Adaptive computing systems (ACSs) and application-specific integrated circuits (ASICs) can serve as flexible hardware accelerators for applications in domains such as image processing and digital signal processing. However, the mapping of applications onto ACSs and ASICs using the traditional methods can take months for a hardware engineer to develop and debug. In this dissertation, a new approach for automatic mapping of software applications onto ACSs and ASICs has been developed, implemented and validated. This dissertation presents the design flow of the software environment called CHAMPION, which is being developed at the University of Tennessee. This environment permits high-level design entry using the Cantata graphical programming software from KRI. Using Cantata as the design entry, CHAMPION hides from the user the low-level details of the hardware architecture and the finer issues of application mapping onto the hardware. Validation of the CHAMPION environment was performed using multiple applications of moderate complexity. In one case, the application mapping time which required six weeks to perform manually took only six minutes for CHAMPION, yet comparable results were produced. Furthermore, the CHAMPION environment was constructed such that retargeting to a new adaptive computing system could be accomplished in just a few hours as opposed to weeks using manual methods. Thus, CHAMPION permits both ACSs and ASICs to be utilized by a wider audience and application development accomplished in less time.
展开▼
