@article {10.3844/jcssp.2020.479.492,
article_type = {journal},
title = {A Novel Method to Predict Processor Performance by Modeling Different Architecture Parameters},
author = {Issa, Joseph},
volume = {16},
number = {4},
year = {2020},
month = {Apr},
pages = {479-492},
doi = {10.3844/jcssp.2020.479.492},
url = {https://thescipub.com/abstract/jcssp.2020.479.492},
abstract = {Predicting processor throughput and performance is one of the essential aspects of computer architecture. It is crucial to model processor performance behavior for future architectures based on the existing data set. Modeling processor performance for a given workload enables architects to enhance processor features to meet specific performance targets for a given benchmark. Developing an estimation method to predict performance using one micro-architecture parameter is limited, given the need to model multiple parameters simultaneously. In this paper, we propose a novel performance prediction method for SPEC CPU 2006 and HDxPRT 2014 benchmarks based on a combination of measured and estimated performance data. The performance project model predicts processor performance while altering multiple microarchitecture parameters simultaneously such as memory speed, number of cores and the core frequency. We also present a detailed timing analysis for each processor sub-component. The model is verified to project performance with less than 5% error margin between projected and measured baseline.},
journal = {Journal of Computer Science},
publisher = {Science Publications}
}