From: Robert Klemme <shortcutter@...>
Date: 2010-02-06T22:40:09+09:00
Subject: Re: Fastest way to iterate over a range in steps

On 02/05/2010 09:19 PM, Intransition wrote:
> 
> On Feb 5, 2:52 pm, Robert Klemme <shortcut...@googlemail.com> wrote:
> 
>> I think we are seeing an effect of the missing call to the block that
>> happens on each iteration for #step approaches.  The while loop does not
>> have this overhead.
> 
> Might consider how #succ plays into this as well.

Good idea!

robert@fussel:~$ ruby19 xx.rb
Rehearsal --------------------------------------------------
while           15.550000   0.060000  15.610000 ( 22.225998)
succ            66.930000   0.340000  67.270000 ( 93.726408)
range step      31.930000   0.080000  32.010000 ( 44.892588)
step            30.650000   0.140000  30.790000 ( 43.183693)
--------------------------------------- total: 145.680000sec

                      user     system      total        real
while           15.430000   0.080000  15.510000 ( 21.889805)
succ            67.170000   0.440000  67.610000 ( 95.958702)
range step      32.090000   0.160000  32.250000 ( 44.213130)
step            30.670000   0.110000  30.780000 ( 43.151260)
robert@fussel:~$ cat xx.rb

require 'benchmark'

LI = 1000000000
ST = 5

Benchmark.bmbm 15 do |b|
   b.report "while" do
     i = 0
     while i < LI
       i += ST
     end
   end

   b.report "succ" do
     i = 0
     while i < LI
       i = i.succ
     end
   end

   b.report "range step" do |b|
     (0...LI).step ST do
     end
   end

   b.report "step" do |b|
     0.step LI, ST do
     end
   end
end
robert@fussel:~$


... or rather not. :-)

Kind regards

	robert

-- 
remember.guy do |as, often| as.you_can - without end
http://blog.rubybestpractices.com/