A strategy for real-time kernel support in application-specific HW/SW embedded architectures

Heterogeneous embedded multiprocessor architectures are becoming more prominent as a key design solution to today's microelectronics design problems. These application-specific architectures integrate multiple software programmable processors and dedicated hardware components together on to a single cost-efficient IC. In contrast to general-purpose computer systems, embedded systems are designed and optimized to provide specific functionality, using possibly a combination of different classes of processors (e.g. DSPs, microcontrollers) from different vendors. While these customized heterogeneous multiprocessor architectures offer designers new possibilities to trade off programmability, processing performance, power dissipation, and design turnaround time, there is currently a lack of tools to support the programming of these architectures. In this paper, we consider the problem of providing real-time kernel support for managing the concurrent software tasks that are distributed over a set of processors in an application-specific multiprocessor architecture. This is complementary to current research activities that aim to provide efficient retargetable code generation for a broad range of embedded processors.