Adaptive Optimization in the Jikes RVM

by Michael Hind

IBM T.J. Watson Research Center

This talk will present an overview of the Jikes Research Virtual Machine (RVM), an open source VM for Java, formerly called Jalapeņo. The Jikes RVM provides a flexible open testbed to prototype new virtual machine technologies. The system includes state-of-the-art techniques for dynamic compilation, adaptive optimization, garbage collection, thread scheduling, and synchronization. The system is implemented in the Java programming language and is self-hosted i.e., its Java code runs on itself without requiring a second virtual machine. The system runs on Linux/IA-32, AIX/PowerPC, and Linux/PowerPC platforms.

In the 16 months since its open source release, the system has been downloaded by over 90 universities around the world. Academic researchers using the system have published 18 papers in leading conferences in 2002. The system has been used in at least 7 university courses.

This presentation will highlight the various components of the Jikes RVM, and focus on results from the adaptive optimization system, a system that uses an online cost/benefit model to determine which methods should be optimized and at what optimization level.

Further information on the Jikes RVM is available at http://www.ibm.com/developerworks/oss/jikesrvm