Search NASA⌕ Search

SEARCH · Search NASA

Results for “hardware accelerators”

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 667 records · Page 37

Using EPICS enabled industrial hardware for upgrading control systems

Los Alamos National Laboratory has been working with National Instruments (NI) and Cosy lab to implement EPICS Input Output Controller (IOC) software that runs directly on NI CompactRIO Real Time Controller (RTC) and communicates with NI LabVIEW through a shared memory interface. In this presentation, we will discuss our current progress in upgrading the control system at the Los Alamos Neutron Science Centre (LANSCE) and what we have learned about integrating CompactRIO into large experimental physics facilities. We will also discuss the implications of using Channel Access Server for LabVIEW which will enable more commercial hardware platforms to be used in upgrading existing facilities or in commissioning new ones.

43 PARTICLE ACCELERATORS↗

Blueprint for DOE Quantum Supercomputing: Ensuring U.S. Leadership in the Quantum Decade

Quantum computing stands at the threshold of a transformative decade, where the field will evolve from small-scale demonstrations toward practical scientific computing at scale. This Blueprint identifies fault-tolerant quantum computers (FTQCs) as a viable, scalable, and broadly applicable path to achieving “quantum scientific utility,” defined as solving scientifically valuable problems beyond the reach of conventional, classical computers. This capability is expected to show scientific demonstrations in the late 2020s and to mature in the early-to-mid 2030s. This Blueprint outlines a strategy to prepare the U.S. Department of Energy (DOE) for FTQCs and their integration into the U.S. national scientific computing infrastructure. Its purpose is to identify the steps, milestones, and research directions necessary for DOE to enable initial deployment of FTQCs in 2028 as a scientific tool for the nation and mature this capability into the 2030s. DOE has a long history of supporting quantum information science and technology, contributing significantly to research advancements, training a quantum-ready workforce, and providing access to early small-scale quantum hardware. Given recent demonstrations of logical operations on error-corrected logical qubits and the advancement of commercial hardware roadmaps, DOE should begin preparations for large-scale, fault-tolerant quantum computing deployment for DOE science missions. This Blueprint proposes that DOE focus on (1) deploying first-generation scientifically relevant quantum computers with at least 100 logical qubits and performing at least 10,000 to 100,000 hard logical operations in scientifically relevant calculations; (2) developing essential FTQC programming competencies, system software, and facility readiness; and (3) investing in cutting edge focused R&D that fosters breakthroughs in scientific applications, algorithms, and logical architectures needed to accelerate the advent of scientific utility. This effort will position DOE to transition to larger systems: production-scale quantum computers that comprise 1,000 to 10,000 logical qubits, perform 1 to 10 billion hard logical operations, and execute scientifically useful computations at scale. Achieving these goals will require DOE facilities to evolve with urgency to support scientific campaigns that integrate quantum and classical computing resources into efficient workflows, novel software and firmware environments for compiling and routing quantum programs on FTQC machines, and suitable infrastructure for quantum hardware. It will also require further development and optimization of scientific applications from the fields of materials science, quantum chemistry, and high-energy and nuclear physics. The Blueprint calls for transformative R&D and collective action to accelerate the advent of scientific quantum utility and bring it within reach by 2028.

97 MATHEMATICS AND COMPUTING↗

Autonomous organic synthesis for redox flow batteries via flexible batch Bayesian optimization

Traditional trial-and-error methods for materials discovery are inefficient to meet the urgent demands posed by the rapid progression of climate change. This urgency has driven the increasing interest in integrating robotics and machine learning into materials research to accelerate experimental learning. However, idealized decision-making frameworks to achieve maximum sampling efficiency are not always compatible with high-throughput experimental workflows inside a laboratory. For multi-step chemical processes, differences in hardware capacities can complicate the digital framework by introducing constraints on the maximum number of samples in each step of the experiment, hence causing varying batch sizes in variable selection within the same batch. Therefore, designing flexible sampling algorithms is necessary to accommodate the multi-step synthesis with practical constraints unique to each high-throughput workflow. In this work, we designed and employed three strategies on a high-throughput robotic platform to optimize the sulfonation reaction of redox-active molecules used in flow batteries. Our strategies adapt to the multi-step experimental workflow, where their formulation and heating steps are separate, causing varying batch size requirements. By strategically sampling using clustering and mixed-variable batch Bayesian optimization, we were able to iteratively identify optimal conditions that maximize the yields. Our work presents a flexible approach that allows tailoring the machine learning decision-making to suit the practical constraints in individual high-throughput experimental platforms, followed by performing resource-efficient yield optimization using available open-source Python libraries.

Tamura, Clara [Univ. of Washington, Seattle, WA (U↗

BeyondFingerprinting: AI-guided discovery of robust materials & processes

BeyondFingerprinting was a 2021-2024 Sandia Grand Challenge LDRD exploring the potential to develop new resilient materials and manufacturing processes by taking an artificial-intelligence (AI)-guided approach that integrates human-subject-matter expertise with algorithms enriched with physics-based constraints to unearth process-structure-property correlations. Such algorithms, trained on high-throughput experiments and simulations, are shown to serve as surrogate models that efficiently detect key “fingerprints” in materials data, prognose material performance, and guide effective process improvements. To accelerate broader adoption across mission areas, this AI-guided approach was demonstrated with three complex process-centric exemplars: electroplating, physical vapor deposition, and laser powder bed fusion. Together, these exemplars impact nearly every hardware component relevant to DOE and NNSA national security missions.

36 MATERIALS SCIENCE↗

Controls system developments for the ERL facility

The BNL Energy Recovery LINAC (ERL) is a high beam current, superconducting RF electron accelerator that is being commissioned to serve as a research and development prototype for a RHIC facility upgrade for electron-ion collision (eRHIC). Key components of the machine include a laser, photocathode, and 5-cell superconducting RF cavity operating at a frequency of 703 MHz. Starting with a foundation based on existing ADO software running on Linux servers and on the VME/VxWorks platforms developed for RHIC, we are developing a controls system that incorporates a wide range of hardware I/O interfaces that are needed for machine R&D. Details of the system layout, specifications, and user interfaces are provided.

43 PARTICLE ACCELERATORS↗

Anomaly Detection in Accelerator Facilities Using Machine Learning

Synchrotron light sources are user facilities and usually run about 5000 hours per year to support many beamlines operations in parallel. Reliability is a key parameter to evaluate machine performance. Even many facilities have achieved >95% beam reliability, there are still many hours of unscheduled downtime and every hour lost is a waste of operation costs along with a big impact on individual scheduled user experiments. Preventive maintenance on subsystems and quick recovery from machine trips are the basic strategies to achieve high reliability, which heavily depends on experts’ dedication. Recently, SLAC, APS, and NSLS-II collaborated to develop machine-learning-based approaches aiming to solve both situations, hardware failure prediction and machine failure diagnosis to find the root sources. In this paper, we report our facility operation status, development progress, and plans.

Accelerator Physics↗

Measurement and analysis of GPU-accelerated applications with HPCToolkit

To address the challenge of performance analysis on the US DOE’s forthcoming exascale supercomputers, Rice University has been extending its HPCToolkit performance tools to support measurement and analysis of GPU-accelerated applications. To help developers understand the performance of accelerated applications as a whole, HPCToolkit’s measurement and analysis tools attribute metrics to calling contexts that span both CPUs and GPUs. To measure GPU-accelerated applications efficiently, HPCToolkit employs a novel wait-free data structure to coordinate monitoring and attribution of GPU performance. To help developers understand the performance of complex GPU code generated from high-level programming models, HPCToolkit constructs sophisticated approximations of call path profiles for GPU computations. To support fine-grained analysis and tuning, HPCToolkit uses PC sampling and instrumentation to measure and attribute GPU performance metrics to source lines, loops, and inlined code. To supplement fine-grained measurements, HPCToolkit can measure GPU kernel executions using hardware performance counters. To provide a view of how an execution evolves over time, HPCToolkit can collect, analyze, and visualize call path traces within and across nodes. Finally, on NVIDIA GPUs, HPCToolkit can derive and attribute a collection of useful performance metrics based on measurements using GPU PC samples. Here, we illustrate HPCToolkit’s new capabilities for analyzing GPU-accelerated applications with several codes developed as part of the Exascale Computing Project.

97 MATHEMATICS AND COMPUTING↗

A Test Bed to Evaluate Advanced Distribution Management Systems for Modern Power Systems: Preprint

This paper presents a test bed that utilities, vendors, and researchers can use to evaluate the performance of existing and emerging advanced distribution management systems (ADMS) or specific applications of such systems. The test bed provides a realistic laboratory test setting, including detailed modeling of the power system, controller and/or power hardware, and industry-standard communication protocols. The test bed can further represent the current state of a specific utility's distribution system or the anticipated state with increased penetrations of distributed energy resources, such as rooftop photovoltaic systems, residential battery energy storage systems, electric vehicles, and loads with smart controls. The paper describes the design and implementation of the test bed. We present two use cases of the test bed and preliminary results for these two use cases. We also discuss upcoming and potential future use cases. Results from test bed experiments can be used to accelerate industry deployment of ADMS technologies and advance development of ADMS capabilities.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Fast scanning probe for the NSTX spherical tokamak

We describe a fast reciprocating Langmuir probe and drive system, which has four main new features: (1) use of high-temperature, vacuum, circuit boards instead of cables to reduce weight and increase to 21 the number of possible connections, (2) rotatable and removable shaft, (3) 10 tip construction with designed hardware bandwidth up to 10 MHz, and (4) a detachable and modular tip assembly for easy maintenance. The probe is mounted in a fast pneumatic drive capable of speeds {approx}7 m/s and {approx}20g's acceleration in order to reach the scrape-off layer (SOL) and pedestal regions and remain inserted long enough to obtain good statistics while minimizing the heat deposition to the tips and head in a power density environment of 1-10 MW/m{sup 2}. The National Spherical Torus Experiment SOL features electron temperature, T{sub e}{approx}10-30 eV, and electron density, n{sub e}{approx}0.1-5x10{sup 12} cm{sup -3} while the pedestal features n{sub e}{approx}0.5-1.5x10{sup 13} cm{sup -3} and T{sub e}{approx}30-150 eV. The probe described here has ten tips which obtain a wide spectrum of plasma parameters: electron temperature profile T{sub e}(r), electron density profile n{sub e}(r) and Mach number profile M(r), floating potential V{sub f}(r), poloidal and radial electric field profiles E{sub {theta}}(r) and E{sub {rho}}(r), saturation current profile I{sub sat}(r), and their fluctuations up to 3 MHz. We describe the probe and show representative radial profiles of various parameters.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Study of gas-puff Z-pinches on COBRA

Gas-puff Z-pinch experiments were conducted on the 1 MA, 200 ns pulse duration Cornell Beam Research Accelerator (COBRA) pulsed power generator in order to achieve an understanding of the dynamics and instability development in the imploding and stagnating plasma. The triple-nozzle gas-puff valve, pre-ionizer, and load hardware are described. Specific diagnostics for the gas-puff experiments, including a Planar Laser Induced Fluorescence system for measuring the radial neutral density profiles along with a Laser Shearing Interferometer and Laser Wavefront Analyzer for electron density measurements, are also described. The results of a series of experiments using two annular argon (Ar) and/or neon (Ne) gas shells (puff-on-puff) with or without an on- (or near-) axis wire are presented. For all of these experiments, plenum pressures were adjusted to hold the radial mass density profile as similar as possible. Initial implosion stability studies were performed using various combinations of the heavier (Ar) and lighter (Ne) gasses. Implosions with Ne in the outer shell and Ar in the inner were more stable than the opposite arrangement. Current waveforms can be adjusted on COBRA and it was found that the particular shape of the 200 ns current pulse affected on the duration and diameter of the stagnated pinched column and the x-ray yield.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

SU-E-T-118: Analysis of Variability and Stability Between Two Water Tank Phantoms Utilizing Water Tank Commissioning Procedures

Purpose: The commissioning criteria of water tank phantoms are essential for proper accuracy and reproducibility in a clinical setting. This study outlines the results of mechanical and dosimetric testing between PTW MP3-M water tank system and the Standard Imaging Doseview 3D water tank system. Methods: Measurements were taken of each axis of movement on the tank using 30 cm calipers at 1, 5, 10, 50, 100, and 200 mm for accuracy and reproducibility of tank movement. Dosimetric quantities such as percent depth dose and dose profiles were compared between tanks using a 6 MV beam from a Varian 23EX LINAC. Properties such as scanning speed effects, central axis depth dose agreement with static measurements, reproducibility of measurements, symmetry and flatness, and scan time between tanks were also investigated. Results: Results showed high geometric accuracy within 0.2 mm. Central axis PDD and in-field profiles agreed within 0.75% between the tanks. These outcomes test many possible discrepancies in dose measurements across the two tanks and form a basis for comparison on a broader range of tanks in the future. Conclusion: Both 3D water scanning phantoms possess a high degree of spatial accuracy, allowing for equivalence in measurements regardless of the phantom used. A commissioning procedure when changing water tanks or upon receipt of a new tank is nevertheless critical to ensure consistent operation before and after the arrival of new hardware.

60 APPLIED LIFE SCIENCES↗

wa-hls4ml: A Benchmark and Surrogate Models for hls4ml Resource and Latency Estimation

As machine learning (ML) is increasingly implemented in hardware to address real-time challenges in scientific applications, the development of advanced toolchains has significantly reduced the time required to iterate on various designs. These advancements have solved major obstacles, but also exposed new challenges. For example, processes that were not previously considered bottlenecks, such as hardware synthesis, are becoming limiting factors in the rapid iteration of designs. To mitigate these emerging constraints, multiple efforts have been undertaken to develop an ML-based surrogate model that estimates resource usage of ML accelerator architectures. We introduce wa-hls4ml, a benchmark for ML accelerator resource and latency estimation, and its corresponding initial dataset of over 680,000 fully connected and convolutional neural networks, all synthesized using hls4ml and targeting Xilinx FPGAs. The benchmark evaluates the performance of resource and latency predictors against several common ML model architectures, primarily originating from scientific domains, as exemplar models, and the average performance across a subset of the dataset. Additionally, we introduce GNN- and transformer-based surrogate models that predict latency and resources for ML accelerators. We present the architecture and performance of the models and find that the models generally predict latency and resources for the 75% percentile within several percent of the synthesized resources on the synthetic test dataset.

Hawks, Benjamin [Fermilab] (ORCID:0000000157000288↗

wa-hls4ml: A Benchmark and Surrogate Models for hls4ml Resource and Latency Estimation

As machine learning (ML) is increasingly implemented in hardware to address real-time challenges in scientific applications, the development of advanced toolchains has significantly reduced the time required to iterate on various designs. These advancements have solved major obstacles, but also exposed new challenges. For example, processes that were not previously considered bottlenecks, such as hardware synthesis, are becoming limiting factors in the rapid iteration of designs. To mitigate these emerging constraints, multipleefforts have been undertaken to develop an ML-based surrogate model that estimates resource usage of synthesized ML accelerator architectures. We introduce wa-hls4ml, a benchmark for ML accelerator resource and latency estimation, and its corresponding initial dataset of over 680 000 fully connected and convolutional neural networks, all synthesized using hls4ml and targeting Xilinx FPGAs. The benchmark evaluates the performance of resource and latency predictors against several common ML model architectures, primarily originating from scientific domains, as exemplar models, and the average performance across a subset of the dataset. Additionally, we introduce GNN- and transformer-based surrogate models that predict latency and resources for ML accelerators. We present the architecture and performance of the models and find that the models generally predict latency and resources for the 75% percentile within several percent of the synthesized resources on the synthetic test dataset.

Hawks, Benjamin G. [Fermilab]↗

COMPENSATION OF FAST KICKER ROLLS WITH SKEW QUADRUPOLES

The development of the third generation light sources lead to the implementation of the top-up operation, when injection occurs while users collect data. The beam excursions due to the non-closure of the injection bump can spoil the data and need to be suppressed. In the horizontal plane compensation can be achieved by adjusting timing and kick amplitudes. The rolls of the kicker magnets create non-closure in the vertical plane and usually there is no means for correction. In the paper we describe proposed compensation scheme utilizing two skew quadrupoles placed inside the injection bump. The third generation light sources implement top-up operation firstly introduced at Advanced Photon Source. In this mode the circulating beam current is supported near constant by frequent injection of small charge, while photon beam is delivered for users. The beam perturbations caused by the mismatched injection bump can provide undesired noise in the user data. Usually the injection trigger is distributed to the users end stations so that those affected would be able to blank data acquisition. Nevertheless, as good operational practice such transients should be suppressed as much as possible. In the horizontal plane (which is commonly used for injection) one can adjust individual kicker strength as well as trigger delay while observing motion of the stored beam centroid. In the vertical plane such means are unavailable in the most cases. The possible solutions include dedicated weak vertical kickers and motorized adjustment of the roll angle of the injection kickers. Both abovementioned approaches are expensive and can significantly deteriorate reliability. We suggest two employ two skew quadrupoles (to correct both angle and position) placed inside the injection bump. In this case the beam position itself serves as measure of the kicker strength (assuming that kickers are well matched) and vertical kicks from the skew quadrupoles will be self synchronized with injection bump. In this paper we will consider the case when injection hardware (kickers and septa) are located in the same straight. Such an approach simplifies consideration but it can be generalized.

43 PARTICLE ACCELERATORS↗

A Test Bed to Evaluate Advanced Distribution Management Systems for Modern Power Systems

This paper presents a test bed that utilities, vendors, and researchers can use to evaluate the performance of existing and emerging advanced distribution management systems (ADMS) or specific applications of such systems. The test bed provides a realistic laboratory test setting, including detailed modeling of the power system, controller and/or power hardware, and industry-standard communication protocols. The test bed can further represent the current state of a specific utility's distribution system or the anticipated state with increased penetrations of distributed energy resources, such as rooftop photovoltaic systems, residential battery energy storage systems, electric vehicles, and loads with smart controls. The paper describes the design and implementation of the test bed. We present two use cases of the test bed and preliminary results for these two use cases. We also discuss upcoming and potential future use cases. Results from test bed experiments are intended to accelerate industry deployment of ADMS technologies and advance development of ADMS capabilities.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Implementation of a Particle Accelerator Beam Dynamics Code on Multi-Node GPUs

Particle accelerators play an important role in a wide range of scientific discoveries and industrial applications. The self-consistent multi-particle simulation based on the particle-in-cell (PIC) method has been used to study charged particle beam dynamics inside those accelerators. However, the PIC simulation is time-consuming and needs to use modern parallel computers for high-resolution applications. In this paper, we implemented a parallel beam dynamics PIC code on multi-node hybrid architecture computers with multiple Graphics Processing Units (GPUs). We used two methods to parallelize the PIC code on multiple GPUs and observed that the replication method is a better choice for moderate problem size and current computer hardware while the domain decomposition method might be a better choice for large problem size and more advanced computer hardware that allows direct communications among multiple GPUs. Using the multi-node hybrid architectures at Oak Ridge Leadership Computing Facility (OLCF), the optimized GPU PIC code achieves a reasonable parallel performance and scales up to 64 GPUs with 16 million particles.

43 PARTICLE ACCELERATORS↗

Introduction: Neuromorphic Materials

The explosive growth in data collection and the need to process it efficiently, as well as the desire to automate increasingly complex tasks in transportation, medical care, manufacturing, security and many other fields have motivated a growing interest in neuromorphic computing. Unlike the binary, transistorbased ON/OFF logic gates and separate logic and memory functionalities employed in digital computing, neuromorphic computing is inspired by animal brains that use interconnected synapses and neurons to perform processing, storage and transmission of information at the same location, while only consuming ~20 W or less of power. Motivated by the brain’s efficiency, adaptability, self-learning and resiliency qualities, neuromorphic computing can be broadly defined as an approach to processing and storing information using hardware and algorithms inspired by models of biological neural systems. Present research in neuromorphic computing encompasses approaches that vary significantly in their degree of neuro-inspiration, from systems that only incorporate features such as asynchronous, event-driven operation or use crossbar arrays of non-volatile memory (NVM) elements to accelerate deep neural networks (DNNs), to designs that embrace the extreme parallelism, sparsity, reconfigurability, adaptability, complexity and stochasticity observed in nervous systems. The term ‘neuromorphic’ computing is often credited to Carver Mead, who in the 1980s investigated Si-based analog electronics to replicate functions of the animal retina. Earlier important advances in this field include the work of Frank Rosenblatt, who proposed the concept of the perceptron, Bernard Widrow, who used this concept to build one of the first analog neural networks, the Adaline and many other researchers (see ref. 6 for an historical perspective on neuromorphic computing). With the recent increase in the use of artificial intelligence and large language models, and rising concerns over the associated energy costs, interest in neuromorphic hardware has expanded rapidly. According to some estimates, driven largely by the drastic growth in the training use of artificial intelligence (AI) models using the current computing architectures, the energy cost of computing is projected to reach the energy supply worldwide by 2045. Furthermore, while this is not a realistic outcome, it means that, if more efficient computing technologies are not developed -- soon -- the world will soon become one where demand for energy and market constraints limit the continued increase of societal access to AI and cloud services from data centers. Data centers used for training and use of these models consume hundreds of terawatt hours of electricity, already past 4% of the US electricity demand.

Circuits↗

RF low-level control for the Linac4 H{sup −} source

The H{sup −} source for the Linac4 accelerator at CERN uses an RF driven plasma for the production of H{sup −}. The RF is supplied by a 2 MHz RF tube amplifier with a maximum power output of 100 kW and a pulse duration of up to 2 ms. The low-level RF signal generation and measurement system has been developed using standard CERN controls electronics in the VME form factor. The RF frequency and amplitude reference signals are generated using separate arbitrary waveform generator channels. The frequency and amplitude are both freely programmable over the duration of the RF pulse, which allows fine-tuning of the excitation. Measurements of the forward and reverse RF power signals are performed via directional couplers using high-speed digitizers, and permit the estimation of the plasma impedance and deposited power via an equivalent circuit model. The low-level RF hardware and software implementations are described, and experimental results obtained with the Linac4 ion sources in the test stand are presented.

43 PARTICLE ACCELERATORS↗