Polynomial-Time Subgraph Enumeration for Automated Instruction Set Extension

This paper proposes a novel algorithm that, given a data-flow graph and an input/output constraint, enumerates all convex subgraphs under the given constraint in polynomial time with respect to the size of the graph. These subgraphs have been shown to represent efficient instruction set extensions for customizable processors. The search space for this problem is inherently polynomial but, this is the first paper to prove this and to present a practical algorithm for this problem with polynomial complexity. The algorithm is based on properties of convex subgraphs that link them to the concept of multiple-vertex dominators. The paper discussed several pruning techniques that, without sacrificing the optimality of the algorithm, make it practical for data-flow graphs of a thousands nodes or more