Efficient Scheduling of Synchronous Dataflow Graphs on Communication Cost-Aware Heterogeneous Platforms
TimeWednesday, July 13th6pm - 7pm PDT
LocationLevel 2 Lobby
Event Type
Networking Reception
Work-in-Progress Poster
DescriptionSynchronous Dataflow Graphs (SDFGs) have been widely used to model and analyze streaming applications, which are often executed on heterogeneous platforms and with a high processing speed requirement. In this paper, we propose an efficient method for scheduling SDFGs on a communication-aware heterogeneous platform to achieve low latency. We further take into account the energy consumption of applications, trying to provide a trade-off between latency and energy used. Our experiments are conducted on 17 benchmark streaming applications. The results show that our algorithm outperforms the state-of-the-art work and is tens of times faster.