Optimal operation scheduling using resource lower bound estimations

Abstract

Presents an accurate resource lower bound estimation technique which leads to an efficient, and optimal solution of time-constrained as well as hardware resource-constrained scheduling problems in high-level synthesis. A given time or hardware constrained scheduling problem is transformed into a cost ordered sequence of feasible scheduling problems where a solution to this new problem is guaranteed to be an optimal solution to the original problem. Efficiency of the approach is demonstrated on large high-level synthesis benchmarks like the elliptical wave filter and the discrete cosine transform.<>