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 325 records · Page 18

High Energy Electromagnetic Particle Transportation on the GPU

We present massively parallel high energy electromagnetic particle transportation through a finely segmented detector on a Graphics Processing Unit (GPU). Simulating events of energetic particle decay in a general-purpose high energy physics (HEP) detector requires intensive computing resources, due to the complexity of the geometry as well as physics processes applied to particles copiously produced by primary collisions and secondary interactions. The recent advent of hardware architectures of many-core or accelerated processors provides the variety of concurrent programming models applicable not only for the high performance parallel computing, but also for the conventional computing intensive application such as the HEP detector simulation. The components of our prototype are a transportation process under a non-uniform magnetic field, geometry navigation with a set of solid shapes and materials, electromagnetic physics processes for electrons and photons, and an interface to a framework that dispatches bundles of tracks in a highly vectorized manner optimizing for spatial locality and throughput. Core algorithms and methods are excerpted from the Geant4 toolkit, and are modified and optimized for the GPU application. Program kernels written in C/C++ are designed to be compatible with CUDA and OpenCL and with the aim to be generic enough for easy porting to future programming models and hardware architectures. To improve throughput by overlapping data transfers with kernel execution, multiple CUDA streams are used. Issues with floating point accuracy, random numbers generation, data structure, kernel divergences and register spills are also considered. Performance evaluation for the relative speedup compared to the corresponding sequential execution on CPU is presented as well.

Canal, P.↗

Binary Complex Neural Network Acceleration on FPGA

Being able to learn from complex data with phase information is imperative for many signal processing applications. Today’s real-valued deep neural networks (DNNs) have shown efficiency in latent information analysis but fall short when applied to the complex domain. Deep complex networks (DCN) , in contrast, can learn from complex data, but have high computational costs; therefore, they cannot satisfy the instant decision making requirements of many deployable systems dealing with short observations or short signal bursts. Recent, Binarized Complex Neural Network (BCNN), which integrates DCNs with binarized neural networks (BNN), shows great potential in classifying complex data in real-time. In this paper, we propose a structural pruning based accelerator of BCNN, which is able to provide more than 5000 frames/s inference throughput on edge devices. The high performance comes from both the algorithm and hardware sides. On the algorithm side, we conduct structural pruning to the original BCNN models and obtain 20 × pruning rates with negligible accuracy loss; on the hardware side, we propose a novel 2D convolution operation accelerator for the binary complex neural network. Experimental results show that the proposed design works with over 90% utilization and is able to achieve the inference throughput of 5882 frames/s and 4938 frames/s for complex NIN-Net and ResNet-18 using CIFAR-10 dataset and Alveo U280 Board.

Peng, Hongwu↗

Bridging Python to Silicon: The SODA Toolchain

Systems performing scientific computing, data analysis, and machine learning tasks have a growing demand for application-specific accelerators that can provide high computational performance while meeting strict size and power requirements. However, the algorithms and applications that need to be accelerated are evolving at a rate that is incompatible with manual design processes based on hardware description languages. Agile hardware design tools based on compiler techniques can help by quickly producing an application-specific integrated circuit (ASIC) accelerator starting from a high-level algorithmic description. Here, we present the software-defined accelerator (SODA) synthesizer, a modular and open-source hardware compiler that provides automated end-to-end synthesis from high-level software frameworks to ASIC implementation, relying on multilevel representations to progressively lower and optimize the input code. Our approach does not require the application developer to write any register-transfer level code, and it is able to reach up to 364 giga floating point operations per second (GFLOPS)/W efficiency (32-bit precision) on typical convolutional neural network operators.

97 MATHEMATICS AND COMPUTING↗

Managing a Real-Time Embedded Linux Platform with Buildroot

Developers of real-time embedded software often need to build the operating system, kernel, tools and supporting applications from source to work with the differences in their hardware configuration. The first attempts to introduce Linux-based real-time embedded systems into the Fermilab accelerator controls system used this approach but it was found to be time-consuming, difficult to maintain and difficult to adapt to different hardware configurations. Buildroot is an open source build system with a menu-driven configuration tool (similar to the Linux kernel build system) that automates this process. A customized Buildroot [1] system has been developed for use in the Fermilab accelerator controls system that includes several hardware configuration profiles (including Intel, ARM and PowerPC) and packages for Fermilab support software. A bootable image file is produced containing the Linux kernel, shell and supporting software suite that varies from 3 to 20 megabytes large – ideal for network booting. The result is a platform that is easier to maintain and deploy in diverse hardware configurations

Diamond, J.↗

The SODA Approach: Leveraging High-Level Synthesis for Hardware/Software Co-design and Hardware Specialization: Invited

Novel "converged" applications combine phases of scientific simulation with data analysis and machine learning. Each computational phase can benefit from specialized accelerators. However, algorithms evolve so quickly that mapping them on existing accelerators is suboptimal or even impossible. This paper presents the SODA (Software Defined Accelerators) framework, a modular, multi-level, open-source, no-human-in-the-loop, hardware synthesizer that enables end-to-end generation of specialized accelerators. SODA is composed of SODA-Opt, a high-level frontend developed in MLIR that interfaces with domain-specific programming frameworks and allows performing system level design, and Bambu, a state-of-the-art high-level synthesis engine that can target different device technologies. The framework implements design space exploration as compiler optimization passes. We show how the modular, yet tight, integration of the high-level optimizer and lower-level HLS tools enables the generation of accelerators optimized for the computational patterns of converged applications. We then discuss some of the research opportunities that such a framework allows, including system-level design, profile driven optimization, and supporting new optimization metrics.

Bohm Agostini, Nicolas↗

High-performance dynamic quantum clustering on graphics processors

Clustering methods in machine learning may benefit from borrowing metaphors from physics. Dynamic quantum clustering associates a Gaussian wave packet with the multidimensional data points and regards them as eigenfunctions of the Schroedinger equation. The clustering structure emerges by letting the system evolve and the visual nature of the algorithm has been shown to be useful in a range of applications. Furthermore, the method only uses matrix operations, which readily lend themselves to parallelization. In this paper, we develop an implementation on graphics hardware and investigate how this approach can accelerate the computations. We achieve a speedup of up to two magnitudes over a multicore CPU implementation, which proves that quantum-like methods and acceleration by graphics processing units have a great relevance to machine learning.

97 MATHEMATICS AND COMPUTING↗

Quantum computing for fusion energy science applications

This is a review of recent research exploring and extending present-day quantum computing capabilities for fusion energy science applications. We begin with a brief tutorial on both ideal and open quantum dynamics, universal quantum computation, and quantum algorithms. Then, we explore the topic of using quantum computers to simulate both linear and nonlinear dynamics in greater detail. Because quantum computers can only efficiently perform linear operations on the quantum state, it is challenging to perform nonlinear operations that are generically required to describe the nonlinear differential equations of interest. In this work, we extend previous results on embedding nonlinear systems within linear systems by explicitly deriving the connection between the Koopman evolution operator, the Perron–Frobenius evolution operator, and the Koopman–von Neumann evolution (KvN) operator. We also explicitly derive the connection between the Koopman and Carleman approaches to embedding. Extension of the KvN framework to the complex-analytic setting relevant to Carleman embedding, and the proof that different choices of complex analytic reproducing kernel Hilbert spaces depend on the choice of Hilbert space metric are covered in the appendixes. Finally, we conclude with a review of recent quantum hardware implementations of algorithms on present-day quantum hardware platforms that may one day be accelerated through Hamiltonian simulation. We discuss the simulation of toy models of wave–particle interactions through the simulation of quantum maps and of wave–wave interactions important in nonlinear plasma dynamics.

Joseph, I. (ORCID:0000000255400840)↗

Surrogate optimization of variational quantum circuits

Variational quantum eigensolvers are touted as a near-term algorithm capable of impacting many applications. However, the potential has not yet been realized, with few claims of quantum advantage and high resource estimates, especially due to the need for optimization in the presence of noise. Finding algorithms and methods to improve convergence is important to accelerate the capabilities of near-term hardware for VQE or more broad applications of hybrid methods in which optimization is required. To this goal, we look to use modern approaches developed in circuit simulations and stochastic classical optimization, which can be combined to form a surrogate optimization approach to quantum circuits. Using an approximate (classical CPU/GPU) state vector simulator as a surrogate model, we efficiently calculate an approximate Hessian, passed as an input for a quantum processing unit or exact circuit simulator. This method will lend itself well to parallelization across quantum processing units. We demonstrate the capabilities of such an approach with and without sampling noise and a proof-of-principle demonstration on a quantum processing unit utilizing 40 qubits.

Gustafson, Erik J. [RIACS, Mtn. View] (ORCID:00000↗

Scalable Second Order Optimization for Machine Learning

Many machine learning (ML) training tasks are essentially optimization processes that would at first glance appear eminently parallelizable and scalable. However, effective acceleration of these tasks with scalable parallel hardware has proven to be elusive. While standard methods for machine learning, e.g., stochastic gradient descent (SGD) for DNNs, tend to be resource efficient, they appear to be fundamentally sequential in nature.

97 MATHEMATICS AND COMPUTING↗

PIC Simulations in Low Energy Part of PIP-II Proton Linac

The front end of PIP-II linac is composed of a 30 keV ion source, low energy beam transport line (LEBT), 2.1 MeV radio frequency quadrupole (RFQ), and medium energy beam transport line (MEBT). This configuration is currently being assembled at Fermilab to support a complete systems test. The front end represents the primary technical risk with PIP-II, and so this step will validate the concept and demonstrate that the hardware can meet the specified requirements. SC accelerating cavities right after MEBT require high quality and well defined beam after RFQ to avoid excessive particle losses. In this paper we will present recent progress of beam dynamic study, using CST PIC simulation code, to investigate partial neutralization effect in LEBT, halo and tail formation in RFQ, total emittance growth and beam losses along low energy part of the linac.

43 PARTICLE ACCELERATORS↗

FastML Science Benchmarks: Accelerating Real-Time Scientific Edge Machine Learning

Applications of machine learning (ML) are growing by the day for many unique and challenging scientific applications. However, a crucial challenge facing these applications is their need for ultra low-latency and on-detector ML capabilities. Given the slowdown in Moore's law and Dennard scaling, coupled with the rapid advances in scientific instrumentation that is resulting in growing data rates, there is a need for ultra-fast ML at the extreme edge. Fast ML at the edge is essential for reducing and filtering scientific data in real-time to accelerate science experimentation and enable more profound insights. To accelerate real-time scientific edge ML hardware and software solutions, we need well-constrained benchmark tasks with enough specifications to be generically applicable and accessible. These benchmarks can guide the design of future edge ML hardware for scientific applications capable of meeting the nanosecond and microsecond level latency requirements. To this end, we present an initial set of scientific ML benchmarks, covering a variety of ML and embedded system techniques.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Beam Synchronous for the Rest of Us!

Fermilab’s Tevatron Clock (TCLK) infrastructure has been an integral part of the accelerator control network since the 1980’s. This 10MHz Manchester encoded protocol has enabled flexible, real-time event distribution for thousands of devices connected to the timing network with a high degree of reliability. Forthcoming upgrades to the Fermilab complex (PIP-II, LBNF, ACORN) necessitate higher levels of precision to maintain inter-bunch timing for Instrumentation and Control purposes. This presents as an opportunity to refine the event distribution protocol for tighter synchronization between machines, experiments, and eventually far-site operations. This paper outlines a method by which beam-synchronous events may be distributed through asynchronous serial protocols via integration with local LLRF and global PPS reference signals. This method is ideal for synchrotron machines with aggressive frequency sweeps (such as Fermilab's 38~53MHz Booster) and allows for precision timing to be maintained across machines without specialized hardware.

43 PARTICLE ACCELERATORS↗

STATUS OF THE NEUTRALIZED DRIFT COMPRESSION EXPERIMENT (NDCX-II)

The Neutralized Drift Compression Experiment (NDCX-II) is an 11 M$ induction accelerator project currently in construction at Lawrence Berkeley National Laboratory for warm dense matter (WDM) experiments investigating the interaction of ion beams with matter at elevated temperature and pressure. The machine consists of a lithium injector, induction accelerator cells, diagnostic cells, a neutralized drift compression line, a final focus solenoid, and a target chamber. The induction cells and some of the pulsed power systems have been reused from the decommissioned Advanced Test Accelerator at Lawrence Livermore National Laboratory after refurbishment and modification. The machine relies on a sequence of acceleration waveforms to longitudinally compress the initial ion pulse from 600 ns to less than 1 ns in {approx} 12 m. Radial confinement of the beam is achieved with 2.5 T pulsed solenoids. In the initial hardware configuration, 50 nC of Li{sup +} will be accelerated to 1.25 MeV and allowed to drift-compress to a peak current of {approx}40 A. The project started in the summer of 2009. Construction of the accelerator will be completed in the fall of 2011 and will provide a worldwide unique opportunity for ion-driven warm dense matter experiments as well as research related to novel beam manipulations for heavy ion fusion drivers.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

An Adaptive Memory Interface Controller for Improving Bandwidth Utilization of Hybrid and Reconfigurable Systems

Emerging applications such as data mining, bioinformatics, knowledge discovery, social network analysis are irregular. They use data structures based on pointers or linked lists, such as graphs, unbalanced trees or unstructures grids, which generates unpredictable memory accesses. These data structures usually are large, but difficult to partition. These applications mostly are memory bandwidth bounded and have high synchronization intensity. However, they also have large amounts of inherent dynamic parallelism, because they potentially perform a task for each one of the element they are exploring. Several efforts are looking at accelerating these applications on hybrid architectures, which integrate general purpose processors with reconfigurable devices. Some solutions, which demonstrated significant speedups, include custom-hand tuned accelerators or even full processor architectures on the reconfigurable logic. In this paper we present an approach for the automatic synthesis of accelerators from C, targeted at irregular applications. In contrast to typical High Level Synthesis paradigms, which construct a centralized Finite State Machine, our approach generates dynamically scheduled hardware components. While parallelism exploitation in typical HLS-generated accelerators is usually bound within a single execution flow, our solution allows concurrently running multiple execution flow, thus also exploiting the coarser grain task parallelism of irregular applications. Our approach supports multiple, multi-ported and distributed memories, and atomic memory operations. Its main objective is parallelizing as many memory operations as possible, independently from their execution time, to maximize the memory bandwidth utilization. This significantly differs from current HLS flows, which usually consider a single memory port and require precise scheduling of memory operations. A key innovation of our approach is the generation of a memory interface controller, which dynamically maps concurrent memory accesses to multiple ports. We present a case study on a typical irregular kernel, Graph Breadth First search (BFS), exploring different tradeoffs in terms of parallelism and number of memories.

BFS↗

High-$Q_0$ treatment of CEBAF 1.5 GHz SRF cavities

The Continuous Electron Beam Accelerator Facility (CEBAF) was the first large-scale accelerator to employ superconducting radiofrequency (SRF) cavities for continuous-wave operation. Ongoing research and development efforts continue to focus on increasing the intrinsic quality factor (Q 0 ) of these cavities in order to reduce cryogenic losses while maintaining operational gradients. Here, in this work, we report on the application of high-Q 0 surface treatments to single-cell and multicell C100 and C75 style 1.5 GHz niobium cavities used in the CEBAF accelerator. Nitrogen infusion and oxygen alloying via medium-temperature baking were applied under heat-treatment constraints relevant to existing cavity hardware. Both processes yielded substantial improvements in Q 0 at moderate accelerating gradients, achieving values of approximately 2 × 10 10 at 2.07 K and 20 MV/m. The effectiveness of nitrogen infusion at reduced annealing temperatures and the successful extension of oxygen alloying to multicell cavities are demonstrated. These results establish viable pathways for implementing high-Q 0 treatments in CEBAF-compatible cavities and support future integration into cryomodules for reduced operational cryogenic load.

Dhakal, Pashupati [Thomas Jefferson National Accel↗

TADSim: Discrete Event-based Performance Prediction for Temperature Accelerated Dynamics

Next-generation high-performance computing will require more scalable and flexible performance prediction tools to evaluate software--hardware co-design choices relevant to scientific applications and hardware architectures. Here, we present a new class of tools called application simulators—parameterized fast-running proxies of large-scale scientific applications using parallel discrete event simulation. Parameterized choices for the algorithmic method and hardware options provide a rich space for design exploration and allow us to quickly find well-performing software--hardware combinations. We demonstrate our approach with a TADSim simulator that models the temperature-accelerated dynamics (TAD) method, an algorithmically complex and parameter-rich member of the accelerated molecular dynamics (AMD) family of molecular dynamics methods. The essence of the TAD application is captured without the computational expense and resource usage of the full code. We accomplish this by identifying the time-intensive elements, quantifying algorithm steps in terms of those elements, abstracting them out, and replacing them by the passage of time. We use TADSim to quickly characterize the runtime performance and algorithmic behavior for the otherwise long-running simulation code. We extend TADSim to model algorithm extensions, such as speculative spawning of the compute-bound stages, and predict performance improvements without having to implement such a method. Validation against the actual TAD code shows close agreement for the evolution of an example physical system, a silver surface. Finally, focused parameter scans have allowed us to study algorithm parameter choices over far more scenarios than would be possible with the actual simulation. This has led to interesting performance-related insights and suggested extensions.

97 MATHEMATICS AND COMPUTING↗

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↗

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↗