PerformanceThis will be a status page for tracking how Kaffe compares, performance-wise, to other Java virtual machines.Remember, many benchmarks only test a specific task, severely stressing a particular area, such as number crunching or string manipulation. But remember, most real-world Java applications occupy the highest levels of the programming abstraction foodchain. Most are primarily concerned with user interfaces (eg. AWT or HTML generation), leaving the serious number crunching to native libraries and applications running in subprocesses. Applications are often bound by user-interface (waiting for user input) or I/O constraints (waiting for network packets, disk I/O, database queries, etc.), so the dramatic variations between different virtual machines you see in the benchmarks are often not reflected in real world performance. As they say, beware of "bench-marketing". If you know of any benchmarking work or articles, send them to jim@kaffe.org. Eventually, we'll build our own performance testsuite.
Benchmarks
Benchmark Competitions
Here are some benchmarking links that don't involve Kaffe: Here are some links to general literature on the problem:
|
Legal:
Kaffe.org is a an independent, free software community project. The Kaffe virtual machine is a clean room implementation of the Java Virtual Machine from Sun Microsystems, Inc. The Kaffe virtual machine is free software, licensed under the terms of the GNU General Public License. Java is a Trademark of Sun Microsystems, Inc. All copyrights are held by their respective owners, please read and respect the individual licenses on the software distributed from this site. |