Adiar Binary Decision Diagrams in External Memory

2021-04-25

International Conference on Tools and Algorithms for Construction and Analysis of Systems

AI Summary

Large-scale binary decision diagrams (BDDs) often exceed main-memory capacity, rendering their construction and manipulation on external storage inefficient and impractical. Method: This paper introduces the first streaming BDD algorithmic framework fully tailored to the external-memory model. It unifies canonical BDD construction and logical operations as I/O-optimal external-memory computations, integrating delayed evaluation, disk-aware scheduling, structured block I/O, and incremental hashing. Contribution/Results: The framework preserves standard BDD semantics and interfaces while eliminating memory spikes during execution. Experiments demonstrate a 99% reduction in peak memory usage for billion-node-scale tasks. It enables, for the first time, formal verification of ultra-large circuits and symbolic model checking—tasks infeasible for conventional in-memory BDD solvers—thereby significantly expanding the applicability of BDDs in industrial-scale formal verification.

Problem

• Optimize BDD manipulation algorithms
• Reduce main memory usage in BDD operations
• Enable BDD manipulation beyond main memory limits

Innovation

• Iterative I/O-efficient BDD algorithms
• Extended BDD operations via Adiar
• Reduced main memory usage significantly