We present a high-level synthesis algorithm solving the combined scheduling, allocation and binding problem minimizing area under both latency and maximum power per clock-cycle constraints. Our approach eliminates the large power spikes, resulting in an increased battery lifetime, a property of outmost importance for battery powered embedded systems. Our approach extends the partial-clique partitioning algorithm of [3] by introducing power awareness through a heuristic algorithm which bounds the design space to those of power feasible schedules. We have applied our algorithm on a set of data flow graphs and investigated the impact on circuit area when applying different power constraints.