ABSTRACT
This chapter provides a background on relevant dataflow concepts and lightweight dataflow (LWDF) programming, and present LWDF-based case studies that demonstrate effective use of dataflow techniques for image processing systems. It focuses on case studies that demonstrate the integration of dataflow programming structures with C, Verilog, and Compute Unified Device Architecture for simulation, Field programmable gate array mapping, and Graphics processing unit implementation, respectively. Dataflow is a well-known computational model and is widely used for expressing the functionality of digital signal processing (DSP) applications, such as audio and video data stream processing, digital communications, and image processing. Synchronous dataflow is a specialized form of dataflow that is useful for an important class of DSP applications, and is used in a variety of commercial design tools. A simulation approach is therefore useful to orthogonalize functional validation of an LWDF design before exploring platform-specific scheduling optimizations—for example, optimizations that exploit the parallelism exposed by the given dataflow representation.
