Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Advanced Supercomputing”

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 55 records · Page 3

Multi-petascale highly efficient parallel supercomputer

A Multi-Petascale Highly Efficient Parallel Supercomputer of 100 petaOPS-scale computing, at decreased cost, power and footprint, and that allows for a maximum packaging density of processing nodes from an interconnect point of view. The Supercomputer exploits technological advances in VLSI that enables a computing model where many processors can be integrated into a single Application Specific Integrated Circuit (ASIC). Each ASIC computing node comprises a system-on-chip ASIC utilizing four or more processors integrated into one die, with each having full access to all system resources and enabling adaptive partitioning of the processors to functions such as compute or messaging I/O on an application by application basis, and preferably, enable adaptive partitioning of functions in accordance with various algorithmic phases within an application, or if I/O or other processors are underutilized, then can participate in computation or communication nodes are interconnected by a five dimensional torus network with DMA that optimally maximize the throughput of packet communications between nodes and minimize latency.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Context-aware learning of hierarchies of low-fidelity models for multi-fidelity uncertainty quantification

Multi-fidelity Monte Carlo methods leverage low-fidelity and surrogate models for variance reduction to make tractable uncertainty quantification even when numerically simulating the physical systems of interest with high-fidelity models is computationally expensive. This work proposes a context-aware multi-fidelity Monte Carlo method that optimally balances the costs of training low-fidelity models with the costs of Monte Carlo sampling. It generalizes the previously developed context-aware bi-fidelity Monte Carlo method to hierarchies of multiple models and to more general types of low-fidelity models. When training low-fidelity models, the proposed approach takes into account the context in which the learned low-fidelity models will be used, namely for variance reduction in Monte Carlo estimation, which allows it to find optimal trade-offs between training and sampling to minimize upper bounds of the mean-squared errors of the estimators for given computational budgets. This is in stark contrast to traditional surrogate modeling and model reduction techniques that construct low-fidelity models with the primary goal of approximating well the high-fidelity model outputs and typically ignore the context in which the learned models will be used in upstream tasks. Further, the proposed context-aware multi-fidelity Monte Carlo method applies to hierarchies of a wide range of types of low-fidelity models such as sparse-grid and deep-network models. Numerical experiments with the gyrokinetic simulation code Gene show speedups of up to two orders of magnitude compared to standard estimators when quantifying uncertainties in small-scale fluctuations in confined plasma in fusion reactors. This corresponds to a runtime reduction from 72 days to four hours on one node of the Lonestar6 supercomputer at the Texas Advanced Computing Center.

42 ENGINEERING↗

Surface Roughness Effects from Additive Manufacturing in High Efficiency Gas Turbine Combustion Systems

In the past, clean combustion systems were characterized by high system development costs due to growing complexity and escalating manufacturing costs from conventional manufacturing processes. As a result, high efficiency concepts are difficult to design and more difficult to be cost-effectively manufactured. In recent years, with the introduction of additive manufacturing (AM) technologies, the rapid prototyping and mass production processes of clean combustion systems are promising to be significantly simplified with significant reduction in terms of engine manufacturing cost and engine energy cost. Compared with conventional manufacturing built parts, the AM process enabled a simpler design to be adopted for the nozzle, reducing the number of required braze and weld joints from twenty-five to just five. The resulting nozzle was 25% lighter and five times more durable and contributed to a 15% reduction in fuel burn in comparison with the previous model produced. Applying these improvements to a fleet of 5,000 turbofan engines at 150 kN thrust would result in fuel cost savings of $ \$ $6B annually (at $ \$ $5/gallon Jet-A fuel price, 700 gallon/hour fuel consumption rate, and 2,300 operational hours per year), and reduce CO2 emissions by more than 25 million tons per year. These advances support core Department of Energy (DOE) missions in energy efficiency, improving productivity, and environmental sustainability. Maximizing the benefit of these new capabilities will require high prediction capability of high speed turbulent flow with wall modeled Large Eddy Simulation (LES). GE Aviation maintains that advanced simulation technology and supercomputing is required in order to provide the appropriate boundary conditions to realize the maximum potential of AM and Ceramic Matrix Composite to reduce cooling flow, a first order penalty on the Brayton cycle. The algorithm developed using the ANSYS/Fluent software would provide the foundation for entirely new avenues of research and development with potential multi-billion dollar impact to the US economy, and significant reduction in carbon based emissions across the aerospace and power generation industries. The formulation of this iWLES (integral wall model for LES) is generic and allows to capture the changes in flow dynamics that have a significant impact on the wall bounded flow characteristics, such as swirler effective area, bulk swirl number, local pressure distribution, exit velocity profile, and turbulence kinetic energy profile. A periodic channel flow with rough flat plate is simulated using LES (Wall-Adapting Local Eddy-viscosity) model to verify the implementation of iWLES in the Fluent User Defined Function (UDF). As the flow fields are highly sensitive due to surface roughness of the wall bounded flows in the combustion systems, there is significant potential to conduct further LES studies focusing on the turbulence boundary layer interaction. Accurately capturing near wall physics will elucidate the impact of rough surfaces on combustor flow and aero-thermal interactions.

42 ENGINEERING↗

Surface Roughness Effects from Additive Manufacturing in High Efficiency Gas Turbine Combustion Systems

In the past, clean combustion systems were characterized by high system development costs due to growing complexity and escalating manufacturing costs from conventional manufacturing processes. As a result, high efficiency concepts are difficult to design and more difficult to be cost-effectively manufactured. In recent years, with the introduction of additive manufacturing (AM) technologies, the rapid prototyping and mass production processes of clean combustion systems are promising to be significantly simplified with significant reduction in terms of engine manufacturing cost and engine energy cost. Compared with conventional manufacturing built parts, the AM process enabled a simpler design to be adopted for the nozzle, reducing the number of required braze and weld joints from twenty-five to just five. The resulting nozzle was 25% lighter and five times more durable and contributed to a 15% reduction in fuel burn in comparison with the previous model produced. Applying these improvements to a fleet of 5,000 turbofan engines at 150 kN thrust would result in fuel cost savings of $6B annually (at $5/gallon Jet-A fuel price, 700 gallon/hour fuel consumption rate, and 2,300 operational hours per year),and reduce CO2 emissions by more than 25 million tons per year. These advances support core Department of Energy (DOE)missions in energy efficiency, improving productivity, and environmental sustainability. Maximizing the benefit of these new capabilities will require high prediction capability of high speed turbulent flow with wall modeled Large Eddy Simulation (LES). GE Aviation maintains that advanced simulation technology and supercomputing is required in order to provide the appropriate boundary conditions to realize the maximum potential of AM and Ceramic Matrix Composite to reduce cooling flow, a first order penalty on the Brayton cycle. The algorithm developed using the ANSYS/Fluent software would provide the foundation for entirely new avenues of research and development with potential multi-billion dollar impact to the US economy, and significant reduction in carbon based emissions across the aerospace and power generation industries. The formulation of this iWLES (integral wall model for LES) is generic and allows to capture the changes in flow dynamics that have a significant impact on the wall bounded flow characteristics, such as swirler effective area, bulk swirl number, local pressure distribution, exit velocity profile, and turbulence kinetic energy profile. A periodic channel flow with rough flat plate is simulated using LES (Wall-Adapting Local Eddy-viscosity) model to verify the implementation of iWLES in the Fluent User Defined Function (UDF). As the flow fields are highly sensitive due to surface roughness of the wall bounded flows in the combustion systems, there is significant potential to conduct further LES studies focusing on the turbulence boundary layer interaction. Accurately capturing near wall physics will elucidate the impact of rough surfaces on combustor flow and aero-thermal interactions.

99 GENERAL AND MISCELLANEOUS↗

Workflows in a secure environment

Petascale simulations on the largest supercomputers in the US require advanced data management techniques in order to optimize the application scien- tist time, and to optimize the time spent on the supercomputers. Researchers in such problems are starting to require workflow automation during their simula- tions in order to monitor the simulations, and in order to automate many of the complex analysis which must take place from the data that is generated from these simulations. Scientific workflows are being used to monitor simulations running on these supercomputers by applying a series of complex analysis, and finally producing images and movies from the variables produced in the simulation, or from the derived quantities produced by the analysis. The typical scenario is where the large calculation runs on the supercomputer, and the auxiliary diagnos- tics/monitors are run on resources, which are either on the local area network of the supercomputer, or over the wide area network. The supercomputers at one of the largest centers are highly secure, and the only method to log into the center is interactive authentication by using One Time Passwords (OTP) that are generated by a security device and expire in half a minute. Therefore, grid certificates are not a current option on these machines in the Department of Energy at Oak Ridge Na- tional Laboratory. In this paper we describe how we have extended the Kepler sci- entific workflow management system to be able to run operations on these supercomputers, how workflows themselves can be executed as batch jobs, and fi- nally, how external data-transfer operations can be utilized when they need to per- form authentication for their own as well.

97 MATHEMATICS AND COMPUTING↗

Proteome-scale Deployment of Protein Structure Prediction Workflows on the Summit Supercomputer

Deep learning has contributed to major advances in the prediction of protein structure from sequence, a fundamental problem in structural bioinformatics. With predictions now approaching the accuracy of crystallographic experiments, and with accelerators like GPUs and TPUs making inference using large models rapid, genome-level structure prediction becomes an obvious aim. Leadership-class computing resources can be used to perform genome-scale protein structure prediction using state-of-the-art deep learning models, providing a wealth of new data for systems biology applications. Here we describe our efforts to efficiently deploy the AlphaFold v.2 program, for full-proteome structure prediction, at scale on the Oak Ridge Leadership Computing Facility's resources, including the Summit supercomputer. We performed inference to produce the predicted structures for 40,526 protein sequences, corresponding to four prokaryotic proteomes and one plant proteome, using under 4,400 total Summit node hours, equivalent to using the majority of the supercomputer for a little over one hour. We also designed an optimized structure refinement that reduced the time for the relaxation stage of the AlphaFold pipeline by over 10X for longer sequences. We demonstrate the types of analyses that can be performed on proteome-scale collections of sequences, including a search for novel quaternary structures and implications for functional annotation.

Gao, Mu↗

SISGR -- Domain Microstructures and Mechanisms for Large, Reversible and Anhysteretic Strain Behaviors in Phase Transforming Ferroelectric Materials

This four-year project (including one-year no-cost extension) aimed to advance fundamental understanding of field-induced strain behaviors of phase transforming ferroelectrics. We performed meso-scale phase field modeling and computer simulation to study domain evolutions, mechanisms and engineering techniques, and developed computational techniques for nanodomain diffraction analysis; to further support above originally planned tasks, we also carried out preliminary first-principles density functional theory calculations of point defects and domain walls to complement meso-scale computations as well as performed in-situ high-energy synchrotron X-ray single crystal diffraction experiments to guide theoretical development (both without extra cost to the project thanks to XSEDE supercomputers and DOE user facility Advanced Photon Source).

36 MATERIALS SCIENCE↗

Cielo Computational Environment Usage Model: With Mappings to ACE Requirements for the General Availability User Environment Capabilities (Release V.1.1)

Cielo is a massively parallel supercomputer funded by the DOE/NNSA Advanced Simulation and Computing (ASC) program, and operated by the Alliance for Computing at Extreme Scale (ACES), a partnership between Los Alamos National Laboratory (LANL) and Sandia National Laboratories (SNL). The primary Cielo compute platform is physically located at Los Alamos National Laboratory. This Cielo Computational Environment Usage Model documents the capabilities and the environment to be provided for the Q1 FY12 Level 2 Cielo Capability Computing (CCC) Platform Production Readiness Milestone. This document describes specific capabilities, tools, and procedures to support both local and remote users. The model is focused on the needs of the ASC user working in the secure computing environments at Lawrence Livermore National Laboratory (LLNL), Los Alamos National Laboratory, or Sandia National Laboratories, but also addresses the needs of users working in the unclassified environment. The Cielo Computational Environment Usage Model maps the provided capabilities to the tri-Lab ASC Computing Environment (ACE) Version 8.0 requirements. The ACE requirements reflect the high performance computing requirements for the Production Readiness Milestone user environment capabilities of the ASC community. A description of ACE requirements met, and those requirements that are not met, are included in each section of this document. The Cielo Computing Environment, along with the ACE mappings, has been issued and reviewed throughout the tri-Lab community.

97 MATHEMATICS AND COMPUTING↗

RAPIDS: Reconciling Availability, Accuracy, and Performance in Managing Geo-Distributed Scientific Data

In modern science, big data plays an increasingly important role. Many scientific applications, such as running simulations on supercomputers or conducting experiments on advanced instruments, produce huge amount of data at unprecedented speed. Analyzing and understanding such big data is the key for scientists to make scientific breakthroughs. However, data might become unavailable for scientists to access when outages or maintenance of the storage system occur, which severely hinders scientific discovery. To improve the data availability, data duplication and erasure coding (EC) are often used. But as the scientific data gets larger, using these two methods can cause considerable storage and network overhead.In this paper, we propose RAPIDS, a hybrid approach that combines the multigrid-based error-bounded lossy compression with erasure coding, to significantly reduce the storage and network overhead required for maintaining high data availability. Our experiments show that RAPIDS reduces the storage overhead by up to 7.5x and network overhead by up to 3x to achieve the same level of availability compared to the regular EC method. We improve RAPIDS by building two models to optimize the fault tolerance configurations and data gathering strategy. We demonstrate that RAPIDS significantly improves performance when running on many CPU cores in parallel or on GPUs.

Wan, Lipeng↗

MASSIVE HYBRID PARALLELISM FOR FULLY IMPLICIT MULTIPHYSICS

As hardware advances continue to modify the supercomputing landscape, traditional scientific software development practices will become more outdated, ineffective, and inefficient. The process of rewriting/retooling existing software for new architectures is a Sisyphean task, and results in substantial hours of development time, effort, and money. Software libraries which provide an abstraction of the resources provided by such architectures are therefore essential if the computational engineering and science communities are to continue to flourish in this modern computing environment. The Multiphysics Object Oriented Simulation Environment (MOOSE) framework enables complex multiphysics analysis tools to be built rapidly by scientists, engineers, and domain specialists, while also allowing them to both take advantage of current HPC architectures, and efficiently prepare for future supercomputer designs. MOOSE employs a hybrid shared-memory and distributed-memory parallel model and provides a complete and consistent interface for creating multiphysics analysis tools. In this paper, a brief discussion of the mathematical algorithms underlying the framework and the internal object-oriented hybrid parallel design are given. Representative massively parallel results from several applications areas are presented, and a brief discussion of future areas of research for the framework are provided.

97 MATHEMATICS AND COMPUTING↗

Massive hybrid parallelism for fully implicit multiphysics

As hardware advances continue to modify the supercomputing landscape, traditional scientific software development practices will become more outdated, ineffective, and inefficient. The process of rewriting/retooling existing software for new architectures is a Sisyphean task, and results in substantial hours of development time, effort, and money. Software libraries which provide an abstraction of the resources provided by such architectures are therefore essential if the computational engineering and science communities are to continue to flourish in this modern computing environment. The Multiphysics Object Oriented Simulation Environment (MOOSE) framework enables complex multiphysics analysis tools to be built rapidly by scientists, engineers, and domain specialists, while also allowing them to both take advantage of current HPC architectures, and efficiently prepare for future supercomputer designs. MOOSE employs a hybrid shared-memory and distributed-memory parallel model and provides a complete and consistent interface for creating multiphysics analysis tools. In this paper, a brief discussion of the mathematical algorithms underlying the framework and the internal object-oriented hybrid parallel design are given. Representative massively parallel results from several applications areas are presented, and a brief discussion of future areas of research for the framework are provided. (authors)

97 MATHEMATICS AND COMPUTING↗

Massively Parallel Near-Linear Scalability Algorithms with Application to Unstructured Video Analysis

This paper describes the use of high performance, massively parallel unstructured data analysis techniques to computationally extract human faces from streaming video data. The use of parallel high-throughput algorithms is essential to maximize the rate at which frames can be analyzed so large volumes of data can be evaluated to answer the question “Have we seen this person’s face before?” This paper will highlight the method used to achieve near-linear scalability according to number of processing cores for principle component analysis (PCA) on the Texas Advanced Computing Center (TACC) Ranger supercomputer. A global summation across processing core operation, accomplished via MPI_Reduce on Ranger, causes a slight deviation from linear scaling behavior. This mapping has been utilized in the past to enable a host of data centric analytic techniques such as neural networks, multi-dimensional scaling (MDS), signal processing and other methods to address, with massively parallel computers, problems of scientific interest in Physics, Mathematics, Biology and Chemistry.

99 GENERAL AND MISCELLANEOUS↗

Ultrahigh performance three-dimensional electromagnetic relativistic kinetic plasma simulation

The algorithms, implementation details, and applications of VPIC, a state-of-the-art first principles 3D electromagnetic relativistic kinetic particle-in-cell code, are discussed. Unlike most codes, VPIC is designed to minimize data motion, as, due to physical limitations (including the speed of light{exclamation_point}), moving data between and even within modern microprocessors is more time consuming than performing computations. As a result, VPIC has achieved unprecedented levels of performance. For example, VPIC can perform {approx}0.17 billion cold particles pushed and charge conserving accumulated per second per processor on IBM's Cell microprocessor--equivalent to sustaining Los Alamos's planned Roadrunner supercomputer at {approx}0.56 petaflop (quadrillion floating point operations per second). VPIC has enabled previously intractable simulations in numerous areas of plasma physics, including magnetic reconnection and laser plasma interactions; next generation supercomputers like Roadrunner will enable further advances.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

NAMD - The Engine for Large-Scale Classical MD Simulations of Biomolecular Systems Based on a Polarizable Force Field (ALCF-2 Early Science Program Technical Report)

Large-scale Molecular dynamics (MD) simulation based on atomic models provide a powerful tool to understand the structure-dynamics-function relationships of important biological systems. Recent advances in computing power, especially in supercomputer, make MD simulation more and more popular in modern scientific community. However, the power of classical MD simulation has been limited mainly by the accuracy of the potential energy function and the efficiency of the dynamic algorithm enabling the adequate configurational sampling. This Early Science Project is aimed to make MD simulation method go beyond current limit using the leadership supercomputer Blue Gene/Q Mira. To address the issue of potential energy accuracy, classical Drude oscillator is developed to take into account the electronic polarizability in molecular systems. In order to develop and test the Drude polarizable force field for biomolecules, extensive MD simulations of typical polypeptides, lipids are needed. This requires efficient dynamic sampling algorithm that allows fast sampling. The sampling issue can be addressed by using advanced strategies based on multiple copies in order to enhance the sampling efficiency of brute-force MD. One such method is called replica-exchange MD (REMD).

Roux, Benoit↗

HPC Molecular Simulation Tries Out a New GPU: Experiences on Early AMD Test Systems for the Frontier Supercomputer

Molecular simulation is an important tool for nu- merous efforts in physics, chemistry, and the biological sciences. Simulating molecular dynamics requires extremely rapid cal- culations to enable sufficient sampling of simulated temporal molecular processes. The Hewlett Packard Enterprise (HPE) Cray EX Frontier supercomputer installed at the Oak Ridge Leadership Computing Facility (OLCF) will provide an exascale resource for open science, and will feature graphics processing units (GPUs) from Advanced Micro Devices (AMD). The future LUMI supercomputer in Finland will be based on an HPE Cray EX platform as well. Here we test the ports of several widely used molecular dynamics packages that have each made substantial use of acceleration with NVIDIA GPUs, on Spock, the early Cray pre-Frontier testbed system at the OLCF which employs AMD GPUs. These programs are used extensively in industry for pharmaceutical and materials research, as well as academia, and are also frequently deployed on high-performance computing (HPC) systems, including national leadership HPC resources. We find that in general, performance is competitive and installation is straightforward, even at these early stages in a new GPU ecosystem. Our experiences point to an expanding arena for GPU vendors in HPC for molecular simulation.

Sedova, Ada↗

Virtual Special Issue on Catalysis at the U.S. Department of Energy’s National Laboratories

Catalysis research at the U.S. Department of Energy's (DOE's) National Laboratories covers a wide range of research topics in heterogeneous catalysis, homogeneous/molecular catalysis, biocatalysis, electrocatalysis, and surface science. Since much of the work at National Laboratories is funded by DOE, the research is largely focused on addressing DOE's mission to ensure America's security and prosperity by addressing its energy, environmental, and nuclear challenges through transformative science and technology solutions. The catalysis research carried out at the DOE National Laboratories ranges from very fundamental catalysis science, funded by DOE's Office of Basic Energy Sciences (BES), to applied research and development (R&D) in areas such as biomass conversion to fuels and chemicals, fuel cells, and vehicle emission control with primary funding from DOE's Office of Energy Efficiency and Renewable Energy.

09 BIOMASS FUELS↗

Parallelization of the Red-Black Algorithm on Solving the Second-Order PN Transport Equation with the Hybrid Finite Element Method

The Red-Black algorithm has been successfully applied on solving the second-order parity transport equation with the PN approximation in angle and the Hybrid Finite Element Method (HFEM) in space, i.e., the Variational Nodal Method (VNM) [1,2,3,4,5]. Any transport solving techniques, including the Red-Black algorithm, need to be parallelized in order to take the advantage of the development of supercomputers with multiple processors for the advanced modeling and simulation. To our knowledge, an attempt [6] was done to parallelize it, but it was devoted only to the z axis plans in three-dimensional calculations. General parallelization of the Red-Black algorithm with the spatial domain decomposition has not been reported in the literature. In this summary, we present our implementation of the parallelization of the Red-Black algorithm and its efficiency results.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗