Performance-Based Numerical Solver Selection in the Lighthouse Framework

Jessup, Elizabeth R.; Motter, Pate; Norris, Boyana; Sood, Kanika

doi:10.1137/15m1028406

Cited by 8 publications

(4 citation statements)

References 39 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The full list of the XAMG library parameters targeted for optimization and their equivalents in hypre are summarized in Table A. 4. The specific parameters included in optimization sets are shown in Table A.5.…”

Section: Appendix a List Of Optimizing Parametersmentioning

confidence: 99%

“…The development of assistive algorithms and tools to simplify the choice of efficient solver configurations based on some specific matrix heuristics has been a popular research topic for decades. The tremendous growth of practical interest was associated with developing the machine learning algorithms, which were realized in a series of publications [1,2,3,4,5], to name but a few examples. The corresponding publications investigate various machine learning algorithms, training datasets, lists of features used to characterize the SLAEs, predefined solver configurations, and others.…”

Section: Introductionmentioning

confidence: 99%

“…The corresponding publications investigate various machine learning algorithms, training datasets, lists of features used to characterize the SLAEs, predefined solver configurations, and others. As a result, several software tools (e.g., SALSA [6] and Lighthouse [4]) were developed to recommend the best solver from a predefined set of configurations. These tools, however, being good assistants for novice users, cannot provide highly optimized method configurations with the modified solver parameters set, tuned for the specific SLAE.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Automated tuning for the parameters of linear solvers

Petrushov¹,

Krasnopolsky²

2023

Preprint

View full text Add to dashboard Cite

Robust iterative methods for solving large sparse systems of linear algebraic equations often suffer from the problem of optimizing the corresponding tuning parameters. To improve the performance for the problem of interest, the specific parameter tuning is required, which in practice can be a timeconsuming and tedious task. The present paper deals with the problem of automating the optimization of the numerical method parameters to improve the performance of the mathematical physics simulations and simplify the modeling process.The paper proposes the hybrid evolution strategy applied to tune the parameters of the Krylov subspace and algebraic multigrid iterative methods when solving a sequence of linear systems with a constant matrix and varying right-hand side. The algorithm combines the traditional evolution strategy with the pre-trained neural network, which filters the individuals in the new generation. The proposed coupling of two optimization approaches allows to integrate the adaptivity properties of the evolution strategy with a priori knowledge realized by the neural network. Furthermore, the use of the neural network as a preliminary filter allows for significant weakening of the prediction accuracy requirements and reusing the pre-trained network with a wide range of linear systems.The detailed algorithm efficiency evaluation is performed for a set of model linear systems, including the ones from the SuiteSparse Matrix Collection and the systems from the turbulent flow simulations. The obtained results show that the pre-trained neural network can be effectively reused to optimize parameters for various linear systems, and a significant speedup in the calculations can be achieved at the cost of about 100 trial solves. The algorithm decreases the calculation time by more than 6 times for the black box matrices from the SuiteSparse Matrix Collection and by a factor of 1.5-1.8 for the turbulent flow simulations considered in the paper.

show abstract

Section: Appendix a List Of Optimizing Parametersmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Automated tuning for the parameters of linear solvers

Petrushov¹,

Krasnopolsky²

2023

Preprint

View full text Add to dashboard Cite

show abstract

“…The work in [20] enhances Petabricks [49] to leverage the accuracy-throughput tradeoffs and to consider input features of the data to be processed, On the other side, [52] and [53] follows the same idea of switching at runtime the configuration according to the input characteristics by applying it to two different application domains, respectively graphics and graph processing. To support the autotuning phase, machine learning techniques have been used in literature [51], [54], [55] to model application metrics and then to predict the best configuration to be applied. This is fundamental when the size of the configuration space is huge, and thus not possible to be entirely profiled, or when the elaboration is heavily data-dependent.…”

Section: Approximation Techniquesmentioning

confidence: 99%

Tunable approximations to control time-to-solution in an HPC molecular docking Mini-App

et al. 2020

View full text Add to dashboard Cite

The drug discovery process involves several tasks to be performed in vivo, in vitro and in silico. Molecular docking is a task typically performed in silico. It aims at finding the three-dimensional pose of a given molecule when it interacts with the target protein binding site. This task is often done for virtual screening a huge set of molecules to find the most promising ones, which will be forwarded to the later stages of the drug discovery process. Given the huge complexity of the problem, molecular docking cannot be solved by exploring the entire space of the ligand poses. State-of-the-art approaches face the problem by sampling the space of the ligand poses to generate results in a reasonable time budget. In this work, we improve the geometric approach to molecular docking by introducing tunable approximations. In particular, we analyzed and enriched the original implementation with tunable software knobs to explore and control the performance-accuracy tradeoffs. We modeled time-to-solution of the virtual screening task as a function of software knobs, input data features, and available computational resources. Therefore, the application can autotune its configuration according to a user-defined time budget. We used a Mini-App derived by LiGenDock -a state-of-the-art molecular docking application -to validate the proposed approach. We run the enhanced Mini-App on an HPC system by using a very large database of pockets and ligands. The proposed approach exposes a time-to-solution interval spanning more than one order of magnitude with accuracy degradation up to 30%, more in general providing different accuracy levels according to the needs of the virtual screening campaign.

show abstract

A New Matrix Feature Selection Strategy in Machine Learning Models for Certain Krylov Solver Prediction

Sun,

Jing,

2024

J Classif

View full text Add to dashboard Cite

Performance-Based Numerical Solver Selection in the Lighthouse Framework

Cited by 8 publications

References 39 publications

Automated tuning for the parameters of linear solvers

Automated tuning for the parameters of linear solvers

Tunable approximations to control time-to-solution in an HPC molecular docking Mini-App

A New Matrix Feature Selection Strategy in Machine Learning Models for Certain Krylov Solver Prediction

Contact Info

Product

Resources

About