摘要
提出了一种基于四叉树结构的高速乘法器自动综合优化算法以提升乘法器运算速度。首先对延时较大的高位积采用四叉树递归直接构建,取代传统部分积进位链,缩短关键路径时延,进而进行分支折合和合并,相邻乘法结果共享部分四叉树,降低硬件开销。算法同时支持不同面积约束下的自动综合。依此算法的乘法器相比基于Booth算法和Wallace树的乘法器速度提高了10%。
This paper used an auto-synthesizing optimized arithmetic for speed up multiplier velocity. It used four-tree recursive method to reduce critical path,instead of traditional part product carry-in chain to generate high part of long-latency multiplier. Then it reduced hardware cost by branch breaking and combing to implement logic share between tree structures. Also it supported auto-synthesize within different area constraints. This new multiplier gets 10% faster than the one based on Booth arithmetic and Wallace tree.
出处
《计算机应用研究》
CSCD
北大核心
2010年第10期3727-3730,共4页
Application Research of Computers
关键词
进位链
延迟
四叉树
分支合并
分支折合
遍历
carry chain
delay
four-tree
branch combining
branch breaking
tree traversal