Analysis & Integrated Modeling of the Performance Evaluation Techniques for Evaluating Parallel Systems.

International Journal of Computer Science and Security

Gurvinder Singh¹  Amit Chhabra¹ 

[1] $$

关键词: Integrated model;    Metrics;    Parallel systems;    Performance;    Evaluation;   

DOI  :  

来源: Computer Science and Security

【 摘 要 】

Parallel computing has emerged as an environment for computing inherently parallel andcomputation intensive applications. Performance is always a key factor in determiningthe success of any system. So parallel computing systems are no exception. Evaluatingand analyzing the performance of parallel systems is an important aspect of parallelcomputing research. Evaluating and analyzing parallel system is difficult due to thecomplex interaction between application characteristics and architectural features.Experimental measurement, Theoretical/Analytical modeling and Simulation arethe most widely used techniques in the performance evaluation of parallel systems.Experimental measurement uses real or synthetic workloads, usually known asbenchmarks, to evaluate and analyze their performance on actual hardware.Theoretical/Analytical models try to abstract details of a parallel system. Simulation andother performance monitoring/visualization tools are extremely popular because they cancapture the dynamic nature of the interaction between applications and architectures. Eachof them has several types. For example, Experimental measurement has software,hardware, and hybrid. Theoretical/Analytical modeling has queueing network, Petri net,etc. and simulation has discrete event, trace/execution driven, Monte Carlo. Each ofthese three techniques has their own pros and cons.The purpose of this paper is firstly to present a qualitative parametric comparativeanalysis of these techniques based on parameters like stage, output statistics, accuracy,cost, resource consumption, time consumption, flexibility, scalability, tools required,trustability and secondly to justify the need for an integrated model combining theadvantages of all these techniques to evaluate the performance of parallel systems andthirdly to present a new integrated model for performance evaluation . This paper alsodiscusses certain issues like selecting an appropriate metric for evaluating parallelsystems.

【 授权许可】

Unknown