Abstract
AbstractIncorporating machine learning into automatic performance analysis and tuning tools is a promising path to tackle the increasing heterogeneity of current HPC applications. However, this introduces the need for generating balanced datasets of parallel applications’ executions and for dealing with natural imbalances for optimizing performance parameters. This work proposes a holistic approach that integrates a methodology for building balanced datasets of OpenMP code-region patterns and a way to use such datasets for tuning performance parameters. The methodology uses hardware performance counters to characterize the execution of a given region and correlation analysis to determine whether it covers an unique part of the pattern input space. Nevertheless, a balanced dataset of region patterns may become naturally imbalanced when used for training a model for tuning any specific performance parameter. For this reason, we have explored several methods for dealing with naturally imbalanced datasets for finding the appropriated way of using them for tuning purposes. Experimentation shows that the proposed methodology can be used to build balanced datasets and that such datasets, plus a combination of Random Forest and binary classification, can be used to train a model able to accurately tune the number of threads of OpenMP parallel regions.
Funder
Ministerio de Ciencia e Innovación
Generalitat de Catalunya
Publisher
Springer Science and Business Media LLC
Subject
Computational Mathematics,Computational Theory and Mathematics,Computer Science Applications,Numerical Analysis,Theoretical Computer Science,Software
Reference31 articles.
1. Alcaraz J, Sikora A, César E (2019) Hardware counters’ space reduction for code region characterization, in Euro-Par 2019, ser. Lecture Notes in Computer Science, R. Yahyapour, Ed., vol. 11725. Springer, 2019, pp 74–86
2. Nguyen GH, Bouzerdoum A, Phung SL (2009) Learning pattern classification tasks with imbalanced data sets, in Pattern Recognition, P.-Y. Yin, Ed. Rijeka: IntechOpen, ch. 10
3. Nath A, Subbiah K (2018) The role of pertinently diversified and balanced training as well as testing data sets in achieving the true performance of classifiers in predicting the antifreeze proteins. Neurocomputing 272:294–305
4. Alcaraz J, Sleder S, TehraniJamsaz A, Sikora A, Jannesari A, Sorribes J, Cesar E (2021) Building representative and balanced datasets of openmp parallel regions, In: 2021 29th Euromicro International Conference on Parallel, Distributed and Network-Based Processing (PDP), pp 67–74
5. Li Z, Jannesari A, Wolf F (2013) Discovery of potential parallelism in sequential programs, In: 42nd international conference on parallel processing, pp 1004–1013
Cited by
2 articles.
订阅此论文施引文献
订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献
1. Power Constrained Autotuning using Graph Neural Networks;2023 IEEE International Parallel and Distributed Processing Symposium (IPDPS);2023-05
2. Pattern-based Autotuning of OpenMP Loops using Graph Neural Networks;2022 IEEE/ACM International Workshop on Artificial Intelligence and Machine Learning for Scientific Applications (AI4S);2022-11