Search NASA⌕ Search

SEARCH · Search NASA

Results for “GPU power”

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 181 records · Page 10
All sourcesNASA NTRSDOE OSTI
DOE OSTI2023

Software-Hardware Co-design of Heterogeneous SmartNIC System for Recommendation Models Inference and Training

Deep Learning Recommendation Models (DLRMs) are critical applications in various domains and have evolved as one of the single largest machine learning applications. Trillions of DLRM parameters exceed the on-chip memory capacity of GPUs. Large-scale multi-node systems are required for distributed DLRM inference and training, which suffer from the all-to-all communication bottleneck, mainly limiting the scalability of ever-growing DLRMs. In recent years, SmartNICs have evolved with coupled computation and communication capabilities providing opportunities for a powerful heterogeneous device in the system. However, there isn't such a distributed system that fully leverages the abundant smartNIC resources that resolve the scalability issue of DLRMs. In this work, we proposed a software-hardware co-design of a heterogeneous smartNIC system that resolves the communication bottleneck of distributed DLRMs, mitigates the memory bandwidth pressure, and improves computation efficiency. We provide a set of smartNIC designs of cache systems (including local cache and remote cache) and smartNIC computation kernels which reduce data movement, relieve memory lookup intensity, and improve the GPU's computation efficiency. In addition, we propose a graph algorithm that improves the data locality of queries within batches which optimizes the overall system performance with higher data reuse. Our evaluation shows that our system achieves 2.1x latency speedup for inference and 1.6x throughput speedup for training.

Guo, Anqi↗
DOE OSTI2024

Machine learning for the redox potential prediction of molecules in organic redox flow battery

Here, organic redox flow batteries (ORFB) are recognized as an innovative technology for the large-scale storage of renewable energy. The redox potential of organic redox-active molecules plays a vital role in their performance. Advanced screening techniques like high-throughput experiment and machine learning (ML) have significantly enhanced organic material performance and transformed the field of ORFB. However, the scarcity of experimental data poses a considerable challenge for ML model development in this domain. In our study, we developed lightweight graph-based Gaussian process regression (GPR) models with GPU-accelerated marginalized graph kernel and hybrid kernel to predict the redox potentials of organic redox-active molecules for ORFBs, specifically focusing on small datasets. To evaluate model accuracy, we created a new experimental database of organic redox-active molecules by the data from hundreds of published papers and assembled previous computational datasets. We also considered some key parameters, such as pH conditions and solvent type, to assess their impact on redox potential prediction. Our GPR model predicted redox potentials with high accuracy across all datasets using minimal training data. The study provides powerful tools for molecule screening and design and delivers valuable guidance on designing training datasets for costly experiments.

25 ENERGY STORAGE↗
DOE OSTI2022

Seismic Waveform Inversion Capability on Resource-Constrained Edge Devices

Seismic full wave inversion (FWI) is a widely used non-linear seismic imaging method used to reconstruct subsurface velocity images, however it is time consuming, has high computational cost and depend heavily on human interaction. Recently, deep learning has accelerated it’s use in several data-driven techniques, however most deep learning techniques suffer from overfitting and stability issues. In this work, we propose an edge computing-based data-driven inversion technique based on supervised deep convolutional neural network to accurately reconstruct the subsurface velocities. Deep learning based data-driven technique depends mostly on bulk data training. In this work, we train our deep convolutional neural network (DCN) (UNet and InversionNet) on the raw seismic data and their corresponding velocity models during the training phase to learn the non-linear mapping between the seismic data and velocity models. The trained network is then used to estimate the velocity models from new input seismic data during the prediction phase. The prediction phase is performed on a resource-constrained edge device such as Raspberry Pi. Raspberry Pi provides real-time and on-device computational power to execute the inference process. In addition, we demonstrate robustness of our models to perform inversion in the presence on noise by performing both noise-aware and no-noise training and feeding the resulting trained models with noise at different signal-to-noise (SNR) ratio values. We make great efforts to achieve very feasible inference times on the Raspberry Pi for both models. Specifically, the inference times per prediction for UNet and InversionNet models on Raspberry Pi were 22 and 4 s respectively whilst inference times for both models on the GPU were 2 and 18 s which are very comparable. Finally, we have designed a user-friendly interactive graphical user interface (GUI) to automate the model execution and inversion process on the Raspberry Pi.

Manu, Daniel (ORCID:0000000154982677)↗
DOE OSTI2025

Accessible, uniform protein property prediction with a scikit-learn based toolset AIDE

Summary Protein property prediction via machine learning with and without labeled data is becoming increasingly powerful, yet methods are disparate and capabilities vary widely over applications. The software presented here, “Artificial Intelligence Driven protein Estimation (AIDE)”, enables instantiating, optimizing, and testing many zero-shot and supervised property prediction methods for variants and variable length homologs in a single, reproducible notebook or script by defining a modular, standardized application programming interface (API), i.e. drop-in compatible with scikit-learn transformers and pipelines. Availability and implementation AIDE is an installable, importable python package inheriting from scikit-learn classes and API and is installable on Windows, Mac, and Linux. Many of the wrapped models internal to AIDE will be effectively inaccessible without a GPU, and some assume CUDA. The newest stable, tested version can be found at https://github.com/beckham-lab/aide_predict and a full user guide and API reference can be found at https://beckham-lab.github.io/aide_predict/. Static versions of both at the time of writing can be found on Zenodo.

36 MATERIALS SCIENCE↗
DOE OSTI2023

A new coupling of a GPU-resident large-eddy simulation code with a multiphysics wind turbine simulation tool

The development of new wind farm control strategies can benefit from combined analysis of flow dynamics in the farm and the behavior of individual turbines within one simulation environment. In this work, we present such an environment by developing a new coupling between the large-eddy simulation (LES) code GRASP and the multiphysics wind turbine simulation tool OpenFAST via an actuator line model (ALM). In addition, the implementation of the recently proposed filtered actuator line model (FALM) within the coupling is described. The new ALM implementation is cross-verified with results from four other commonly used research LES codes. The results for the blade loads and the near wake obtained with the new coupling are consistent with the other codes. Deviations are observed in the far wake. The results further indicate that the FALM is able to reduce the lift and power overprediction from which the traditional ALM suffers on coarse LES grids. This new simulation environment paves the way for future wind farm simulations under realistic weather conditions by leveraging GRASP's ability to impose data from large-scale meteorological models as boundary conditions.

17 WIND ENERGY↗
DOE OSTI2023

First principles reaction discovery: from the Schrodinger equation to experimental prediction for methane pyrolysis

Our recent success in exploiting graphical processing units (GPUs) to accelerate quantum chemistry computations led to the development of the ab initio nanoreactor, a computational framework for automatic reaction discovery and kinetic model construction. In this work, we apply the ab initio nanoreactor to methane pyrolysis, from automatic reaction discovery to path refinement and kinetic modeling. Elementary reactions occurring during methane pyrolysis are revealed using GPU-accelerated ab initio molecular dynamics simulations. Subsequently, these reaction paths are refined at a higher level of theory with optimized reactant, product, and transition state geometries. Reaction rate coefficients are calculated by transition state theory based on the optimized reaction paths. The discovered reactions lead to a kinetic model with 53 species and 134 reactions, which is validated against experimental data and simulations using literature kinetic models. We highlight the advantage of leveraging local brute force and Monte Carlo sensitivity analysis approaches for efficient identification of important reactions. Both sensitivity approaches can further improve the accuracy of the methane pyrolysis kinetic model. The results in this work demonstrate the power of the ab initio nanoreactor framework for computationally affordable systematic reaction discovery and accurate kinetic modeling.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2026

Data-driven analysis to understand GPU hardware resource usage of optimizations

With heterogeneous systems, the number of GPUs per chip increases to provide computational capabilities for solving science at a nanoscopic scale. However, low utilization for single GPUs defies the need to invest more money in expensive accelerators. Although related work develops optimizations to improve application performance, none studies how these optimizations impact hardware resource usage or average GPU utilization. Here, this paper takes a data-driven analysis approach in addressing this gap by (1) characterizing how hardware resource usage affects device utilization, execution time, or both, (2) presenting a multiobjective metric to identify important application-device interactions that can be optimized to improve device utilization and application performance jointly, (3) studying hardware resource usage behaviors of several optimizations for a benchmark application, and finally (4) identifying optimization opportunities for several scientific proxy applications based on their hardware resource usage behaviors. Furthermore, we demonstrate the applicability of our methodology by applying the identified optimizations to a proxy application, which improves the execution time, device utilization, and power consumption by up to 29.6%, 5.3% and 26.5% respectively.

Computer science↗
DOE OSTI2023

Enhancing Adaptive Physics Refinement Simulations Through the Addition of Realistic Red Blood Cell Counts

Simulations of cancer cell transport require accurately modeling mm-scale and longer trajectories through a circulatory system containing trillions of deformable red blood cells, whose intercellular interactions require submicron fidelity. Using a hybrid CPU-GPU approach, we extend the advanced physics refinement (APR) method to couple a finely-resolved region of explicitly-modeled red blood cells to a coarsely-resolved bulk fluid domain. We further develop algorithms that: capture the dynamics at the interface of differing viscosities, maintain hematocrit within the cell-filled volume, and move the finely-resolved region and encapsulated cells while tracking an individual cancer cell. Comparison to a fully-resolved fluid-structure interaction model is presented for verification. Finally, we use the advanced APR method to simulate cancer cell transport over a mm-scale distance while maintaining a local region of RBCs, using a fraction of the computational power required to run a fully-resolved model.

Roychowdhury, Sayan↗
DOE OSTI2024

A Performance and Energy Study of GPU-Resident Preconditioners for Conjugate Gradient Solvers: In the Context of Existing and Novel Approaches

Optimizing a particular subprogram out of the set of Basic (sparse) Linear Algebra Subprograms (BLAS) for a given architecture is a common topic of research. In applications, however, these BLAS functions rarely appear in isolation; usually, many of them are used together, in various combinations and with varying inputs. As the need to solve a large, sparse linear system is ubiquitous throughout HPC applications, linear solvers constitute a realistic, sufficiently complex and well-defined representative use case for composite BLAS routines. To this end, based on a representative set of matrices drawn from a diverse set of fields, we present a framework to study, from the performance and energy perspective, the efficacy of GPU- resident parallel Conjugate Gradient (CG) linear solver with different preconditioner options, including Gauss-Seidel, Jacobi, and incomplete Cholesky. We also propose a novel GPU-based preconditioner, in which the triangular solves are approximated by an iterative process. The development of this preconditioner was motivated by solving large graph Laplacian linear systems, for which the existing preconditioners either perform slow on GPU-based platforms or are not applicable. We compare the performance of these preconditioners on different hardware accelerator architectures, i.e., AMD MI250X, MI100, Nvidia A100, V100, and Jetson. Our experiments reveal performance trade-offs and provide information on how to select the best strategy for the given linear system, dictated by its properties, and the platform of interest. We demonstrate the application of our novel preconditioner for solving CG and graph Laplacian systems. Overall, the framework can be utilized as a benchmark to guide informed decisions in choosing a specific preconditioner, i.e., whether it is better to rely on the performance of a triangular solver or on the performance of sparse matrix-vector product. Finally, by considering power consumption to solve the linear systems, we report the energy footprint for the solvers.

Preconditioned Conjugate Gradient, GPUs, iterative↗
DOE OSTI2020

Real-time Electromagnetic Transient Simulation of Multi-Terminal HVDC-AC Grids based on GPU

High-fidelity electromagnetic transient (EMT) simulation plays a critical role in understanding the dynamic behavior and fast transients involved in operation, control, and protection of Multi-Terminal dc (MTdc) grids. Here, this paper proposes a cost-effective high-performance real-time EMT simulation platform for large-scale cross-continental MTdc grids based on graphics processing unit (GPU). The proposed simulation platform: i) assembles detailed EMT models of all components within an MTdc-ac grid into a single platform. This setup provides a complete simulation solution to capture fast transient signals required for high-bandwidth controller design and protection studies without any compromise; ii) implements the first GPU-based simulation architecture and corresponding algorithms for MTdc-ac grids with real-time performance at scales of 1s; iii) is highly-efficient and balances the high utilization of GPU resources and low latency required for the simulation; and iv) outperforms the existing central processing unit (CPU)- or digital signal processor (DSP)/field-programmable gate array (FPGA)-based simulators in terms of its higher scalability on large-scale MTdc-ac grids and superior price-performance ratio on the hardware. Accuracy and performance of the proposed platform are evaluated with respect to the reference results from PSCAD/EMTDC environment.

24 POWER TRANSMISSION AND DISTRIBUTION↗
DOE OSTI2021

Large-Scale, GPU-Enhanced DFTB Approaches for Probing Multi-Component Alloys

The major goals of the project are to develop and apply GPU-enhanced density functional theory tight binding (DFTB) calculations for efficient simulations of structural materials of alloy systems. The figure below depicts the GPU algorithmic developments on the left that will be carried out, and the solid-state structures on the right are representative of the complex alloys that can be computed with the GPU-enhanced DFTB approach. The methodology and tools developed in this project encompass accurate intermolecular potentials and GPU enhancements to the density functional tight binding (DFTB) approach for high-throughput ab initio molecular dynamics calculations of multi-component alloys at elevated temperatures. While classical molecular dynamics can handle hundreds of thousands of atoms, it cannot provide a first-principles based description of large, multi-component alloys at the predictive quantum level. At the other extreme, conventional Kohn-Sham DFT methods can probe the true quantum nature of chemical systems; however, these methods cannot tackle the large sizes relevant to these multi-component systems. The DFTB-based ab initio molecular dynamics approach (coupled with our in-house GPU capabilities for enhanced speed) used in this project provides a viable approach for probing these large systems at a quantum level of detail that is significantly faster than current first-principle methods Collectively, the capabilities developed in this project directly respond to DOE HBCU-OMI, AOI 2-1 initiatives by (1) enabling accurate and efficient predictions and (2) bringing a fundamental understanding of structural interactions in these complex systems at elevated temperatures.

20 FOSSIL-FUELED POWER PLANTS↗
DOE OSTI2024

Efficient distributed continual learning for steering experiments in real-time

Deep learning has emerged as a powerful method for extracting valuable information from large volumes of data. However, when new training data arrives continuously (i.e., is not fully available from the beginning), incremental training suffers from catastrophic forgetting (i.e., new patterns are reinforced at the expense of previously acquired knowledge). Training from scratch each time new training data becomes available would result in extremely long training times and massive data accumulation. Rehearsal-based continual learning has shown promise for addressing the catastrophic forgetting challenge, but research to date has not addressed performance and scalability. To fill this gap, we propose an approach based on a distributed rehearsal buffer that efficiently complements data-parallel training on multiple GPUs to achieve high accuracy, short runtime, and scalability. It leverages a set of buffers (local to each GPU) and uses several asynchronous techniques for updating these local buffers in an embarrassingly parallel fashion, all while handling the communication overheads necessary to augment input minibatches using unbiased, global sampling. We further propose a generalization of rehearsal buffers to support both classification and generative learning tasks, as well as more advanced rehearsal strategies (notably Dark Experience Replay, leveraging knowledge distillation). We illustrate this approach with a real-life HPC streaming application from the domain of ptychographic image reconstruction. Furthermore, we run extensive experiments on up to 128 GPUs of the ThetaGPU supercomputer to compare our approach with baselines representative of training-from-scratch (the upper bound in terms of accuracy) and incremental training (the lower bound). Results show that rehearsal-based continual learning achieves a top-5 validation accuracy close to the upper bound, while simultaneously exhibiting a runtime close to the lower bound.

Asynchronous data management↗
DOE OSTI2024

Towards reverse mode automatic differentiation of Kokkos-based codes

Derivative computation is a key component of optimization, sensitivity analysis, uncertainty quantification, and the solving of nonlinear problems. Automatic differentiation (AD) is a powerful technique for evaluating such derivatives, and in recent years, has been integrated into programming environments such as Jax, PyTorch, and TensorFlow to support derivative computations needed for training of machine learning models, facilitating wide-spread use of these technologies. The C++ language has become the de facto standard for scientific computing due to numerous factors, yet language complexity has made the wide-spread adoption of AD technologies for C++ difficult, hampering the incorporation of powerful differentiable programming approaches into C++ scientific simulations. This is exacerbated by the increasing emergence of architectures, such as GPUs, with limited memory capabilities and requiring massive thread-level concurrency. C++ AD tools must effectively use these environments to bring novel scientific simulations to next-generation DOE experimental and observational facilities. In this project, we investigated source transformation-based automatic differentiation using LLVM compiler infrastructure to automatically generate portable and efficient gradient computations of Kokkos-based code. We have demonstrated that our proposed strategy is feasible by investigating the usage of a prototype LLVM-based source transformation tool to generate gradients of simple functions made of sequences of simple Kokkos parallel regions. Speedups of up to 500x compared to Sacado were observed on NVIDIA V100 GPU.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2021

Scaling Inference Using Triton to Accelerate Particle Physics at the LHC and DUNE

DUNE and the LHC experiments consist of unique and cutting-edge particle detectors that create massive, complex, and rich datasets with billions of events. They require sophisticated algorithms to reconstruct and interpret the data. Modern machine learning algorithms provide a powerful toolset to detect and classify particles, from familiar image processing convolutional neural networks to newer graph neural network architectures. A full reconstruction of these particle collisions requires novel approaches to handle the computing challenge of processing so much raw data. In a series of studies, physicists from Fermilab, CERN, and university groups explored how to accelerate their data processing using the Triton Inference Server.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2023

Fuel Injection Dynamics and Composition Effects on RDE Performance

Rotating detonation engines (RDEs) provide a promising route to substantially increasing cycle efficiency in stationary gas turbines. Much of this increase relies on the ability to achieve consistent pressure gain within the combustor. In particular, the design of injectors that feed fuel and air into the detonation channel plays a crucial role. Such injectors have to ensure proper mixing of fuel and oxidizer, while minimizing backflow of detonation products into the feed plenums, and reduce susceptibility to the complex wave structures that exist within the combustor. From a practical perspective, such RDEs also need to operate with variable fuel composition. When fuel mixtures with components that possess vastly different oxidation pathways and time-scales are used, there could be additional losses through deflagrative burning instead of detonation-driven heat release. Such sensitivity to the complete flow path is akin to the physics of thermoacoustic instabilities in conventional gas turbines. In this sense, RDEs pose a unique research challenge: the performance of the device relies on the small-scale heat release process, which is highly dependent on the flow interactions within the full-scale system. As a result, canonical flow configurations, instrumented with detailed diagnostics or modeled using high-fidelity tools, but only focus on the small-scale processes will not contain the key system-level interactions. At the same time, macroscopic measurements and models that only capture system-level performance will not provide insight into the key sources of pressure losses. These couplings and sensitivities provide a formidable challenge to both experimental and simulation studies of the effects Thus, a joint experimental/computational program designed specifically to address these challenges was undertaken in this program. The focus of this program was on two key topics: a) the interaction between injector flow and the overall wave dynamics within the combustor, and b) the deflagration/detonation structure in multi-component fuels that are of practical interest. Both topics involve interaction of small-scale heat release processes with the geometry-dependent wave structure. Studies focused on the study of full-scale RDE systems, based on a 6-inch conventional annular geometry. Experimentally RDEs were studied using a combination of diagnostics. A combination of optical diagnostics and aero-thermo-acoustic analysis based on a combination of spectral and mode decomposition analysis was used to identify the dynamics of the detonation wave and other secondary waves that exist in the system. These studies have helped the identification and investigation of injector and detonation dynamics arising from coupling, and how they affect RDE mixing, detonation structure, operability and performance. Performance of RDEs was investigated through thrust stand measurements, which was used to evaluate the effective pressure gain generated by the system through the concept of equivalent available pressure. Optical diagnostics were developed and implemented to investigate the distribution of heat release, across the detonation wave. Novel optical diagnostics of NIR imaging was also developed and applied to investigate the high temperature / high pressure distribution across the detonation wave. In order to complement the experiments, the computational tools were geared to simulate the full experimental setup. GPU-based acceleration of the models and computations were developed to enable rapid simulation of the full system. In addition, the use of adaptive mesh refinement, and unstructured grid formulation, enabled the investigation of realistic geometries studied in the laboratory. The simulations produced a wealth of detail on the structure of the detonation wave under different operating conditions. Emphasis was placed on quantifying mixture pre-burning and the impact on wave propagation and structure.

03 NATURAL GAS↗
DOE OSTI2022

General-Purpose Unsupervised Cyber Anomaly Detection via Non-Negative Tensor Factorization

Distinguishing malicious anomalous activities from unusual but benign activities is a fundamental challenge for cyber defenders. Prior studies have shown that statistical user behavior analysis yields accurate detections by learning behavior profiles from observed user activity. These unsupervised models are able to generalize to unseen types of attacks by detecting deviations from normal behavior, without knowledge of specific attack signatures. However, approaches proposed to date based on probabilistic matrix factorization are limited by the information conveyed in a two-dimensional space. Non-negative tensor factorization, on the other hand, is a powerful unsupervised machine learning method that naturally models multi-dimensional data, capturing complex and multi-faceted details of behavior profiles. Herein, our new unsupervised statistical anomaly detection methodology matches or surpasses state-of-the-art supervised learning baselines across several challenging and diverse cyber application areas, including detection of compromised user credentials, botnets, spam e-mails, and fraudulent credit card transactions.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2022

Adaptive Fault Detection Based on Neural Networks and Multiple Sampling Points for Distribution Networks and Microgrids

Smart networks such as microgrid (MG) and active distribution networks (ADN) have been recently playing an important role in power system operation. The design and implementation of appropriate protection systems for such networks must be addressed, which imposes new technical challenges. This paper presents the implementation and validation aspects of an adaptive fault detection strategy based on neural networks (NNs) and multiple sampling points for ADN and MG. The solution is implemented on an edge device. Artificial NNs are used to derive a data-driven model that uses only local measurements to detect fault states of the network without the need for communication infrastructure. Multiple sampling points are used to derive a data-driven model, which allows the generalization considering the implementation in physical systems. The adaptive fault detector model is implemented on a Jetson Nano system, which is a single-board computer (SBC) with a small Graphic Processing Unit (GPU) intended to run machine learning loads at the edge. The proposed method is tested in a physical, real-life, low-voltage network located at Universidad del Norte, Colombia. This testing network is based on the IEEE-13 Node Test Feeder scaled down to 220 V. The validation in a simulation environment shows the accuracy and dependability above 99.6%, while the real-time tests show the accuracy and dependability of 95.5% and 100%, respectively. Without hard-to-derive parameters, the easy-to-implement embedded model highlights the potential for real-life applications.

42 ENGINEERING↗
DOE OSTI2020

Discovery of a Gamma-Ray Black Widow Pulsar by GPU-accelerated Einstein@Home

We report the discovery of 1.97 ms period gamma-ray pulsations from the 75 minute orbital-period binary pulsar now named PSR J1653-0158. The associated Fermi Large Area Telescope gamma-ray source 4FGL J1653.6-0158 has long been expected to harbor a binary millisecond pulsar. Despite the pulsar-like gamma-ray spectrum and candidate optical/X-ray associations—whose periodic brightness modulations suggested an orbit—no radio pulsations had been found in many searches. The pulsar was discovered by directly searching the gamma-ray data using the GPU-accelerated Einstein@Home distributed volunteer computing system. The multidimensional parameter space was bounded by positional and orbital constraints obtained from the optical counterpart. More sensitive analyses of archival and new radio data using knowledge of the pulsar timing solution yield very stringent upper limits on radio emission. Any radio emission is thus either exceptionally weak, or eclipsed for a large fraction of the time. The pulsar has one of the three lowest inferred surface magnetic-field strengths of any known pulsar with B surf ≈ 4 × 10 7 G. The resulting mass function, combined with models of the companion star's optical light curve and spectra, suggests a pulsar mass gsim 2 M ⊙ . The companion is lightweight with mass ~0.01 M ⊙ , and the orbital period is the shortest known for any rotation-powered binary pulsar. This discovery demonstrates the Fermi Large Area Telescope's potential to discover extreme pulsars that would otherwise remain undetected.

79 ASTRONOMY AND ASTROPHYSICS↗