Design Space Exploration using Answer Set Programming

Abstract

This paper proposes a design space exploration DSE method which integrates an application in dataflow and an architecture with satisfying a given latency constraint. In order to meet the latency, the allocation of tasks onto processing elements which is called mapping and the order of execution of the tasks called scheduling should be determined in the design space exploration. Since the problem is NP complete and new design factors that will be available should be considered, a flexible DSE algorithm platform is required. In this paper, we propose a DSE algorithm platform that determines mapping and scheduling statically using answer set programming (ASP). In order to solve the DSE formulation rapidly, we propose formulation optimization techniques in ASP. Experimental results show that the solving time for an optimized formulation is orders of magnitude faster than a non-optimized one.