Search NASA⌕ Search

SEARCH · Search NASA

Results for “accelerator performance”

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 91 records · Page 5
All sourcesNASA NTRSDOE OSTI
DOE OSTI2024

XaaS: Acceleration as a Service to Enable Productive High-Performance Cloud Computing

High-performance computing (HPC) and the cloud have evolved independently, specializing their innovations into performance or productivity. Acceleration as a Service (XaaS) is a recipe to empower both fields with a shared execution platform that provides transparent access to computing resources, regardless of the underlying cloud or HPC service provider. Bridging HPC and cloud advancements, XaaS presents a unified architecture built on performance-portable containers. Here, our converged model concentrates on low-overhead, high-performance communication and computing, targeting resource-intensive workloads from climate simulations to machine learning. XaaS lifts the restricted allocation model of Function as a Service (FaaS), allowing users to benefit from the flexibility and efficient resource utilization of serverless computing while supporting long-running and performance-sensitive workloads from HPC.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2022

Challenges of Future Accelerators for Particle Physics Research

For over half a century, high-energy particle accelerators have been a major enabling technology for particle and nuclear physics research as well as sources of X-rays for photon science research in material science, chemistry and biology. Particle accelerators for energy and intensity Frontier research in particle and nuclear physics continuously push the accelerator community to invent ways to increase the energy and improve the performance of accelerators, reduce their cost, and make them more power efficient. The accelerator community has demonstrated imagination and creativity in developing a plethora of future accelerator ideas and proposals. The technical maturity of the proposed facilities ranges from shovel-ready to those that are still largely conceptual. At this time, over 100 contributed papers have been submitted to the Accelerator Frontier of the US particle physics decadal community planning exercise known as Snowmass’2021. These papers cover a broad spectrum of topics: beam physics and accelerator education, accelerators for neutrinos, colliders for Electroweak/Higgs studies and multi-TeV energies, accelerators for Physics Beyond Colliders and rare processes, advanced accelerator concepts, and accelerator technology for Radio Frequency cavities (RF), magnets, targets and sources. This paper provides an overview of the present state of accelerators for particle physics and gives a brief description of some of the major facilities that have been proposed, their perceived advantages and some of the remaining challenges.

43 PARTICLE ACCELERATORS↗
DOE OSTI2020

Liquid helium fluid dynamics studies. Final Technical Report

Future high energy physics accelerators depend on a number of advanced technologies to open the many doors of scientific discovery. Among these advanced technologies, superconducting magnets and superconducting radio frequency (SRF) cavities are the backbone of the accelerator and detector systems. But all these low temperature systems depend critically on successful and reliable operation of their supporting technologies, among which the liquid helium cooling system is of the utmost importance. To improve the quality of these systems both in terms of efficiency and reliability, a robust helium cryogenics research and development (R&D) effort is required. The proposed research to be conducted by the FSU cryogenics group aims to produce fundamental knowledge that meets this R&D need. The projects that we have completed over the past grant period at Florida State University consist of experimental research on liquid helium fluid dynamics and heat transfer problems relevant to the development of future superconducting particle physics accelerators. Liquid helium is the coolant used in all such facilities and in many of these facilities He II (the low temperature phase of liquid helium also known as superfluid helium) is preferred due to its outstanding heat transfer characteristics. The work consists of two main experimental studies that probe both fundamental as well as practical aspects of liquid helium cooling. The first is a broad and fundamental study of the heat and mass transfer processes that can occur during a sudden catastrophic loss of vacuum (SCLV) incident in a superconducting accelerator. SCLV refers to the remote but extremely critical accident scenario where atmospheric pressure air is allowed to flood into the insulating vacuum system and impinge on the liquid helium cooled surfaces in the accelerator. Safe performance and recovery from such accidents is essential to the reliable operation of superconducting accelerators. The dynamics of this process is quite complex and so our approach is to conduct a series of well-orchestrated experiments that probe the various physical phenomena that can occur during an SCLV event. The experiments are coupled with analytic and numerical analysis in an effort to develop a general understanding of the process and to assist with future accelerator design and development. The second activity is directed toward fundamental understanding of heat and mass transfer in He II, which is essential to the design of superconducting magnets and radio frequency cavities in accelerators. The work consists of flow visualization of the dynamics of He II using laser assisted techniques. Two complementary techniques are used to study the fundamentals of the turbulent state. The first technique uses neutrally buoyant solid hydrogen particles to probe the flow fields of the superfluid and normal fluid components. The other technique uses laser excited He2* molecules as tracers of the normal fluid motion within the He II. The activities also included an effort to use visualization techniques to locate transient hot spots in radio frequency superconducting cavities. Such work provides valuable information about the heat transfer process in He II and its impact on the performance of superconducting devices. The research effort at Florida State University is not directly in support of a specific high energy physics experiment or facility. Rather, the work is general and coordinated with HEP accelerator laboratories to provide valuable insight that can assist with future accelerator development.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗
DOE OSTI2020

Online accelerator optimization with a machine learning-based stochastic algorithm

Abstract Online optimization is critical for realizing the design performance of accelerators. Highly efficient stochastic optimization algorithms are needed for many online accelerator optimization problems in order to find the global optimum in the non-linear, coupled parameter space. In this study, we propose to use the multi-generation Gaussian process optimizer for online accelerator optimization and demonstrate that the algorithm is significantly more efficient than other stochastic algorithms that are commonly used in the accelerator community.

Zhang, Zhe (ORCID:0000000281430381)↗
DOE OSTI2023

HipBone: A performance-portable graphics processing unit-accelerated C++ version of the NekBone benchmark

We present hipBone, an open-source performance-portable proxy application for the Nek5000 (and NekRS) computational fluid dynamics applications. HipBone is a fully GPU-accelerated C++ implementation of the original NekBone CPU proxy application with several novel algorithmic and implementation improvements which optimize its performance on modern fine-grain parallel GPU accelerators. Our optimizations include a conversion to store the degrees of freedom of the problem in assembled form in order to reduce the amount of data moved during the main iteration and a portable implementation of the main Poisson operator kernel. We demonstrate near-roofline performance of the operator kernel on three different modern GPU accelerators from two different vendors. We present a novel algorithm for splitting the application of the Poisson operator on GPUs which aggressively hides MPI communication required for both halo exchange and assembly. Our implementation of nearest-neighbor MPI communication then leverages several different routing algorithms and GPU-Direct RDMA capabilities, when available, which improves scalability of the benchmark. We demonstrate the performance of hipBone on three different clusters housed at Oak Ridge National Laboratory, namely, the Summit supercomputer and the Frontier early-access clusters, Spock and Crusher. Our tests demonstrate both portability across different clusters and very good scaling efficiency, especially on large problems.

Computer Science↗
DOE OSTI2024

Toward a Holistic Performance Evaluation of Large Language Models Across Diverse AI Accelerators

Artificial intelligence (AI) methods have become critical in scientific applications to help accelerate scientific discovery. Large language models (LLMs) are being considered a promising approach to address some challenging problems because of their superior generalization capabilities across domains. The effectiveness of the models and the accuracy of the applications are contingent upon their efficient execution on the underlying hardware infrastructure. Specialized Al accelerator hardware systems have recently become available for accelerating Al applications. However, the comparative performance of these AI accelerators on large language models has not been previously studied. In this paper, we systematically study LLMs on multiple AI accelerators and GPUs and evaluate their performance characteristics for these models. We evaluate these systems with (i) a micro-benchmark using a core transformer block, (ii) a GPT-2 model, and (iii) an 1,I,M-driven science use case, GenSLM. We present our findings and analyses of the models' performance to better understand the intrinsic capabilities of AI accelerators. Furthermore, our analysis takes into account key factors such as sequence lengths, scaling behavior, and sensitivity to gradient accumulation steps.

Emani, Murali↗
DOE OSTI2024

Application of performance portability solutions for GPUs and many-core CPUs to track reconstruction kernels

Next generation High-Energy Physics (HEP) experiments are presented with significant computational challenges, both in terms of data volume and processing power. Using compute accelerators, such as GPUs, is one of the promising ways to provide the necessary computational power to meet the challenge. The current programming models for compute accelerators often involve using architecture-specific programming languages promoted by the hardware vendors and hence limit the set of platforms that the code can run on. Developing software with platform restrictions is especially unfeasible for HEP communities as it takes significant effort to convert typical HEP algorithms into ones that are efficient for compute accelerators. Multiple performance portability solutions have recently emerged and provide an alternative path for using compute accelerators, which allow the code to be executed on hardware from different vendors. We apply several portability solutions, such as Kokkos, SYCL, C++17 std::execution::par and Alpaka, on two mini-apps extracted from the mkFit project: p2z and p2r. These apps include basic kernels for a Kalman filter track fit, such as propagation and update of track parameters, for detectors at a fixed z or fixed r position, respectively. The two mini-apps explore different memory layout formats. We report on the development experience with different portability solutions, as well as their performance on GPUs and many-core CPUs, measured as the throughput of the kernels from different GPU and CPU vendors such as NVIDIA, AMD and Intel.

Kwok, Ka Martin↗
DOE OSTI2024

Application of performance portability solutions for GPUs and many-core CPUs to track reconstruction kernels

Next generation High-Energy Physics (HEP) experiments are presented with significant computational challenges, both in terms of data volume and processing power. Using compute accelerators, such as GPUs, is one of the promising ways to provide the necessary computational power to meet the challenge. The current programming models for compute accelerators often involve using architecture-specific programming languages promoted by the hardware vendors and hence limit the set of platforms that the code can run on. Developing software with platform restrictions is especially unfeasible for HEP communities as it takes significant effort to convert typical HEP algorithms into ones that are efficient for compute accelerators. Multiple performance portability solutions have recently emerged and provide an alternative path for using compute accelerators, which allow the code to be executed on hardware from different vendors. We apply several portability solutions, such as Kokkos, SYCL, C++17 std::execution::par, Alpaka, and OpenMP/OpenACC, on two mini-apps extracted from the mkFit project: p2z and p2r. These apps include basic kernels for a Kalman filter track fit, such as propagation and update of track parameters, for detectors at a fixed z or fixed r position, respectively. The two mini-apps explore different memory layout formats. We report on the development experience with different portability solutions, as well as their performance on GPUs and many-core CPUs, measured as the throughput of the kernels from different GPU and CPU vendors such as NVIDIA, AMD and Intel.

Kwok, Ka Hei Martin↗
DOE OSTI2022

Data Processing Unit Services Module

The Data Processing Services Module (DPUSM) provides the ability to perform pluggable compression, erasure coding, checksuming and other important file system operations within the Linux kernel. The pluggable provider interface allows for the use of hardware acceleration of those services. In-kernel file systems are then able to use these functions to use these accelerators to perform operations that are normally run on the processor, resulting in improved file system performance. Third parties will register "providers" with the DPUSM to communicate with their respective accelerators. Providers will implement functions with DPUSM API signatures so that the DPUSM can translate the data inputted by users of the DPUSM into data that providers recognize.

Lee, Jason↗
DOE OSTI2023

LANSCE CCL Performance Limits

This report summarizes the performance limits of the LANSCE Coupled-Cavity Linac (CCL). These results are captured or summarized directly from the references cited. This report was written in support of the LANSCE Modernization Project (LAMP). Many factors contribute to the performance limits of the CCL accelerator system. One performance limit is set by the CCL mechanical structure and available cooling. Heating of the structure is ultimately linked to the operating RF duty factor of the CCL. Another performance limit is set by the maximum RF and beam duty factors that can be supported by the 805-MHz klystrons and associated HVDC power supplies. The RF and HVDC systems also set limits for the maximum peak beam current that can be accelerated in the CCL. And finally, a maximum beam current limit is set by the beam dynamics determined by the details of the CCL physics design (number of cells per CCL tank, accelerating gradients, magnetic focusing lattice, beam losses, etc.). This limit can be informed by both simulation results and beam measurement data, if available. A detailed explanation of each performance limit is given in the sections below. The results are summarized in the table below. In all cases, it is assumed that the CCL and the RF system are operating in their nominal beam production configuration – nominal magnet set points and nominal cavity fields, unless otherwise specified.

43 PARTICLE ACCELERATORS↗
DOE OSTI2020

Development of a CPU/GPU portable software library for Lagrangian–Eulerian simulations of liquid sprays

The Lagrangian–Eulerian method is widely used for simulations of fuel sprays in turbulent combustion because of the advantage of treating the spray droplets as discrete points. One challenge of the Lagrangian–Eulerian method is the intense computational requirement when tracking the large number of Lagrangian particles needed for high fidelity. We have developed a performance-portable library, Grit, to track the Lagrangian particles in parallel on central processing unit (CPU) and graphics processing unit (GPU) accelerated high performance computing (HPC) architectures. Grit is a C++ library which employs Message Passing Interface (MPI) for distributed memory parallelism and Kokkos programming model for on-node shared memory parallelism with performance portability across different architectures of GPUs and multi-core/manycore CPUs. The parallel algorithms, key parallel kernels, and their performances on the pre-exascale supercomputer, Summit, are presented. Grit is coupled with a direct numerical simulation (DNS) solver, S3D, for multiphase simulations. A conservative formulation has been developed and implemented in Grit for phase coupling with thermodynamic consistency. The formulation separates the conservation of mass, momentum and energy from the physical models to prevent accidental violation of conservation laws due to inconsistent models. The formulation also enforces consistent definitions of enthalpies of the fuel for both phases and the latent heat of evaporation. Finally, simulations of turbulent particle-laden flow and the evaporation of dilute turbulent spray jet are performed to verify the software implementation, and to demonstrate the scalability of Grit for large-scale multiphase simulations.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2021

Cooperative Research and Development Agreement among National Energy Technology Laboratory, Advanced Manufacturing LLC, and Connecticut Center for Advanced Technology [Abstract]

The National Energy Technology Laboratory (NETL), Advanced Manufacturing LLC (AMLLC) and Connecticut Center for Advanced Technology (CCAT) will collaborate to develop and manufacture cost-effective, oxide dispersion-strengthened, NiCrFeCo-rich high entropy alloys that are superior to Nickel-based superalloys (e.g. IN740) for repair or replacement service in extreme environments. The new alloys can result in higher efficiency and reduced cost of advanced power systems such as advanced ultra-supercritical (AUSC) boilers and gas turbines. High throughput, multiscale computational modeling will be performed to accelerate alloy discovery by interrogating the intrinsic properties of the alloys including thermodynamic, kinetic, and mechanical properties. Validation and evaluation will be performed, including additive manufacturing, microstructure characterization, and mechanical properties tests.

36 MATERIALS SCIENCE↗
DOE OSTI2022

Third-Order Resonance Compensation at the FNAL Recycler Ring

The Recycler Ring (RR) at the Fermilab Accelerator Complex performs slip-stacking on 8 GeV protons, doubling the beam intensity delivered to the Main Injector (MI). At MI, the beam is accelerated to 120 GeV and delivered to the high energy neutrino experiments. Femilab’s Proton Improvement Plan II (PIP-II) will require the Recycler to store 50% more beam. Simulations have shown that the space charge tune shift at this new intensity will lead to the excitation of multiple resonance lines. Specifically, this study looks at normal sextupole lines 3 Q_{x}=76 and Q_{x}+2Q_{y}=74, plus skew sextupole lines 3 Q_{y}=73 and 2 Q_{x}+Q_{y}=75. Dedicated normal and skew sextupoles have been installed in order to compensate for these resonance lines. By measuring and calculating the Resonance Driving Terms (RDT), this study shows how each of the resonance lines can be compensated independently. Furthermore, this study shows and discusses initial investigations into compensating multiple lines simultaneously.

43 PARTICLE ACCELERATORS↗
DOE OSTI2023

Towards On-Chip Learning for Low Latency Reasoning with End-to-End Synthesis

The Software Defined Architectures (SODA) Synthesizer is an open-source compiler-based tool able to automatically generate domain-specialized systems targeting Application-Specific Integrated Circuits (ASICs) or Field Programmable Gate Arrays (FPGAs) starting from high-level programming. SODA is composed of a frontend, SODA-OPT, which leverages the multilevel intermediate representation (MLIR) framework to interface with productive programming tools (e.g., machine learning frame-works), identify kernels suitable for acceleration, and perform high-level optimizations, and of a state-of-the-art high-level synthesis backend, Bambu from the PandA framework, to generate custom accelerators. One specific application of the SODA Synthesizer is the generation of accelerators to enable ultra-low latency inference and control on autonomous systems for scientific discovery (e.g., electron microscopes, sensors in particle accelerators, etc.). This paper provides an overview of the flow in the context of the generation of accelerators for edge processing to be integrated in transmission electron microscopy (TEM) devices, focusing on use cases from precision material synthesis. We show the tool in action with an example of design space exploration for inference on reconfigurable devices with a conventional deep neural network model (LeNet). Finally, we discuss the research directions and opportunities enabled by SODA in the area of autonomous control for scientific experimental workflows.

Castellana, Vito G.↗
DOE OSTI2024

Towards Automated Generation of Chiplet-Based Systems

The Software Defined Architectures (SODA) Synthesizer is an open-source compiler-based tool able to automatically generate domain-specialized systems targeting Application- Specific Integrated Circuits (ASICs) or Field Programmable Gate Arrays (FPGAs) starting from high-level programming. SODA is composed of a high-level frontend, SODA-OPT, which leverages the multilevel intermediate representation (MLIR) framework to interface with productive programming tools (e.g., machine learning frameworks), identify kernels suitable for acceleration, and perform high-level optimizations, and of a state-of-the-art high-level synthesis backend, Bambu from the PandA framework, to generate custom accelerators. One specific application of the SODA Synthesizer is the generation of accelerators to enable ultra-low latency inference and control on autonomous systems for scientific discovery (e.g., electron microscopes, sensors in particle accelerators, etc.). This talk will discuss ongoing work on the SODA synthesizer to enable no-human-in-the-loop generation and design space exploration of the chiplets for highly specialized artificial intelligence accelerators. Connecting these highly specialized chiplets to general-purpose cores or programmable accelerators will allow to quickly deploy autonomous systems for scientific discovery.

Limaye, Ankur M.↗
DOE OSTI2022

Effect of Commercial Gas Diffusion Layers on Catalyst Durability of Polymer Electrolyte Fuel Cells in Varied Cathode Gas Environment

Gas diffusion layers (GDLs) play a crucial role in heat transfer and water management of cathode catalyst layers in polymer electrolyte fuel cells (PEFCs). Thermal and water gradients can accelerate electrocatalyst degradation and therefore the selection of GDLs can have a major influence on PEFC durability. Currently, the role of GDLs in electrocatalyst degradation is poorly studied. Here, we perform electrocatalyst accelerated stress test studies on membrane electrode assemblies (MEAs) prepared using three most commonly used GDLs. The effect of GDLs on electrocatalyst degradation is evaluated in both nitrogen (non-reactive) and air (reactive) gas environments at 100% relative humidity. In situ electrochemical characterization and extensive physical characterization is performed to understand the subtle differences in electrocatalyst degradation and correlated to the use of different GDLs. Overall, no difference is observed in the electrocatalyst degradation due to GDLs based on polarization curves at the end of life. But interestingly, MEA with a cracked microporous layer (MPL) in the GDL exhibited a higher electrocatalyst loading loss, which resulted in a lower and more heterogeneous increase in the average electrocatalyst nanoparticle size.

25 ENERGY STORAGE↗
DOE OSTI2020

Accelerators for HEP: Challenges and R&D

Particle accelerators are arguably the most effective tools for the particle physics research. The physics needs continuously push us to invent novel ways to increase energy and improve performance of accelerators, reduce their cost and make them more power efficient. Here we briefly overview three main families of modern and future accelerators for HEP – high intensity accelerators for neutrino research, particle factories for the electro-weak physics and Higgs studies, and post-LHC energy frontier colliders – and discuss corresponding ongoing or planned accelerator R&D topics and objectives.

43 PARTICLE ACCELERATORS↗
DOE OSTI2025

Testing and Expertise for Marine Energy (TEAMER) Program Support: MRE Dynamic Seals Performance Investigation (CRADA Final Report)

This TEAMER project allows NREL to use and augment a custom special-purpose testing system to perform accelerated-life testing of rotating seals for use in the Marine Renewable Energy (“MRE”) industry. The requirements of this industry are different from most applications including marine, and relevant performance information is lacking. In order to achieve low costs with extremely effective and reliable operation over long periods deployed underwater, more study of the best seals and arrangements of seals must be performed. By developing a rigorous framework for such testing, NREL hopes to greatly improve understanding of the relevant factors and provide significant data to the industry. Further, this work can inform standards developments for this critical component of MRE systems, leading to maturation and improved acceptance of the industry in commerce.

16 TIDAL AND WAVE POWER↗