Optimal static scheduling of real-time dependent tasks on reconfigurable hardware devices

Abstract

Reconfigurable hardware devices are increasingly being adopted in the design of emerging complex embedded systems since they offer novel reconfiguration capabilities and a high density of heterogeneous resources conducting to improve the system performance. Key issues of designing such systems are interconnecting, memory management and task placement and scheduling. This paper addresses the problem of placement and scheduling of real-time DAG-structured applications in these emergent reconfigurable hardware devices. This paper presents a scheduling/placement methodology that uses mixed integer programming to establish the optimal static scenario for executing a set of periodic real-time inter-dependent tasks on reconfigurable devices. Experiments are conducted on a DAG composed of heterogeneous hardware tasks to evaluate the placement/scheduling quality. The results demonstrate a resource gain of 27 % achieved by the run-time reconfiguration concept compared to a static design. The configuration overhead is reduced to 1 % of the total running time. Due to pipelined scheduling, the task graph spanning is minimized by 4 % compared to sequential execution of the graph and the waiting time is reduced up to 7 % of the overall running time.