| 注册
首页|期刊导航|高技术通讯|处理器硅前性能评估仿真点全局贪心分配方法

处理器硅前性能评估仿真点全局贪心分配方法

韩晨吉 薛峰 吴钰轩 汪文祥 张福新

高技术通讯2026,Vol.36Issue(1):29-40,12.
高技术通讯2026,Vol.36Issue(1):29-40,12.DOI:10.3772/j.issn.1002-0470.2026.01.003

处理器硅前性能评估仿真点全局贪心分配方法

Global greedy allocation for simulation points in processor pre-silicon performance evaluation

韩晨吉 1薛峰 1吴钰轩 2汪文祥 2张福新3

作者信息

  • 1. 处理器芯片全国重点实验室(中国科学院计算技术研究所) 北京 100190||中国科学院大学 北京 100049
  • 2. 龙芯中科技术有限公司 北京 100190
  • 3. 处理器芯片全国重点实验室(中国科学院计算技术研究所) 北京 100190
  • 折叠

摘要

Abstract

SimPoint,as a representative sampling technique,is widely employed in pre-silicon performance evaluation of processors.SimPoint determines the number of simulation points for each program based on the Bayesian informa-tion criterion,yet it fails to recognize the complexity variations among different programs.However,within a stand-ard test suite,different programs exhibit varying levels of behavioral complexity,necessitating different numbers of simulation points to accurately characterize their behaviors.Given a fixed total number of simulation points,Sim-Point is incapable of allocating more simulation points to more complex programs to reduce their performance evaluation errors while assigning fewer simulation points to simpler programs without compromising their perform-ance evaluation accuracy.Due to the lack of a global view of the test suite,although SimPoint can provide relatively accurate average performance evaluation errors,certain behaviorally complex programs still exhibit significant per-formance evaluation errors.To address this issue,this paper optimizes the local allocation of simulation points with-in the SimPoint method and proposes a global greedy allocation method,GreedyPoint.This method abstracts the simulation point allocation problem as a constrained optimization problem,utilizes microarchitecture-independent characteristics to quantify represent errors,and employs a global greedy algorithm to solve the optimization prob-lem.Experimental data indicate that,under the same simulation overhead,compared to SimPoint,GreedyPoint reduces the average performance evaluation error of the SPEC CPU 2017 test suite from 3.23%to 2.08%,and sig-nificantly decreases the maximum performance evaluation error from 21.22%to 7.01%.

关键词

处理器硅前性能评估/代表性采样/程序微架构无关特征

Key words

processor pre-silicon performance evaluation/representative sampling/microarchitecture-inde-pendent characteristics

引用本文复制引用

韩晨吉,薛峰,吴钰轩,汪文祥,张福新..处理器硅前性能评估仿真点全局贪心分配方法[J].高技术通讯,2026,36(1):29-40,12.

基金项目

中国科学院战略性先导科技专项(C类)(XDC05020100)资助项目. (C类)

高技术通讯

1002-0470

访问量0
|
下载量0
段落导航相关论文