Jackal is a Java DSM: its goals are to run unmodified (multithreaded)
Java programs EFFICIENTLY on a cluster of workstations.
Our approach is to use extensive compiler analysis and optimization in combination with supporting runtime optimizations.
Jackal builds upon the manta project.
As such we use a fast network (myrinet) and a native compiler (.java to assembler in one go).
Related projects are
Manta (fast RMI, native compilation, dynamic byte code compilation)
Treadmarks (de facto page based DSM), CVM, etc
Shasta (instruments existing alpha executables for DSM execution)
Blizzard/Sirocco (like shasta, although with a range of different implementation techniques and for SPARC machines)
Javagrande 2001: Runtime
Optimizations for a Java DSM implementation
PPoPP 2001: Jackal: A Compiler-Supported Distributed Shared Memory implementation of Java
HPCN 2001: Optimizing Java specific overheads, Java at the speed of C ?
Work in progress: Parallel Garbage Collection and Memory Management for a Java DSM
Work in progress: Data Prefeching Techniques for a Java DSM
Work in progress: On demand access check generation
Work in progress: My Thesis