Portfolio-Based Selection of Robust Dynamic Loop Scheduling Algorithms Using Machine Learning

Abstract

The execution of computationally intensive parallel applications in heterogeneous environments, where the quality and quantity of computing resources available to a single user continuously change, often leads to irregular behavior, in general due to variations of algorithmic and systemic nature. To improve the performance of scientific applications, loop scheduling algorithms are often employed for load balancing of their parallel loops. However, it is a challenge to select the most robust scheduling algorithms for guaranteeing optimized performance of scientific applications on large-scale computing systems that comprise resources which are widely distributed, highly heterogeneous, often shared among multiple users, and have computing availabilities that cannot always be guaranteed or predicted. To address this challenge, in this work we focus on a portfolio-based approach to enable the dynamic selection and use of the most robust dynamic loop scheduling (DLS) algorithm from a portfolio of DLS algorithms, depending on the given application and current system characteristics including workload conditions. Thus, in this paper we provide a solution to the algorithm selection problem and experimentally evaluate its quality. We propose the use of supervised machine learning techniques to build empirical robustness prediction models that are used to predict DLS algorithm's robustness for given scientific application characteristics and system availabilities. Using simulated scientific applications characteristics and system availabilities, along with empirical robustness prediction models, we show that the proposed portfolio-based approach enables the selection of the most robust DLS algorithm that satisfies a user-specified tolerance on the given application's performance obtained in the particular computing system with a certain variable availability. We also show that the portfoliobased approach offers higher guarantees regarding the robust performance of the application using the automatically selected DLS algorithms when compared to the robust performance of the same application using a manually selected DLS algorithm.

Design methodology for workload‐aware loop scheduling strategies based on genetic algorithm and simulation

“…To address this situation, Sukhija et al proposed an approach that uses supervised machine learning techniques to predict the most robust loop scheduling strategy for a target application/platform [11]. To address this situation, Sukhija et al proposed an approach that uses supervised machine learning techniques to predict the most robust loop scheduling strategy for a target application/platform [11].…”

“…With this emerging variety of loop scheduling strategies, the task of selecting the most adequate one for a given application becomes challenging. To address this situation, Sukhija et al proposed an approach that uses supervised machine learning techniques to predict the most robust loop scheduling strategy for a target application/platform . They showed that their proposed approach (i) enables the selection of the most robust loop scheduling algorithm that satisfies a user‐specified tolerance on the given application's performance and (ii) offers higher guarantees regarding the performance of the application using the automatically selected loop scheduling algorithms, when compared with the performance of the same application using an empirically selected loop scheduling algorithm.…”

Investigating the Resilience of Dynamic Loop Scheduling in Heterogeneous Computing Systems

“…In earlier work [3], a theoretical foundation of measuring the resilience value of a set of DLS methods with respect to variation in the processor failures has been described. In the latest related work [4], the flexibility analysis of DLS techniques in the presence of fluctuating system load for scheduling scientific applications onto computing systems was performed. To the best of our knowledge, this is the first work to analyze the resilience of the DLS algorithms using various combinations of problem sizes, system sizes and different types of application and systemic characteristics.…”

SimAS: A simulation‐assisted approach for the scheduling algorithm selection under perturbations

“…It provides ready to use application programming interfaces (APIs) to represent applications and computing systems through different interfaces: MSG (SG-MSG), SimDag (SG-SD), and SMPI (SG-SMPI). SG uses a simple and fast CPU computation model and verified and more complex network models [18], which render it well suited for the study of computationallyintensive parallel and distributed scientific applications.Various studies have used SG to evaluate the performance of applications with DLS techniques in different scenarios [12,10]. To attain high trustworthiness in the performance results obtained with SG, the implementation of the nonadaptive DLS techniques in SG-SD has been verified [19] by reproducing the results presented in the work that introduced factoring [5].…”

“…Various studies have used SG to evaluate the performance of applications with DLS techniques in different scenarios [12,10]. To attain high trustworthiness in the performance results obtained with SG, the implementation of the nonadaptive DLS techniques in SG-SD has been verified [19] by reproducing the results presented in the work that introduced factoring [5].…”