Dataflow-based co-design strategies: the Multi-Dataflow Composer Tool.14/12/16 KB L.T. 1.5Published: Wednesday, 09 November 2016
Dataflow-based co-design strategies: the Multi-Dataflow Composer Tool.14/12/16 KB L.T. 1.5 Dr Francesca Palumbo (IEU, University of Sassari)
Dataflow-based co-design strategies: the Multi-Dataflow Composer Tool.14/12/16 KB L.T. 1.5
Dr Francesca Palumbo (IEU, University of Sassari)
Host: Dr. Kung Kiu Lau
Power reduction in modern embedded systems is a challenging issue exacerbated by their design complexity and heterogeneity, normally colliding with the need of cutting down designer effort to achieve shorter time to market. Reconfigurable Video Coding (RVC) adopts challenge those issues. Dynamic and incremental configuration and reconfiguration are favored by the native formalism modularity, though required the definition of dedicated methodologies and tools (e.g. Orcc, Xronos, etc.) constituting, at the moment, the MPEG-RVC framework.
In this scenario, my research has led to the definition and incremental extension of the Multi-Dataflow Composer (MDC) tool, born within MPEG-RVC studies. The baseline MDC tool addresses automatic generation and management of runtime reconfigurable coarse-grained heterogenous platforms. It combines the dataflow high-level formalism and the coarse-grained reconfigurable approach to deploy multi-functions systems, where flexibility and area minimization are realistically challenged. MDC is capable of assembly any type of reconfigurable platform, without any knowledge of the Functional Units (Fus) granularity/functionality, maximizing resource sharing among the given input dataflow specifications and implementing single-cycle reconfiguration. Different extensions enrich the MDC design environment to provide: high-level profiling and system characterization, power-management and hardware accelerators deployment.