Hardware/software partitioning aiming at fulfilment of real-time constraints

摘要

Whether a real-time system meets its timing constraints depends to a large extent on which hardware architecture is used, and how the functionality is distributed on it. This article discusses how such design decisions can be supported by hardware/software codesign. An implementation quality metric - the minimal required speedup — is presented, which indicates how well the timing constraints are fulfilled, and it is shown how to calculate this using an extension of fixed-priority scheduling analysis. A partitioning algorithm is developed, which assigns tasks to architecture components in an attempt to meet the timing constraints at a minimal cost.