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 487 records · Page 27

Understanding GPU errors on large-scale HPC systems and the implications for system design and operation

Increase in graphics hardware performance and improvements in programmability has enabled GPUs to evolve from a graphics-specific accelerator to a general-purpose computing device. Titan, the world's second fastest supercomputer for open science in 2014, consists of more dum 18,000 GPUs that scientists from various domains such as astrophysics, fusion, climate, and combustion use routinely to run large-scale simulations. Unfortunately, while the performance efficiency of GPUs is well understood, their resilience characteristics in a large-scale computing system have not been fully evaluated. We present a detailed study to provide a thorough understanding of GPU errors on a large-scale GPU-enabled system. Our data was collected from the Titan supercomputer at the Oak Ridge Leadership Computing Facility and a GPU cluster at the Los Alamos National Laboratory. We also present results from our extensive neutron-beam tests, conducted at Los Alamos Neutron Science Center (LANSCE) and at ISIS (Rutherford Appleron Laboratories, UK), to measure the resilience of different generations of GPUs. We present several findings from our field data and neutron-beam experiments, and discuss the implications of our results for future GPU architects, current and future HPC computing facilities, and researchers focusing on GPU resilience.

Tiwari, Devesh↗

Building Scalable PGAS Communication Subsystem on Blue Gene/Q

This paper presents a design of scalable Partitioned Global Address Space (PGAS) communication subsystems on recently proposed Blue Gene/Q architecture. The proposed design provides an in-depth modeling of communication infrastructure using Parallel Active Messaging Interface (PAMI). The communication infrastructure is used to design time-space efficient communication protocols for frequently used data-types (contiguous, uniformly non-contiguous) using Remote Direct Memory Access (RDMA) get/put primitives. The proposed design accelerates load balance counters by using asynchronous threads, which are required due to the missing network hardware support for Atomic Memory Operations (AMOs). Under the proposed design, the synchronization traffic is reduced by tracking conflicting memory accesses in distributed space with slightly increment in space complexity. An evaluation with simple communication benchmarks show a adjacent node get latency of 2.89$\mu$s and peak bandwidth of 1775 MB/s resulting in $\approx$ 99\% communication efficiency.The evaluation shows a reduction in the execution time by up to 30\% for NWChem self consistent field calculation on 4096 processes using the proposed asynchronous thread based design.

Vishnu, Abhinav↗

Towards a study protocol: A data-driven workflow to identify error sources in direct ink write mechatronics

Abstract Using Direct Ink Write (DIW) technology in a rapid and large-scale production requires reliable quality control for printed parts. Data streams generated during printing, such as print mechatronics, are massive and diverse which impedes extracting insights. In our study protocol approach, we developed a data-driven workflow to understand the behavior of sensor-measured X- and Y- axes positional errors with process parameters, such as print velocity and velocity control. We uncovered patterns showing that instantaneous changes in the velocity, when the build platform accelerates and decelerates, largely influence the positional errors, especially in the X- axis due to the hardware architecture. Since DIW systems share similar mechatronic inputs and outputs, our study protocol approach is broadly applicable and scalable across multiple systems. Graphical abstract

36 MATERIALS SCIENCE↗

Vehicle Systems Integration Laboratory Accelerates Powertrain Development

ORNL's Vehicle Systems Integration (VSI) Laboratory accelerates the pace of powertrain development by performing prototype research and characterization of advanced systems and hardware components. The VSI Lab is capable of accommodating a range of platforms from advanced light-duty vehicles to hybridized Class 8 powertrains with the goals of improving overall system efficiency and reducing emissions.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Accelerating Transformer-based Deep Learning Models on FPGAs using Column Balanced Block Pruning

Although Transformer-based language representations achieve state-of-the-art accuracy on various natural language processing (NLP) tasks, the large model size has been challenging the resource constrained computing platforms. Weight pruning, as a popular and effective technique in reducing the number of weight parameters and accelerating the Transformer, has been investigated on GPUs. However, the Transformer acceleration using weight pruning on field-programmable gate array (FPGAs) remains unexplored. This paper investigates the column balanced block-wise pruning on Transformer and designs an FPGA acceleration engine to customize the balanced blockwise matrix multiplication. We implement the Transformer model with proper hardware scheduling, and the experiments show that the Transformer inference on FPGA achieves 10.35 ms latency with the batch size of 32, which is 10.96 × speed up comparing to CPU platform and 2.08 × speed up comparing to GPU platform.

Peng, Hongwu↗

Offshore Wind to Hydrogen - Modeling, Analysis, Testing and International Collaboration Work

This project explores electrolytic hydrogen production from an offshore wind turbine by: Modeling: Simulate an offshore wind turbine and generate power output profiles; Analysis: Analyze offshore conditions and determine operational modes of the stack; Testing: Perform hardware testing to evaluate dynamic characteristics; and International Collaboration Work: Share learnings and publish. Potential Outcomes: Accelerate development of an integrated, in-turbine offshore wind hydrogen system; and Support two DOE Energy Earthshots - Hydrogen and Floating Offshore Wind.

electrolyzer↗

The IPNS second harmonic RF upgrade.

The intense pulsed neutron source (IPNS) rapid cycling synchrotron (RCS) is used to accelerate protons from 50 MeV to 450 MeV, at a repetition rate of 30 Hz. The original ring design included two identical rf systems, each consisting of an accelerating cavity, cavity bias supply, power amplifiers and low-level analog electronics. The original cavities are located 180 degrees apart in the ring and provide a total peak accelerating voltage of {approx}21 kV over the 2.21-MHz to 5.14-MHz revolution frequency sweep. A third rf system has been constructed and installed in the RCS. The third rf system is capable of operating at the fundamental revolution frequency for the entire acceleration cycle, providing an additional peak accelerating voltage of up to {approx}11 kV, or at the second harmonic of the revolution frequency for the first {approx}4 ms of the acceleration cycle, providing an additional peak voltage of up to {approx}11 kV for bunch shape control. We describe here the hardware implementation and operation to date of the third rf cavity in the second harmonic mode.

43 PARTICLE ACCELERATORS↗

Concurrent Relaxation through Accelerated Deep Learning

CRADL captures performance metrics of machine learning algorithms operating on mesh data from multiphysics codes This proxy application is a tool to explore scalability of inference on HPC platforms, and also gather performance metrics for inference on new machine learning specific hardware. CRADL is designed to give users as fine a control as possible over an inference simulation. Users may select the number of cycles, amount of data, and batch size to pass to the accelerator of choice. Additionally the user may select a number of performance optimization libraries and flags. CRADL comes packaged with a repository of anonymized multi-physics simulation data, as well as a pretrained model for inference. The code allows a user to load their own pre-trained model and data if they wish. The code can operate in multiple parallelization schemes, with performance enhancing options such as half-precision libraries, PyTorch benchmarking, and pinned memory with non-blocking data transfers.

Zieb, KristoferJ.↗

Accelerating Scientific Applications With SambaNova Reconfigurable Dataflow Architecture

Here, our exploratory work finds that the SambaNova Reconfigurable Dataflow Architecture (RDA) along with the SambaFlow software stack provides for an attractive system and solution to accelerate AI for science workloads. We have observed the efficacy of using the system with a diverse set of science applications and reasoned their suitability for performance gains over traditional hardware. As the Data-Scale system provides for a very large memory capacity, the system can be used to train models that typically do not fit in a GPU. The architecture also provides for deeper integration with upcoming supercomputers at the Argonne Leadership Computing Facility (ALCF), a US Department of Energy Office of Science user facility, to help advance science insights.

97 MATHEMATICS AND COMPUTING↗

Production, Characterization, and Acceleration of Optical Microbunches

Optical microbunches with a spacing of 800 nm have been produced for laser acceleration research. The microbunches are produced using a inverse Free-Electron-Laser (IFEL) followed by a dispersive chicane. The microbunched electron beam is characterized by coherent optical transition radiation (COTR) with good agreement to the analytic theory for bunch formation. In a second experiment the bunches are accelerated in a second stage to achieve for the first time direct net acceleration of electrons traveling in a vacuum with visible light. This dissertation presents the theory of microbunch formation and characterization of the microbunches. It also presents the design of the experimental hardware from magnetostatic and particle tracking simulations, to fabrication and measurement of the undulator and chicane magnets. Finally, the dissertation discusses three experiments aimed at demonstrating the IFEL interaction, microbunch production, and the net acceleration of the microbunched beam. At the close of the dissertation, a separate but related research effort on the tight focusing of electrons for coupling into optical scale, Photonic Bandgap, structures is presented. This includes the design and fabrication of a strong focusing permanent magnet quadrupole triplet and an outline of an initial experiment using the triplet to observe wakefields generated by an electron beam passing through an optical scale accelerator.

43 PARTICLE ACCELERATORS↗

Linear solvers for power grid optimization problems: A review of GPU-accelerated linear solvers

The linear equations that arise in interior methods for constrained optimization are sparse symmetric indefinite, and they become extremely ill-conditioned as the interior method converges. These linear systems present a challenge for existing solver frameworks based on sparse LU or LDL T decompositions. Here, we benchmark five well known direct linear solver packages on CPU- and GPU-based hardware, using matrices extracted from power grid optimization problems. The achieved solution accuracy varies greatly among the packages. None of the tested packages delivers significant GPU acceleration for our test cases. For completeness of the comparison we include results for MA57, which is one of the most efficient and reliable CPU solvers for this class of problem.

97 MATHEMATICS AND COMPUTING↗

Leveraging Computational Storage Devices in Campaign Storage [Slides]

Computational storage provides new ways of accelerating data-intensive applications. In-drive data management schemes matter (O_DIRECT, clustered index). Layer violation: “cheating” one filesystem may be possible; cheating multiple layers of filesystems is hard (FS internal load balancing, fail over, compression, concurrency control). The future directions include block-based acceleration to object-based acceleration.

97 MATHEMATICS AND COMPUTING↗

Towards performance portability in the Spark astrophysical magnetohydrodynamics solver in the Flash-X simulation framework

Simulations of core-collapse supernovae, and other astrophysical phenomena, are quintessential extreme-scale computing challenges. For core-collapse supernova simulations to be carried out by the ExaStar project under the Exascale Computing Project umbrella, a robust, efficient, and state-of-the-art magnetohydrodynamics solver is a critical requirement. In Flash-X, the primary software instrument for ExaStar, a new magnetohydrodynamics solver has been designed and implemented from the ground up to achieve accuracy and efficiency for simulations of complex astrophysical flows. This new solver, dubbed Spark, uses high-order spatial reconstruction, Runge-Kutta time integration, and an efficient cell-centered approach to satisfying the divergence-free condition for the magnetic fields. Spark was written to be optimized for data locality in cache hierarchy of CPUs. Since data locality optimizations for cache hierarchy are not directly compatible with those of accelerators, we have taken the approach of using program synthesis to avoid massive amounts of code replication that would be necessary if we were to maintain two different versions of the solver. Our program synthesis relies on a simple key-dictionary approach, implemented in python, that enables us to assemble the version of the solver suitable for the target hardware from code fragments identified by specific keys. In this work, we describe the data locality optimizations of the solver for CPUs and accelerators and the program synthesis tools that enable this portability. We also detail the parallel performance of Spark for both CPUs and accelerators.

97 MATHEMATICS AND COMPUTING↗

Evaluating Proxy Apps with FPGAs

Field-programmable gate arrays (FPGAs) are becoming more promising for high performance computing (HPC). Unlike fixed hardware specialization, reconfigurability allows FPGAs to be customized for a wide variety of HPC workloads. Our ultimate goal is to understand the viability of FPGAs as accelerators for some of the key HPC codes at LLNL and identify application characteristics that can benefit from the reconfigurability of FPGAs. More specifically, this LDRD project conducted a detailed performance study of FPGAs using sparse matrix computations and identified key optimization techniques to effectively use FPGA hardware resources.

97 MATHEMATICS AND COMPUTING↗

Investigations of shot reproducibility for the SMP diode at 4.5 MV.

In experiments conducted on the RITS-6 accelerator, the SMP diode exhibits sig- ni cant shot-to-shot variability. Speci cally, for identical hardware operated at the same voltage, some shots exhibit a catastrophic drop in diode impedance. A study is underway to identify sources of shot-to-shot variations which correlate with diode impedance collapse. To remove knob emission as a source, only data from a shot series conducted with a 4.5-MV peak voltage are considered. The scope of this report is limited to sources of variability which occur away from the diode, such as power ow emission and trajectory changes, variations in pulsed power, dustbin and transmission line alignment, and di erent knob shapes. We nd no changes in the transmission line hardware, alignment, or hardware preparation methods which correlate with impedance collapse. However, in classifying good versus poor shots, we nd that there is not a continuous spectrum of diode impedance behavior but that the good and poor shots can be grouped into two distinct impedance pro les. This result forms the basis of a follow-on study focusing on the variability resulting from diode physics. 3

24 POWER TRANSMISSION AND DISTRIBUTION↗

An introduction to neuromorphic computing and its potential impact for unattended ground sensors

Neuromorphic computers are hardware systems that mimic the brain’s computational process phenomenology. This is in contrast to neural network accelerators, such as the Google TPU or the Intel Neural Compute Stick, which seek to accelerate the fundamental computation and data flows of neural network models used in the field of machine learning. Neuromorphic computers emulate the integrate and fire neuron dynamics of the brain to achieve a spiking communication architecture for computation. While neural networks are brain-inspired, they drastically oversimplify the brain’s computation model. Neuromorphic architectures are closer to the true computation model of the brain (albeit, still simplified). Neuromorphic computing models herald a 1000x power improvement over conventional CPU architectures. Sandia National Labs is a major contributor to the research community on neuromorphic systems by performing design analysis, evaluation, and algorithm development for neuromorphic computers. Space-based remote sensing development has been a focused target of funding for exploratory research into neuromorphic systems for their potential advantage in that program area; SNL has led some of these efforts. Recently, neuromorphic application evaluation has reached the NA-22 program area. This same exploratory research and algorithm development should penetrate the unattended ground sensor space for SNL’s mission partners and program areas. Neuromorphic computing paradigms offer a distinct advantage for the SWaP-constrained embedded systems of our diverse sponsor-driven program areas.

97 MATHEMATICS AND COMPUTING↗

CAPTAN: A hardware architecture for integrated data acquisition, control, and analysis for detector development

The Electronic Systems Engineering Department of the Computing Division at the Fermi National Accelerator Laboratory has developed a data acquisition system flexible and powerful enough to meet the needs of a variety of high energy physics applications. The system described in this paper is called CAPTAN (Compact And Programmable daTa Acquisition Node) and its architecture and capabilities are presented in detail here. The three most important characteristics of this system are flexibility, versatility and scalability. These three main features are supported by key architectural features; a vertical bus that permits the user to stack multiple boards, a gigabit Ethernet link that permits high speed communications to the system and the core group of boards that provide specific capabilities for the system. In this paper, we describe the system architecture, give an overview of its capabilities and point out possible applications.

99 GENERAL AND MISCELLANEOUS↗