面向算术单元的FPGA工艺映射算法

Arithmetic Operation Oriented FPGA Technology Mapping Algorithm

摘要: 本文提出了一种针对算术单元的FPGA工艺映射算法ArithM.实验结果表明,与公认ABC中的黑盒子映射算法相比,本文算法能平均减少逻辑单元面积7%,减少电路关键路径延时5%.ArithM采用了单元共享、平衡算术链以及吸收邻近节点三种方法来优化算术资源.

Abstract: This paper proposes arithmetic mapping algorithm, ArithM.Experiments show that, compared with accepted ABC black box mapping algorithm, this algorithm can reduce area of logic blocks by 7%, and decrease circuit delay by 5%.AithM adopts three optimizations-components sharing, balancing arithmetic list and absorbing abutting resource, to reduce the mapping cost of area and delay.