Optimizing Reconfigurable Hardware Resource Usage in System-on-a-Programmable-Chip with Location-Aware Genetic Algorithm

This paper presents static task scheduling using location-aware genetic algorithm techniques to schedule task systems to finite amounts of reconfigurable hardware. This research optimizes the use of limited reconfigurable resources. This scheduling algorithm is built upon our previous work [12- 14]. In this paper, the genetic algorithm has been expanded to include a feature to assign selected tasks to specific functional units. In this reconfigurable hardware environment, multiple sequential processing elements (soft core processors such as Xilinx MicroBlaze [22] or Altera Nios-II [1]), task-specific core (application specific hardware), and communication network within the reconfigurable hardware can be used (such a system is called system-on-a-programmable-chip, SoPC). This paper shows that by pre-assigning (manually or randomly) a percentage of tasks to the desired functional units, the search algorithm is capable of finding acceptable schedules and maintaining high resource utilization (\u3e93 percent, with two processors configuration)