A Tool for Design Space Exploration Using Multi-Core, FPGA and GPU in Programmable Automation Controllers

摘要

While diversity in signal processing hardware has continued to grow, the available programming paradigms have remained largely unchanged. Historically, both hardware- and software- centric paradigms required non-trivial knowledge of the physical architecture to achieve acceptable performance. These requirements burden the developer, who must become an expert in a wide variety of software systems as well as hardware architectures. A tool is needed to assist the developer to make informed choices during algorithm composition that will implement efficiently on a chosen hardware architecture, or conversely, to suggest an appropriate hardware architecture given an algorithm specification. Such a tool is a benefit to scientists and researchers given the task of developing control and data acquisition system components but who are not specifically trained with the skills needed to program such systems.