Abstract

• Relational data plus explicit dependencies allows aggressive parallelization . • Datalog + Negation + Timestamps is a useful parallel programming language . • JStar enables parallelization alternatives without changing the program source. • JStar separates the roles of application and parallelization programmers. This paper introduces the JStar parallel programming language, which is a Java-based declarative language aimed at discouraging sequential programming, encouraging massively parallel programming, and giving the compiler and runtime maximum freedom to try alternative parallelisation strategies. We describe the execution semantics and runtime support of the language, several optimisations and parallelism strategies, with some benchmark results.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call