Frontera: The Evolution of Leadership Computing at the National Science Foundation

Stanzione, Dan; West, John E.; Evans, R. T.; Minyard, Tommy; Ghattas, Omar; Panda, Dhabaleswar K.

doi:10.1145/3311790.3396656

Cited by 114 publications

(80 citation statements)

References 3 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We conduct our experimental evaluation using the Frontera supercomputer [55] at TACC, of which each GPU node is equipped with two Intel E5-2620 v4 CPUs and four NVIDIA Quadro RTX 5000 GPUs [1], interconnected by FDR InfiniBand. We use NVIDIA CUDA 10.1 and its default profiler for time measurement.…”

Section: Methodsmentioning

confidence: 99%

ClickTrain

Zhang

Yuan

Niu

et al. 2021

Proceedings of the ACM International Conference on Supercomputing

View full text Add to dashboard Cite

Convolutional neural networks (CNNs) are becoming increasingly deeper, wider, and non-linear because of the growing demand on prediction accuracy and analysis quality. The wide and deep CNNs, however, require a large amount of computing resources and processing time. Many previous works have studied model pruning to improve inference performance, but little work has been done for effectively reducing training cost. In this paper, we propose Click-Train: an efficient and accurate end-to-end training and pruning framework for CNNs. Different from the existing pruning-duringtraining work, ClickTrain provides higher model accuracy and compression ratio via fine-grained architecture-preserving pruning. By leveraging pattern-based pruning with our proposed novel accurate weight importance estimation, dynamic pattern generation and selection, and compiler-assisted computation optimizations, Click-Train generates highly accurate and fast pruned CNN models for direct deployment without any extra time overhead, compared with the baseline training. ClickTrain also reduces the end-to-end time cost of the pruning-after-training method by up to 2.3× with comparable accuracy and compression ratio. Moreover, compared with the state-of-the-art pruning-during-training approach, ClickTrain provides significant improvements both accuracy and compression ratio on the tested CNN models and datasets, under similar limited training time. CCS CONCEPTS• Computing methodologies → Neural networks.

show abstract

Section: Methodsmentioning

confidence: 99%

ClickTrain

Zhang

Yuan

Niu

et al. 2021

Proceedings of the ACM International Conference on Supercomputing

View full text Add to dashboard Cite

show abstract

“…Gabriele Bozzola is supported by by the Frontera Fellowship by the Texas Advanced Computing Center (TACC). Frontera (Stanzione et al, 2020)…”

Section: Acknowledgmentsmentioning

confidence: 99%

kuibit: Analyzing Einstein Toolkit simulations with Python

Bozzola¹

2021

JOSS

View full text Add to dashboard Cite

kuibit 1 is a Python library for analyzing simulations performed with the Einstein Toolkit 2 Löffler et al. ( 2012), a free and open-source code for numerical relativity and relativistic astrophysics. Over the past years, numerical simulations like the ones enabled by the Einst ein Toolkit have become a critical tool in modeling, predicting, and understanding several astrophysical phenomena, including binary black hole or neutron star mergers. As a result of the recent detections of gravitational waves by the LIGO-Virgo collaboration, these studies are at the forefront of scientific research. The package presented in this paper, kuibit, provides an intuitive infrastructure to read and represent the output of the Einstein Toolkit. This simplifies analyzing simulations and significantly lowers the barrier in learning how to use the tool.

show abstract

“…This work benefited also from access to the Joliot-Curie supercomputer of the TGCC (CEA) in France as part of a "grand challenge" project (project id: gch413) awarded by GENCI (Grand Equipement National de Calcul Intensif). We would also like to thank the Texas Advanced Computing Center (TACC) at the University of Texas at Austin for providing HPC resources on Frontera [147] that have contributed to the results in this paper. We acknowledge the facilities of the USQCD Collaboration used for this research in part, which are funded by the Office of Science of the U.S. Department of Energy.…”

Section: Jhep11(2021)024mentioning

confidence: 99%

The continuum and leading twist limits of parton distribution functions in lattice QCD

Karpie

Orginos

Radyushkin

et al. 2021

J. High Energ. Phys.

View full text Add to dashboard Cite

In this study, we present continuum limit results for the unpolarized parton distribution function of the nucleon computed in lattice QCD. This study is the first continuum limit using the pseudo-PDF approach with Short Distance Factorization for factorizing lattice QCD calculable matrix elements. Our findings are also compared with the pertinent phenomenological determinations. Inter alia, we are employing the summation Generalized Eigenvalue Problem (sGEVP) technique in order to optimize our control over the excited state contamination which can be one of the most serious systematic errors in this type of calculations. A crucial novel ingredient of our analysis is the parameterization of systematic errors using Jacobi polynomials to characterize and remove both lattice spacing and higher twist contaminations, as well as the leading twist distribution. This method can be expanded in further studies to remove all other systematic errors.

show abstract

Frontera: The Evolution of Leadership Computing at the National Science Foundation

Cited by 114 publications

References 3 publications

ClickTrain

ClickTrain

kuibit: Analyzing Einstein Toolkit simulations with Python

The continuum and leading twist limits of parton distribution functions in lattice QCD

Contact Info

Product

Resources

About