Comments on: clocking jruby1.1 http://www.matusiak.eu/numerodix/blog/index.php/2008/04/21/clocking-jruby11/ A blog about nothing Tue, 31 Jan 2012 20:14:03 -0500 http://wordpress.org/?v=2.8.6 hourly 1 By: numerodix http://www.matusiak.eu/numerodix/blog/index.php/2008/04/21/clocking-jruby11/#comment-52846 numerodix Wed, 23 Apr 2008 21:29:02 +0000 http://www.matusiak.eu/numerodix/blog/?p=966#comment-52846 Right. I tried to apply a little more rigor and redo the test on slightly more data to get longer runtimes. Bug is filed: http://jira.codehaus.org/browse/JRUBY-2436 Right. I tried to apply a little more rigor and redo the test on slightly more data to get longer runtimes. Bug is filed: http://jira.codehaus.org/browse/JRUBY-2436

]]> By: Charles Oliver Nutter http://www.matusiak.eu/numerodix/blog/index.php/2008/04/21/clocking-jruby11/#comment-52808 Charles Oliver Nutter Wed, 23 Apr 2008 13:53:25 +0000 http://www.matusiak.eu/numerodix/blog/?p=966#comment-52808 URL to report JRuby bugs: http://jira.codehaus.org/browse/JRUBY URL to report JRuby bugs: http://jira.codehaus.org/browse/JRUBY

]]> By: Charles Oliver Nutter http://www.matusiak.eu/numerodix/blog/index.php/2008/04/21/clocking-jruby11/#comment-52806 Charles Oliver Nutter Wed, 23 Apr 2008 13:38:06 +0000 http://www.matusiak.eu/numerodix/blog/?p=966#comment-52806 Looks like you have a potential perf bug for us :) If you can sort out what's actually slower, we'd love to have it reported. My guess is that it might be IO, which we haven't done a lot to optimize. But if you've changed anything since the original results, like tweaked the regex or added a duplicate checker, that could be the culprity too. At any rate, for a run this long, we should be faster. So you found a perf bug. Looks like you have a potential perf bug for us :) If you can sort out what’s actually slower, we’d love to have it reported. My guess is that it might be IO, which we haven’t done a lot to optimize. But if you’ve changed anything since the original results, like tweaked the regex or added a duplicate checker, that could be the culprity too.

At any rate, for a run this long, we should be faster. So you found a perf bug.

]]> By: numerodix http://www.matusiak.eu/numerodix/blog/index.php/2008/04/21/clocking-jruby11/#comment-52801 numerodix Wed, 23 Apr 2008 12:45:46 +0000 http://www.matusiak.eu/numerodix/blog/?p=966#comment-52801 Hello Charlie, just last night I was watching your talk at Fosdem :) I have java version 1.6.0_03. Well I reran the tests and to make it a more controlled environment I dd'ed 1gb of disk data to search in. The results were quite different, and I ran it twice to make sure it wasn't a fluke: jruby -J-server real 3m56.149s user 3m36.662s sys 0m17.017s jruby real 3m47.196s user 3m30.149s sys 0m16.497s ruby real 1m11.586s user 1m6.280s sys 0m3.324s On the upshot, the output was identical this time in all cases. :D Hello Charlie, just last night I was watching your talk at Fosdem :)

I have java version 1.6.0_03.

Well I reran the tests and to make it a more controlled environment I dd’ed 1gb of disk data to search in. The results were quite different, and I ran it twice to make sure it wasn’t a fluke:

jruby -J-server
real 3m56.149s
user 3m36.662s
sys 0m17.017s

jruby
real 3m47.196s
user 3m30.149s
sys 0m16.497s

ruby
real 1m11.586s
user 1m6.280s
sys 0m3.324s

On the upshot, the output was identical this time in all cases. :D

]]> By: Charles Oliver Nutter http://www.matusiak.eu/numerodix/blog/index.php/2008/04/21/clocking-jruby11/#comment-52763 Charles Oliver Nutter Wed, 23 Apr 2008 03:21:26 +0000 http://www.matusiak.eu/numerodix/blog/?p=966#comment-52763 Not bad for JRuby, but run with -J-server and see what sort of numbers you get. I'll wager it will be faster again. Also, what Java version? Java 6 (or prerelease Java 7/OpenJDK builds) is recommended for best performance. Not bad for JRuby, but run with -J-server and see what sort of numbers you get. I’ll wager it will be faster again.

Also, what Java version? Java 6 (or prerelease Java 7/OpenJDK builds) is recommended for best performance.

]]>