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 253 records · Page 14

The Fermilab accelerator control system

For many years the Fermilab physics program has been dominated by the superconducting Tevatron accelerator producing beams for many fixed target and the proton-antiproton colliding beam experiments CDF and D0. More recently, major experiments have used beam from intermediate accelerators. The MiniBooNE and MINOS experiments use 8 and 120 GeV beam respectively for neutrino oscillation studies. Several other experiments and test beams have also used 120 GeV beam. This paper describes the control system for the accelerator complex that was originally developed for the start of Tevatron operation in 1983. This system is common to all accelerators in the chain, and has been successfully evolved to accommodate new hardware, new software platforms, new accelerators, and increasingly complex modes of operation.

43 PARTICLE ACCELERATORS↗

From Atoms to Wheels: The Role of Multi-Scale Modeling in the Future of Transportation Electrification

Traditionally, prototype hardware is built for validation testing to ensure battery systems design changes meet vehicle-level requirements, which is expensive both in cost and time. Virtual engineering (VE) of battery systems for electric vehicle (EV) propulsion offers a reduced-cost alternative to the traditional development process and uses multi-scale modeling to virtually probe the impact of design changes in a particular part on the overall performance of the system. This allows for rapid iteration over multiple design spaces, without committing to build hardware. This perspective article discusses current trends in VE for EV applications and proposes improvements to accelerate EV adoption.

Garrick, Taylor R. (ORCID:0000000322518129)↗

ARCHER{sub RT} – A GPU-based and photon-electron coupled Monte Carlo dose computing engine for radiation therapy: Software development and application to helical tomotherapy

Purpose: Using the graphical processing units (GPU) hardware technology, an extremely fast Monte Carlo (MC) code ARCHER{sub RT} is developed for radiation dose calculations in radiation therapy. This paper describes the detailed software development and testing for three clinical TomoTherapy® cases: the prostate, lung, and head and neck. Methods: To obtain clinically relevant dose distributions, phase space files (PSFs) created from optimized radiation therapy treatment plan fluence maps were used as the input to ARCHER{sub RT}. Patient-specific phantoms were constructed from patient CT images. Batch simulations were employed to facilitate the time-consuming task of loading large PSFs, and to improve the estimation of statistical uncertainty. Furthermore, two different Woodcock tracking algorithms were implemented and their relative performance was compared. The dose curves of an Elekta accelerator PSF incident on a homogeneous water phantom were benchmarked against DOSXYZnrc. For each of the treatment cases, dose volume histograms and isodose maps were produced from ARCHER{sub RT} and the general-purpose code, GEANT4. The gamma index analysis was performed to evaluate the similarity of voxel doses obtained from these two codes. The hardware accelerators used in this study are one NVIDIA K20 GPU, one NVIDIA K40 GPU, and six NVIDIA M2090 GPUs. In addition, to make a fairer comparison of the CPU and GPU performance, a multithreaded CPU code was developed using OpenMP and tested on an Intel E5-2620 CPU. Results: For the water phantom, the depth dose curve and dose profiles from ARCHER{sub RT} agree well with DOSXYZnrc. For clinical cases, results from ARCHER{sub RT} are compared with those from GEANT4 and good agreement is observed. Gamma index test is performed for voxels whose dose is greater than 10% of maximum dose. For 2%/2mm criteria, the passing rates for the prostate, lung case, and head and neck cases are 99.7%, 98.5%, and 97.2%, respectively. Due to specific architecture of GPU, modified Woodcock tracking algorithm performed inferior to the original one. ARCHER{sub RT} achieves a fast speed for PSF-based dose calculations. With a single M2090 card, the simulations cost about 60, 50, 80 s for three cases, respectively, with the 1% statistical error in the PTV. Using the latest K40 card, the simulations are 1.7–1.8 times faster. More impressively, six M2090 cards could finish the simulations in 8.9–13.4 s. For comparison, the same simulations on Intel E5-2620 (12 hyperthreading) cost about 500–800 s. Conclusions: ARCHER{sub RT} was developed successfully to perform fast and accurate MC dose calculation for radiotherapy using PSFs and patient CT phantoms.

60 APPLIED LIFE SCIENCES↗

Mixed-precision numerics in scientific applications: survey and perspectives

The explosive demand for artificial intelligence (AI) workloads has led to a significant increase in silicon area dedicated to lower-precision computations on recent high-performance computing hardware designs. However, mixed-precision capabilities, which can achieve performance improvements of up to 8x compared to double-precision in extreme compute-intensive workloads, remain largely untapped in most scientific applications. A growing number of efforts have shown that mixed-precision algorithmic innovations can deliver superior performance without sacrificing accuracy. These developments should prompt computational scientists to seriously consider whether their scientific modeling and simulation applications could benefit from the acceleration offered by new hardware and mixed-precision algorithms. In this survey, we (1) review progress across diverse scientific domains—fluid dynamics, weather and climate, quantum chemistry, and computational genomics—that have begun adopting mixed-precision strategies; (2) examine state-of-the-art algorithmic techniques such as iterative refinement, splitting and emulation schemes, and adaptive precision solvers; (3) assess their implications for accuracy, performance, and resource utilization; and (4) survey the emerging software ecosystem that enables mixed-precision methods at scale. We conclude with perspectives and recommendations on cross-cutting opportunities, domain-specific challenges, and the role of co-design between application scientists, numerical analysts, and computer scientists. Collectively, this survey underscores that mixed-precision numerics can reshape computational science by aligning algorithms with the evolving landscape of hardware capabilities.

Graphics processing units↗

Impacts of floating-point non-associativity on reproducibility for HPC and deep learning applications

Run to run variability in parallel programs caused by floating-point non-associativity has been known to significantly affect reproducibility in iterative algorithms, due to accumulating errors. Non-reproducibility can critically affect the efficiency and effectiveness of correctness testing for stochastic programs. Recently, the sensitivity of deep learning training and inference pipelines to floating-point non-associativity has been found to sometimes be extreme. It can prevent certification for commercial applications, accurate assessment of robustness and sensitivity, and bug detection. New approaches in scientific computing applications have coupled deep learning models with high-performance computing, leading to an aggravation of debugging and testing challenges. Here we perform an investigation of the statistical properties of floating-point non-associativity within modern parallel programming models, and analyze performance and productivity impacts of replacing atomic operations with deterministic alternatives on GPUs. We examine the recently-added deterministic options in PyTorch within the context of GPU deployment for deep learning, uncovering and quantifying the impacts of input parameters triggering run to run variability and reporting on the reliability and completeness of the documentation. Finally, we evaluate the strategy of exploiting automatic determinism that could be provided by deterministic hardware, using the Groq LPUTM accelerator for inference portions of the deep learning pipeline. We demonstrate the benefits that a hardware-based strategy can provide within reproducibility and correctness efforts.

Shanmugavelu, Sanjif↗

Build and Execute Environment

At exascale, the challenge becomes to develop applications that run at scale and use exascale platforms reliably, efficiently, and flexibly. Workflows become much more complex because they must seamlessly integrate simulation and data analytics. They must include down-sampling, post-processing, feature extraction, and visualization. Power and data transfer limitations require these analysis tasks to be run in-situ or in-transit. We expect successful workflows will comprise multiple linked simulations along with tens of analysis routines. Users will have limited development time at scale and, therefore, must have rich tools to develop, debug, test, and deploy applications. At this scale, successful workflows will compose linked computations from an assortment of reliable, well-defined computation elements, ones that can come and go as required, based on the needs of the workflow over time. We propose a novel framework that utilizes both virtual machines (VMs) and software containers to create a workflow system that establishes a uniform build and execution environment (BEE) beyond the capabilities of current systems. In this environment, applications will run reliably and repeatably across heterogeneous hardware and software. Containers, both commercial (Docker and Rocket) and open-source (LXC and LXD), define a runtime that isolates all software dependencies from the machine operating system. Workflows may contain multiple containers that run different operating systems, different software, and even different versions of the same software. We will run containers in open-source virtual machines (KVM) and emulators (QEMU) so that workflows run on any machine entirely in user-space. On this platform of containers and virtual machines, we will deliver workflow software that provides services, including repeatable execution, provenance, checkpointing, and future proofing. We will capture provenance about how containers were launched and how they interact to annotate workflows for repeatable and partial re-execution. We will coordinate the physical snapshots of virtual machines with parallel programming constructs, such as barriers, to automate checkpoint and restart. We will also integrate with HPC-specific container runtimes to gain access to accelerators and other specialized hardware to preserve native performance. Containers will link development to continuous integration. When application developers check code in, it will automatically be tested on a suite of different software and hardware architectures.

Guan, Qiang↗

Build and Execute Environment

At exascale, the challenge becomes to develop applications that run at scale and use exascale platforms reliably, efficiently, and flexibly. Workflows become much more complex because they must seamlessly integrate simulation and data analytics. They must include down-sampling, post-processing, feature extraction, and visualization. Power and data transfer limitations require these analysis tasks to be run in-situ or in-transit. We expect successful workflows will comprise multiple linked simulations along with tens of analysis routines. Users will have limited development time at scale and, therefore, must have rich tools to develop, debug, test, and deploy applications. At this scale, successful workflows will compose linked computations from an assortment of reliable, well-defined computation elements, ones that can come and go as required, based on the needs of the workflow over time. We propose a novel framework that utilizes both virtual machines (VMs) and software containers to create a workflow system that establishes a uniform build and execution environment (BEE) beyond the capabilities of current systems. In this environment, applications will run reliably and repeatably across heterogeneous hardware and software. Containers, both commercial (Docker and Rocket) and open-source (LXC and LXD), define a runtime that isolates all software dependencies from the machine operating system. Workflows may contain multiple containers that run different operating systems, different software, and even different versions of the same software. We will run containers in open-source virtual machines (KVM) and emulators (QEMU) so that workflows run on any machine entirely in user-space. On this platform of containers and virtual machines, we will deliver workflow software that provides services, including repeatable execution, provenance, checkpointing, and future proofing. We will capture provenance about how containers were launched and how they interact to annotate workflows for repeatable and partial re-execution. We will coordinate the physical snapshots of virtual machines with parallel programming constructs, such as barriers, to automate checkpoint and restart. We will also integrate with HPC-specific container runtimes to gain access to accelerators and other specialized hardware to preserve native performance. Containers will link development to continuous integration. When application developers check code in, it will automatically be tested on a suite of different software and hardware architectures.

Guan, Qiang↗

Generation of virtual monochromatic CBCT from dual kV/MV beam projections

Purpose: To develop a novel on-board imaging technique which allows generation of virtual monochromatic (VM) cone-beam CT (CBCT) with a selected energy from combined kilovoltage (kV)/megavoltage (MV) beam projections. Methods: With the current orthogonal kV/MV imaging hardware equipped in modern linear accelerators, both MV projections (from gantry angle of 0°–100°) and kV projections (90°–200°) were acquired as gantry rotated a total of 110°. A selected range of overlap projections between 90° to 100° were then decomposed into two material projections using experimentally determined parameters from orthogonally stacked aluminum and acrylic step-wedges. Given attenuation coefficients of aluminum and acrylic at a predetermined energy, one set of VM projections could be synthesized from two corresponding sets of decomposed projections. Two linear functions were generated using projection information at overlap angles to convert kV and MV projections at nonoverlap angles to approximate VM projections for CBCT reconstruction. The contrast-to-noise ratios (CNRs) were calculated for different inserts in VM CBCTs of a CatPhan phantom with various selected energies and compared with those in kV and MV CBCTs. The effect of overlap projection number on CNR was evaluated. Additionally, the effect of beam orientation was studied by scanning the CatPhan sandwiched with two 5 cm solid-water phantoms on both lateral sides and an electronic density phantom with two metal bolt inserts. Results: Proper selection of VM energy [30 and 40 keV for low-density polyethylene (LDPE), polymethylpentene, 2 MeV for Delrin] provided comparable or even better CNR results as compared with kV or MV CBCT. An increased number of overlap kV and MV projection demonstrated only marginal improvements of CNR for different inserts (with the exception of LDPE) and therefore one projection overlap was found to be sufficient for the CatPhan study. It was also evident that the optimal CBCT image quality was achieved when MV beams penetrated through the heavy attenuation direction of the object. Conclusions: A novel technique was developed to generate VM CBCTs from kV/MV projections. This technique has the potential to improve CNR at selected VM energies and to suppress artifacts at appropriate beam orientations.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

A Heterogeneity-Aware Task Scheduler for Spark

Big data processing systems such as Spark are employed in an increasing number of diverse applications-such as machine learning, graph computation, and scientific computing-each with dynamic and different resource needs. These applications increasingly run on heterogeneous hardware, e.g., with out-of-core accelerators. However, big data platforms do not factor in the multi-dimensional heterogeneity of applications and hardware. This leads to a fundamental mismatch between the application and hardware characteristics, and the resource scheduling adopted in big data platforms. For example, Hadoop and Spark consider only data locality when assigning tasks to nodes, and typically disregard the hardware capabilities and suitability to specific application requirements. In this paper, we present RUPAM, a heterogeneity-aware task scheduling system for big data platforms, which considers both task-level resource characteristics and underlying hardware characteristics, as well as preserves data locality. RUPAM adopts a simple yet effective heuristic to decide the dominant scheduling factor (e.g., CPU, memory, or I/O), given a task in a particular stage. Our experiments show that RUPAM is able to improve the performance of representative applications by up to 62.3% compared to the standard Spark scheduler.

Computer Science↗

Development of FELIX based readout system for HV-CMOS sensor testbeam

The High Voltage CMOS (HV-CMOS) sensors are extensively investigated by the ATLAS collaboration for the High-Luminosity LHC (HL-LHC) upgrade of the Inner Tracker (ITk) detector. The HV-CMOS technology is a commercial integrated circuit process with the advantages of monolithic readout and fast charge collection. A front-end test platform of CaRIBOu (Control and Readout Itk BOard) and a testbeam Telescope based on the ATLAS IBL (Insertable B-Layer) silicon pixel modules, have been developed for the HV-CMOS sensor study and characterization in testbeam experiments. The Front-End LInk eXchange (FELIX) system is a newapproach to function as the gateway between detector front-end electronics and the commodity switched network in the ATLAS upgrade. A FELIX based readout system has been developed for both the CaRIBOu and the testbeam Telescope. It has been deployed successfully in the testbeam at theCERNSPS H8 beamline in August 2017. The test results show that the readout system is capable of sensor calibration and readout of a high-density pixel detector with high trigger rate in testbeam experiments.

47 OTHER INSTRUMENTATION↗

Embedded EPICS server for PowerPMAC motion controllers

An embedded server layer of Experimental Physics and Industrial Control System (EPICS) for PowerPMAC motion controllers has been developed and deployed at two undulator beamlines of the National Institute of General Medical Sciences and the National Cancer Institute (GM/CA) Structural Biology Facility at the Advanced Photon Source (APS). This compact, open source solution makes the power and versatility of PowerPMAC motion controls directly accessible to distributed EPICS clients. At GM/CA the system controls about 200 servo and stepper motors — both encoded and unencoded — and multiple digital and analog I/O accessories. The server stack comprises two sublayers: a lower-level driver and database that communicates directly with PowerPMAC, and a facility-specific soft sublayer built on top. The paper describes installing EPICS on PowerPMAC, the implementation of both layers and client examples, including on-the-fly scanning.

EPICS↗

Caspian: A Neuromorphic Development Platform

Current neuromorphic systems often may be difficult to use and costly to deploy. There exists a need for a simple yet flexible neuromorphic development platform which can allow researchers to quickly prototype ideas and applications. Caspian offers a high-level API along with a fast spiking simulator to enable the rapid development of neuromorphic solutions. It further offers an FPGA architecture that allows for simplified deployment -- particularly in SWaP (size, weight, and power) constrained environments. Leveraging both software and hardware, Caspian aims to accelerate development and deployment while enabling new researchers to quickly become productive with a spiking neural network system.

Mitchell, Parker↗

Shaker Table Test Plan

Currently, spent nuclear fuel (SNF) is stored in onsite independent spent fuel storage facilities (ISFSIs), which is a dry storage facility, at 55 nuclear power plant sites. The majority of SNF in dry storage is in welded metal canisters (2,917 canisters at the end of 2019). The canisters are loaded for storage in storage overpacks (vertical casks or horizontal storage modules) and placed on outdoor concrete pads. Because the SNF will be stored at ISFSIs for an extended period of time, there is growing concern with regards to the behavior of the SNF within these dry storage systems during earthquakes. To address these concerns, the SFWST program is considering conducting an earthquake shaker table test. The goal of this test is to determine the strains and accelerations on fuel assembly hardware and cladding during earthquakes of different magnitudes to better quantify the potential damage an earthquake could inflict on spent nuclear fuel rods. The seismic integrity of the storage system has been addressed in the past by the US Nuclear Regulatory Commission and is not the focus of this potential test. Instead the DOE would benefit from knowing the condition of the fuel cladding from storage, transportation, to disposal so that it can ascertain repository performance for the fuel and packaging in its final state. A seismic event is part of the possible loading events that the fuel could experience in its lifetime. This report proposes several earthquake shaker table tests with different degrees of complexity. Alternative 1 was defined in the FY20 work scope. Alternatives 2 and 3 were recently developed to take advantage of the NUHOMS 32PTH dry storage canister that may be available in FY21 for this test at a minimum cost to the project. The selection of the alternative(s) will depend on the available budget and the SFWST program priorities for the near future.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗

Develop a Fast Analysis Solver for Welding Sequence Optimization

During the shipbuilding manufacturing process, materials are exposed to significant stresses, as induced both thermally and mechanically, that alter the intended design and significantly affect the production schedule, labor hours (fitting, welding, rework, etc.), and material structural performance. The type and magnitude of deformation of a given structure depends on many factors such as the material, thickness and quality of components, the process heat input, preheat and inter-pass temperatures, type and size of welds, welding sequence and direction, location, sequence, and degree of fixturing. Numerical simulations using finite element analysis (FEA) have long been used to analyze welding-induced structural distortion. For large assemblies, transient thermal elastic-plastic analysis (TEPA) can take days or weeks to run, and optimization of welding sequence is not feasible. Simplified analysis methods were developed to reduce computational time. However, it is challenging to use these techniques to fully optimize welding sequencing because of their applied simplifications in modeling weld details. A fast analysis solver that could be used by the shipbuilding industry is being developed for optimizing welding sequences by taking full advantage of modern GPU-based HPC hardware and incorporating patented acceleration schemes. The accelerated processing factors are up to 2200 times greater for large, multi-pass welded structures.

Yang, Yu-Ping↗

OpenCGRA: An Open-Source Unified Framework for Modeling,Testing, and Evaluating CGRAs

Coarse-grained reconfigurable arrays (CGRAs),loosely defined as arrays of functional units (e.g, adder, sub-tractor, multiplier, divider, or larger multi-operation units, butsmaller than a general-purpose core) interconnected through aNetwork-on-Chip, provide higher flexibility than domain-specificASIC accelerators while offering increased hardware efficiencywith respect to fine-grained reconfigurable devices, such as FieldProgrammable Gate Arrays (FPGAs). The fast evolving fieldsof machine learning and edge computing, which are seeing acontinuous flow of novel algorithms and larger models, makeCGRAs ideal target architectures to allow domain specializationwithout loosing too much generality. They also generally offerquicker and more effective reconfigurability than FPGAs, po-tentially allowing adaptation during actual algorithm execution,and implement a dataflow programming paradigm that adaptswell to these emerging workloads. Designing and generating aCGRA, however, still requires to define the type and number ofthe specific functional units, implement their interconnect andthe network topology, and perform its simulation and validation,given a variety of workloads of interest.In this paper, we propose OpenCGRA, a Python-based unifiedframework that integrates generation, modeling, testing and eval-uation for CGRAs. OpenCGRA is the first open-source integratedframework able to support the full top-to-bottom design flow forspecializing and implementing CGRAs: modeling at different ab-straction levels (functional level, cycle level, register-transfer level),generation, simulation, testing at different granularities (unit test-ing, integration testing, property-based testing), and characteriza-tion (area, power, and timing). OpenCGRAs will be made availableon GitHub.

CGRA, synthesis↗