摘要
本文提出了通过基于预处理和逻辑转换的并行Sticky位的计算方法。该方法已经成功地应用到64位高性能CPU的浮点部件设计中,能有效提高浮点部件的延时性能。
The paper presents a parallel method of computing Sticky bits based upon preprocessing and logic conversion, which may improve the performance of floating point units and has been applied to the floating point units of high-performance 64-bit CPUs.
出处
《计算机工程与科学》
CSCD
2006年第4期124-125,129,共3页
Computer Engineering & Science
基金
国家自然科学基金资助项目(9020701