值预测技术中基本值预测模型的功耗分析  

A Power Consumption Analysis of the Basic Value Prediction Models in Value Prediction Techniques

作　　者：冀蓉[1] 张民选[1] 陈怒兴[1] 

机构地区：[1]国防科技大学计算机学院,湖南长沙410073

出　　处：《计算机工程与科学》2006年第4期126-129,共4页Computer Engineering & Science

基　　金：国家自然科学基金资助项目(60273069)

摘　　要：值预测是通过推断执行来提高ILP的有效技术之一,但实现其预测模型的硬件成本和功耗却是制约它的两个主要因素。本文从实现值预测的底层电路入手,着重探究了启用值预测时功耗的主要来源,并从性能与功耗比率的角度对目前值预测中采用的几种预测模型进行了比较和评估。Value prediction is one of the effective techniques that increase ILP by speculative execution, but its hardware cost and power consumption are two main limitations. This paper focuses on studying the main sources of power consumption during the use of value prediction, and compare several value prediction models from the angle of power consumption.

关 键 词：ILP VP 值预测 功耗 IPC Last预测 Stride预测 Context预测 

分 类 号：TP303[自动化与计算机技术—计算机系统结构]