利用进位选择和超前进位逻辑的快速双分区混合加法器
Fast Bipartitioned Hybrid Adder Utilizing Carry Select and Carry Lookahead Logic
AI总结:
提出一种结合超前进位与进位选择逻辑的快速双分区混合加法器FBHA,并在28-nm CMOS下验证其相对多种32位高速加法器的延迟、面积和功耗优势。
摘要:
我们提出了一种新颖的快速双分区混合加法器(FBHA),它利用进位选择和超前进位逻辑。所提出的FBHA是一种精确加法器,由一个较高有效部分和一个较低有效部分通过进位信号连接而成。在一个N位FBHA中,K位较低有效部分使用超前进位加法器逻辑实现,而(N-K)位有效部分使用进位选择加法器逻辑实现。本工作以32位加法作为示例运算。从慢速的行波进位加法器到快速的并行前缀Kogge-Stone加法器以及所提出的加法器,许多32位加法器均使用28-nm CMOS标准单元库进行综合,并比较了它们的设计指标。与高速同类加法器相比,经过良好优化的FBHA在设计指标上实现了显著优化,部分示例如下:(a)与超前进位加法器相比,延迟降低19.8%;(b)与进位选择加法器相比,延迟降低19.8%、面积降低24.4%、功耗降低19.4%;(c)与条件求和加法器相比,延迟降低45.6%、功耗降低13.5%;以及(d)与Kogge-Stone加法器相比,面积降低46.5%、功耗降低29.3%。
英文摘要:
We present a novel fast bipartitioned hybrid adder (FBHA) that utilizes carry-select and carry-lookahead logic. The proposed FBHA is an accurate adder with a significant part and a less significant part joined together by a carry signal. In an N-bit FBHA, the K-bit less significant part is realized using carry-lookahead adder logic, and the (N-K)-bit significant part is realized using carry-select adder logic. The 32-bit addition was considered as an example operation for this work. Many 32-bit adders ranging from the slow ripple carry adder to the fast parallel-prefix Kogge-Stone adder and the proposed adder were synthesized using a 28-nm CMOS standard cell library and their design metrics were compared. A well-optimized FBHA achieved significant optimizations in design metrics compared to its high-speed adder counterparts and some examples are mentioned as follows: (a) 19.8% reduction in delay compared to a carry-lookahead adder; (b) 19.8% reduction in delay, 24.4% reduction in area, and 19.4% reduction in power compared to a carry-select adder; (c) 45.6% reduction in delay, and 13.5% reduction in power compared to a conditional sum adder; and (d) 46.5% reduction in area, and 29.3% reduction in power compared to the Kogge-Stone adder.
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。