SCHEDULING OF CAL ACTOR NETWORKS BASED ON DYNAMIC CODE ANALYSIS
Parallel Software Implementation of DSP Algorithms
Presented by: Jani Boutellier, Author(s): Jani Boutellier, University of Oulu, Finland; Mickaël Raulet, INSA Rennes, France; Olli Silvén, University of Oulu, Finland
CAL is a dataflow oriented language for writing high-level specifications of signal processing applications. The language has recently been standardized and selected for the new MPEG Reconfigurable Video Coding standard. Application specifications written in CAL can be transformed into executable implementations through development tools. Unfortunately, the present tools provide no way to schedule the CAL entities efficiently at run-time. This paper proposes an automated approach to analyze specifications written in CAL, and produce run-time schedules that perform on average 1.45x faster than implementations relying on default scheduling. The approach is based on quasi-static scheduling, which reduces conditional execution in the run-time system.