From: Robert Klemme <bob.news@...>
Date: 2004-09-15T17:34:54+09:00
Subject: Re: Groovy speeds

------=_NextPart_000_003C_01C49B0F.86C5A110
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: 7bit


"Jim Weirich" <jim@weirichhouse.org> schrieb im Newsbeitrag
news:4146E923.2050601@weirichhouse.org...
> > Not exactly but there is Groovy - if you need the looks of Ruby and
the
> > speed of Java.  See http://groovy.codehaus.org/
>
> I've seen several posts that hint that Groovy runs at Java-like speeds.
>   I haven't found this to be true at all.  You can find one of my (very
> incomplete and probably totally useless) benchmarks here:
> http://onestepback.org/articles/groovy/groovyspeed.html

That's interesting, although I think you didn't choose optimal
implementations:

10:16:21 [source]: ruby /c/temp/ruby/fact-bm.rb
Ruby:   54.1 MicroSeconds per call
Ruby:   14.7 MicroSeconds per call
Ruby:   13.5 MicroSeconds per call

and

10:33:25 [groovy]: groovy fact.gr
Groovy: 67.2 MicroSeconds per call
Groovy: 59.4 MicroSeconds per call (HotSpot)
Groovy: 57.9 MicroSeconds per call
Groovy: 57.8 MicroSeconds per call (HotSpot)
Groovy: 57.8 MicroSeconds per call
Groovy: 57.8 MicroSeconds per call (HotSpot)

Darn, I didn't expect groovy to be *that* slow.

Kind regards

    robert

------=_NextPart_000_003C_01C49B0F.86C5A110
Content-Type: application/octet-stream;
	name="fact.rb"
Content-Transfer-Encoding: 7bit
Content-Disposition: attachment;
	filename="fact.rb"


def fact1(n)
  n == 1 ? 1 : n*fact1(n-1)
end

def fact2(n)
  f = 1

  for i in 1..n
    f *= i
  end

  f
end

require 'benchmark'
include Benchmark

N=700

bm do |x|
  x.report("recursive") do
    for n in 1..N
      fact1 n
    end
  end

  x.report("iterative") do
    for n in 1..N
      fact2 n
    end
  end
end

------=_NextPart_000_003C_01C49B0F.86C5A110
Content-Type: application/octet-stream;
	name="fact.gr"
Content-Transfer-Encoding: quoted-printable
Content-Disposition: attachment;
	filename="fact.gr"

REPEAT =3D 10000=0A=
=0A=
benchmark =3D { n, closure |=0A=
    start =3D System.currentTimeMillis()=0A=
    /* n.times { closure.call } */=0A=
    for ( i in 1 .. n ) { closure.call }=0A=
    delta =3D System.currentTimeMillis() - start=0A=
    ((float)delta) / n=0A=
}=0A=
=0A=
def fact(n) {=0A=
  (2..n).inject(1) { f, i | f * i }=0A=
}=0A=
=0A=
t =3D benchmark(REPEAT) { fact(10) }=0A=
println "Groovy: ${1000*t} MicroSeconds per call"=0A=
=0A=
t =3D benchmark(REPEAT) { fact(10) }=0A=
println "Groovy: ${1000*t} MicroSeconds per call (HotSpot)"=0A=
=0A=
def fact2(n) {=0A=
  f =3D 1=0A=
  for ( i in 2..n )=0A=
    f *=3D i=0A=
  end=0A=
  f=0A=
}=0A=
=0A=
t =3D benchmark(REPEAT) { fact2(10) }=0A=
println "Groovy: ${1000*t} MicroSeconds per call"=0A=
=0A=
t =3D benchmark(REPEAT) { fact2(10) }=0A=
println "Groovy: ${1000*t} MicroSeconds per call (HotSpot)"=0A=
=0A=
def fact3(n) {=0A=
  f =3D n=0A=
  while ( n !=3D 1 ) {=0A=
    f *=3D (n -=3D 1)=0A=
  }=0A=
  f=0A=
}=0A=
=0A=
t =3D benchmark(REPEAT) { fact3(10) }=0A=
println "Groovy: ${1000*t} MicroSeconds per call"=0A=
=0A=
t =3D benchmark(REPEAT) { fact3(10) }=0A=
println "Groovy: ${1000*t} MicroSeconds per call (HotSpot)"=0A=

------=_NextPart_000_003C_01C49B0F.86C5A110--