From: Sean O'Halpin <sean.ohalpin@...>
Date: 2011-08-08T06:08:14+09:00
Subject: Re: Methods defined with module_eval are... faster?

Hi,

When I run this using Benchmark.bmbm (to account for GC) I don't see
any appreciable difference between the two methods.

I used this code with ruby 1.9.2:

TIMES=4000000
Benchmark.bmbm {|x| x.report("M.kk") { TIMES.times { M.kk(123) ; M.kk(4) } } }
Benchmark.bmbm {|x| x.report("M.aa") { TIMES.times { M.aa(123) ; M.aa(4) } } }
Benchmark.bmbm {|x| x.report("M.kk") { TIMES.times { M.kk(123) ; M.kk(4) } } }
Benchmark.bmbm {|x| x.report("M.aa") { TIMES.times { M.aa(123) ; M.aa(4) } } }

and got these results:

M.kk   3.850000   0.010000   3.860000 (  4.149469)
M.aa   3.960000   0.010000   3.970000 (  3.995387)
M.kk   3.900000   0.020000   3.920000 (  4.154616)
M.aa   3.900000   0.010000   3.910000 (  4.201047)

I then re-ran the original code but changed the order of the tests. I
got these results:

Original order:
M.kk: 3.75673246383667
M.aa: 3.9386367797851562
M.kk: 3.531486988067627
M.aa: 3.7896509170532227

Swapped:
M.aa: 4.207374095916748
M.kk: 4.187262296676636
M.aa: 3.915318727493286
M.kk: 4.421187162399292

Apart from showing that I have a ~much~ slower machine than yours, I
don't see anything significant in these results.
I think what you're seeing is an artefact of garbage collection.

Regards,
Sean