Sa based power efficient FPGA LUT mapping

  • Authors:
  • Richard Dobson;Kathleen Steinhöfel

  • Affiliations:
  • King's College London, London, United Kingdom;King's College London, London, United Kingdom

  • Venue:
  • Proceedings of the 15th annual conference companion on Genetic and evolutionary computation
  • Year:
  • 2013

Quantified Score

Hi-index 0.00

Visualization

Abstract

Look up Table (LUT) based Field Programmable Gate Arrays (FPGAs) are commonly used in mobile devices due to their efficient signal processing capabilities and flexibility to be reprogrammed in situ. However the mechanisms which enable a FPGA to be re-programmable make it require more power than an Application Specific Integrated Circuit. In this paper we consider the power reduction of a FPGA by optimising the mapping the underlying boolean circuit onto the LUT based FPGA with respect to cumulative switching. We formulate the power minimisation problem as a combinatorial optimisation problem. To tackle this NP hard problem we propose the application of a local search method. Here we introduce a complete a neighborhood function and apply heuristic simulated annealing in conjunction with the objective function from [20] 'cumulative switching'. Our experimental results show a 42.96% average reduction in power consumption compared to SIS based mapping and 27.44% average reduction in power consumption compared to a genetic algorithm.