From: Tomas Matousek <Tomas.Matousek@...>
Date: 2010-07-01T01:41:23+09:00
Subject: [ruby-core:30977] Re: ironruby vs ruby

Note that Antonio's benchmark compares 64bit IronRuby build against 32bit 1.8.7 MRI and thus favoring MRI. 
Our 32bit build is doing better simply due to the fact that pointers are half in size.

We have adapted versions of some of Antonio's microbenchmarks in our depo that I use for IronRuby perf analysis. The results I get on the current build are below.
As you can see we are slower in some faster in others. We know the reasons and how to speed it up for most of them. We just haven't had time to do so yet. And keep in mind these are just microbenchmarks so they don't really reflect perf thruput of real apps. If you would like to see the exact source code let me know.

C:\Temp>rbr -X:NoAdaptiveCompilation run.rb
./bm_app_factorial.rb
  Ruby 1.9: 2.777/2.777
  IronRuby: 1.342/1.326 -> 2.094x faster
./bm_app_mandelbrot.rb
  Ruby 1.9: 2.246/2.231
  IronRuby: 2.730/2.496 -> 1.119x SLOWER
./bm_fractal.rb
  Ruby 1.9: 6.209/6.334
  IronRuby: 1.451/1.576 -> 4.280x faster
./bm_hilbert_matrix.rb
  Ruby 1.9: 0.016/0.031
  IronRuby: 0.437/0.047 -> 3.000x SLOWER
./bm_knucleotide.rb
  Ruby 1.9: 0.889/0.827
  IronRuby: 1.264/1.123 -> 1.358x SLOWER
./bm_list_part1.rb
  Ruby 1.9: 2.153/1.966
  IronRuby: 3.900/3.838 -> 1.952x SLOWER
./bm_list_part2.rb
  Ruby 1.9: 1.092/1.061
  IronRuby: 2.948/2.636 -> 2.485x SLOWER
./bm_mandelbrot.rb
  Ruby 1.9: 1.934/1.810
  IronRuby: 1.576/1.513 -> 1.196x faster
./bm_nsieve.rb
  Ruby 1.9: 2.933/2.948
  IronRuby: 2.028/1.997 -> 1.469x faster
./bm_nsieve_bits.rb
  Ruby 1.9: 6.084/6.100
  IronRuby: 3.385/3.307 -> 1.840x faster
./bm_primes.rb
  Ruby 1.9: 19.578/19.469
  IronRuby: 29.297/28.923 -> 1.486x SLOWER
./bm_simple_connect.rb
  Ruby 1.9: 19.937/19.734
  IronRuby: 25.179/25.241 -> 1.276x SLOWER
./bm_simple_server.rb
  Ruby 1.9: 3.416/3.401
  IronRuby: 3.104/3.026 -> 1.124x faster
./bm_socket_transfer_1mb.rb
  Ruby 1.9: 2.028/1.950
  IronRuby: 0.374/0.374 -> 5.208x faster
./bm_so_array.rb
  Ruby 1.9: 7.831/7.831
  IronRuby: 5.600/5.538 -> 1.414x faster
./bm_so_matrix.rb
  Ruby 1.9: 2.215/2.215
  IronRuby: 1.576/1.498 -> 1.479x faster
./bm_sudoku.rb
  Ruby 1.9: 8.112/8.112
  IronRuby: 6.427/6.380 -> 1.271x faster
./bm_sum_file.rb
  Ruby 1.9: 0.624/0.640
  IronRuby: 0.546/0.515 -> 1.212x faster

Tomas

-----Original Message-----
From: Luis Lavena [mailto:luislavena@gmail.com] 
Sent: Wednesday, June 30, 2010 7:46 AM
To: ruby-core@ruby-lang.org
Subject: [ruby-core:30976] Re: ironruby vs ruby

On Wed, Jun 30, 2010 at 6:25 AM, C.E. Thornton <admin@hawthorne-press.com> wrote:
> Matz,
>
> �Information Week claims that IronRuby runs two to four times faster 
> than the "main implementation"
> of Ruby ...
>
> �I would really like to know how those tests were run?

That sounds unreal.

Do you have a link to the article?

Antonio Cangiano latest shootout has proven IronRuby is still par to 1.8.7, and in some tests, even slower.

http://programmingzen.com/2010/06/28/the-great-ruby-shootout-windows-edition/

Please provide a link to the article.
--
Luis Lavena
AREA 17
-
Perfection in design is achieved not when there is nothing more to add, but rather when there is nothing more to take away.
Antoine de Saint-Exup�ry