[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [stem-dev] 64 Bit STEM
- From: Stefan Edlund <edlund@xxxxxxxxxxxxxxx>
- Date: Fri, 3 Apr 2009 15:35:57 -0700
- Delivered-to: firstname.lastname@example.org
Cool, how many concurrent worker threads did you use to run the simulation? I think the default is 2 if you didn't change it.
On my x3500 64 bit Ubuntu quad core 16 GB memory running the ant script takes 1:58 minutes too. Pretty good :)
Public Health and Computer Science Research
IBM Almaden Research Center
(408) 927-1766 edlund@xxxxxxxxxxxxxxx
Daniel Ford ---04/03/2009 03:00:52 PM---I've been experimenting with running STEM on a 64 bit machine with the 64 bit version of Java 6 and various Heap sizes and garb
Daniel Ford <daford@xxxxxxxxxxxxxxx>
Sent by: stem-dev-bounces@xxxxxxxxxxx
04/03/2009 03:00 PM
Please respond to
STEM developer mailing list <stem-dev@xxxxxxxxxxx>
I've been experimenting with running STEM on a 64 bit machine with the 64 bit version of Java 6 and various Heap sizes and garbage collection options.
This is running on a Mac Pro with 2 x 2.8 GHz Quad-Core Intel Xeon processors and 8 GB of RAM.
I configured it to run with up to 6GB of Heap and ran the built-in USA Level 0,1,2 Scenario starting in Florida.
I configured for three different GC options, default, Parallel and concurrent Mark and Sweep.
Summary: No surprise, it was perceptively faster, with crude stop watch measurements of multiple cycles of less than 500 milliseconds per cycle. The different GC options didn't seem to make too much difference, but no more than 2 GB's of the heap was used for the first two, and roughly 500 megabytes for the later (M&S). This was with the delay preference switched off.
As a matter of interest, running the update.xml script for the entire project on the 64 bit Java 6 JRE takes just over two minutes to complete. On my (older) windows machine (dual processors, 4GB, 32bit Java 1.5) running XP, the same operation takes over 18 minutes.
stem-dev mailing list