From: Michael Selig <michael.selig@...>
Date: 2006-11-09T07:24:17+09:00
Subject: Re: Ruby performanmce improvements

Hi Ed,

Thanks for another benchmark to try. I got around a 30% performance  
increase! Obviously not as good as YARV, but not too shabby for a few days  
work, I'd say. I am hoping that a couple of things I have done can also be  
applied to the YARV version.

FYI, it was run on an Athlon 2600+, under Ubuntu 6.06 Linux (kernel  
2.6.15-27-686). Both versions of ruby were compiled using GCC 4.0.3 with  
"-O3".

Mike.

Standard 1.8.4:

/home/mike> ruby-1.8.4 -v MatrixBenchmark.rb 256
ruby 1.8.4 (2005-12-24) [i686-linux]
/usr/lib/ruby/1.8/mathn.rb:19: warning: method redefined; discarding old  
gcd2
/usr/lib/ruby/1.8/mathn.rb:109: warning: discarding old /
/usr/lib/ruby/1.8/mathn.rb:113: warning: discarding old /
/usr/lib/ruby/1.8/mathn.rb:119: warning: method redefined; discarding old  
inspect
/usr/lib/ruby/1.8/mathn.rb:232: warning: method redefined; discarding old  
sqrt
Matrix of dimension 256 times its inverse = identity? true
  71.370000   0.000000  71.370000 ( 71.441071)

Optimized 1.8.4:
/home/mike> ruby-1.8.4-opt -v MatrixBenchmark.rb 256
ruby 1.8.4-OPT (2005-12-24) [i686-linux]
/usr/lib/ruby/1.8/mathn.rb:19: warning: method redefined; discarding old  
gcd2
/usr/lib/ruby/1.8/mathn.rb:109: warning: discarding old /
/usr/lib/ruby/1.8/mathn.rb:113: warning: discarding old /
/usr/lib/ruby/1.8/mathn.rb:119: warning: method redefined; discarding old  
inspect
/usr/lib/ruby/1.8/mathn.rb:232: warning: method redefined; discarding old  
sqrt
Matrix of dimension 256 times its inverse = identity? true
  49.930000   0.000000  49.930000 ( 49.966208)

On Thu, 09 Nov 2006 02:36:37 +1100, M. Edward (Ed) Borasky  
<znmeb@cesmail.net> wrote:

> Could you run my matrix benchmark against these and post the results  
> here? The benchmark is at
>
> http://rubyforge.org/cgi-bin/viewvc.cgi/MatrixBenchmark/?root=cougar
>
> I got a 4X improvement by running with YARV the last time I tested this.
>
> Also, did you use "gprof"? How do your profiles compare with the ones I  
> posted for the matrix benchmark?