一种高精度、高指令密集度开放式异常处理方法  

An Open Exception Processing Mechanism With High-precision and High-instruction-density Features

作　　者：田杭沛[1] 曾宇永[2] 佘建功[2] 吴松林[1] 曾翀[3] 

机构地区：[1]西北工业大学计算机学院 [2]湘南学院 [3]清华大学计算机系

出　　处：《中国集成电路》2008年第8期57-60,56,共5页China lntegrated Circuit

摘　　要：根据IEEE754/854标准,微处理器浮点单元的异常类型繁多,产生异常的原因和处理方式也比较复杂。因此浮点单元的设计中经常面临异常难以捕获的问题。本文提出一种高精度、高指令密集度开放式异常处理方法,保证了异常检测的完备性,减少了面积和功耗,加快了指令执行速度。According to the IEEE754/854 standard, there is a diversity of microprocessor floating-point unit exceptions, and the exception generation and processing mechanism are also complex. For these reasons the exceptions in the floating-point unit are often difficult to capture. In this paper an open exception processing mechanism with high-precision and high-instruction-density features is presented. With this mechanism, the completeness of exception checking is guaranteed, and the area-power and instruction processing speed also benefit.

关 键 词：浮点异常 松耦合关联 异常处理 

分 类 号：TP332[自动化与计算机技术—计算机系统结构]