Mapping of streaming applications considering alternative application specifications

Abstract

Streaming applications often require a parallel Model of Computation (MoC) to specify their application behavior and to facilitate mapping onto Multi-Processor System-on-Chip (MPSoC) platforms. Various performance requirements and resource budgets of embedded systems ask for an efficient design space exploration (DSE) approach to select the best design from a design space consisting of a large number of design choices. However, existing DSE approaches explore the design space that includes only architecture and mapping alternatives for an initial application specification given by the application designer. In this article, we first show that a design often might not be optimal if alternative specifications of a given application are not taken into account. We further argue that the best alternative specification consists of only independent and load-balanced application tasks. Based on the Polyhedral Process Network (PPN) MoC, we present an approach to analyze and transform an initial PPN to an alternative one that contains only independent processes if possible. Finally, by prototyping real-life applications on both FPGA-based MPSoCs and desktop multi-core platforms, we demonstrate that mapping the alternative application specification results in a large performance gain compared to those approaches, in which alternative application specifications are not taken into account. ACM Reference Format:Zhai, J. T., Nikolov, H., and Stefanov, T. 2013. Mapping of streaming applications considering alternative application specifications.

DAEDALUS: System-Level Design Methodology for Streaming Multiprocessor Embedded Systems on Chips

“…30.4. By means of automated (polyhedral) transformations [49,59], PNGEN is also capable of producing alternative input-output equivalent PPNs, in which, for example, the degree of parallelism can be varied. Such transformations enable functional-level design space exploration.…”

Exploiting just-enough parallelism when mapping streaming applications in hard real-time systems

“…The platform constraints can be the number of available PEs and temporal scheduling of tasks on the PEs. On the one hand in [22,5,20], the authors determine an unfolding factor for each task such that the obtained alternative graph exposes the maximum data-level parallelism, without considering the platform constraints. However, unfolding a task too many times reveals more parallelism than needed.…”