Digital signal processing (DSP) and streaming systems are naturally modeled using rigorous, formal models of computation such as synchronous data flow graphs that are conducive to static analysis and optimization techniques. However, mapping such descriptions into tightly constrained real-time implementations requires optimizating of resource sharing, buffering, and scheduling across a multidimensional latency-throughput-area objective space. Explore new approaches to these design trade-offs to achieve a successful system design.
