Scheduling Optimization Focus On Area-Performance Adaption On Reconfigurable Hard-ware

In traditional architecture, the processing unit can be divided into two categories, one is general-purpose processing unit, which achieve high generality but with low performance. The second category is ASIC which is with high performance but lack of generality. In order to compromise the high performance with the generality, reconfigurable computing was proposed since the reconfigurable logic can be reconfigured for different tasks. With the growing capacity of the hardware, reconfigurable device has the ability to support multi-task.Previous studies on scheduling only focused on reconfigurable logic area allocation while ignoring the efficiency of the area. Studies have shown that for different tasks, same area of the reconfigurable logic brought in different performance improving. According to that, this paper presents a scheduling optimization strategy with area-performance adaption.In this paper, the model of hardware and software of reconfigurable computing plat-form is established. According to the model, we turned the scheduling optimization prob-lem into a standard linear programming problem, and proposed a scheduling optimization strategy with a simplified version.The simulation results show that the strategy can be implemented on a variety of scheduling policies. The greatest improvement on average is 10.33%and the reconfigu-ration overhead is an important fact that affects the performance of our strategy.In summary, this paper presents a scheduling optimization strategy based on area-performance adaption for reconfigurable computing platform, improving the over-all per-formance of the reconfigurable computing platform.