Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Computational Performance”

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 541 records · Page 30

Coordinated Fault-Tolerance for High-Performance Computing Final Project Report

With the Coordinated Infrastructure for Fault Tolerance Systems (CIFTS, as the original project came to be called) project, our aim has been to understand and tackle the following broad research questions, the answers to which will help the HEC community analyze and shape the direction of research in the field of fault tolerance and resiliency on future high-end leadership systems. Will availability of global fault information, obtained by fault information exchange between the different HEC software on a system, allow individual system software to better detect, diagnose, and adaptively respond to faults? If fault-awareness is raised throughout the system through fault information exchange, is it possible to get all system software working together to provide a more comprehensive end-to-end fault management on the system?

97 MATHEMATICS AND COMPUTING↗

Advancing the Adoption of High Performance Computing for Time Domain Simulation

Time domain transient stability simulation of large interconnected power systems is becoming increasingly computationally intensive. With the addition of high levels of variable generation that can aggregately ramp over many hours, both the complexity and time horizon for which power system simulations will need to be conducted are increasing. Additionally, extended-term time domain simulation is required to analyze power system security when cascading events across the system span minutes or even hours. Analysis of these occurrences requires modeling power system dynamics over tens of minutes to many hours rather than tens of seconds. Because of the required high computational intensity, there are presently no stability modeling/simulation platforms that combine the needed computational capabilities with the required power system models relevant for representing and analyzing the potential reliability implications of phenomena that occurs over both short- and extended-term dynamic timeframes.

29 ENERGY PLANNING, POLICY, AND ECONOMY↗

HPGMG 1.0: A Benchmark for Ranking High Performance Computing Systems

This document provides an overview of the benchmark - HPGMG - for ranking large scale general purpose computers for use on the Top500 list. We provide a rationale for the need for a replacement for the current metric HPL, some background of the Top500 list and the challenges of developing such a metric; we discuss our design philosophy and methodology, and an overview of the specification of the benchmark. The primary documentation with maintained details on the specification can be found at hpgmg.org and the Wiki and benchmark code itself can be found in the repository https://bitbucket.org/hpgmg/hpgmg.

97 MATHEMATICS AND COMPUTING↗

Using High Performance Computing to Examine the Processes of Neurogenesis Underlying Pattern Separation and Completion of Episodic Information

Adult neurogenesis in the hippocampus region of the brain is a neurobiological process that is believed to contribute to the brain's advanced abilities in complex pattern recognition and cognition. Here, we describe how realistic scale simulations of the neurogenesis process can offer both a unique perspective on the biological relevance of this process and confer computational insights that are suggestive of novel machine learning techniques. First, supercomputer based scaling studies of the neurogenesis process demonstrate how a small fraction of adult-born neurons have a uniquely larger impact in biologically realistic scaled networks. Second, we describe a novel technical approach by which the information content of ensembles of neurons can be estimated. Finally, we illustrate several examples of broader algorithmic impact of neurogenesis, including both extending existing machine learning approaches and novel approaches for intelligent sensing.

Aimone, James Bradley↗

Energy Performance Testing of Asetek's RackCDU System at NREL's High Performance Computing Data Center

In this study, we report on the first tests of Asetek's RackCDU direct-to-chip liquid cooling system for servers at NREL's ESIF data center. The system was simple to install on the existing servers and integrated directly into the data center's existing hydronics system. The focus of this study was to explore the total cooling energy savings and potential for waste-heat recovery of this warm-water liquid cooling system. RackCDU captured up to 64% of server heat into the liquid stream at an outlet temperature of 89 degrees F, and 48% at outlet temperatures approaching 100 degrees F. This system was designed to capture heat from the CPUs only, indicating a potential for increased heat capture if memory cooling was included. Reduced temperatures inside the servers caused all fans to reduce power to the lowest possible BIOS setting, indicating further energy savings potential if additional fan control is included. Preliminary studies manually reducing fan speed (and even removing fans) validated this potential savings but could not be optimized for these working servers. The Asetek direct-to-chip liquid cooling system has been in operation with users for 16 months with no necessary maintenance and no leaks.

97 MATHEMATICS AND COMPUTING↗

Analysis of Alternatives (AoA) of Open Colllaboration and Research Capabilities Collaboratipon in Research and Engineering in Advanced Technology and Education and High-Performance Computing Innovation Center (HPCIC) on the LVOC.

The Livermore Valley Open Campus (LVOC), a joint initiative of the National Nuclear Security Administration (NNSA), Lawrence Livermore National Laboratory (LLNL), and Sandia National Laboratories (SNL), enhances the national security missions of NNSA by promoting greater collaboration between world-class scientists at the national security laboratories, and their partners in industry and academia. Strengthening the science, technology, and engineering (ST&E) base of our nation is one of the NNSA’s top goals. By conducting coordinated and collaborative programs, LVOC enhances both the NNSA and the broader national science and technology base, and helps to ensure the health of core capabilities at LLNL and SNL. These capabilities must remain strong to enable the laboratories to execute their primary mission for NNSA.

99 GENERAL AND MISCELLANEOUS↗

Using High Performance Computing to Support Water Resource Planning

In recent years, decision support modeling has embraced deliberation-withanalysis— an iterative process in which decisionmakers come together with experts to evaluate a complex problem and alternative solutions in a scientifically rigorous and transparent manner. Simulation modeling supports decisionmaking throughout this process; visualizations enable decisionmakers to assess how proposed strategies stand up over time in uncertain conditions. But running these simulation models over standard computers can be slow. This, in turn, can slow the entire decisionmaking process, interrupting valuable interaction between decisionmakers and analytics.

54 ENVIRONMENTAL SCIENCES↗

Validation of MCNP6 Version 1.0 with the ENDF/B-VII.1 Cross Section Library for Uranium Metal, Oxide, and Solution Systems on the High Performance Computing Platform Moonlight

In this document, the code MCNP is validated with ENDF/B-VII.1 cross section data under the purview of ANSI/ANS-8.24-2007, for use with uranium systems. MCNP is a computer code based on Monte Carlo transport methods. While MCNP has wide reading capability in nuclear transport simulation, this validation is limited to the functionality related to neutron transport and calculation of criticality parameters such as k eff .

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

The Livermore Brain: Massive Deep Learning Networks Enabled by High Performance Computing

The proliferation of inexpensive sensor technologies like the ubiquitous digital image sensors has resulted in the collection and sharing of vast amounts of unsorted and unexploited raw data. Companies and governments who are able to collect and make sense of large datasets to help them make better decisions more rapidly will have a competitive advantage in the information era. Machine Learning technologies play a critical role for automating the data understanding process; however, to be maximally effective, useful intermediate representations of the data are required. These representations or “features” are transformations of the raw data into a form where patterns are more easily recognized. Recent breakthroughs in Deep Learning have made it possible to learn these features from large amounts of labeled data. The focus of this project is to develop and extend Deep Learning algorithms for learning features from vast amounts of unlabeled data and to develop the HPC neural network training platform to support the training of massive network models. This LDRD project succeeded in developing new unsupervised feature learning algorithms for images and video and created a scalable neural network training toolkit for HPC. Additionally, this LDRD helped create the world’s largest freely-available image and video dataset supporting open multimedia research and used this dataset for training our deep neural networks. This research helped LLNL capture several work-for-others (WFO) projects, attract new talent, and establish collaborations with leading academic and commercial partners. Finally, this project demonstrated the successful training of the largest unsupervised image neural network using HPC resources and helped establish LLNL leadership at the intersection of Machine Learning and HPC research.

97 MATHEMATICS AND COMPUTING↗

Create full-scale predictive economic models on ROI and innovation with performance computing

The U.S. Department of Energy (DOE), the world's largest buyer and user of supercomputers, awarded IDC Research, Inc. a grant to create two macroeconomic models capable of quantifying, respectively, financial and non-financial (innovation) returns on investments in HPC resources. Following a 2013 pilot study in which we created the models and tested them on about 200 real-world HPC cases, DOE authorized us to conduct a full-out, three-year grant study to collect and measure many more examples, a process that would also subject the methodology to further testing and validation. A secondary, "stretch" goal of the full-out study was to advance the methodology from association toward (but not all the way to) causation, by eliminating the effects of some of the other factors that might be contributing, along with HPC investments, to the returns produced in the investigated projects.

97 MATHEMATICS AND COMPUTING↗

Optimization and Control of Burning Plasmas Through High Performance Computing

This project has revived the FACETS code, that has been developed under SciDAC fund- ing in 2008-2012. The code has been dormant for a number of years after the SciDAC funding stopped. FACETS depends on external packages. The external packages and libraries such as PETSc, FFTW, HDF5 and NETCDF that are included in FACETS have evolved during these years. Some packages in FACETS are also parts of other codes such as PlasmaState, NUBEAM, GACODES, and UEDGE. These packages have been also evolved together with their host codes which include TRANSP, TGYRO and XPTOR. Finally, there is also a set of packages in FACETS that are being developed and maintained by Tech-X. These packages include BILDER, SciMake, and FcioWrappers. Many of these packages evolved significantly during the last several years and FACETS had to be updated to synchronize with the re- cent progress in the external packages. The PI has introduced new changes to the BILDER package to support the updated interfaces to the external modules. During the last year of the project, the FACETS version of the UEDGE code has been extracted from FACETS as a standalone package. The PI collaborates with the scientists from LLNL on the updated UEDGE model in FACETS. Drs. T. Rognlien, M. Umansky and A. Dimits from LLNL are contributing to this task.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Open Collaboration and Research Capabilities on the LVOC/CREATE/High Performance Computing Innovation Center

While Lawrence Livermore National Laboratory (LLNL) and Sandia National Laboratories (SNL) have remained committed to their core nuclear weapons (NW) mission, the world has changed greatly since the opening of these national labs in the mid-20th century. During this era, U.S. universities were ranked as the top science centers in the world in essentially all fields, and the national laboratories were able to recruit the top graduates - who were predominantly U.S. citizens - from the nation's best academic institutions.

97 MATHEMATICS AND COMPUTING↗