Reconfigurable Computing And Task Scheduling Algorithm Research

With the development of society, the amount of data which need to be deal by computer system keep growing, this leads that the traditional CPU system can’t satisfy the requirements of the applications. Although the current mainstream multi-core processors can integrate many cores on single chip, and there will be more and more cores integrated on single chip, this means can’t satisfy the requirements of high performance computing which is used in compute-intensive applications as the generality of processors. Though ASIC (Application Specific Integrated Circuit) can raise the throughout of system significantly, and when it is be integrated in the processor systems, the systems obtain quicker operating rate for special applications. However, this means lack flexibility for different applications.As a semi custom circuit in the field of application-specific integrated circuit. FPGA overcomes the disadvantages of the ASIC flexibility and the disadvantages of other programmable devices’ less gate number, and retain the advantage of the hardware’s high performance. With the development of manufacture technology, FPGA can be applied to varied applications from accessories to the processor. The emergence of FPGA based on SRAM has laid the foundation of the realization of reconfigurable computing systems. Reconfigurable computing system often has two parts:general processors and reconfigurable processing unit commonly (FPGA). Reconfigurable computing system can both retain the hardware’s high effect and the software’s flexibility.Reconfigurable computing system plays a more and more important role in the area of high performance computing and embedded intelligent terminal because of its software’s flexibility and hardware’s high performance. Different from software task scheduling, in the process of reconfigurable computing system’s task scheduling it will produce extra reconfiguration process, and this process will produce extra time and energy consumption, these consumption lead to different consideration between reconfigurable computing system’s task scheduling and software’s task scheduling. Therefore, how to efficiently schedule tasks in the reconfigurable computing system is very important. Relevant literature indicated the current studies for this problem is focused on time optimization. This paper proposes an algorithm, namely TEO (Time and Energy Optimization) algorithm, which considers both time and energy optimization. In this algorithm, we can change the scheduling results by changing the parameters of TEO algorithm reasonably, this also increase the flexibility of the algorithm.The experimental results show that TEO algorithm can reduce 22.1% energy consumption and only increase 15.7% executing time.