摘要
设计了一种符合IEEE-754标准的32bits高速CMOS浮点乘法器。该乘法器采用MBA算法和基于4:2compressor的树型结构完成CarrySave形式的部分积压缩,再由高速CarrySelect加法器求得乘积。电路设计采用了新型的高速加法运算单元。乘法器采用0.35μm制程,内含19,197个晶体管。3.3V工作电压下(室温),乘法器延迟时间为3.807ns,功耗为107mW@100MHz。
A high-performance lEEE-compliant 32-bits floating-point multiplier design has been presented in this paper. Modified booth's algorithm and the 4:2 tree reduce the carry save partial products to sum and carry vectors, then a fast final carry select adder convert the sum and carry vectors to conventional format. We also present a novel adder cell which offers faster operation. This multiplier was simulated by 0.35μm CMOS technology. The number of transistors is 19,197. The operating cycle time is 3.807ns at the supply voltage of 3.3V and room temperature. Only 107mW power dissipation needed at the operating frequency of 100MHz.
出处
《电路与系统学报》
CSCD
北大核心
2005年第6期6-11,共6页
Journal of Circuits and Systems