Computer comparisons in the presence of performance variation  

作　　者：Samuel IRVING Bin LI Shaoming CHEN Lu PENG Weihua ZHANG Lide DUAN 

机构地区：[1]Louisiana State University,Baton Rouge,LA 70803,USA [2]Shanghai Institute of Intelligent Electronics&Systems,Shanghai 201203,China [3]Software School,Fudan University,Shanghai 201203,China [4]Shanghai Key Laboratory of Data Science,Fudan University,Shanghai 200433,China [5]University of Texas at San Antonio,San Antonio,TX 78249,USA

出　　处：《Frontiers of Computer Science》2020年第1期21-41,共21页中国计算机科学前沿（英文版）

基　　金：This work was supported in part by the National High Technology Research and Development Program of China(2015AA015303);the National Natural Science Foundation of China(Grant No.61672160);Shanghai Science and Technology Development Funds(17511102200);National Science Foundation(NSF)(CCF-1017961,CCF-1422408,and CNS-1527318);We acknowledge the computing resources provided by the Louisiana Optical Network Initiative(LONI)HPC team.Finally,we appreciate invaluable comments from anonymous reviewers.

摘　　要：Performance variability,stemming from nondeterministic hardware and software behaviors or deterministic behaviors such as measurement bias,is a well-known phenomenon of computer systems which increases the difficulty of comparing computer performance metrics and is slated to become even more of a concern as interest in Big Data analytic increases.Conventional methods use various measures(such as geometric mean)to quantify the performance of different benchmarks to compare computers without considering this variability which may lead to wrong conclusions.In this paper,we propose three resampling methods for performance evaluation and comparison:a randomization test for a general performance comparison between two computers,bootstrapping confidence estimation,and an empirical distribution and five-number-summary for performance evaluation.The results show that for both PARSEC and highvariance BigDataBench benchmarks 1)the randomization test substantially improves our chance to identify the difference between performance comparisons when the difference is not large;2)bootstrapping confidence estimation provides an accurate confidence interval for the performance comparison measure(e.g.,ratio of geometric means);and 3)when the difference is very small,a single test is often not enough to reveal the nature of the computer performance due to the variability of computer systems.We further propose using empirical distribution to evaluate computer performance and a five-number-summary to summarize computer performance.We use published SPEC 2006 results to investigate the sources of performance variation by predicting performance and relative variation for 8,236 machines.We achieve a correlation of predicted performances of 0.992 and a correlation of predicted and measured relative variation of 0.5.Finally,we propose the utilization of a novel biplotting technique to visualize the effectiveness of benchmarks and cluster machines by behavior.We illustrate the results and conclusion through detailed Monte Carlo simulation st

关 键 词：PERFORMANCE of SYSTEMS VARIATION PERFORMANCE attributes measurement evaluation modeling simulation of multiple-processor SYSTEMS experimental design BIG Data 

分 类 号：TP39[自动化与计算机技术—计算机应用技术]