Search NASA⌕ Search

SEARCH · Search NASA

Results for “Accelerators, Computing”

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

In-Depth Optimization with the OpenACC-to-FPGA Framework on an Arria 10 FPGA

The reconfigurable computing paradigm that uses field programmable gate arrays (FPGAs) has received renewed interest in the high-performance computing field due to FPGAs' unique combination of performance and energy efficiency. However, difficulties in programming and optimizing FPGAs have prevented them from being widely accepted as general-purpose computing devices. In accelerator-based heterogeneous computing, portability across diverse heterogeneous devices is also an important issue, but the unique architectural features in FPGAs make this difficult to achieve. To address these issues, a directive-based, high-level FPGA programming and optimization framework was previously developed. In this work, developed optimizations were combined holistically using the directive-based approach to show that each individual benchmark requires a unique set of optimizations to maximize performance. The relationships between FPGA resource usages and runtime performance were also explored.

Lambert, Jacob↗

Manycore Parallel Computing for a Hybridizable Discontinuous Galerkin Nested Multigrid Method

In this work, we present a parallel computing strategy for a hybridizable discontinuous Galerkin (HDG) nested geometric multigrid (GMG) solver. Parallel GMG solvers require a combination of coarse-grain and fine-grain parallelism to improve time-to-solution performance. In this work we focus on fine-grain parallelism. We use Intel's second generation Xeon Phi (Knights Landing) manycore processor. The GMG method achieves ideal convergence rates of 0.2 or less, for high polynomial orders. A matrix free (assembly free) technique is exploited to save considerable memory usage and increase arithmetic intensity. HDG enables static condensation, and due to the discontinuous nature of the discretization, we developed a matrix vector multiply routine that does not require any costly synchronizations or barriers. Our algorithm is able to attain 80% of peak bandwidth performance for higher order polynomials. This is possible due to the data locality inherent in the HDG method. Very high performance is realized for high order schemes, due to good arithmetic intensity, which declines as the order is reduced.

97 MATHEMATICS AND COMPUTING↗

HPC Innovation Center Feasibility Assessment, CRADA No. TC02172.0 (Final Report)

This was a collaborative effort between Lawrence Livermore National Security, LLC as manager and operator of Lawrence Livermore National Laboratory (LLNL) and IBM TJ Watson Research Center to research, assess feasibility and develop an implementation plan for a High Performance Computing Innovation Center (HPCIC) in the Livermore Valley Open Campus (LVOC). The ultimate goal of this work was to help advance the State of California and U.S. commercial competitiveness in the arena of High Performance Computing (HPC) by accelerating the adoption of computational science solutions, consistent with recent DOE strategy directives. The desired result of this CRADA was a well-researched, carefully analyzed market evaluation that would identify those firms in core sectors of the US economy seeking to adopt or expand their use of HPC to become more competitive globally, and to define how those firms could be helped by the HPCIC with IBM as an integral partner.

East, D. R.↗

HPC Innovation Center Feasibility Assessment Final Report CRADA No. TC02172.0

This was a collaborative effort between Lawrence Livermore National Security, LLC as manager and operator of Lawrence Livermore National Laboratory (LLNL) and IBM TJ Watson Research Center to research, assess feasibility and develop an implementation plan for a High Performance Computing Innovation Center (HPCIC) in the Livermore Valley Open Campus (LVOC). The ultimate goal of this work was to help advance the State of California and U.S. commercial competitiveness in the arena of High Performance Computing (HPC) by accelerating the adoption of computational science solutions, consistent with recent DOE strategy directives. The desired result of this CRADA was a well-researched, carefully analyzed market evaluation that would identify those firms in core sectors of the US economy seeking to adopt or expand their use of HPC to become more competitive globally, and to define how those firms could be helped by the HPCIC with IBM as an integral partner.

97 MATHEMATICS AND COMPUTING↗

Facts About XLDB-2011

This note provides details of the 5th Extremely Large Databases Conference and Invitational Workshop that were held in 2011 on 18-19 October and 20 October, respectively, at the SLAC National Accelerator Laboratory in Menlo Park, California. The main goals of the conference were: (1) Encourage and accelerate the exchange of ideas between users trying to build extremely large databases worldwide and database solution providers; (2) Share lessons, trends, innovations, and challenges related to building extremely large databases; (3) Facilitate the development and growth of practical technologies for extremely large databases; and (4) Strengthen, expand, and engage the XLDB community.

43 PARTICLE ACCELERATORS↗

A simple eigenfunction convergence acceleration method for Monte Carlo

Monte Carlo transport codes typically use a power iteration method to obtain the fundamental eigenfunction. The standard convergence rate for the power iteration method is the ratio of the first two eigenvalues, that is, k{sub 2}/k{sub 1}. Modifications to the power method have accelerated the convergence by explicitly calculating the subdominant eigenfunctions as well as the fundamental. Calculating the subdominant eigenfunctions requires using particles of negative and positive weights and appropriately canceling the negative and positive weight particles. Incorporating both negative weights and a {+-} weight cancellation requires a significant change to current transport codes. This paper presents an alternative convergence acceleration method that does not require modifying the transport codes to deal with the problems associated with tracking and cancelling particles of {+-} weights. Instead, only positive weights are used in the acceleration method.

97 MATHEMATICS AND COMPUTING↗

Differentiable Neural Architecture, Mixed Precision and Accelerator Co-Search

Quantization, effective Neural Network architecture, and efficient accelerator hardware are three important design paradigms to maximize accuracy and efficiency. Mixed Precision Quantization is a process of assigning different precision to different Neural Network layers for optimized inference. Neural Architecture Search (NAS) is a process of automatically designing the neural network for a task and can also be extended to search for the precision of each weight and activation matrix. In this paper, we develop the following three methods: (i) Fast Differentiable Hardware-aware Mixed Precision Quantization Search method to find optimal precision, (ii) Joint Differentiable hardware-aware Architecture and Mixed Precision Quantization Co-search, (iii) Joint Accelerator, Architecture, and Precision triple co-search to find best possibilities in all the three worlds. We demonstrate the effectiveness of our proposed methods targeting Bitfusion accelerator by searching mixed precision models on MobilenetV2. We achieve better accuracy-latency trade-off models than the manually designed and previously proposed search methods.

97 MATHEMATICS AND COMPUTING↗

Building an Integrated Ecosystem of Computational and Observational Facilities to Accelerate Scientific Discovery

Future scientific discoveries will rely on flexible ecosystems that incorporate modern scientific instruments, high performance computing resources, parallel distributed data storage, and performant networks across multiple, independent facilities. In addition to connecting physical resources, such an ecosystem presents many challenges in logistics and accessibility, especially in orchestrating computations and experiments that span across leadership computing systems and experimental instruments. Past efforts have typically been application-specific or limited to interfaces for computing resources. This paper proposes a general framework for integrating computation resources and instrument operations, addressing challenges in code development/execution, data staging and collection, software stack, control mechanisms, resource authorization and governance, and hardware integration. We also describe a demonstration use case wherein a Bayesian optimization algorithm running on an edge computing resource guides a scanning probe microscope to autonomously and intelligently characterize a material sample. This science edge ecosystem framework will provide a blueprint for federating multi-institutional, disparate resources and orchestrating scientific workflows across them to enable next-generation discoveries.

Somnath, Suhas↗

Convergence acceleration for vector sequences and applications to computational fluid dynamics

Some recent developments in acceleration of convergence methods for vector sequences are reviewed. The methods considered are the minimal polynomial extrapolation, the reduced rank extrapolation, and the modified minimal polynomial extrapolation. The vector sequences to be accelerated are those that are obtained from the iterative solution of linear or nonlinear systems of equations. The convergence and stability properties of these methods as well as different ways of numerical implementation are discussed in detail. Based on the convergence and stability results, strategies that are useful in practical applications are suggested. Two applications to computational fluid mechanics involving the three dimensional Euler equations for ducted and external flows are considered. The numerical results demonstrate the usefulness of the methods in accelerating the convergence of the time marching techniques in the solution of steady state problems.

Sidi, Avram↗

Convergence acceleration for vector sequences and applications to computational fluid dynamics

Some recent developments in acceleration of convergence methods for vector sequences are reviewed. The methods considered are the minimal polynomial extrapolation, the reduced rank extrapolation, and the modified minimal polynomial extrapolation. The vector sequences to be accelerated are those that are obtained from the iterative solution of linear or nonlinear systems of equations. The convergence and stability properties of these methods as well as different ways of numerical implementation are discussed in detail. Based on the convergence and stability results, strategies that are useful in practical applications are suggested. Two applications to computational fluid mechanics involving the three dimensional Euler equations for ducted and external flows are considered. The numerical results demonstrate the usefulness of the methods in accelerating the convergence of the time marching techniques in the solution of steady state problems.

Sidi, Avram↗

A nonvoxel-based dose convolution/superposition algorithm optimized for scalable GPU architectures

Purpose: Real-time adaptive planning and treatment has been infeasible due in part to its high computational complexity. There have been many recent efforts to utilize graphics processing units (GPUs) to accelerate the computational performance and dose accuracy in radiation therapy. Data structure and memory access patterns are the key GPU factors that determine the computational performance and accuracy. In this paper, the authors present a nonvoxel-based (NVB) approach to maximize computational and memory access efficiency and throughput on the GPU. Methods: The proposed algorithm employs a ray-tracing mechanism to restructure the 3D data sets computed from the CT anatomy into a nonvoxel-based framework. In a process that takes only a few milliseconds of computing time, the algorithm restructured the data sets by ray-tracing through precalculated CT volumes to realign the coordinate system along the convolution direction, as defined by zenithal and azimuthal angles. During the ray-tracing step, the data were resampled according to radial sampling and parallel ray-spacing parameters making the algorithm independent of the original CT resolution. The nonvoxel-based algorithm presented in this paper also demonstrated a trade-off in computational performance and dose accuracy for different coordinate system configurations. In order to find the best balance between the computed speedup and the accuracy, the authors employed an exhaustive parameter search on all sampling parameters that defined the coordinate system configuration: zenithal, azimuthal, and radial sampling of the convolution algorithm, as well as the parallel ray spacing during ray tracing. The angular sampling parameters were varied between 4 and 48 discrete angles, while both radial sampling and parallel ray spacing were varied from 0.5 to 10 mm. The gamma distribution analysis method (γ) was used to compare the dose distributions using 2% and 2 mm dose difference and distance-to-agreement criteria, respectively. Accuracy was investigated using three distinct phantoms with varied geometries and heterogeneities and on a series of 14 segmented lung CT data sets. Performance gains were calculated using three 256 mm cube homogenous water phantoms, with isotropic voxel dimensions of 1, 2, and 4 mm. Results: The nonvoxel-based GPU algorithm was independent of the data size and provided significant computational gains over the CPU algorithm for large CT data sizes. The parameter search analysis also showed that the ray combination of 8 zenithal and 8 azimuthal angles along with 1 mm radial sampling and 2 mm parallel ray spacing maintained dose accuracy with greater than 99% of voxels passing the γ test. Combining the acceleration obtained from GPU parallelization with the sampling optimization, the authors achieved a total performance improvement factor of >175 000 when compared to our voxel-based ground truth CPU benchmark and a factor of 20 compared with a voxel-based GPU dose convolution method. Conclusions: The nonvoxel-based convolution method yielded substantial performance improvements over a generic GPU implementation, while maintaining accuracy as compared to a CPU computed ground truth dose distribution. Such an algorithm can be a key contribution toward developing tools for adaptive radiation therapy systems.

60 APPLIED LIFE SCIENCES↗

ChemComp: A Compilation Framework for Computing with Chemical Reaction Networks

The acceleration of scientific computation, data analytics, and artificial intelligence is driving a surge in computational requirements. Yet, state-of-the-art high-performance computing systems are approaching physical limitations that impede further significant improvements in energy efficiency. As we move towards post-exascale computing systems, innovative approaches are necessary to overcome this barrier in power consumption. Novel analog and hybrid digital-analog architectures hold promise for enhancing energy efficiency by several orders of magnitude. Biochemical computation stands out among the various solutions being explored due to its potential to enable new classes of devices with immense computational capabilities. These devices can capitalize on the inherent efficacy of biological cells in solving optimization problems and are scalable through increasing reaction system size or vessel capacity, potentially satisfying scientific computing's high-performance requirements. Nonetheless, several theoretical and practical limitations persist, including problem formulation and mapping to chemical reaction networks (CRNs) and implementation of actual CRN devices. In this paper, we propose a framework for biochemical computation using systems chemistry. We present the initial components of our approach: an abstract chemical reaction dialect implemented as a multi-level intermediate representation (MLIR) compiler extension and a pathway to represent mathematical problems with CRNs. To showcase the potential of this approach, we emulate a simplified chemical reservoir device. This work lays the groundwork for leveraging chemistry's computing potential in creating energy-efficient, high-performance computing systems tailored to contemporary computational needs.

artificial intelligence↗

A New Control Room for SLAC Accelerators

We are planning to construct a new control room at SLAC to unify and improve the operation of the LCLS, SPEAR3, and FACET accelerator facilities, and to provide the space and flexibility needed to support the LCLS-II and proposed new test beam facilities. The existing control rooms for the linac and SPEAR3 have been upgraded in various ways over the last decade, but their basic features have remained unchanged. We propose to build a larger modern Accelerator Control Room (ACR) in the new Research Support Building (RSB) which is currently under construction at SLAC. Shifting the center of control for the accelerator facilities entails both technical and administrative challenges. In this paper, we describe the history, concept, and status of this project.

43 PARTICLE ACCELERATORS↗

Fingerprinting Anomalous Computation with RNN for GPU-accelerated HPC Machines

This paper presents a workload classification framework that discriminates illicit computation from authorized workloads on GPU-accelerated HPC systems. As such systems become more and more powerful, they are exploited by attackers to run malicious and for-profit programs that typically require extremely high computing ability to be successful. Our classification framework leverages the distinctive signatures between illicit and authorized workloads, and explore machine learning methods to learn the workloads and classify them. The framework uses lightweight, non-intrusive workload profiling to collect model input data, and explores multiple machine learning methods, particularly recurrent neural network (RNN) that is suitable for online anomalous workload detection. Evaluation results on four generations of GPU machines demonstrate that the workload classification framework can tell apart the illicit authorized workloads with a high accuracy of over 95%.

Zou, Pengfei↗

Detecting Anomalous Computation with RNNs on GPU-Accelerated HPC Machines

This paper presents a workload classification framework that discriminates illicit computation from authorized workloads on GPU-accelerated HPC systems. As such systems become more and more powerful, they are exploited by attackers to run malicious and for-profit programs that typically require extremely high computing ability to be successful. Our classification framework leverages the distinctive signatures between illicit and authorized workloads, and explore machine learning methods to learn the workloads and classify them. The framework uses lightweight, non-intrusive workload profiling to collect model input data, and explores multiple machine learning methods, particularly recurrent neural network (RNN) that is suitable for online anomalous workload detection. Evaluation results on three generations of GPU machines demonstrate that the workload classification framework can tell apart the illicit authorized workloads with a high accuracy of over 95%.

Pengfei, Zou↗

Energetics and energy scaling of quasi-monoenergetic protons in laser radiation pressure acceleration

Theoretical and computational studies of the ion energy scaling of the radiation pressure acceleration of an ultra-thin foil by short pulse intense laser irradiation are presented. To obtain a quasi-monoenergetic ion beam with an energy spread of less than 20%, two-dimensional particle-in-cell simulations show that the maximum energy of the quasi-monoenergetic ion beam is limited by self-induced transparency at the density minima caused by the Rayleigh-Taylor instability. For foils of optimal thickness, the time over which Rayleigh-Taylor instability fully develops and transparency occurs is almost independent of the laser amplitude. With a laser power of about one petawatt, quasi-monogenetic protons with 200 MeV and carbon ions with 100 MeV per nucleon can be obtained, suitable for particle therapy applications.

43 PARTICLE ACCELERATORS↗

SU-F-BRD-09: A Random Walk Model Algorithm for Proton Dose Calculation

Purpose: To develop a random walk model algorithm for calculating proton dose with balanced computation burden and accuracy. Methods: Random walk (RW) model is sometimes referred to as a density Monte Carlo (MC) simulation. In MC proton dose calculation, the use of Gaussian angular distribution of protons due to multiple Coulomb scatter (MCS) is convenient, but in RW the use of Gaussian angular distribution requires an extremely large computation and memory. Thus, our RW model adopts spatial distribution from the angular one to accelerate the computation and to decrease the memory usage. From the physics and comparison with the MC simulations, we have determined and analytically expressed those critical variables affecting the dose accuracy in our RW model. Results: Besides those variables such as MCS, stopping power, energy spectrum after energy absorption etc., which have been extensively discussed in literature, the following variables were found to be critical in our RW model: (1) inverse squared law that can significantly reduce the computation burden and memory, (2) non-Gaussian spatial distribution after MCS, and (3) the mean direction of scatters at each voxel. In comparison to MC results, taken as reference, for a water phantom irradiated by mono-energetic proton beams from 75 MeV to 221.28 MeV, the gamma test pass rate was 100% for the 2%/2mm/10% criterion. For a highly heterogeneous phantom consisting of water embedded by a 10 cm cortical bone and a 10 cm lung in the Bragg peak region of the proton beam, the gamma test pass rate was greater than 98% for the 3%/3mm/10% criterion. Conclusion: We have determined key variables in our RW model for proton dose calculation. Compared with commercial pencil beam algorithms, our RW model much improves the dose accuracy in heterogeneous regions, and is about 10 times faster than MC simulations.

60 APPLIED LIFE SCIENCES↗