Time a function: Difference between revisions
No edit summary |
No edit summary |
||
Line 779: | Line 779: | ||
call silly |
call silly |
||
say 'function SILLY took' format(time("E"),,2) 'seconds.' |
say 'function SILLY took' format(time("E"),,2) 'seconds.' |
||
/* The above 2 for the FORMAT function displays the time */ |
|||
/* with 2 decimal digits (past the decimal point). Using */ |
|||
/* a 0 (zero) would round the time to whole seconds. */ |
|||
exit |
exit |
||
silly: procedure /*subroutine chews up CPU time doing silly stuff*/ |
silly: procedure /*subroutine chews up CPU time doing silly stuff*/ |
Revision as of 08:13, 31 October 2010
You are encouraged to solve this task according to the task description, using any language you may know.
Write a program which uses a timer (with the least granularity available on your system) to time how long a function takes to execute.
Whenever possible, use methods which measure only the processing time used by the current process; instead of the difference in system time between start and finish, which could include time used by other processes on the computer.
This task is intended as a subtask for Measure relative performance of sorting algorithms implementations.
Ada
<lang ada>with Ada.Calendar; use Ada.Calendar; with Ada.Text_Io; use Ada.Text_Io;
procedure Query_Performance is
type Proc_Access is access procedure(X : in out Integer); function Time_It(Action : Proc_Access; Arg : Integer) return Duration is Start_Time : Time := Clock; Finis_Time : Time; Func_Arg : Integer := Arg; begin Action(Func_Arg); Finis_Time := Clock; return Finis_Time - Start_Time; end Time_It; procedure Identity(X : in out Integer) is begin X := X; end Identity; procedure Sum (Num : in out Integer) is begin for I in 1..1000 loop Num := Num + I; end loop; end Sum; Id_Access : Proc_Access := Identity'access; Sum_Access : Proc_Access := Sum'access;
begin
Put_Line("Identity(4) takes" & Duration'Image(Time_It(Id_Access, 4)) & " seconds."); Put_Line("Sum(4) takes:" & Duration'Image(Time_It(Sum_Access, 4)) & " seconds.");
end Query_Performance;</lang>
Example
Identity(4) takes 0.000001117 seconds. Sum(4) takes: 0.000003632 seconds.
AutoHotkey
System time
Uses system time, not process time <lang AutoHotkey>MsgBox % time("fx") Return
fx() {
Sleep, 1000
}
time(function, parameter=0) {
SetBatchLines -1 ; don't sleep for other green threads StartTime := A_TickCount %function%(parameter) Return ElapsedTime := A_TickCount - StartTime . " milliseconds"
}</lang>
C
On some system (like GNU/Linux) to be able to use the clock_gettime function you must link with the rt (RealTime) library.
<lang c>#include <stdio.h>
- include <time.h>
int identity(int x) { return x; }
int sum(int s) {
int i; for(i=0; i < 1000000; i++) s += i; return s;
}
- define CLOCKTYPE CLOCK_MONOTONIC
/* this one should be appropriate to avoid errors on multiprocessors systems */
double time_it(int (*action)(int), int arg) {
struct timespec tsi, tsf;
clock_gettime(CLOCKTYPE, &tsi); action(arg); clock_gettime(CLOCKTYPE, &tsf);
double elaps_s = difftime(tsf.tv_sec, tsi.tv_sec); long elaps_ns = tsf.tv_nsec - tsi.tv_nsec; return elaps_s + ((double)elaps_ns) / 1.0e9;
}
int main() {
printf("identity (4) takes %lf s\n", time_it(identity, 4)); printf("sum (4) takes %lf s\n", time_it(sum, 4)); return 0;
}</lang>
C++
<lang cpp>#include <ctime>
- include <iostream>
using namespace std;
int identity(int x) { return x; } int sum(int num) {
for (int i = 0; i < 1000000; i++) num += i; return num;
}
double time_it(int (*action)(int), int arg) {
clock_t start_time = clock(); action(arg); clock_t finis_time = clock(); return ((double) (finis_time - start_time)) / CLOCKS_PER_SEC;
}
int main() {
cout << "Identity(4) takes " << time_it(identity, 4) << " seconds." << endl; cout << "Sum(4) takes " << time_it(sum, 4) << " seconds." << endl; return 0;
}</lang>
Example
Identity(4) takes 0 seconds. Sum(4) takes 0.01 seconds.
C#
Using Stopwatch.
<lang csharp>using System; using System.Linq; using System.Threading; using System.Diagnostics;
class Program {
static void Main(string[] args) { Stopwatch sw = new Stopwatch();
sw.Start(); DoSomething(); sw.Stop();
Console.WriteLine("DoSomething() took {0}ms.", sw.Elapsed.TotalMilliseconds); }
static void DoSomething() { Thread.Sleep(1000);
Enumerable.Range(1, 10000).Where(x => x % 2 == 0).Sum(); // Sum even numers from 1 to 10000 }
}</lang>
Using DateTime.
<lang csharp>using System; using System.Linq; using System.Threading;
class Program {
static void Main(string[] args) { DateTime start, end;
start = DateTime.Now; DoSomething(); end = DateTime.Now;
Console.WriteLine("DoSomething() took " + (end - start).TotalMilliseconds + "ms"); }
static void DoSomething() { Thread.Sleep(1000);
Enumerable.Range(1, 10000).Where(x => x % 2 == 0).Sum(); // Sum even numers from 1 to 10000 }
}</lang>
Output:
DoSomething() took 1071,5408ms
Common Lisp
Common Lisp provides a standard utility for performance measurement, time:
<lang lisp>CL-USER> (time (reduce #'+ (make-list 100000 :initial-element 1))) Evaluation took:
0.151 seconds of real time 0.019035 seconds of user run time 0.01807 seconds of system run time 0 calls to %EVAL 0 page faults and 2,400,256 bytes consed.</lang>
(The example output here is from SBCL.)
However, it merely prints textual information to trace output, so the information is not readily available for further processing (except by parsing it in a CL-implementation-specific manner).
The functions get-internal-run-time and get-internal-real-time may be used to get time information programmatically, with at least one-second granularity (and usually more). Here is a function which uses them to measure the time taken for one execution of a provided function:
<lang lisp>(defun timings (function)
(let ((real-base (get-internal-real-time)) (run-base (get-internal-run-time))) (funcall function) (values (/ (- (get-internal-real-time) real-base) internal-time-units-per-second) (/ (- (get-internal-run-time) run-base) internal-time-units-per-second))))
CL-USER> (timings (lambda () (reduce #'+ (make-list 100000 :initial-element 1)))) 17/500 7/250</lang>
D
<lang D> import tango.io.Stdout; import tango.time.Clock;
int identity (int x) {
return x;
}
int sum (int num) {
for (int i = 0; i < 1000000; i++) num += i; return num;
}
double timeIt(int function(int) func, int arg) {
long before = Clock.now.ticks; func(arg); return (Clock.now.ticks - before) / cast(double)TimeSpan.TicksPerSecond;
}
void main () {
Stdout.format("Identity(4) takes {:f6} seconds",timeIt(&identity,4)).newline; Stdout.format("Sum(4) takes {:f6} seconds",timeIt(&sum,4)).newline;
} </lang>
E
— E has no standardized facility for CPU time measurement; this
.
<lang e>def countTo(x) { println("Counting...") for _ in 1..x {} println("Done!") }
def MX := <unsafe:java.lang.management.makeManagementFactory> def threadMX := MX.getThreadMXBean() require(threadMX.isCurrentThreadCpuTimeSupported()) threadMX.setThreadCpuTimeEnabled(true)
for count in [10000, 100000] { def start := threadMX.getCurrentThreadCpuTime() countTo(count) def finish := threadMX.getCurrentThreadCpuTime() println(`Counting to $count takes ${(finish-start)//1000000}ms`) }</lang>
Factor
<lang factor>[ 10000 iota sum drop ] time</lang> Output:
Running time: 0.002888635 seconds Additional information was collected. dispatch-stats. - Print method dispatch statistics gc-events. - Print all garbage collection events gc-stats. - Print breakdown of different garbage collection events gc-summary. - Print aggregate garbage collection statistics
Forth
<lang forth>: time: ( "word" -- )
utime 2>R ' EXECUTE utime 2R> D- <# # # # # # # [CHAR] . HOLD #S #> TYPE ." seconds" ;
1000 time: MS \ 1.000081 seconds ok</lang>
Haskell
<lang haskell>import System.CPUTime
-- We assume the function we are timing is an IO monad computation timeIt :: (Fractional c) => (a -> IO b) -> a -> IO c timeIt action arg =
do startTime <- getCPUTime action arg finishTime <- getCPUTime return $ fromIntegral (finishTime - startTime) / 1000000000000
-- Version for use with evaluating regular non-monadic functions timeIt' :: (Fractional c) => (a -> b) -> a -> IO c timeIt' f = timeIt (\x -> f x `seq` return ())</lang>
Example
*Main> :m + Text.Printf Data.List *Main Data.List Text.Printf> timeIt' id 4 >>= printf "Identity(4) takes %f seconds.\n" Identity(4) takes 0.0 seconds. *Main Data.List Text.Printf> timeIt' (\x -> foldl' (+) x [1..1000000]) 4 >>= printf "Sum(4) takes %f seconds.\n" Sum(4) takes 0.248015 seconds.
HicEst
<lang HicEst>t_start = TIME() ! returns seconds since midnight SYSTEM(WAIT = 1234) ! wait 1234 milliseconds t_end = TIME()
WRITE(StatusBar) t_end - t_start, " seconds"</lang>
Ioke
<lang ioke>use("benchmark")
func = method((1..50000) reduce(+))
Benchmark report(1, 1, func)</lang>
Icon and Unicon
Icon
The function 'timef' takes as argument a procedure name and collects performance and timing information including run time (in milliseconds), garbage collection, and memory usage by region.
<lang Icon>procedure timef(f) #: time a function f local gcol,alloc,used,size,runtime,header,x,i
title := ["","total","static","string","block"] # headings collect() # start with collected memory (before baseline) every put(gcol := [], -&collections) # baseline collections count every put(alloc := [], -&allocated) # . total allocated space by region every put(used := [], -&storage) # . currently used space by region - no total every put(size := [], -®ions) # . current size of regions - no total
write("Performance and Timing measurement for ",image(f),":") runtime := &time # base time f() write("Execution time=",&time-runtime," ms.")
every (i := 0, x := &collections) do gcol[i +:= 1] +:= x every (i := 0, x := &allocated ) do alloc[i +:= 1] +:= x every (i := 0, x := &storage ) do used[i +:= 1] +:= x every (i := 0, x := ®ions ) do size[i +:= 1] +:= x
push(gcol,"garbage collections:") push(alloc,"memory allocated:") push(used,"N/A","currently used:") push(size,"N/A","current size:")
write("Memory Region and Garbage Collection Summary (delta):") every (i := 0) <:= *!(title|gcol|alloc|used|size) every x := (title|gcol|alloc|used|size) do {
f := left every writes(f(!x,i + 3)) do f := right write() }
write("Note: static region values should be zero and may not be meaningful.") return end</lang>
Sample usage:<lang Icon>procedure main() timef(perfectnumbers) end
procedure perfectnumbers() ...</lang>
Sample output (from the Perfect Numbers task):
Performance and Timing measurement for procedure perfectnumbers: Perfect numbers from 1 to 10000: 6 28 496 8128 Done. Execution time=416 ms. Memory Region and Garbage Collection Summary (delta): total static string block garbage collections: 2 0 0 2 memory allocated: 1247012 0 24 1246988 currently used: N/A 0 0 248040 current size: N/A 0 0 0 Note: static region values should be zero and may not be meaningful.
Unicon
The Icon solution works in Unicon.
J
Time and space requirements are tested using verbs obtained through the Foreign conjunction (!:). 6!:2 returns time required for execution, in floating-point measurement of seconds. 7!:2 returns a measurement of space required to execute. Both receive as input a sentence for execution.
When the Memoize feature or similar techniques are used, execution time and space can both be affected by prior calculations.
Example
<lang j> (6!:2,7!:2) '|: 50 50 50 $ i. 50^3' 0.00387912 1.57414e6</lang>
Java
<lang java>import java.lang.management.ManagementFactory; import java.lang.management.ThreadMXBean;
public class TimeIt { public static void main(String[] args) { final ThreadMXBean threadMX = ManagementFactory.getThreadMXBean(); assert threadMX.isCurrentThreadCpuTimeSupported(); threadMX.setThreadCpuTimeEnabled(true);
long start, end; start = threadMX.getCurrentThreadCpuTime(); countTo(100000000); end = threadMX.getCurrentThreadCpuTime(); System.out.println("Counting to 100000000 takes "+(end-start)/1000000+"ms"); start = threadMX.getCurrentThreadCpuTime(); countTo(1000000000L); end = threadMX.getCurrentThreadCpuTime(); System.out.println("Counting to 1000000000 takes "+(end-start)/1000000+"ms");
}
public static void countTo(long x){ System.out.println("Counting..."); for(long i=0;i<x;i++); System.out.println("Done!"); } }</lang>
Measures real time rather than CPU time:
<lang java> public static void main(String[] args){ long start, end; start = System.currentTimeMillis(); countTo(100000000); end = System.currentTimeMillis(); System.out.println("Counting to 100000000 takes "+(end-start)+"ms"); start = System.currentTimeMillis(); countTo(1000000000L); end = System.currentTimeMillis(); System.out.println("Counting to 1000000000 takes "+(end-start)+"ms");
}</lang> Output:
Counting... Done! Counting to 100000000 takes 370ms Counting... Done! Counting to 1000000000 takes 3391ms
Lua
<lang lua>function Test_Function()
for i = 1, 10000000 do local s = math.log( i ) s = math.sqrt( s ) end
end
t1 = os.clock()
Test_Function()
t2 = os.clock()
print( os.difftime( t2, t1 ) )</lang>
Logo
on a Unix system
This is not an ideal method; Logo does not expose a timer (except for the WAIT command) so we use the Unix "date" command to get a second timer.
<lang logo>to time
output first first shell "|date +%s|
end to elapsed :block
localmake "start time run :block (print time - :start [seconds elapsed])
end
elapsed [wait 300] ; 5 seconds elapsed</lang>
Mathematica
<lang Mathematica>AbsoluteTiming[x];</lang> where x is an operation. Example calculating a million digits of Sqrt[3]: <lang Mathematica>AbsoluteTiming[N[Sqrt[3], 10^6]]</lang> gives: <lang Mathematica>{0.000657, 1.7320508075688772935274463......}</lang> First elements if the time in seconds, second elements if the result from the operation. Note that I truncated the result.
OCaml
<lang ocaml>let time_it action arg =
let start_time = Sys.time () in ignore (action arg); let finish_time = Sys.time () in finish_time -. start_time</lang>
Example
# Printf.printf "Identity(4) takes %f seconds.\n" (time_it (fun x -> x) 4);; Identity(4) takes 0.000000 seconds. - : unit = () # let sum x = let num = ref x in for i = 0 to 999999 do num := !num + i done; !num;; val sum : int -> int = <fun> # Printf.printf "Sum(4) takes %f seconds.\n" (time_it sum 4);; Sum(4) takes 0.084005 seconds. - : unit = ()
Oz
<lang oz>declare
%% returns milliseconds fun {TimeIt Proc} Before = {Now} in {Proc} {Now} - Before end
fun {Now} {Property.get 'time.total'} end
in
{Show {TimeIt proc {$} {FoldL {List.number 1 1000000 1} Number.'+' 4 _} end} }</lang>
Perl
Example of using the built-in Benchmark core module - it compares two versions of recursive factorial functions: <lang perl>use Benchmark; use Memoize;
sub fac1 {
my $n = shift; return $n == 0 ? 1 : $n * fac1($n - 1);
} sub fac2 {
my $n = shift; return $n == 0 ? 1 : $n * fac2($n - 1);
} memoize('fac2');
my $result = timethese(100000, {
'fac1' => sub { fac1(50) }, 'fac2' => sub { fac2(50) },
}); Benchmark::cmpthese($result);</lang> Output:
Benchmark: timing 100000 iterations of fac1, fac2... fac1: 6 wallclock secs ( 5.45 usr + 0.00 sys = 5.45 CPU) @ 18348.62/s (n=100000) fac2: 1 wallclock secs ( 0.84 usr + 0.00 sys = 0.84 CPU) @ 119047.62/s (n=100000) Rate fac1 fac2 fac1 18349/s -- -85% fac2 119048/s 549% --
Example without using Benchmark: <lang perl>sub cpu_time {
my ($user,$system,$cuser,$csystem) = times; $user + $system
}
sub time_it {
my $action = shift; my $startTime = cpu_time(); $action->(@_); my $finishTime = cpu_time(); $finishTime - $startTime
}
printf "Identity(4) takes %f seconds.\n", time_it(sub {@_}, 4);
- outputs "Identity(4) takes 0.000000 seconds."
sub sum {
my $x = shift; foreach (0 .. 999999) { $x += $_; } $x
}
printf "Sum(4) takes %f seconds.\n", time_it(\&sum, 4);
- outputs "Sum(4) takes 0.280000 seconds."</lang>
PL/I
<lang PL/I> declare (start_time, finish_time) float (18);
start_time = secs();
do i = 1 to 10000000;
/* something to be repeated goes here. */
end; finish_time = secs();
put skip edit ('elapsed time=', finish_time - start_time, ' seconds')
(A, F(10,3), A); /* gives the result to thousandths of a second. */
/* Note: using the SECS function takes into account the clock */ /* going past midnight. */ </lang>
PicoLisp
There is a built-in function 'bench' for that. However, it measures wall-clock time, because for practical purposes the real time needed by a task (including I/O and communication) is more meaningful. There is another function, 'tick', which also measures user time, and is used by the profiling tools. <lang PicoLisp>: (bench (do 1000000 (* 3 4))) 0.080 sec -> 12</lang>
PureBasic
Built in timer
This version uses the built in timer, on Windows it has an accuracy of ~10-15 msec. <lang Purebasic>Procedure Foo(Limit)
Protected i, palindromic, String$ For i=0 To Limit String$=Str(i) If String$=ReverseString(String$) palindromic+1 EndIf Next ProcedureReturn palindromic
EndProcedure
If OpenConsole()
Define Start, Stop, cnt PrintN("Starting timing of a calculation,") PrintN("for this we test how many of 0-1000000 are palindromic.") Start=ElapsedMilliseconds() cnt=Foo(1000000) Stop=ElapsedMilliseconds() PrintN("The function need "+Str(stop-Start)+" msec,") PrintN("and "+Str(cnt)+" are palindromic.") Print("Press ENTER to exit."): Input()
EndIf</lang>
Starting timing of a calculation, for this we test how many of 0-1000000 are palindromic. The function need 577 msec, and 1999 are palindromic. Press ENTER to exit.
Hi-res version
This version uses a hi-res timer, but it is Windows only. <lang PureBasic>If OpenConsole()
Define Timed.f, cnt PrintN("Starting timing of a calculation,") PrintN("for this we test how many of 0-1000000 are palindromic.") ; Dependent on Droopy-library If MeasureHiResIntervalStart() ; Same Foo() as above... cnt=Foo(1000000) Timed=MeasureHiResIntervalStop() EndIf PrintN("The function need "+StrF(Timed*1000,3)+" msec,") PrintN("and "+Str(cnt)+" are palindromic.") Print("Press ENTER to exit."): Input()
EndIf</lang>
Starting timing of a calculation, for this we test how many of 0-1000000 are palindromic. The function need 604.341 msec, and 1999 are palindromic. Press ENTER to exit.
This version still relies on the Windows API but does not make use of any additional libraries.
<lang PureBasic>Procedure.f ticksHQ(reportIfPresent = #False)
Static maxfreq.q Protected T.q If reportIfPresent Or maxfreq = 0 QueryPerformanceFrequency_(@maxfreq) If maxfreq ProcedureReturn 1.0 Else ProcedureReturn 0 EndIf EndIf QueryPerformanceCounter_(@T) ProcedureReturn T / maxfreq ;Result is in milliseconds
EndProcedure
If OpenConsole()
Define timed.f, cnt PrintN("Starting timing of a calculation,") PrintN("for this we test how many of 0-1000000 are palindromic.") ; Dependent on Windows API If ticksHQ(#True) timed = ticksHQ() ;start time ; Same Foo() as above... cnt = Foo(1000000) timed = ticksHQ() - timed ;difference EndIf PrintN("The function need " + StrF(timed * 1000, 3) + " msec,") PrintN("and " + Str(cnt) + " are palindromic.") Print("Press ENTER to exit."): Input()
EndIf</lang> Sample output:
Starting timing of a calculation, for this we test how many of 0-1000000 are palindromic. The function need 174.811 msec, and 1999 are palindromic.
Python
Given function and arguments return a time (in microseconds) it takes to make the call.
Note: There is an overhead in executing a function that does nothing. <lang python>import sys, timeit def usec(function, arguments):
modname, funcname = __name__, function.__name__ timer = timeit.Timer(stmt='%(funcname)s(*args)' % vars(), setup='from %(modname)s import %(funcname)s; args=%(arguments)r' % vars()) try: t, N = 0, 1 while t < 0.2: t = min(timer.repeat(repeat=3, number=N)) N *= 10 microseconds = round(10000000 * t / N, 1) # per loop return microseconds except: timer.print_exc(file=sys.stderr) raise
def nothing(): pass def identity(x): return x</lang>
Example
>>> print usec(nothing, []) 1.7 >>> print usec(identity, [1]) 2.2 >>> print usec(pow, (2, 100)) 3.3 >>> print [usec(qsort, (range(n),)) for n in range(10)] [2.7, 2.8, 31.4, 38.1, 58.0, 76.2, 100.5, 130.0, 149.3, 180.0]
using qsort() from Quicksort. Timings show that the implementation of qsort() has quadratic dependence on sequence length N for already sorted sequences (instead of O(N*log(N)) in average).
R
R has a built-in function, system.time, to calculate this. <lang R># A task foo <- function() {
for(i in 1:10) { mat <- matrix(rnorm(1e6), nrow=1e3) mat^-0.5 }
}
- Time the task
timer <- system.time(foo())
- Extract the processing time
timer["user.self"]</lang> For a breakdown of processing time by function, there is Rprof. <lang R>Rprof() foo() Rprof(NULL) summaryRprof()</lang>
Retro
Retro has a time function returning the current time in seconds. This can be used to build a simple timing function:
<lang Retro>: .runtime ( a- ) time push do time pop - cr putn cr ;
- test 20000 fori putn space nexti ;
&test .runtime</lang>
Finer measurements are not possible with the standard implementation.
REXX
REXX doesn't have a language feature for true CPU time, but it does have a built-in function for elapsed time(s). <lang rexx> /*REXX program to show the elapsed time for a function. */
call time 'E' call silly say 'function SILLY took' format(time("E"),,2) 'seconds.'
/* The above 2 for the FORMAT function displays the time */ /* with 2 decimal digits (past the decimal point). Using */ /* a 0 (zero) would round the time to whole seconds. */
exit
silly: procedure /*subroutine chews up CPU time doing silly stuff*/
do j=1 for 100000 a.j=random() date() time() digits() fuzz() form() xrange() queued() end
return </lang> Output:
function SILLY took 3.54 seconds.
Ruby
Ruby's Benchmark module provides a way to generate nice reports (numbers are in seconds): <lang ruby>require 'benchmark'
Benchmark.bm(8) do |x|
x.report("nothing:") { } x.report("sum:") { (1..1_000_000).inject(4) {|sum, x| sum + x} }
end</lang> Output:
user system total real nothing: 0.000000 0.000000 0.000000 ( 0.000014) sum: 2.700000 0.400000 3.100000 ( 3.258348)
You can get the total time as a number for later processing like this: <lang ruby>Benchmark.measure { whatever }.total</lang>
Scheme
<lang scheme>(time (some-function))</lang>
Standard ML
<lang sml>fun time_it (action, arg) = let
val timer = Timer.startCPUTimer () val _ = action arg val times = Timer.checkCPUTimer timer
in
Time.+ (#usr times, #sys times)
end</lang>
Example
- print ("Identity(4) takes " ^ Time.toString (time_it (fn x => x, 4)) ^ " seconds.\n"); Identity(4) takes 0.000 seconds. val it = () : unit - fun sum (x:IntInf.int) = let fun loop (i, sum) = if i >= 1000000 then sum else loop (i + 1, sum + i) in loop (0, x) end; val sum = fn : IntInf.int -> IntInf.int - print ("Sum(4) takes " ^ Time.toString (time_it (sum, 4)) ^ " seconds.\n"); Sum(4) takes 0.220 seconds. val it = () : unit
Tcl
The Tcl time
command returns the real time elapsed averaged over a number of iterations.
<lang tcl>proc sum_n {n} {
for {set i 1; set sum 0.0} {$i <= $n} {incr i} {set sum [expr {$sum + $i}]} return [expr {wide($sum)}]
}
puts [time {sum_n 1e6} 100] puts [time {} 100]</lang> Results in
163551.0 microseconds per iteration 0.2 microseconds per iteration
UNIX Shell
<lang bash>$ time sleep 1
real 0m1.074s user 0m0.001s</lang>
sys 0m0.006s