Abstract

Reconfigurable computing involves adapting hardware resources to the specific needs of applications in order to obtain performance benefits. This emerging architectural paradigm holds some promise for delivering significant speedups to compute-bound applications. However, many challenges need to be overcome before reconfigurable computing becomes mainstream. Significantly, supporting applications design and providing a convenient run-time environment create opportunities to propose and in-vestigate new ways of managing chip resources. In examining the potential benefits of providing operating systems support for reconfigurable processors, this paper identifies opportunities for the development of policies for dynamic hardware management and strategies for communicating design ideas to the run-time management system.

Citations