面向滑动窗口应用,提出一种设计空间探索算法,通过判断芯片面积、存储带宽和存储资源3种片上资源约束的上限,实现充分利用片上资源.实验结果表明,该算法可以将存储资源利用率提高85%以上;同时,与相关工作相比,文中的阵列分块方法可以将访存次数降低2%~20%.
This paper presents a design space exploration algorithm for sliding-window operations. By finding the three upper bounds according to area constraints, memory bandwidth constraints and on-chip memory constraints, the block structure of the design which can fully utilize the available resources on the board is determined. Experimental results show that the efficiency of memory usage can be increased to 85%. In comparison with related work of array partition, the number of memory accesses can be reduced by 2% to 20%.