On Sat, 11 Oct 2008 12:14:23 +1100, SASADA Koichi <ko1 / atdot.net> wrote:

> I committed such an optimization last night :)   please check it.

Hi,

That's fantastic - the performances of "for" and "each" are now quite  
similar, and as I suspected, for a very small loop like this benchmark,  
"for" is just marginally slower due to the extra assignment of the dynamic  
variable to the local.

Now:
ruby 1.9.0 (2008-10-11 revision 19756) [i686-linux]
                           user     system      total        real
for                   2.490000   0.000000   2.490000 (  2.508887)
each                  2.370000   0.000000   2.370000 (  2.381762)
times                 2.380000   0.010000   2.390000 (  2.420313)

Was:
ruby 1.9.0 (2008-10-10 revision 19741) [i686-linux]
                           user     system      total        real
for                   6.030000   0.020000   6.050000 (  6.074545)
each                  2.380000   0.000000   2.380000 (  2.384460)
times                 2.380000   0.000000   2.380000 (  2.394605)

Much better :)

Thanks very much
Mike