The following routines will be exported into your namespace
if you use the Benchmark module:
timeit(COUNT, CODE)
Arguments: COUNT is the number of times to run the loop, and CODE is
the code to run. CODE may be either a code reference or a string to
be eval'd; either way it will be run in the caller's package.
Returns: a Benchmark object.
timethis ( COUNT, CODE, [ TITLE, [ STYLE ]] )
Time COUNT iterations of CODE. CODE may be a string to eval or a
code reference; either way the CODE will run in the caller's package.
Results will be printed to STDOUT as TITLE followed by the times.
TITLE defaults to ``timethis COUNT'' if none is provided. STYLE
determines the format of the output, as described for timestr() below.
The COUNT can be zero or negative: this means the minimum number of
CPU seconds to run. A zero signifies the default of 3 seconds. For
example to run at least for 10 seconds:
timethis(-10, $code)
or to run two pieces of code tests for at least 3 seconds:
timethese(0, { test1 => '...', test2 => '...'})
CPU seconds is, in UNIX terms, the user time plus the system time of
the process itself, as opposed to the real (wallclock) time and the
time spent by the child processes. Less than 0.1 seconds is not
accepted (-0.01 as the count, for example, will cause a fatal runtime
exception).
Note that the CPU seconds is the minimum time: CPU scheduling and
other operating system factors may complicate the attempt so that a
little bit more time is spent. The benchmark output will, however,
also tell the number of $code runs/second, which should be a more
interesting number than the actually spent seconds.
Returns a Benchmark object.
timethese ( COUNT, CODEHASHREF, [ STYLE ] )
The CODEHASHREF is a reference to a hash containing names as keys
and either a string to eval or a code reference for each value.
For each (KEY, VALUE) pair in the CODEHASHREF, this routine will
call
timethis(COUNT, VALUE, KEY, STYLE)
The routines are called in string comparison order of KEY.
The COUNT can be zero or negative, see timethis().
Returns a hash of Benchmark objects, keyed by name.
timediff ( T1, T2 )
Returns the difference between two Benchmark times as a Benchmark
object suitable for passing to timestr().
timestr ( TIMEDIFF, [ STYLE, [ FORMAT ] ] )
Returns a string that formats the times in the TIMEDIFF object in
the requested STYLE. TIMEDIFF is expected to be a Benchmark object
similar to that returned by timediff().
STYLE can be any of 'all', 'none', 'noc', 'nop' or 'auto'. 'all' shows
each of the 5 times available ('wallclock' time, user time, system time,
user time of children, and system time of children). 'noc' shows all
except the two children times. 'nop' shows only wallclock and the
two children times. 'auto' (the default) will act as 'all' unless
the children times are both zero, in which case it acts as 'noc'.
'none' prevents output.
FORMAT is the printf(3)-style format specifier (without the
leading '%') to use to print the times. It defaults to '5.2f'.
Optional Exports
The following routines will be exported into your namespace
if you specifically ask that they be imported:
clearcache ( COUNT )
Clear the cached time for COUNT rounds of the null loop.
clearallcache ( )
Clear all cached times.
cmpthese ( COUT, CODEHASHREF, [ STYLE ] )
cmpthese ( RESULTSHASHREF )
Optionally calls timethese(), then outputs comparison chart. This
chart is sorted from slowest to fastest, and shows the percent
speed difference between each pair of tests. Can also be passed
the data structure that timethese() returns:
Returns the data structure returned by timethese() (or passed in).
countit(TIME, CODE)
Arguments: TIME is the minimum length of time to run CODE for, and CODE is
the code to run. CODE may be either a code reference or a string to
be eval'd; either way it will be run in the caller's package.
TIME is not negative. countit() will run the loop many times to
calculate the speed of CODE before running it for TIME. The actual
time run for will usually be greater than TIME due to system clock
resolution, so it's best to look at the number of iterations divided
by the times that you are concerned with, not just the iterations.
Returns: a Benchmark object.
disablecache ( )
Disable caching of timings for the null loop. This will force Benchmark
to recalculate these timings for each new piece of code timed.
enablecache ( )
Enable caching of timings for the null loop. The time taken for COUNT
rounds of the null loop will be calculated only once for each
different COUNT used.
timesum ( T1, T2 )
Returns the sum of two Benchmark times as a Benchmark object suitable
for passing to timestr().
NOTES
The data is stored as a list of values from the time and times
functions:
The time of the null loop (a loop with the same
number of rounds but empty loop body) is subtracted
from the time of the real loop.
The null loop times can be cached, the key being the
number of rounds. The caching can be controlled using
calls like these:
clearcache($key);
clearallcache();
disablecache();
enablecache();
Caching is off by default, as it can (usually slightly) decrease
accuracy and does not usually noticably affect runtimes.
EXAMPLES
For example,
use Benchmark;$x=3;cmpthese(-5,{a=>sub{$x*$x},b=>sub{$x**2}})
outputs something like this:
Benchmark: running a, b, each for at least 5 CPU seconds...
a: 10 wallclock secs ( 5.14 usr + 0.13 sys = 5.27 CPU) @ 3835055.60/s (n=20210743)
b: 5 wallclock secs ( 5.41 usr + 0.00 sys = 5.41 CPU) @ 1574944.92/s (n=8520452)
Rate b a
b 1574945/s -- -59%
a 3835056/s 144% --
while
use Benchmark;
$x=3;
$r=timethese(-5,{a=>sub{$x*$x},b=>sub{$x**2}},'none');
cmpthese($r);
outputs something like this:
Rate b a
b 1559428/s -- -62%
a 4152037/s 166% --
INHERITANCE
Benchmark inherits from no other class, except of course
for Exporter.
CAVEATS
Comparing eval'd strings with code references will give you
inaccurate results: a code reference will show a slightly slower
execution time than the equivalent eval'd string.
The real time timing is done using time(2) and
the granularity is therefore only one second.
Short tests may produce negative figures because perl
can appear to take longer to execute the empty loop
than a short test; try:
timethis(100,'1');
The system time of the null loop might be slightly
more than the system time of the loop with the actual
code and therefore the difference might end up being < 0.
March 28th, 1997; by Hugo van der Sanden: added support for code
references and the already documented 'debug' method; revamped
documentation.
April 04-07th, 1997: by Jarkko Hietaniemi, added the run-for-some-time
functionality.
September, 1999; by Barrie Slaymaker: math fixes and accuracy and
efficiency tweaks. Added cmpthese(). A result is now returned from
timethese(). Exposed countit() (was runfor()).