Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “supercomputer”

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 1,675 records · Page 93

SST-GPU: An Execution -Driven CUDA Kernel Scheduler and Streaming-Multiprocessor Compute Model

Programmable accelerators have become commonplace in modern computing systems. Advances in programming models and the availability of massive amounts of data have created a space for massively parallel acceleration where the context for thousands of concurrent threads are resident on-chip. These threads are grouped and interleaved on a cycle-by-cycle basis among several massively parallel computing cores. The design of future supercomputers relies on an ability to model the performance of these massively parallel cores at scale. To address the need for a scalable, decentralized GPU model that can model large GPUs, chiplet-based GPUs and multi-node GPUs, this report details the first steps in integrating the open-source, execution driven GPGPU-Sim into the SST framework. The first stage of this project, creates two elements: a kernel scheduler SST element accepts work from SST CPU models and schedules it to an SM-collection element that performs cycle-by-cycle timing using SSTs Mem Hierarchy to model a flexible memory system.

97 MATHEMATICS AND COMPUTING↗

Swift/T Guide

Swift/T is a hierarchical, parallel programming system that allows you to rapidly develop many-task workflows that call into sequential code concurrently. It is implemented with a compiler and runtime system; the Swift/Turbine Compiler (STC) allows you to write Swift programs and run them using Turbine. At runtime, Swift/T programs are MPI programs capable of utilizing very large supercomputers.

97 MATHEMATICS AND COMPUTING↗

A Non-hydrostatic Variable Resolution Atmospheric Model in ACME

The Accelerated Climate Modeling for Energy (ACME) Project is designed to perform coupled climate simulations at high resolution on current and next generation DOE leadership class supercomputers. Prior to this project, the ACME atmospheric model employed the primitive equations of motion, which constrained its maximum resolution and limited the physical phenomena that it could resolve. Our team removed this constraint by implementing a non-hydrostatic (NH) atmospheric model in the ACME Framework. Collaboration with SciDAC FASTMath institutes facilitated improvements to key technologies needed for the non-hydrostatic model. This model was tested in a variable resolution framework, which enables high resolution simulation regions at reasonable computational costs. The correct performance of the NH model was verified with standard dynamical-core tests, and by demonstrating that it reproduced expected climate statistics in the hydrostatic regime. The completed model was applied to evaluate the limitations of existing turbulent mountain stress and gravity wave drag parameterizations, in order to demonstrate the power of the new nonhydrostatic model to model new phenomena and enhance the quality of present day 1/4 degree simulations.

Hall, David Matthew↗

CRADA Final Report: Development of Opposed-Piston Variable Compression Ratio Engine for Automotive Applications

Pinnacle Engines, Inc., and researchers at Oak Ridge National Laboratory (ORNL) collaborated under the DOE Small Business Voucher program to perform high-fidelity engine simulations of Pinnacle’s unique opposed-piston engine. In addition to the opposed-piston architecture, Pinnacle’s design includes features such as variable valve timing and compression ratio which provide the potential for improved performance, efficiency, and emissions but also introduce many degrees of freedom in design and operation. The impact and complex interactions of these design and operating parameters must be well understood to direct and optimize the final design. Using DOE’s flagship supercomputing resource, Titan, the team was able to perform an extensive set of numerical engine simulations to explore the impact of these parameters on predicted engine performance. Based on the learnings from this study, Pinnacle will build and experimentally evaluate a prototype engine for light-duty gasoline applications.

42 ENGINEERING↗

Report of the DOE Workshop on Management, Analysis, and Visualization of Experimental and Observational data – The Convergence of Data and Computing

The Department of Energy (DOE) Office of Science (SC) operates dozens of national science user facilities that span many different disciplines. These facilities include accelerators, colliders, supercomputers, light sources and neutron sources, as well as facilities for studying the nanoworld, the environment, and the atmosphere. In Fiscal Year 2014 over 33,000 researchers from academia, industry, and government laboratories, spanning all fifty states and the District of Columbia, utilized these unique facilities to perform new scientific research. Each of these facilities generates vast amounts of scientific data, and the rate, size, and complexity of this data is rapidly increasing, thanks to advances in technology. A growing concern, which motivates this workshop, is the likely significant adverse impact on science programs that will result without significant advances in the capabilities needed to manage and gain knowledge from these collections of data. The purpose of this workshop, held 29 September 2015 through 1 October 2015 in Bethesda, MD, is to help the Advanced Scientific Computing Research (ASCR) and research community better understand needs related to the management, analysis, and visualization of experimental and observational data (EOD) collected and generated by experimental and observational science projects (EOS) at Office of Science user facilities.

97 MATHEMATICS AND COMPUTING↗

Full Trinity Run with SPARTA

A heterogeneous run on the full Trinity supercomputer at LANL was performed using SPARTA during March 9-12, 2018. Over 19,000 nodes (9200+ Haswell and 9900+ KNL) and 1.2 million MPI processes were used. The run was successful, with SPARTA running for several hours with good performance (better than the same simulation running on full Sequoia). However, several challenges were encountered, and some unresolved issues remain.

97 MATHEMATICS AND COMPUTING↗

Review of V&V Documentation

Consortium for Advance Simulation of Light Water Reactors (CASL) is a Department of Energy Innovation Hub whose mission is the following, "CASL is a collaboration of the nation's leading scientists, institutions, and supercomputers, with an aggressive 10-year mission to confidently predict the performance of existing and next-generation commercial nuclear reactors through comprehensive, science-based modeling and simulation." The CASL program to date has focused on developing the necessary predictive capability. Rightly so, it is characterized by many as a research project. As a matter of intent, the first 6 years of CASL focused on developing and demonstrating the prediction capability of a suite of independent physics codes: MPACT (neutronics), CTH (thermal hydraulics in the core), BISON (fuel performance), and MAMBA (CRUD and boron uptake on fuel rod surfaces) The last 4 years focused on initial attempts to couple the codes and to demonstrate those capabilities through a series of challenge problems aligned to 3 key issues of interest to the nuclear power industry.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

MagMicro with Ultra-High Energy Resolution (Progress Report Q4 FY18)

We are performing extensive simulations of the MMC detector response to optimize their energy resolution, detection efficiency and count rate based on the data we took in FY17/18. The optimized layout separates the MMC detector chip from the SQUID preamplifier chips coupled to the MMCs through Al wire bonds to avoid heating the MMC with the power dissipation of the SQUID. The goal is to design MMC gamma detectors with a limiting energy resolution below ~25 eV FWHM up to >100 keV and ~1 ms decay times for maximum count rates of several tens of counts/s per pixel. The simulations, which were done at the Supercomputing Center of the University of New Mexico, are taking significantly longer than expected. The difficulty is that the device performance depends intricately on the design parameters, and each simulation is quite involved because it needs to cover the frequency range well into the GHz range to account for the ac Josephson effect. This is delaying the start of the detector array fabrication, and we have not been able to complete the fabrication and the testing of the 32-pixel MMC arrays by September 30. We have therefore requested a no-cost extension of the project to March 31, 2019.

42 ENGINEERING↗

Basic Research Needs for Quantum Materials (Brochure)

In a world of technologies built on quantum materials, computers will work a million times faster than today’s supercomputers while using a fraction as much energy. Electricity will be generated, transmitted, and used with almost no losses. Smart building materials, fabrics, and electronic devices will change their properties on demand to satisfy our needs. The astounding effects of quantum mechanics will enable materials properties and functions that seem like science fiction today.

36 MATERIALS SCIENCE↗

Balar: A SST GPU Component for Performance Modeling and Profiling

Programmable accelerators have become commonplace in modern computing systems. Advances in programming models and the availability of massive amounts of data have created a space for massively parallel accelerators capable of maintaining context for thousands of concurrent threads resident on-chip. These threads are grouped and interleaved on a cycle-by-cycle basis among several massively parallel computing cores. One path for the design of future supercomputers relies on an ability to model the performance of these massively parallel cores at scale. The SST framework has been proven to scale up to run simulations containing tens of thousands of nodes. A previous report described the initial integration of the open-source, execution-driven GPU simulator, GPGPU-Sim, into the SST framework. This report discusses the results of the integration and how to use the new GPU component in SST. It also provides examples of what it can be used to analyze and a correlation study showing how closely the execution matches that of a Nvidia V100 GPU when running kernels and mini-apps.

97 MATHEMATICS AND COMPUTING↗

Harnessing the power of ab initio calculations, distributed computing and machine learning to efficiently locate extreme molecules for use in carbon-based solar cells (Final Technical Report)

The use of high-throughput virtual screening (HTVS) tools is a powerful tool to expedite the materials discovery of commercially relevant materials. In previous years, our group has developed a molecular discovery platform to generate libraries in order to obtain suitable candidates for different applications, starting from the Harvard Clean Energy Project. This platform is suitable to test in-silico on traditional supercomputing clusters and shared resources, for example, in the IBM World Community Grid. In this project, we used the molecular discovery platform to create and screen a library of candidates of organic photovoltaic (OPVs) molecules. Based on a set of candidates created with combinations of molecular moieties, we were able to filter, by conformation stability, the energy of electronic orbitals and approximated power conversion efficiencies (PCE). To improve the predictions of orbital energies calculated and the PCEs, we used Gaussian Process regression and two sets of molecules. These sets correspond to electronic structure calculations of a higher level of theory and experimental PCE values, respectively. Finally, we selected a subset of the best candidates (molecules with a PCE higher than 10%) to understand its absorbance properties with TD-DFT. This project has demonstrated the capabilities of our molecular discovery platform for HTVS. Finally, machine learning can help us to introduce more complex effects included in bulk conditions and computational intensive calculations on models.

14 SOLAR ENERGY↗

Sparse Recovery for Scientific Data

We have built build and analyzed fundamental mathematical tools that enable sparse recovery of scientific data for high performance computing, experimental facilities and sensor measurement. The proposed work will impact many areas of research across the DOE. We have designed, developed and analyzed fast scalable algorithms for the sparse recovery of scientific data sets that may be extremely large, generated very quickly and distributed across multiple storage devices. The algorithms we developed will be used to process the copious amounts of scientific data generated by the DOE supercomputer simulations and by DOE experimental and observational facilities. They leverage the fact that while the amount of data produced can be extremely large, the actual quantities of interest (QoI's) are often sparse, either in space time, or in some feature space. Mathematically, notions of compressive sensing, including l1 regularized and total variation regularized optimization have played an important part in this effort.

97 MATHEMATICS AND COMPUTING↗

Center for Computing Research Highlights

Sandia has a legacy of leadership in the advancement of high performance computing (HPC) at extreme scales. First-of-a-kind scalable distributed-memory parallel platforms such as the Intel Paragon, ASCI Red (the world’s first teraflops computer), and Red Storm (co-developed with Cray) helped form the basis for one of the most successful supercomputer product lines ever: the Cray XT series. Sandia also has pioneered system software elements—including lightweight operating systems, the Portals network programming interface, advanced interconnection network designs, and scalable I/O— that are critical to achieving scalability on large computing systems.

97 MATHEMATICS AND COMPUTING↗

Evaluating the Opportunities for Multi-Level Memory – An ASC 2016 L2 Milestone

The next two Advanced Technology platforms for the ASC program will feature complex memory hierarchies – in the Trinity supercomputer being deployed in 2016, Intel’s Knights Landing processors will feature 16GB of on-package, high-bandwidth memory, combined with a larger capacity DDR4 memory and in 2018, the Sierra machine deployed at Lawrence Livermore National Laboratory will feature powerful compute nodes containing POWER9 processors with large capacity memories and an array of coherent GPU accelerators also with high bandwidth memories.

97 MATHEMATICS AND COMPUTING↗

Performance Portability of the Aeras Atmosphere Model to Next Generation Architectures using Kokkos

The subject of this report is the performance portability of the Aeras global atmosphere dynamical core (implemented within the Albany multi-physics code) to new and emerging architecture machines using the Kokkos library and programming model. We describe the process of refactoring the finite element assembly process for the 3D hydrostatic model in Aeras and highlight common issues associated with development on GPU architectures. After giving detailed build and execute instructions for Aeras with MPI, OpenMP and CUDA on the Shannon cluster at Sandia National Laboratories and the Titan supercomputer at Oak Ridge National Laboratory, we evaluate the per- formance of the code on a canonical test case known as the baroclinic instability problem. We show a speedup of up to 4 times on 8 OpenMP threads, but we were unable to achieve a speedup on the GPU due to memory constraints. We conclude by providing methods for improving the performance of the code for future optimization.

97 MATHEMATICS AND COMPUTING↗

Quantum-Classical Science for the Plasma-Material Interface in NSTXU (Final Technical Report)

Plasma facing surfaces (PFS) have a profound influence on plasma performance and choice of material for PFSs is a burning issue for tokamak reactors. Using lithium wall conditioning in NSTX has resulted in a beneficial influence on plasma behavior including control of hydrogen recycling and improvements in energy confinement, longer confinement times, edge density control, and the elimination of Edge Localized Modes. A major challenge for the NSTX-U (increased power flux, increased campaign time) as well as for other magnetic fusion devices is to extend high-performance plasmas for very long durations, and to integrate this high performance with plasma facing components (PFCs) that can withstand very high heat and particle fluxes while maintaining structural integrity with minimal retention of fusion fuel. This project directly contributed to filling up the knowledge gaps and data for the dynamics of the plasma-facing interfaces in the NSTX-U at nano and nano-to-micro scales. This was achieved by both deciphering the poorly understood nanoscale phenomenology of plasma-surface interactions and providing the connection and comparison with the experiments which provide understanding at the mesoscopic scales. We modelled the dynamic response and evolution of a mixed material surfaces (D, Li, C, B, O) to bombardment by plasma particles, and investigated microstructure changes, erosion, surface chemistry, deuterium implantation and permeation. The main objectives of the project are i) a comparison of Li and B effects when deposited on carbon, and ii) the role of oxygen and other impurities (e.g. carbon, boron) in the lithium performance. In addition to predicting and understanding the phenomenology of the processes, the project achieved deliverables on the i) plasma induced erosion rates of PFCs (outer wall, divertor, etc.), including chemical and physical sputtering yields at various temperatures (300-700K), ii) deuterium uptake/recycling rates, and iii) accurate probability rates for the processes and reactions in i-ii which were used to understand the experimental results of the J.P. Allain and B. Koel NSTX-U projects. Lithium was a distinctive focus of the project. It easily charges in contact with other fusion material atoms, giving partially or fully its external electron cloud to other atoms in its vicinity and polarizing the mixed material. Our research strategy used experimentally validated multiscale theory applied to the large, multicomponent material surface systems. With the use of US supercomputing facilities we simulated evolution of atomic charges, chemistry, retention of deuterium, sputtering and morphology by introducing the corrections to the classical molecular dynamics, like are electronegativity equalization method (EEM) and quantum-classical molecular dynamics (QCMD). The goal was to understand and quantify the dynamics of the creation and evolution of the plasma-material interface under irradiation by atoms and molecules at boronized carbon, lithiated carbon, and boronized-lithiated carbon for both solid and self-healing liquid-metal divertor.

36 MATERIALS SCIENCE↗

Argonne Leadership Computing Facility (2018 Annual Report)

The Argonne Leadership Computing Facility enables breakthroughs in science and engineering by providing supercomputing resources and expertise to the research community. In 2018, we continued to build up programs and efforts that reinforce our overall simulation, learning, and data analysis capabilities as a facility. We also enhanced our services to better meet our users’ needs and workflow preferences.

96 KNOWLEDGE MANAGEMENT AND PRESERVATION↗

Report from the Inaugural Meeting on the Opportunities at the SEEMS Facility June 11-12, 2019

The Inaugural Meeting on the Opportunities at the Single Event Effects—Muon Spectroscopy (SEEMS) facility was held June 11–12, 2019, at Oak Ridge National Laboratory (ORNL). Capabilities of the SEEMS facility concept were reviewed, with emphasis on particle beams for single event effects (SEE) testing of electronic devices (integrated circuits) and systems—a topic of growing interest and concern in aerospace avionics, satellites, supercomputing, autonomous cars, unmanned aerial vehicles and more. Participants from these industries and their regulators attended. A clear consensus on the growing and urgent need for SEE test capabilities like those offered by SEEMS was expressed. Subsequent to the meeting, letters of support for SEEMS were sent by The Boeing Company and Honeywell to the ORNL Laboratory Director, Thomas Zacharia.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗