Mini-shootout on Windows

3 views
Skip to first unread message

Antonio Cangiano

unread,
Aug 3, 2009, 1:06:29 PM8/3/09
to ruby-bench...@googlegroups.com
In case you missed it. :)

http://antoniocangiano.com/2009/08/03/performance-of-ironruby-ruby-on-windows/ :)
--
High-Quality Programming Screencasts: http://ThinkCode.TV
Zen and the Art of Programming: http://AntonioCangiano.com
Mathematics is wonderful: http://math-blog.com
Aperiodico di resistenza informatica: http://stacktrace.it/
Author of "Ruby on Rails for Microsoft Developers": http://bit.ly/rorforms

rogerdpack

unread,
Aug 3, 2009, 1:29:41 PM8/3/09
to Ruby Benchmark Suite
> In case you missed it. :)
>
> http://antoniocangiano.com/2009/08/03/performance-of-ironruby-ruby-on...)

Nice. Surprising to see ironruby come up so quickly in speed.

I'll admit I'm with both you and Luis on the mingw issue--mingw is
indeed faster, and indeed is far more obscure and only used by a
handful of developers currently [myself being one].

There are some [not yet released] installers for it on
http://rubyinstaller.org/downloads

Unfortunately many gems don't have precompiled binaries for mingw it
so you'd have to install the devkit, too.
http://programming-gone-awry.blogspot.com/2009/05/ruby-19-one-click-installer.html

Now that I think about it, 1.9.x compiled by mingw is faster than the
(msvc 8) binaries offered at garbagecollect.xxx
I.e. if you redid your benchmarks and used mingw for the 1.9 distro,
you'd get better results for MRI.
I should tell Luis to do a benchmark between all possible windows
builds using the RBS. That would be interesting :)

Also a comparison with jruby would be kind. And rails tests :)

Thanks much.

=r

Antonio Cangiano

unread,
Aug 3, 2009, 1:50:18 PM8/3/09
to ruby-bench...@googlegroups.com
On Mon, Aug 3, 2009 at 1:29 PM, rogerdpack <rogerp...@gmail.com> wrote:
I should tell Luis to do a benchmark between all possible windows
builds using the RBS.  That would be interesting :)

Tomorrow I will publish a comparison between Ruby 1.8 (One-Click Installer), Ruby 1.8 (mingw) and Ruby 1.9 (mingw).

Cheers,
Antonio

rogerdpack

unread,
Aug 3, 2009, 3:02:13 PM8/3/09
to Ruby Benchmark Suite
> Tomorrow I will publish a comparison between Ruby 1.8 (One-Click Installer),
> Ruby 1.8 (mingw) and Ruby 1.9 (mingw).

Nice. Perhaps throwing in 1.9 mswin and/or jruby would be useful, as
well :)

Thanks much!
=r

Antonio Cangiano

unread,
Aug 3, 2009, 3:04:29 PM8/3/09
to ruby-bench...@googlegroups.com

I just discovered that I have a couple of days off. If I can get them to work, I will do "The Great Ruby Shootout (Windows Edition)".

Jim Deville

unread,
Aug 3, 2009, 3:20:45 PM8/3/09
to ruby-bench...@googlegroups.com

Antonio, If I push on the guys to fix the bugs blocking some of those from running, can I get a redo when IronRuby can run at least as many of the tests as 1.9?

 

JD

 

…there is no try

http://blog.jredville.com

Roger Pack

unread,
Aug 3, 2009, 3:24:09 PM8/3/09
to ruby-bench...@googlegroups.com
>> Nice.  Perhaps throwing in 1.9 mswin and/or jruby would be useful, as
>> well :)
>
> I just discovered that I have a couple of days off. If I can get them to
> work, I will do "The Great Ruby Shootout (Windows Edition)".

Nice. One thing might be to not include
say

ironruby mswin
test_x FAIL 200

not include that 200s in the total for mswin, so that the total stats
are a little more comparable.

Another nicety would be to include rails tests :)
GL.
=r

Antonio Cangiano

unread,
Aug 3, 2009, 3:30:58 PM8/3/09
to ruby-bench...@googlegroups.com
On Mon, Aug 3, 2009 at 3:20 PM, Jim Deville <jdev...@microsoft.com> wrote:

Antonio, If I push on the guys to fix the bugs blocking some of those from running, can I get a redo when IronRuby can run at least as many of the tests as 1.9?

You can ask me anytime there is a significant change that will affect performance.

Cheers,
Antonio

Antonio Cangiano

unread,
Aug 3, 2009, 3:32:41 PM8/3/09
to ruby-bench...@googlegroups.com
On Mon, Aug 3, 2009 at 3:24 PM, Roger Pack <roger...@gmail.com> wrote:

           ironruby   mswin
test_x      FAIL       200

not include that 200s in the total for mswin, so that the total stats
are a little more comparable.

If any of the implementations failed a test, that test was not included in the total.

Cheers,
Antonio

Roger Pack

unread,
Aug 3, 2009, 5:58:47 PM8/3/09
to ruby-bench...@googlegroups.com
> If any of the implementations failed a test, that test was not included in
> the total.

Nice.
Re: windows benchmarks: throwing in a cygwin build might be interesting.
I have also heard that rubinius runs on windows, though I've never
attempted such myself.
You could try it by installing a mingw build environment [1] and
attempting a build.
GL.
=r
[1] http://programming-gone-awry.blogspot.com/2009/05/how-to-setup-complete-mingw-msys-build.html

Roger Pack

unread,
Aug 4, 2009, 1:56:50 PM8/4/09
to ruby-bench...@googlegroups.com
> Re: windows benchmarks: throwing in a cygwin build might be interesting.

Here's a more exotic build if you're interested:

[ruby 19 mingw compiled using GCC 4.4 [1]]

http://all.faithpromotingstories.org/ruby19_mingw_44.tgz

GL.
=r
[1] http://www.tdragon.net/recentgcc

Reply all
Reply to author
Forward
0 new messages