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 1,009 records · Page 56

CG-Kit: Code Generation Toolkit for performant and maintainable variants of source code applied to Flash-X hydrodynamics simulations

CG-Kit is a new Code Generation tool-Kit that we have developed as a part of the solution for portability and maintainability for multiphysics computing applications. The development of CG-Kit is rooted in the urgent need created by the shifting landscape of high-performance computing platforms and the algorithmic complexities of a particular large-scale multiphysics application: Flash-X. To efficiently use computing resources on a heterogeneous node, an application must have a map of computation to resources and a mechanism to move the data and computation to the resources according to the map. Most existing performance portability solutions are focussed on abstracting the expression of computations so that a unified source code can be specialized to run on different resources. However, such an approach is insufficient for a code like Flash-X, which has a multitude of code components that can be assembled in various permutations and combinations to form different instances of applications. Similar challenges apply to any code that has composability, where a single specified way of apportioning work among devices may not be optimal. Additionally, use cases arise where the optimal control flow of computation may differ for different devices while the underlying numerics remain identical. This combination leads to unique challenges including handling an existing large code base in Fortran and/or C/C++, subdivision of code into a great variety of units supporting a wide range of physics and numerical methods, different parallelization techniques for distributed and shared memory systems and accelerator devices, and heterogeneity of computing platforms requiring coexisting variants of parallel algorithms. All of these challenges demand that scientific software developers apply existing knowledge about domain applications, algorithms, and computing platforms to determine custom abstractions and granularity for code generation. There is a critical lack of tools to tackle those problems. CG-Kit is designed to fill this gap by providing a user with the ability to express their desired control flow and computation-to-resource map in the form a pseudocode-like recipe. It consists of standalone tools that can be combined into highly specific and, we argue, highly effective portability and maintainability toolchains. Here we present the design of our new tools: parametrized source trees, control flow graphs, and recipes. The tools are implemented in Python. They are agnostic to the programming language of the source code targeted for code generation. In conclusion, we demonstrate the capabilities of the toolkit with two examples, first, multithreaded variants of the basic AXPY operation, and second, variants of parallel algorithms within a hydrodynamics solver, called Spark, from Flash-X that operates on block-structured adaptive meshes.

Algorithmic portability↗

Evaluating Near-Term Adiabatic Quantum Computing

This report summarizes the first year’s effort on the Enceladus project, under which Sandia was asked to evaluate the potential advantages of adiabatic quantum computing for analyzing large data sets in the near future, 5-to-10 years from now. We were not specifically evaluating the machine being sold by D-Wave Systems, Inc; we were asked to anticipate what future adiabatic quantum computers might be able to achieve. While realizing that the greatest potential anticipated from quantum computation is still far into the future, a special purpose quantum computing capability, Adiabatic Quantum Optimization (AQO), is under active development and is maturing relatively rapidly; indeed, D-Wave Systems Inc. already offers an AQO device based on superconducting flux qubits. The AQO architecture solves a particular class of problem, namely unconstrained quadratic Boolean optimization. Problems in this class include many interesting and important instances. Because of this, further investigation is warranted into the range of applicability of this class of problem for addressing challenges of analyzing big data sets and the effectiveness of AQO devices to perform specific analyses on big data. Further, it is of interest to also consider the potential effectiveness of anticipated special purpose adiabatic quantum computers (AQCs), in general, for accelerating the analysis of big data sets. The objective of the present investigation is an evaluation of the potential of AQC to benefit analysis of big data problems in the next five to ten years, with our main focus being on AQO because of its relative maturity. We are not specifically assessing the efficacy of the D-Wave computing systems, though we do hope to perform some experimental calculations on that device in the sequel to this project, at least to provide some data to compare with our theoretical estimates.

97 MATHEMATICS AND COMPUTING↗

Machine learning for accuracy in density functional approximations

Machine learning techniques have found their way into computational chemistry as indispensable tools to accelerate atomistic simulations and materials design. In addition, machine learning approaches hold the potential to boost the predictive power of computationally efficient electronic structure methods, such as density functional theory, to chemical accuracy and to correct for fundamental errors in density functional approaches. In this paper, recent progress in applying machine learning to improve the accuracy of density functional and related approximations is reviewed. Promises and challenges in devising machine learning models transferable between different chemistries and materials classes are discussed with the help of examples applying promising models to systems far outside their training sets.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

PRESSURE DROP EVALUATION OF THE HYDROGEN CIRCULATION SYSTEM FOR JSNS

In J-PARC, an intense spallation neutron source (JSNS) driven by a proton beam of 1 MW has selected supercritical hydrogen with a temperature of around 20 K and the pressure of 1.5 MPa as a moderator material. A hydrogen-circulation system, which consists of two pumps, an ortho-para hydrogen converter, a heater, an accumulator and a helium-hydrogen heat exchanger, has been designed to provide supercritical hydrogen to the moderators and remove the nuclear heating there. A hydrogen-circulation system is cooled through the heat exchanger by a helium refrigerator with the refrigeration power of 6.45 kW at 15.5 K. It is important for the cooling design of the hydrogen-circulation system to understand the pressure drops through the equipments. In this work, the pressure drop through each component was analyzed by using a CFD code, STAR-CD. The correlation of the pressure drops through the components that can describe the analytical results within 14% differences has been derived. It is confirmed that the pressure drop in the hydrogen circulation system would be estimated to be 37 kPa for the circulation flow rate of 160 g/s by using the correlations derived here, and is sufficiently lower than the allowable pump head of 100 kPa.

42 ENGINEERING↗

CHARM-SYCL: New Unified Programming Environment for Multiple Accelerator Types

Addressing performance portability across diverse accelerator architectures has emerged as a major challenge in the development of application and programming systems for high-performance computing environments. Although recent programming systems that focus on performance portability have significantly improved productivity in an effort to meet this challenge, the problem becomes notably more complex when compute nodes are equipped with multiple accelerator types—each with unique performance attributes, optimal data layout, and binary formats. To navigate the intricacies of multi-accelerator programming, we propose CHARM-SYCL as an extension of our CHARM multi-accelerator execution environment [27]. This environment will combine our SYCL-based performance-portability programming front end with a back end for extremely heterogeneous architectures as implemented with the IRIS runtime from Oak Ridge National Laboratory. Our preliminary evaluation indicates potential productivity boost and reasonable performance compared to vendor-specific programming system and runtimes.

Fujita, Norihisa↗

Application Experiences on a GPU-Accelerated Arm-based HPC Testbed

This paper assesses and reports the experience of ten teams working to port, validate, and benchmark several High Performance Computing applications on a novel GPU-accelerated Arm testbed system. The testbed consists of eight NVIDIA Arm HPC Developer Kit systems, each one equipped with a server-class Arm CPU from Ampere Computing and two data center GPUs from NVIDIA Corp. The systems are connected together using InfiniBand interconnect. The selected applications and mini-apps are written using several programming languages and use multiple accelerator-based programming models for GPUs such as CUDA, OpenACC, and OpenMP offloading. Working on application porting requires a robust and easy-to-access programming environment, including a variety of compilers and optimized scientific libraries. The goal of this work is to evaluate platform readiness and assess the effort required from developers to deploy well-established scientific workloads on current and future generation Arm-based GPU-accelerated HPC systems. The reported case studies demonstrate that the current level of maturity and diversity of software and tools is already adequate for large-scale production deployments.

Elwasif, Wael↗

Computational Design of Batteries from Materials to Systems

Computer models are helping to accelerate the design and validation of next generation batteries and provide valuable insights not possible through experimental testing alone. Validated 3-D physics-based models exist for predicting electrochemical performance, thermal and mechanical response of cells and packs under normal and abuse scenarios. The talk describes present efforts to make the models better suited for engineering design, including improving their computation speed, developing faster processes for model parameter identification including under aging, and predicting the performance of a proposed electrode material recipe a priori using microstructure models.

25 ENERGY STORAGE↗

Understanding the Design Space of Sparse/Dense Multiphase Dataflows for Mapping Graph Neural Networks on Spatial Accelerators

Graph Neural Networks (GNNs) have garnered a lot of recent interest because of their success in learning representations from graph-structured data across several critical applications in cloud and HPC. Owing to their unique compute and memory characteristics that come from an interplay between dense and sparse phases of computations, the emergence of reconfigurable dataflow (aka spatial) accelerators offers promise for acceleration by mapping optimized dataflows (i.e., computation order and parallelism) for both phases. The goal of this work is to characterize and understand the design-space of dataflow choices for running GNNs on spatial accelerators in order for the compilers to optimize the dataflow based on the workload. Specifically, we propose a taxonomy to describe all possible choices for mapping the dense and sparse phases of GNNs spatially and temporally over a spatial accelerator, capturing both the intra-phase dataflow and the inter-phase (pipelined) dataflow. Using this taxonomy, we do deep-dives into the cost and benefits of several dataflows and perform case studies on implications of hardware parameters for dataflows and value of flexibility to support pipelined execution.

97 MATHEMATICS AND COMPUTING↗

Comparison of pseudo-static limit equilibrium and elastic wave equation analyses of dynamic earth pressures on retaining structures

The seismic earth pressure increment is typically computed using either pseudo-static limit equilibrium methods or elastic wave equation analyses of the interaction between a retaining structure and backfill material, yet current interpretations of the two methods provide conflicting recommendations. The focus of this study is to compare the seismic earth pressure increment computed using the two methods. This approach is demonstrated by subjecting an initially uniform prototype site selected from standard site classifications to harmonic excitation in one-dimensional equivalent linear analyses. Then, the seismic earth pressure resultant for a rigid wall is computed using the two methods. The limit equilibrium approach utilizes the acceleration records from the equivalent linear analysis to compute a seismic coefficient, whereas the elastic solution incorporates the reduced modulus and damping from the final iteration of the analysis, as well as the relative displacement records. The results presented herein corroborate the findings of recent centrifuge experiments and associated analyses.

58 GEOSCIENCES↗

Coupling of Mechanical Behavior of Cell Components to Electrochemical-Thermal Models for Computer-Aided Engineering of Batteries under Abuse (Presentation)

The EV Everywhere Grand Challenge aims to produce plug-in electric vehicles as affordable and convenient for the American family as gasoline-powered vehicles by 2022. Among the requirements set by the challenge, electric vehicles must be as safe as conventional vehicles, and EV batteries must not lead to unsafe situations under abuse conditions. NREL's project started in October 2013, based on a proposal in response to the January 2013 DOE VTO FOA, with the goal of developing computer aided engineering tools to accelerate the development of safer lithium ion batteries.

25 ENERGY STORAGE↗

GALARIO: a GPU accelerated library for analysing radio interferometer observations

We present GALARIO, a computational library that exploits the power of modern graphical processing units (GPUs) to accelerate the analysis of observations from radio interferometers like Atacama Large Millimeter and sub-millimeter Array or the Karl G. Jansky Very Large Array. GALARIO speeds up the computation of synthetic visibilities from a generic 2D model image or a radial brightness profile (for axisymmetric sources). On a GPU, GALARIO is 150 faster than standard PYTHON and 10 times faster than serial C++ code on a CPU. Highly modular, easy to use, and to adopt in existing code, GALARIO comes as two compiled libraries, one for Nvidia GPUs and one for multicore CPUs, where both have the same functions with identical interfaces. GALARIO comes with PYTHON bindings but can also be directly used in C or C++. The versatility and the speed of GALARIO open new analysis pathways that otherwise would be prohibitively time consuming, e.g. fitting high-resolution observations of large number of objects, or entire spectral cubes of molecular gas emission. It is a general tool that can be applied to any field that uses radio interferometer observations. The source code is available online at http://github.com/mtazzari/galario under the open source GNU Lesser General Public License v3.

Astronomy & Astrophysics↗

Overview of the U.S. DOE’s Carbon Capture Simulation Initiative for Accelerating the Commercialization of CCS Technology

The Carbon Capture Simulation Initiative (CCSI) is a partnership among national laboratories, industry and academic institutions that will develop and deploy state-of-theart computational modeling and simulation tools to accelerate the commercialization of carbon capture technologies from discovery to development, demonstration, and ultimately the widespread deployment to hundreds of power plants. The CCSI Toolset will provide end users in industry with a comprehensive, integrated suite of scientifically validated models with uncertainty quantification, optimization, risk analysis and decision making capabilities. The CCSI Toolset will incorporate commercial and open-source software currently in use by industry and will also develop new software tools as necessary to fill technology gaps identified during execution of the project. The CCSI Toolset will (1) enable promising concepts to be more quickly identified through rapid computational screening of devices and processes; (2) reduce the time to design and troubleshoot new devices and processes; (3) quantify the technical risk in taking technology from laboratory-scale to commercial-scale; and (4) stabilize deployment costs more quickly by replacing some of the physical operational tests with a virtual power plant.

Miller, David↗

The CINDER'90 transmutation code package for use in accelerator applications in combination with MCNPX

CINDER'90, a nuclear inventory code originated at the Bettis Atomic Power Laboratory for reactor irradiation calculations and extended for use of in accelerator dr iven systems and high-energy applications at Los Alamos National Laboratory, has been released as a code package for distribution through the Radiation Safety Information Computational Center (RSICC). The code package and its updated data libraries come with several scripts that allow calculations of multi-cell problems in combination with the radiation transport code MCNPX. A script was developed that manages all the pre-processing steps extracting the necessary information from MCNPX output or from one input file, and that runs the CINDER’90 code for a requested list of MCNPX cells and for a requested time history. A second script was developed that extracts the decay photon sources from CINDER’90 output for a requested list of cells and for a requested irradiation or decay time step and builds source deck for subsequent MCNPX calculations. Since the package release, improvements to CINDER’90 are underway in algorithms, libraries, and interfaces to transport codes.

43 PARTICLE ACCELERATORS↗

A Case Study with the HACCmk Kernel in SYCL

The SYCL standard specifies a cross-platform abstraction layer that enables programming of heterogeneous computing systems using standard C++ [1]. In the Open Computing Language (OpenCL) programming model, host and device codes are written in different languages [2 ]. To improve development productivity and performance portability, the SYCL programming model combines host and device codes for an application in a type-safe way. In this report, we are interested in applying the SYCL programming model to a computationally intensive routine derived from the Hardware Accelerated Cosmology Code (HACC) framework for a study on performance portability on a heterogeneous computing device. We use Intel® OneAPI toolkit [3] to build the OpenCL and SYCL programs, and evaluate the performance of both implementations on Intel® integrated GPUs. We find that the SYCL implementation can achieve the same performance as the OpenCL implementation after we specify a target backend for the compiler to build the SYCL program in offline compilation. Without the specification of a target backend, the runtime will have to compile intermediate device codes for a target platform. Such runtime overhead may become significant compared to the execution of a kernel on a target device. As the kernel routine is compute-bound, we evaluate the impact of the number of compute units upon the kernel’s raw performance using two GPUs. When the hardware resource is not underutilized, we can obtain almost linear performance speedup from 48 compute units to 72 compute units in offline compilation, which indicates that the number of compute units on a GPU are important to improving the raw performance of a compute-bound kernel. The experimental results show that SYCL is a promising programming model for heterogeneous computing. The remainder of the report is organized as follows: In Section II, we describe the SYCL programming model, the steps to map an OpenCL program to a SYCL program, and the architecture of an integrated GPU. Section III introduces the kernel routine, and presents the OpenCL and SYCL implementations of the kernel. In Section IV, we evaluate the performance of the implementations on the GPUs. Section V concludes the report.

97 MATHEMATICS AND COMPUTING↗

Kinetics of particles with short-range interactions

Self-assembly is one of the grand challenges of the 21st century – as the devices and materials we would like to build become too complex or small-scale for top-down manufacturing to be efficient, it is increasingly important to find ways to create these through bottom-up, dynamical approaches. Many particles used in self-assembly have very short-ranged attractive interactions, making simulations expensive or impossible. This proposal develops a set of conceptual and computational tools to study the dynamics of self-assembly for particles with short-ranged interactions, harnessing ideas in differential and computational geometry, and stochastic analysis, to accelerate simulations.

74 ATOMIC AND MOLECULAR PHYSICS↗

Computationally efficient evaluation of optimum homogeneous charge compression ignition operating range with accelerated multizone engine cycle simulation

The very intensive calculations necessary to define a performance map requiring evaluation of over a hundred individual operating points can be efficiently conducted with accelerated multizone for engine cycle simulation, leading to a definition of regions of acceptable and optimum homogeneous charge compression ignition operation. Accelerated multizone for engine cycle simulation has the virtue of enabling accurate evaluation of many operating conditions based on thermal stratification data from a single fluid mechanics run at motored conditions. This is possible because thermal stratification is more sensitive to engine geometry than to operating conditions. In this article, accuracy of accelerated multizone for engine cycle simulation is demonstrated by comparison with experimental data for iso-octane homogeneous charge compression ignition operation over a broad range of lean equivalence ratios (0.14–0.28). The validated accelerated multizone for engine cycle simulation model is then applied to generating a performance map for an engine controlled by appropriately adjusting equivalence ratio and internal exhaust gas recirculation. Regions of acceptable and optimum combustion are identified. It is finally demonstrated that while indicated mean effective pressure remains low for optimum homogeneous charge compression ignition operation (1–4 bar), this is sufficient for a large fraction of typical driving in light-duty vehicles. Much driving including idle can therefore be done in homogeneous charge compression ignition mode at high efficiency and low (essentially zero) NOx and particulate matter emissions.

30 DIRECT ENERGY CONVERSION↗

Software package for modeling spin–orbit motion in storage rings

A software package providing a graphical user interface for computer experiments on the motion of charged particle beams in accelerators, as well as analysis of obtained data, is presented. The software package was tested in the framework of the international project on electric dipole moment measurement JEDI (Jülich Electric Dipole moment Investigations). The specific features of particle spin motion imply the requirement to use a cyclic accelerator (storage ring) consisting of electrostatic elements, which makes it possible to preserve horizontal polarization for a long time. Computer experiments study the dynamics of 10{sup 6}–10{sup 9} particles in a beam during 10{sup 9} turns in an accelerator (about 10{sup 12}–10{sup 15} integration steps for the equations of motion). For designing an optimal accelerator structure, a large number of computer experiments on polarized beam dynamics are required. The numerical core of the package is COSY Infinity, a program for modeling spin–orbit dynamics.

43 PARTICLE ACCELERATORS↗

Compiler-based code generation and autotuning for geometric multigrid on GPU-accelerated supercomputers

GPUs, with their high bandwidths and computational capabilities are an increasingly popular target for scientific computing. Unfortunately, to date, harnessing the power of the GPU has required use of a GPU-specific programming model like CUDA, OpenCL, or OpenACC. Thus, in order to deliver portability across CPU-based and GPU-accelerated supercomputers, programmers are forced to write and maintain two versions of their applications or frameworks. In this paper, we explore the use of a compiler-based autotuning framework based on CUDA-CHiLL to deliver not only portability, but also performance portability across CPU- and GPU-accelerated platforms for the geometric multigrid linear solvers found in many scientific applications. We also show that with autotuning we can attain near Roofline (a performance bound for a computation and target architecture) performance across the key operations in the miniGMG benchmark for both CPU- and GPU-based architectures as well as for a multiple stencil discretizations and smoothers. We show that our technology is readily interoperable with MPI resulting in performance at scale equal to that obtained via hand-optimized MPI+CUDA implementation.

97 MATHEMATICS AND COMPUTING↗