .= sprintf(" @ %$f/s (n=$n)", $n / ( $pu + $ps )) if $n && $pu+$ps; |
a0d0e21e LW |
578 | $s; |
579 | } | |
523cc92b CS |
580 | |
581 | sub timedebug { | |
a0d0e21e | 582 | my($msg, $t) = @_; |
53aa2791 | 583 | print STDERR "$msg",timestr($t),"\n" if $Debug; |
a0d0e21e LW |
584 | } |
585 | ||
a0d0e21e LW |
586 | # --- Functions implementing low-level support for timing loops |
587 | ||
53aa2791 MS |
588 | $_Usage{runloop} = <<'USAGE'; |
589 | usage: runloop($number, [$string | $coderef]) | |
590 | USAGE | |
591 | ||
a0d0e21e LW |
592 | sub runloop { |
593 | my($n, $c) = @_; | |
4aa0a1f7 AD |
594 | |
595 | $n+=0; # force numeric now, so garbage won't creep into the eval | |
523cc92b | 596 | croak "negative loopcount $n" if $n<0; |
53aa2791 | 597 | confess usage unless defined $c; |
a0d0e21e LW |
598 | my($t0, $t1, $td); # before, after, difference |
599 | ||
600 | # find package of caller so we can execute code there | |
523cc92b CS |
601 | my($curpack) = caller(0); |
602 | my($i, $pack)= 0; | |
a0d0e21e LW |
603 | while (($pack) = caller(++$i)) { |
604 | last if $pack ne $curpack; | |
605 | } | |
606 | ||
3f943bd9 GS |
607 | my ($subcode, $subref); |
608 | if (ref $c eq 'CODE') { | |
609 | $subcode = "sub { for (1 .. $n) { local \$_; package $pack; &\$c; } }"; | |
610 | $subref = eval $subcode; | |
611 | } | |
612 | else { | |
613 | $subcode = "sub { for (1 .. $n) { local \$_; package $pack; $c;} }"; | |
614 | $subref = _doeval($subcode); | |
615 | } | |
4aa0a1f7 | 616 | croak "runloop unable to compile '$c': $@\ncode: $subcode\n" if $@; |
53aa2791 | 617 | print STDERR "runloop $n '$subcode'\n" if $Debug; |
a0d0e21e | 618 | |
3c6312e9 BS |
619 | # Wait for the user timer to tick. This makes the error range more like |
620 | # -0.01, +0. If we don't wait, then it's more like -0.01, +0.01. This | |
621 | # may not seem important, but it significantly reduces the chances of | |
622 | # getting a too low initial $n in the initial, 'find the minimum' loop | |
431d98c2 | 623 | # in &countit. This, in turn, can reduce the number of calls to |
bba8fca5 BS |
624 | # &runloop a lot, and thus reduce additive errors. |
625 | my $tbase = Benchmark->new(0)->[1]; | |
277427cf | 626 | while ( ( $t0 = Benchmark->new(0) )->[1] == $tbase ) {} ; |
a0d0e21e | 627 | &$subref; |
6ee623d5 | 628 | $t1 = Benchmark->new($n); |
a0d0e21e | 629 | $td = &timediff($t1, $t0); |
a0d0e21e LW |
630 | timedebug("runloop:",$td); |
631 | $td; | |
632 | } | |
633 | ||
53aa2791 MS |
634 | $_Usage{timeit} = <<'USAGE'; |
635 | usage: $result = timeit($count, 'code' ); or | |
636 | $result = timeit($count, sub { code } ); | |
637 | USAGE | |
a0d0e21e LW |
638 | |
639 | sub timeit { | |
640 | my($n, $code) = @_; | |
641 | my($wn, $wc, $wd); | |
642 | ||
53aa2791 MS |
643 | die usage unless defined $code and |
644 | (!ref $code or ref $code eq 'CODE'); | |
645 | ||
646 | printf STDERR "timeit $n $code\n" if $Debug; | |
3c6312e9 | 647 | my $cache_key = $n . ( ref( $code ) ? 'c' : 's' ); |
53aa2791 MS |
648 | if ($Do_Cache && exists $Cache{$cache_key} ) { |
649 | $wn = $Cache{$cache_key}; | |
523cc92b | 650 | } else { |
6bf773bc | 651 | $wn = &runloop($n, ref( $code ) ? sub { } : '' ); |
3c6312e9 BS |
652 | # Can't let our baseline have any iterations, or they get subtracted |
653 | # out of the result. | |
654 | $wn->[5] = 0; | |
53aa2791 | 655 | $Cache{$cache_key} = $wn; |
a0d0e21e LW |
656 | } |
657 | ||
658 | $wc = &runloop($n, $code); | |
659 | ||
660 | $wd = timediff($wc, $wn); | |
a0d0e21e LW |
661 | timedebug("timeit: ",$wc); |
662 | timedebug(" - ",$wn); | |
663 | timedebug(" = ",$wd); | |
664 | ||
665 | $wd; | |
666 | } | |
667 | ||
6ee623d5 GS |
668 | |
669 | my $default_for = 3; | |
670 | my $min_for = 0.1; | |
671 | ||
3c6312e9 | 672 | |
53aa2791 MS |
673 | $_Usage{countit} = <<'USAGE'; |
674 | usage: $result = countit($time, 'code' ); or | |
675 | $result = countit($time, sub { code } ); | |
676 | USAGE | |
677 | ||
431d98c2 BS |
678 | sub countit { |
679 | my ( $tmax, $code ) = @_; | |
6ee623d5 | 680 | |
53aa2791 MS |
681 | die usage unless @_; |
682 | ||
6ee623d5 GS |
683 | if ( not defined $tmax or $tmax == 0 ) { |
684 | $tmax = $default_for; | |
685 | } elsif ( $tmax < 0 ) { | |
686 | $tmax = -$tmax; | |
687 | } | |
688 | ||
431d98c2 | 689 | die "countit($tmax, ...): timelimit cannot be less than $min_for.\n" |
6ee623d5 GS |
690 | if $tmax < $min_for; |
691 | ||
3c6312e9 | 692 | my ($n, $tc); |
6ee623d5 | 693 | |
bba8fca5 | 694 | # First find the minimum $n that gives a significant timing. |
3c6312e9 BS |
695 | for ($n = 1; ; $n *= 2 ) { |
696 | my $td = timeit($n, $code); | |
697 | $tc = $td->[1] + $td->[2]; | |
698 | last if $tc > 0.1; | |
699 | } | |
6ee623d5 | 700 | |
3c6312e9 BS |
701 | my $nmin = $n; |
702 | ||
703 | # Get $n high enough that we can guess the final $n with some accuracy. | |
704 | my $tpra = 0.1 * $tmax; # Target/time practice. | |
705 | while ( $tc < $tpra ) { | |
706 | # The 5% fudge is to keep us from iterating again all | |
707 | # that often (this speeds overall responsiveness when $tmax is big | |
708 | # and we guess a little low). This does not noticably affect | |
709 | # accuracy since we're not couting these times. | |
710 | $n = int( $tpra * 1.05 * $n / $tc ); # Linear approximation. | |
711 | my $td = timeit($n, $code); | |
c5d57293 A |
712 | my $new_tc = $td->[1] + $td->[2]; |
713 | # Make sure we are making progress. | |
714 | $tc = $new_tc > 1.2 * $tc ? $new_tc : 1.2 * $tc; | |
6ee623d5 GS |
715 | } |
716 | ||
3c6312e9 BS |
717 | # Now, do the 'for real' timing(s), repeating until we exceed |
718 | # the max. | |
719 | my $ntot = 0; | |
720 | my $rtot = 0; | |
721 | my $utot = 0.0; | |
722 | my $stot = 0.0; | |
723 | my $cutot = 0.0; | |
724 | my $cstot = 0.0; | |
725 | my $ttot = 0.0; | |
726 | ||
727 | # The 5% fudge is because $n is often a few % low even for routines | |
728 | # with stable times and avoiding extra timeit()s is nice for | |
729 | # accuracy's sake. | |
730 | $n = int( $n * ( 1.05 * $tmax / $tc ) ); | |
731 | ||
732 | while () { | |
733 | my $td = timeit($n, $code); | |
734 | $ntot += $n; | |
735 | $rtot += $td->[0]; | |
736 | $utot += $td->[1]; | |
737 | $stot += $td->[2]; | |
6ee623d5 GS |
738 | $cutot += $td->[3]; |
739 | $cstot += $td->[4]; | |
3c6312e9 BS |
740 | $ttot = $utot + $stot; |
741 | last if $ttot >= $tmax; | |
6ee623d5 | 742 | |
c5d57293 | 743 | $ttot = 0.01 if $ttot < 0.01; |
3c6312e9 | 744 | my $r = $tmax / $ttot - 1; # Linear approximation. |
bba8fca5 | 745 | $n = int( $r * $ntot ); |
6ee623d5 | 746 | $n = $nmin if $n < $nmin; |
6ee623d5 GS |
747 | } |
748 | ||
749 | return bless [ $rtot, $utot, $stot, $cutot, $cstot, $ntot ]; | |
750 | } | |
751 | ||
a0d0e21e LW |
752 | # --- Functions implementing high-level time-then-print utilities |
753 | ||
6ee623d5 GS |
754 | sub n_to_for { |
755 | my $n = shift; | |
756 | return $n == 0 ? $default_for : $n < 0 ? -$n : undef; | |
757 | } | |
758 | ||
53aa2791 MS |
759 | $_Usage{timethis} = <<'USAGE'; |
760 | usage: $result = timethis($time, 'code' ); or | |
761 | $result = timethis($time, sub { code } ); | |
762 | USAGE | |
763 | ||
a0d0e21e LW |
764 | sub timethis{ |
765 | my($n, $code, $title, $style) = @_; | |
53aa2791 MS |
766 | my($t, $forn); |
767 | ||
768 | die usage unless defined $code and | |
769 | (!ref $code or ref $code eq 'CODE'); | |
6ee623d5 GS |
770 | |
771 | if ( $n > 0 ) { | |
772 | croak "non-integer loopcount $n, stopped" if int($n)<$n; | |
773 | $t = timeit($n, $code); | |
774 | $title = "timethis $n" unless defined $title; | |
775 | } else { | |
53aa2791 | 776 | my $fort = n_to_for( $n ); |
431d98c2 | 777 | $t = countit( $fort, $code ); |
6ee623d5 GS |
778 | $title = "timethis for $fort" unless defined $title; |
779 | $forn = $t->[-1]; | |
780 | } | |
523cc92b | 781 | local $| = 1; |
523cc92b | 782 | $style = "" unless defined $style; |
3c6312e9 | 783 | printf("%10s: ", $title) unless $style eq 'none'; |
53aa2791 | 784 | print timestr($t, $style, $Default_Format),"\n" unless $style eq 'none'; |
6ee623d5 GS |
785 | |
786 | $n = $forn if defined $forn; | |
523cc92b | 787 | |
a0d0e21e LW |
788 | # A conservative warning to spot very silly tests. |
789 | # Don't assume that your benchmark is ok simply because | |
790 | # you don't get this warning! | |
791 | print " (warning: too few iterations for a reliable count)\n" | |
53aa2791 | 792 | if $n < $Min_Count |
a0d0e21e | 793 | || ($t->real < 1 && $n < 1000) |
53aa2791 | 794 | || $t->cpu_a < $Min_CPU; |
a0d0e21e LW |
795 | $t; |
796 | } | |
797 | ||
53aa2791 MS |
798 | |
799 | $_Usage{timethese} = <<'USAGE'; | |
800 | usage: timethese($count, { Name1 => 'code1', ... }); or | |
801 | timethese($count, { Name1 => sub { code1 }, ... }); | |
802 | USAGE | |
803 | ||
a0d0e21e LW |
804 | sub timethese{ |
805 | my($n, $alt, $style) = @_; | |
53aa2791 MS |
806 | die usage unless ref $alt eq 'HASH'; |
807 | ||
523cc92b CS |
808 | my @names = sort keys %$alt; |
809 | $style = "" unless defined $style; | |
3c6312e9 | 810 | print "Benchmark: " unless $style eq 'none'; |
6ee623d5 GS |
811 | if ( $n > 0 ) { |
812 | croak "non-integer loopcount $n, stopped" if int($n)<$n; | |
3c6312e9 | 813 | print "timing $n iterations of" unless $style eq 'none'; |
6ee623d5 | 814 | } else { |
3c6312e9 | 815 | print "running" unless $style eq 'none'; |
6ee623d5 | 816 | } |
3c6312e9 | 817 | print " ", join(', ',@names) unless $style eq 'none'; |
6ee623d5 GS |
818 | unless ( $n > 0 ) { |
819 | my $for = n_to_for( $n ); | |
df7779cf T |
820 | print ", each" if $n > 1 && $style ne 'none'; |
821 | print " for at least $for CPU seconds" unless $style eq 'none'; | |
6ee623d5 | 822 | } |
3c6312e9 | 823 | print "...\n" unless $style eq 'none'; |
523cc92b CS |
824 | |
825 | # we could save the results in an array and produce a summary here | |
a0d0e21e | 826 | # sum, min, max, avg etc etc |
3c6312e9 | 827 | my %results; |
4dbb2df9 | 828 | foreach my $name (@names) { |
3c6312e9 | 829 | $results{$name} = timethis ($n, $alt -> {$name}, $name, $style); |
4dbb2df9 | 830 | } |
3c6312e9 BS |
831 | |
832 | return \%results; | |
a0d0e21e LW |
833 | } |
834 | ||
53aa2791 MS |
835 | |
836 | $_Usage{cmpthese} = <<'USAGE'; | |
837 | usage: cmpthese($count, { Name1 => 'code1', ... }); or | |
838 | cmpthese($count, { Name1 => sub { code1 }, ... }); or | |
839 | cmpthese($result, $style); | |
840 | USAGE | |
841 | ||
3c6312e9 | 842 | sub cmpthese{ |
53aa2791 MS |
843 | my ($results, $style); |
844 | ||
845 | if( ref $_[0] ) { | |
846 | ($results, $style) = @_; | |
847 | } | |
848 | else { | |
849 | my($count, $code) = @_[0,1]; | |
850 | $style = $_[2] if defined $_[2]; | |
851 | ||
852 | die usage unless ref $code eq 'HASH'; | |
853 | ||
854 | $results = timethese($count, $code, ($style || "none")); | |
855 | } | |
3c6312e9 | 856 | |
d1083c7a | 857 | $style = "" unless defined $style; |
3c6312e9 BS |
858 | |
859 | # Flatten in to an array of arrays with the name as the first field | |
860 | my @vals = map{ [ $_, @{$results->{$_}} ] } keys %$results; | |
861 | ||
862 | for (@vals) { | |
863 | # The epsilon fudge here is to prevent div by 0. Since clock | |
864 | # resolutions are much larger, it's below the noise floor. | |
865 | my $rate = $_->[6] / ( $_->[2] + $_->[3] + 0.000000000000001 ); | |
866 | $_->[7] = $rate; | |
867 | } | |
868 | ||
869 | # Sort by rate | |
870 | @vals = sort { $a->[7] <=> $b->[7] } @vals; | |
871 | ||
872 | # If more than half of the rates are greater than one... | |
873 | my $display_as_rate = $vals[$#vals>>1]->[7] > 1; | |
874 | ||
875 | my @rows; | |
876 | my @col_widths; | |
877 | ||
878 | my @top_row = ( | |
879 | '', | |
880 | $display_as_rate ? 'Rate' : 's/iter', | |
881 | map { $_->[0] } @vals | |
882 | ); | |
883 | ||
884 | push @rows, \@top_row; | |
885 | @col_widths = map { length( $_ ) } @top_row; | |
886 | ||
887 | # Build the data rows | |
888 | # We leave the last column in even though it never has any data. Perhaps | |
889 | # it should go away. Also, perhaps a style for a single column of | |
890 | # percentages might be nice. | |
891 | for my $row_val ( @vals ) { | |
892 | my @row; | |
893 | ||
894 | # Column 0 = test name | |
895 | push @row, $row_val->[0]; | |
896 | $col_widths[0] = length( $row_val->[0] ) | |
897 | if length( $row_val->[0] ) > $col_widths[0]; | |
898 | ||
899 | # Column 1 = performance | |
900 | my $row_rate = $row_val->[7]; | |
901 | ||
902 | # We assume that we'll never get a 0 rate. | |
53aa2791 | 903 | my $rate = $display_as_rate ? $row_rate : 1 / $row_rate; |
3c6312e9 BS |
904 | |
905 | # Only give a few decimal places before switching to sci. notation, | |
906 | # since the results aren't usually that accurate anyway. | |
907 | my $format = | |
53aa2791 | 908 | $rate >= 100 ? |
3c6312e9 | 909 | "%0.0f" : |
53aa2791 | 910 | $rate >= 10 ? |
3c6312e9 | 911 | "%0.1f" : |
53aa2791 | 912 | $rate >= 1 ? |
3c6312e9 | 913 | "%0.2f" : |
53aa2791 | 914 | $rate >= 0.1 ? |
3c6312e9 BS |
915 | "%0.3f" : |
916 | "%0.2e"; | |
917 | ||
918 | $format .= "/s" | |
919 | if $display_as_rate; | |
53aa2791 MS |
920 | |
921 | my $formatted_rate = sprintf( $format, $rate ); | |
922 | push @row, $formatted_rate; | |
923 | $col_widths[1] = length( $formatted_rate ) | |
924 | if length( $formatted_rate ) > $col_widths[1]; | |
3c6312e9 BS |
925 | |
926 | # Columns 2..N = performance ratios | |
927 | my $skip_rest = 0; | |
928 | for ( my $col_num = 0 ; $col_num < @vals ; ++$col_num ) { | |
929 | my $col_val = $vals[$col_num]; | |
930 | my $out; | |
931 | if ( $skip_rest ) { | |
932 | $out = ''; | |
933 | } | |
934 | elsif ( $col_val->[0] eq $row_val->[0] ) { | |
935 | $out = "--"; | |
936 | # $skip_rest = 1; | |
937 | } | |
938 | else { | |
939 | my $col_rate = $col_val->[7]; | |
940 | $out = sprintf( "%.0f%%", 100*$row_rate/$col_rate - 100 ); | |
941 | } | |
942 | push @row, $out; | |
943 | $col_widths[$col_num+2] = length( $out ) | |
944 | if length( $out ) > $col_widths[$col_num+2]; | |
945 | ||
946 | # A little wierdness to set the first column width properly | |
947 | $col_widths[$col_num+2] = length( $col_val->[0] ) | |
948 | if length( $col_val->[0] ) > $col_widths[$col_num+2]; | |
949 | } | |
950 | push @rows, \@row; | |
951 | } | |
952 | ||
d1083c7a BS |
953 | return \@rows if $style eq "none"; |
954 | ||
3c6312e9 BS |
955 | # Equalize column widths in the chart as much as possible without |
956 | # exceeding 80 characters. This does not use or affect cols 0 or 1. | |
957 | my @sorted_width_refs = | |
958 | sort { $$a <=> $$b } map { \$_ } @col_widths[2..$#col_widths]; | |
959 | my $max_width = ${$sorted_width_refs[-1]}; | |
960 | ||
277427cf | 961 | my $total = @col_widths - 1 ; |
3c6312e9 BS |
962 | for ( @col_widths ) { $total += $_ } |
963 | ||
964 | STRETCHER: | |
965 | while ( $total < 80 ) { | |
966 | my $min_width = ${$sorted_width_refs[0]}; | |
967 | last | |
968 | if $min_width == $max_width; | |
969 | for ( @sorted_width_refs ) { | |
970 | last | |
971 | if $$_ > $min_width; | |
972 | ++$$_; | |
973 | ++$total; | |
974 | last STRETCHER | |
975 | if $total >= 80; | |
976 | } | |
977 | } | |
978 | ||
979 | # Dump the output | |
980 | my $format = join( ' ', map { "%${_}s" } @col_widths ) . "\n"; | |
981 | substr( $format, 1, 0 ) = '-'; | |
982 | for ( @rows ) { | |
983 | printf $format, @$_; | |
984 | } | |
985 | ||
d1083c7a | 986 | return \@rows ; |
3c6312e9 BS |
987 | } |
988 | ||
989 | ||
a0d0e21e | 990 | 1; |