Jochen Theodorou schrieb: [...]

I did some tests using

java -version java version "1.7.0-ea" Java(TM) SE Runtime Environment (build 1.7.0-ea-b24) Java HotSpot(TM) 64-Bit Server VM (build 12.0-b01, mixed mode)

uname -a Linux darkangel 2.6.24-21-generic #1 SMP Mon Aug 25 16:57:51 UTC 2008 x86_64 GNU/Linux

CPU is a Q6600

anyway... I give now first the command line, then the time before my commit then the time after my commit and then the percentage-wise change (new/old)-1. So + means new is faster, - means the new version is slower. If the command line starts with time, then I sued the command line tool time to make the measurement. Each test was run 10 times, removing the best and worst measurement.

target/install/bin/groovy ../alioth/binarytrees.groovy 12 1202ms 1181ms +2% target/install/bin/groovy ../alioth/fannkuch.groovy 436ms 442ms -1% target/install/bin/groovy ../alioth/partialsums.groovy 10299 330ms 348ms -5% time target/install/bin/groovy ../alioth/recursive.groovy 6 3024ms 2980ms +1% time target/install/bin/groovy ../alioth/ray 10 10 3706ms 3787ms -2% target/install/bin/groovy ../alioth/spectralnorm.groovy 559ms 564ms -1%

that means within a range of 1-2% old and new are at the same speed with the exception of the recusive test which has 5%. I think these 1-2 are well in the measurement range, probably I only had bad luck to get the 5%.

Frankly that is much better than I expected, because especially when the test takes more memory I expected the call back reference manager to kick in, which can have a bad performance. but it seems not to be the case.

bye blackdrag