engleski

A Framework for Allocation of Software Components onto a Heterogeneous Computing System

A recent development of heterogeneous platforms (i.e. those containing different types of computing units such as multicore CPUs, GPUs, and FPGAs) has enabled significant improvements in performance of real-time data processing. However, due to increased development efforts for such platforms, they are not fully exploited. To use the full potential of such platforms, it is necessary to develop new frameworks and methods in order to capture the optimal configuration of a software. Different configurations, i.e. allocations of software components to different computing unit types can be essential for getting the maximal utilization of the platform. For more complex systems it is difficult to find ad--hoc, good enough or the best configuration. This research suggests the application of component based software engineering (CBSE) principles, by which it is possible to achieve the same functionality of software components across various computing units of different types, however with different extra- functional properties (EFP). The objective of this research is to construct a framework which optimizes the allocation of software components on a heterogeneous computing platform with respect to specified extra-functional requirements. The two main extra-functional requirements which are examined in this work are the average power consumption and the average execution time.

heterogeneous computing platform ; optimal software allocation ; cyber–physical system ; formal model ; software performance measurement

nije evidentirano