Sylvan: Multi-Core Decision Diagrams

Decision diagrams such as binary decision diagrams and multi-valued decision diagrams play an important role in various fields, including symbolic model checking. An ongoing challenge is to develop datastructures and algorithms for modern multi-core architectures. The BDD package Sylvan provides one contribution by implementing parallelized BDD operations and thus allowing sequential algorithms to exploit the power of multi-core machines.

We present several extensions to Sylvan. We implement parallel operations on list decision diagrams, a variant of multi-valued decision diagrams that is useful for symbolic model checking. We also substitute several core components of Sylvan by new designs, such as the work-stealing framework, the unique table and the operation cache. Furthermore, we combine parallel operations with parallelization on a higher level, by partitioning the transition relation. We show that this results in an improved speedup using the model checking toolset

ltsmin

. We also demonstrate that the parallelization of symbolic model checking for explicit-state modeling languages with an on-the-fly next-state function, as supported by

ltsmin

, scales well.