14 Aug
2009
14 Aug
'09
8:21 a.m.
Hi Niels,
Cool.
If you come to conclusive results it might be great to creata a "JVM
Tuning" section on
http://platform.xwiki.org/xwiki/bin/view/AdminGuide/Performances with
the result of your findings.
Thanks
-Vincent
On Aug 13, 2009, at 9:24 PM, Niels Mayer wrote:
The latest Sun JDK 6
<http://java.sun.com/javase/6/webnotes/
6u14.html> has
optional performance optimization for threading. Since Xwiki running
in a
web-container is thread-intensive, I'm wondering whether
performance improvements like Escape
Analysis<http://en.wikipedia.org/wiki/Escape_analysis> will
be noticed for Xwiki. How about for its increasing amount of work
happening
out of Groovy scripts?
Much attention has been given by Sun, IBM, BEA and others to
optimize lock
http://java.dzone.com/articles/escape-analysis-java-6-update
<http://java.dzone.com/articles/escape-analysis-java-6-update>
Niels
http://nielsmayer.com
PS: I need to try this experiment again with the non-early access
version
http://nielsmayer.com/xwiki/bin/view/fedora10/64bitJavaAppletIn64BitFirefox…
PPS: in clojure land (will this also apply to Groovy?)
http://www.mail-archive.com/clojure@googlegroups.com/msg17221.html
management and synchronization in their
respective Java 6 virtual
machine
offerings. Features like biased locking, lock coarsening, lock
elision by
escape analysis and adaptive spin locking are all designed to
increase
concurrency by allowing more effective sharing amongst application
threads.
http://www.infoq.com/articles/java-threading-optimizations-p1
http://www.infoq.com/articles/java-threading-optimizations-p2
http://blog.xebia.com/2007/12/21/did-escape-analysis-escape-from-java-6/
http://blog.xebia.com/2009/02/18/jvm-threading-optimizations-
revisited/
http://blog.juma.me.uk/2008/12/17/objects-with-no-allocation-overhead/
<http://blog.xebia.com/2007/12/21/did-escape-analysis-escape-from-java-6/
I'm
getting a very significant performance improvement by adding a
couple
of JVM parameters (using jdk 1.6.0_14). They are: -XX:+DoEscapeAnalysis
-XX:+UseBiasedLocking (I think the -server flag
is required for
those two
flags to do anything).
...
My runtime with n = 5,000,000 goes from ~7.5
seconds to ~4.5 seconds.
...
I actually just confirmed that, on my computer at least, those flags
have
basically no effect on the java version and a big
difference on the
clojure
one.
...
-XX:+AggressiveOpts improves another 5-10%.