Heuristic approaches to energy optimization in deep submicron bus design through QAP formulation

Abstract

As modern microprocessors become increasingly interconnected, energy dissipation within digital circuits has emerged as a critical challenge, particularly in deep submicron (DSM) technologies where parasitic effects are significant. This paper addresses the energy reduction problem in DSM bus design by formulating it as a Quadratic Assignment Problem (QAP), an NP-hard combinatorial optimization problem. By leveraging this framework, optimal wire permutations that minimize energy dissipation are systematically pursued.Given the intractability of solving large-scale QAPs exactly, we evaluate heuristic algorithms for their effectiveness in approximating near-optimal solutions within a feasible timeframe. The study compares the performance of various heuristic approaches. Results demonstrate that certain methods achieve rapid convergence and significant energy reduction, proving their suitability for real-time applications. The analysis highlights the potential of heuristic methods as practical solutions for complex energy optimization problems in DSM bus systems.