Evaluating machine learning techniques and simulations to create efficient scheduling heuristics

• Grant number: 22/14673-6
• Support Opportunities: Scholarships abroad - Research Internship - Scientific Initiation
• Start date: February 01, 2023
• End date: March 31, 2023
• Field of knowledge: Physical Sciences and Mathematics - Computer Science
• Principal Investigator: Alfredo Goldman vel Lejbman
• Grantee: Lucas de Sousa Rosa
• Supervisor: Denis Trystram
• Host Institution: Instituto de Matemática e Estatística (IME). Universidade de São Paulo (USP). São Paulo , SP, Brazil
• Institution abroad: Université Grenoble Alpes (UGA), France
• Associated to the scholarship: 22/06906-0 - On limits of Machine Learning techniques in the learning of scheduling policies, BP.IC

Abstract

High-Performance Computing (HPC) systems are used to solve a number of complex issues in different fields of knowledge. However, these platforms have been rapidly evolving in size and complexity, and ensuring efficiency in managing applications (jobs) has become a challenge. Typically, this management involves scheduling heuristics that consist of functions to order jobs in an execution queue. The goal of this proposal is to investigate machine learning techniques, specifically regression methods, to create efficient scheduling heuristics. We also intend to use simulations and workload traces to determine the characteristics of HPC jobs that lead to a reduction in the mean slowdown of the jobs in the queue. The research visit will be carried out at the Université Grenoble-Alpes (UGA), in France, under the supervision of Professor Denis Trystram. During the visit, we will set up an experimental campaign that ranges from obtaining the data to validating the heuristics through performance tests. (AU)