Using Machine Learning in Trace-driven Energy-Aware Simulations of High-Throughput Computing Systems

McGough, A. S; Al Moubayed, N; M, Forshaw

Abstract

When performing a trace-driven simulation of a High Throughput Computing system we are limited to the knowledge which should be available to the system at the current point within the simulation. However, the trace-log contains information we would not be privy to during the simulation. Through the use of Machine Learning we can extract the latent patterns within the trace-log allowing us to accurately predict characteristics of tasks based only on the information we would know. These characteristics will allow us to make better decisions within simulations allowing us to derive better policies for saving energy. We demonstrate that we can accurately predict (up-to 99% accuracy), using oversampling and deep learning, those tasks which will complete while at the same time provide accurate predictions for the task execution time and memory footprint using Random Forest Regression.

Citation

McGough, A. S., Al Moubayed, N., & M, F. (2017, April). Using Machine Learning in Trace-driven Energy-Aware Simulations of High-Throughput Computing Systems. Presented at ENERGY-SIM 2017, L'Aqua

• Presentation Conference Type: Conference Paper (published)
• Conference Name: ENERGY-SIM 2017
• Start Date: Apr 23, 2017
• Acceptance Date: Mar 7, 2017
• Online Publication Date: Apr 18, 2017
• Publication Date: Apr 18, 2017
• Deposit Date: Mar 19, 2017
• Publicly Available Date: Mar 22, 2017
• Pages: 55-60
• Series Title: ICPE '17 Companion
• Book Title: Proceedings of the 8th ACM/SPEC on International Conference on Performance Engineering Companion (ICPE '17 Companion), April 22 - 26, 2017, L’Aquila, Italy.
• DOI: https://doi.org/10.1145/3053600.3053612
• Public URL: https://durham-repository.worktribe.com/output/1146620