Search NASA⌕ Search

SEARCH · Search NASA

Results for “HPC training”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 19 records

A Hands-On Curriculum for Training in HPC Cluster Deployment and Management

This paper presents the design, methodology, and outcomes of the High-Performance Computing Technologies (HPCT) course, a hands-on training program focused on the system-side of HPC cluster deployment and administration. Delivered as part of the Master in High Performance Computing (MHPC) program, the course introduces students to key concepts in cluster configuration, including networking, software stack provisioning, job scheduling, and monitoring. Initially taught in person, the course was transitioned to an online format during the COVID-19 pandemic. This shift led to the development of openly available instructional material and a flipped-classroom approach that continues to support both in-person and hybrid delivery. All course materials are publicly available at www.hpc.temple.edu/mhpc/hpc-technology/index.html. By documenting the structure, infrastructure, and evolution of HPCT, this paper offers a model for accessible HPC system training that supports workforce development in computational science.

Posada Correa, Fernando [ORNL] (ORCID:000000022565↗

Pretraining Billion-Scale Geospatial Foundational Models on Frontier

As AI workloads increase in scope, generalization capability becomes challenging for small task-specific models and their demand for large amounts of labeled training samples increases. On the contrary, Foundation Models (FMs) are trained with internet-scale unlabeled data via self-supervised learning and have been shown to adapt to various tasks with minimal fine-tuning. Although large FMs have demonstrated significant impact in natural language processing and computer vision, efforts toward FMs for geospatial applications have been restricted to smaller size models, as pretraining larger models requires very large computing resources equipped with state-of-the-art hardware accelerators. Current satellite constellations collect 100+TBs of data a day, resulting in images that are billions of pixels and multimodal in nature. Such geospatial data poses unique challenges opening up new opportunities to develop FMs. We investigate billion scale FMs and HPC training profiles for geospatial applications by pretraining on publicly available data. We studied from end-to-end the performance and impact in the solution by scaling the model size. Our larger 3B parameter size model achieves up to 30% improvement in top1 scene classification accuracy when comparing a 100M parameter model. Moreover, we detail performance experiments on the Frontier supercomputer, America's first exascale system, where we study different model and data parallel approaches using PyTorch's Fully Sharded Data Parallel library. Specifically, we study variants of the Vision Transformer architecture (ViT), conducting performance analysis for ViT models with size up to 15B parameters. By discussing throughput and performance bottlenecks under different parallelism configurations, we offer insights on how to leverage such leadership-class HPC resources when developing large models for geospatial imagery applications.

Tsaris, Aristeidis (aris)↗

The Livermore Brain: Massive Deep Learning Networks Enabled by High Performance Computing

The proliferation of inexpensive sensor technologies like the ubiquitous digital image sensors has resulted in the collection and sharing of vast amounts of unsorted and unexploited raw data. Companies and governments who are able to collect and make sense of large datasets to help them make better decisions more rapidly will have a competitive advantage in the information era. Machine Learning technologies play a critical role for automating the data understanding process; however, to be maximally effective, useful intermediate representations of the data are required. These representations or “features” are transformations of the raw data into a form where patterns are more easily recognized. Recent breakthroughs in Deep Learning have made it possible to learn these features from large amounts of labeled data. The focus of this project is to develop and extend Deep Learning algorithms for learning features from vast amounts of unlabeled data and to develop the HPC neural network training platform to support the training of massive network models. This LDRD project succeeded in developing new unsupervised feature learning algorithms for images and video and created a scalable neural network training toolkit for HPC. Additionally, this LDRD helped create the world’s largest freely-available image and video dataset supporting open multimedia research and used this dataset for training our deep neural networks. This research helped LLNL capture several work-for-others (WFO) projects, attract new talent, and establish collaborations with leading academic and commercial partners. Finally, this project demonstrated the successful training of the largest unsupervised image neural network using HPC resources and helped establish LLNL leadership at the intersection of Machine Learning and HPC research.

97 MATHEMATICS AND COMPUTING↗

MARBLE: A Multi-GPU Aware Job Scheduler for Deep Learning on HPC Systems

Deep learning (DL) has become a key tool for solving complex scientific problems. However, managing the multi-dimensional large-scale data associated with DL, especially atop extant multiple graphics processing units (GPUs) in modern supercomputers poses significant challenges. Moreover, the latest high-performance computing (HPC) architectures bring different performance trends in training throughput compared to the existing studies. Existing DL optimizations such as larger batch size and GPU locality-aware scheduling have little effect on improving DL training throughput performance due to fast CPU-to-GPU connections. Additionally, DL training on multiple GPUs scales sublinearly. Thus, simply adding more GPUs to a system is ineffective. To this end, we design MARBLE, a first-of-its-kind job scheduler, which considers the non-linear scalability of GPUs at the intra-node level to schedule an appropriate number of GPUs per node for a job. By sharing the GPU resources on a node with multiple DL jobs, MARBLE avoids low GPU utilization in current multi-GPU DL training on HPC systems. Our comprehensive evaluation in the Summit supercomputer shows that MARBLE is able to improve DL training performance by up to 48.3% compared to the popular Platform Load Sharing Facility (LSF) scheduler. Compared to the state-of-the-art of DL scheduler, Optimus, MARBLE reduces the job completion time by up to 47%.

Han, Jingoo↗

Intro to HPC Bootcamp: Engaging New Communities Through Energy Justice Projects

The U.S. Department of Energy (DOE) is a long-standing leader in research and development of high-performance computing (HPC) in the pursuit of science. However, we face daunting challenges in fostering a robust and diverse HPC workforce. Basic HPC is not typically taught at early stages of students' academic careers, and the capacity and knowledge of HPC at many institutions are limited. Even so, such topics are prerequisites for advanced training programs, internships, graduate school, and ultimately for careers in HPC. To help address this challenge, as part of the DOE Exascale Computing Project's Broadening Participation Initiative, we recently launched the Introduction to HPC Training and Workforce Pipeline Program to provide accessible introductory material on HPC, scalable AI, and analytics. We describe the Intro to HPC Bootcamp, an immersive program designed to engage students from underrepresented groups as they learn foundational HPC skills. Here, the program takes a novel approach to HPC training by turning the traditional curriculum upside down. Instead of focusing on technology and its applications, the bootcamp focuses on energy justice to motivate the training of HPC skills through project-based pedagogy and real-life science stories. Additionally, the bootcamp prepares students for internships and future careers at DOE labs. The first bootcamp, hosted by the advanced computing facilities at Argonne, Lawrence Berkeley, and Oak Ridge National Labs and organized by Sustainable Horizons Institute, took place in August 2023.

96 KNOWLEDGE MANAGEMENT AND PRESERVATION↗

Experience Paper: Software Engineering and Community Codes Track in ATPESC

Argonne Training Program in Extreme Scale Computing (ATPESC) was started by the Argonne National Laboratory with the objective of expanding the ranks of better prepared users of high performance computing (HPC) machines. One of the unique aspects of the program was inclusion of software engineering and community codes track. The inclusion was motivated by the observation that the projects with a good scientific and software process were better able to meet their scientific goals. In this paper we present our experience of running the software track from the beginning of the program until now. We discuss the motivations, the reception, and the evolution of the track over the years. We welcome discussion and input from the community to enhance the track in ATPESC, and also to facilitate inclusion of similar tracks in other HPC oriented training programs.

Dubey, Anshu↗

PowerGridworld: A Framework for Multi-Agent Reinforcement Learning in Power Systems [SWR-22-07]

NREL's PowerGridworld provides a modular simulation environment for training heterogenous, grid-aware, multi-agent reinforcement learning (RL) policies at scale. The package enables the user to create component gym environments that can be composed into more complex agents. For example, a grid interactive building environment can be created by composing together component environments each encapsulating the building, PV, and battery physics. These multi-component environments can then be combined into multi-agent simulation where each agent's power consumption/injection becomes an input for solving the optimal power flow on a distribution feeder modeled in OpenDSS. Information from OpenDSS, such as bus voltages and line flows, can be included in the agents' observation spaces to enable grid-aware rewards. The default API for the PowerGridworld simulator conforms to RLLib's MultiAgent API and thus enables distributed training using HPC and cloud resources.

Biagioni, David↗

PPO And Friends

PPO and Friends (PPOAF) is a pytorch implementation of proximal policy optimization for single- and multi-agent reinforcement learning (the PPO), along with several optimizations and add-ons (the Friends) to enable efficient MPI-parallelized model training on HPC clusters.

Maguire, AlisterO↗

Scalable training of graph convolutional neural networks for fast and accurate predictions of HOMO-LUMO gap in molecules

Abstract Graph Convolutional Neural Network (GCNN) is a popular class of deep learning (DL) models in material science to predict material properties from the graph representation of molecular structures. Training an accurate and comprehensive GCNN surrogate for molecular design requires large-scale graph datasets and is usually a time-consuming process. Recent advances in GPUs and distributed computing open a path to reduce the computational cost for GCNN training effectively. However, efficient utilization of high performance computing (HPC) resources for training requires simultaneously optimizing large-scale data management and scalable stochastic batched optimization techniques. In this work, we focus on building GCNN models on HPC systems to predict material properties of millions of molecules. We use HydraGNN, our in-house library for large-scale GCNN training, leveraging distributed data parallelism in PyTorch. We use ADIOS, a high-performance data management framework for efficient storage and reading of large molecular graph data. We perform parallel training on two open-source large-scale graph datasets to build a GCNN predictor for an important quantum property known as the HOMO-LUMO gap. We measure the scalability, accuracy, and convergence of our approach on two DOE supercomputers: the Summit supercomputer at the Oak Ridge Leadership Computing Facility (OLCF) and the Perlmutter system at the National Energy Research Scientific Computing Center (NERSC). We present our experimental results with HydraGNN showing (i) reduction of data loading time up to 4.2 times compared with a conventional method and (ii) linear scaling performance for training up to 1024 GPUs on both Summit and Perlmutter.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Shaping the FutureWorkforce: Challenges and Lessons Learned in HPC Education from National Labs and Computing Centers

Workforce training at national laboratories and computing centers is essential and typically falls into two categories: foundational training for newcomers and advanced training for experienced users. Foundational topics—such as version control, build systems, and basic HPC usage—are largely transferable across institutions, while cluster-specific training varies due to differences in hardware, job schedulers, and local workflows. Training on emerging technologies is split between hardware-specific content and broadly applicable programming paradigms. Here, to reduce redundancy and increase impact, national labs, computing centers, and vendors are collaborating through initiatives like the HPC Training Working Group to share best practices, co-develop materials, and broaden outreach. These coordinated efforts aim to make HPC training more accessible, scalable, and consistent across the community.

HPC↗

Effect of Exercise Training and +Gz Acceleration Training on Men

Countermeasures for reduction in work capacity (maximal oxygen uptake and strength) during spaceflight and enhanced orthostatic intolerance during re-entry, landing and egress from the return vehicle are continuing problems. The purpose for this study was to test the hypothesis that passive-acceleration training; supine, interval, exercise plus acceleration training and exercise combined with acceleration training would improve orthostatic tolerance in ambulatory men; and that addition of the aerobic exercise conditioning would not alter this improved tolerance from that of passive-acceleration training. Seven men (24-38 yr) underwent "Passive" training on the Ames human-powered centrifuge (HPC) for 30 min, "Exercise" training on the cycle ergometer with constant +Gz acceleration; and "Combined" exercise training at 40% to 90% of the HPC +Gz(max) exercise level. Maximal supine exercise loads increased significant (P<0.05) by 8.3% (Passive), 12.6% (Exercise), and by 15.4% (Combined) after training, but their post-training maximal oxygen uptakes and maximal heart rates were unchanged. Maximal time to fatigue (endurance) was unchanged with Passive was increased (P<0.05) with Exercise and Combined training. Thus, the exercise in the Exercise and Combined training Phases resulted in greater maximal loads and endurance without effect on maximal oxygen uptake or heart rate. There was a 4% to 6% increase (P<0.05) in all four quadriceps muscle volumes (right and left) after post-Combined training. Resting pre-tilt heart rate was elevated by 12.9% (P<0.05) only after Passive training suggesting that the exercise training attenuated the HR response. Plasma volume (% Delta) was uniformly decreased by 8% to 14% (P<0.05) at tilt-tolerance pre- vs. post-training indicating essentially no effect of training on the level of hypovolemia. Post-training tilt-tolerance time and heart rate were increased (P<0.05) only with Passive training by 37.8% and by 29.1%, respectively. Thus, addition of exercise training appeared to attenuate the increased Passive tilt-tolerance.

Greenleaf, John E.↗

Tandem Predictions for HPC Jobs

At the core of the predictive analytics applied to High Performance Computing (HPC), the most prominent tasks are the prediction of job runtimes and the prediction of job queue times, both of which have the potential for informing HPC users during their every-day decision making. Accurate runtime predictions can help users better choose so-called wallclock times at job submission, decreasing the odds of their jobs waiting in queues longer than necessary. The accurate and timely queue time predictions offered for the available partitions can inform the favorable selection of partitions for running jobs. This potential is well understood as we see in the abundance of research studies that propose solutions for these tasks, including the work published in the last several years. These tasks are seemingly receptive to the Machine Learning (ML) solutions, considering that there is no shortage of training data where HPC centers over time run millions and millions of jobs. However, we study the existing research literature, as well as look for examples in the toolchains supported on the exemplar HPC facilities, and, surprisingly, do not find any practical solutions that are ready to be adopted. We interpret this as a manifestation of the shortage of UX/UI efforts that support HPC analytics and also as a sign that the research has not come to the consensus on solving these tasks. In this study, we aim to shed new light on the long-running task of job queue time prediction by exploring the utility of runtime predictions in improving prediction accuracy and, actually, predicting these two metrics together, in tandem. In other words, we show how runtime predictions become valuable input in the queue time modeling. We challenge the existing approaches to feature engineering for the queue time prediction and describe promising results we obtained for a large dataset of HPC jobs from a supercomputer at the National Renewable Energy Laboratory.

HPC↗

Tandem Predictions for HPC Jobs: Preprint

At the core of the predictive analytics applied to High Performance Computing (HPC), the most prominent tasks are the prediction of job runtimes and the prediction of job queue times, both of which have the potential for informing HPC users during their every-day decision making. Accurate runtime predictions can help users better choose so-called wallclock times at job submission, decreasing the odds of their jobs waiting in queues longer than necessary. The accurate and timely queue time predictions offered for the available partitions can inform the favorable selection of partitions for running jobs. This potential is well understood as we see in the abundance of research studies that propose solutions for these tasks, including the work published in the last several years. These tasks are seemingly receptive to the Machine Learning (ML) solutions, considering that there is no shortage of training data where HPC centers over time run millions and millions of jobs. However, we study the existing research literature, as well as look for examples in the toolchains supported on the exemplar HPC facilities, and, surprisingly, do not find any practical solutions that are ready to be adopted. We interpret this as a manifestation of the shortage of UX/UI efforts that support HPC analytics and also as a sign that the research has not come to the consensus on solving these tasks. In this study, we aim to shed new light on the long-running task of job queue time prediction by exploring the utility of runtime predictions in improving prediction accuracy and, actually, predicting these two metrics together, in tandem. In other words, we show how runtime predictions become valuable input in the queue time modeling. We challenge the existing approaches to feature engineering for the queue time prediction and describe promising results we obtained for a large dataset of HPC jobs from a supercomputer at the National Renewable Energy Laboratory.

97 MATHEMATICS AND COMPUTING↗

High-resolution hourly surrogate modeling framework for physics-based large-scale building stock modeling

Surrogate modeling can play a key role in reducing high computational burdens for large-scale physics-based modeling and uncertainty quantification. With the rapid development of large-scale building stock energy modeling, surrogate modeling has also begun to be widely applied in this field; however, most existing surrogate models lack hourly time resolution for regional-scale modeling, which is essential for understanding building demand profiles and grid impacts. Further, there is generally a lack of necessary data and feature engineering frameworks specific to building modeling for efficiently managing large datasets and complex computations. Furthermore, this paper proposes a modeling framework for large-scale (city-/region-scale), high-resolution, high-fidelity surrogate building stock energy models. Our developed framework consists of six modules: (1) building stock energy modeling (ComStock TM and ResStock TM ), (2) data engineering for large simulation data, (3) high performance computing workflow, (4) feature engineering, (5) machine learning model development, and (6) model performance evaluation. Two case studies apply the developed framework in both residential and commercial building stock analysis to demonstrate its computational efficiency and surrogate modeling accuracies. Results show that surrogate models, when efficiently trained using the HPC workflow module, reach a high level of modeling accuracy for two case studies.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Oak Ridge Computing Academy: An HPC cluster deployment and management pilot

The High Performance Computing Technologies (HPCT) course is a hands-on High Performance Computing (HPC) cluster deployment and management training program offered as part of the International School for Advanced Studies (SISSA) and the International Center for Theoretical Physics (ICTP) Master in High Performance Computing (MHPC) specialization. Here, this training program introduces students to key concepts in cluster configuration. which include networking, software stack provisioning, job scheduling, and monitoring. The publicly available course materials feature several examples and underlying methods that are broadly applicable to cluster deployment and management. This paper discusses the design of a new workforce development program at the Oak Ridge National Laboratory that is based on HPCT, the Oak Ridge Computing Academy (ORCA). The ORCA pilot program was hosted by the Oak Ridge Leadership Computing Facility (OLCF) in Summer 2025. As a part of this discussion, HPCT and ORCA course contents and infrastructure are outlined, ORCA participant experiences are detailed, and potential opportunities for improvement are discussed.

Education↗