Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “PROCESSORS”

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 1,747 records · Page 97

Multilevel Combinatorial Optimization across Quantum Architectures

Emerging quantum processors provide an opportunity to explore new approaches for solving traditional problems in the post Moore’s law supercomputing era. However, the limited number of qubits makes it infeasible to tackle massive real-world datasets directly in the near future, leading to new challenges in utilizing these quantum processors for practical purposes. Furthermore, hybrid quantum-classical algorithms that leverage both quantum and classical types of devices are considered as one of the main strategies to apply quantum computing to large-scale problems. In this article, we advocate the use of multilevel frameworks for combinatorial optimization as a promising general paradigm for designing hybrid quantum-classical algorithms. To demonstrate this approach, we apply this method to two well-known combinatorial optimization problems, namely, the Graph Partitioning Problem, and the Community Detection Problem. We develop hybrid multilevel solvers with quantum local search on D-Wave’s quantum annealer and IBM’s gate-model based quantum processor. We carry out experiments on graphs that are orders of magnitude larger than the current quantum hardware size, and we observe results comparable to state-of-the-art solvers in terms of quality of the solution.

97 MATHEMATICS AND COMPUTING↗

VTK-m: Visualization for the Exascale Era and Beyond

A recent trend in modern high-performance computing is the increasing use of hybrid architectures, where the vast majority of performance comes from accelerators. Modern accelerators are based on Graphics Processing Units (GPU) that contain many low power cores that in their aggregate provides an extremely high computation rate. Current and future CPU processors are requiring more explicit parallelism as each successive version of the hardware packs in more cores, and technologies like hyperthreading and vector operations require even more parallel processing to leverage each core’s full potential. As an example, the Frontier supercomputer installed at Oak Ridge National Laboratories recently hit a record breaking 1.1 exaflops1 on the LINPACK HPC benchmark [Shoemaker 2022]. The system contains 37632 AMD MI250x GPUs which requires more than half a billion threads to keep the system fully utilized [Khizeran 2022].VTK-m is a toolkit of scientific visualization algorithms for these emerging processor architectures. VTK-m supports the fine-grained concurrency for data analysis and visualization algorithms required to drive extreme scale computing by providing abstract models for data and execution that can be applied to a variety of algorithms across many different processor architectures.

Bolstad, Mark↗

HamPerf: A Hamiltonian-Oriented Approach to Quantum Benchmarking

Quantum computing technologies are undergoing rapid development. The different qubit modalities being considered for quantum computing each have their strengths and weaknesses, making it challenging to compare their performance relative to each other and the state-of-the-art in classical high-performance computing. To better understand the utility of a given quantum processor and to assess when and how it will be able to advance the frontiers of computational science, researchers need a robust approach to quantum benchmarking. A variety of approaches have been proposed, many of which characterize the presence of noise in current quantum devices. These efforts include component-level performance metrics, such as randomized benchmarking and gate set tomography; high-level application-dependent metrics; and devicelevel metrics, such as the Quantum Volume. However, it remains unclear how low-level metrics, such as fidelities and decoherence times, and global device metrics, such as Quantum Volume, relate to the computational utility and practical limitations of quantum processors to solve useful problems. In this paper, we describe our Hamiltonian-oriented approach to quantum benchmarking called HamPerf. Where previous application-dependent approaches specify a suite of benchmarking circuits inspired by applications, we place the problem Hamiltonian at the center. Our strategy allows us to probe the computational performance of a quantum processor on standardized and relevant problem sets, agnostic of the algorithms and hardware used to solve them; it also provides fundamental insights into how device characteristics correlate with computational utility.

Butko, Anastasiia↗

Domi

Domi (Greek for "structure") is a Trilinos package that provides support for multi-dimensional data arrays distributed over multiple processors of a high performance computing system. The software supports data arrays of arbitrary number of dimensions mapped to a multi-dimensional grid of processors. Domi provides local access to the data, as well as the ability to compute global quantities such as norms, maxima and minima. Data arrays can be constructed with boundary and communication padding, necessary for finite difference algorithms, and provides high-level methods for communicating data between processors to update these buffers. The distributed maps and vectors in Domi can be converted to the equivalent data structures in Epetra and Tpetra (both of which assume unstructured data), using view semantics when appropriate, for use with other Trilinos solvers.

Spotz, WilliamF.↗

fftMPI, a library for performing 2d and 3d FFTs in parallel

The fftMPI library enables two and three dimensional Fast Fourier Transforms (FFTs) to be performed efficiently in parallel on a variety of platforms from desktop machines to supercomputers. It allows for flexibility in the initial distribution of the FFT grid across processors, as well as the final output. It handles the data communication and reordering necessary to perform multi-dimensional FFTs in parallel. It relies on open-source or vendor-provided libraries to perform one dimensional FFTs on a single processor. It uses MPI for inter-processor communication, which is a standard distributed memory message-passing library.

Plimpton, Steven↗

slaclab/smurf-processor-base-docker

This docker image, named smurf-processor-base contains smurf-processor and additional tools used by the SMuRF project. It is based on the smurf-rogue docker image and contains smurf-processor.

Vasquez, Jesus↗

Strategies for Reducing Moisture Content in Forest Residues at the Harvest Site

Abstract. The moisture content (MC) of biomass derived from forest residues can pose a challenge to biomass utilization. It plays a significant role in determining the cost of transportation and subsequent market price. Additionally, emerging biomass conversion technologies, such as gasification, torrefaction, and briquetting, have very narrow specifications for the MC (e.g., <15%) in their feedstocks. The goal of this study was to develop strategies for reducing moisture content by evaluating different arrangement patterns of forest residues and its effect on MC reduction at the harvest site. The study compared four different arrangement patterns including criss-cross, teepees, traditional piling (processor piled), and scattered residues in three different timber harvest units in northern California. Two of the arrangement patterns (criss-cross and processor piled) were also covered with a plastic cover. Samples were collected from each treatment using a transect method and were recorded for 12 months. There was an overall drop of MC from 52% (freshly cut) to 12% between all arrangements over the study period. The cost of construction per pile, averaged $37, $41, and $48 for teepees, criss-cross, and processor piles, respectively. Even though, there was no significant difference in MC reduction between piles (except scattered), each pile arrangement of forest residues directly affected biomass feedstock operations, logistics, and costs. Keywords: Feedstock quality, Logging slash, Transect sampling method, Woody biomass energy.

Agriculture↗

Advanced Control Systems for Wave Energy Converters

This submission contains several papers, a final report, descriptions of a theoretical framework for two types of control systems, and descriptions of eight real-time flap load control policies with the objective of assessing the potential improvement of annual average capture efficiency at a reference site on an MHK device developed by Resolute Marine Energy, Inc. (RME). The submission also contains an LCOE model that estimates the performance and related energy cost improvements that each advanced control system might provide and recommendations for improving DOE's LCOE model. The two types of control systems are for wave energy converters which transform data into commands that, in the case of RME's OWSC wave energy converter, provide real-time adjustments to damping forces applied to the prime mover via the power take-off system (PTO). The control theories developed were: 1) Model Predictive Control (MPC) or so-called "non-causal" control whereby sensors deployed seaward of a wave energy converter measure incoming wave characteristics and transmit that information to a data processor which issues commands to the PTO to adjust the damping force to an optimal value; and 2) "Causal" control which utilizes local sensors on the wave energy converter itself to transmit information to a data processor which then issues appropriate commands to the PTO. The two advanced control policies developed by Scruggs and Re Vision were then compared to a simple control policy, Coulomb damping, which was utilized by RME during the two rounds of ocean trials it had conducted prior to the commencement of this project. The project work plan initially included a provision for RME to conduct hardware-in-the-loop (HIL) testing of the data processors and configurations of valves, sensors and rectifiers needed to implement the two advanced control systems developed by Scruggs and Re Vision Consulting but the funding for that aspect of the project was cut at the conclusion of Budget Period 1. Accordingly, more work needs to be done to determine: a) means and feasibility of implementing real-time control; and b) added costs associated with such implementation taking into account estimated effects on system availability in addition to component costs.

16 TIDAL AND WAVE POWER↗

NEW EPICS/RTEMS IOC BASED ON ALTERA SOC AT JEFFERSON LAB

A new EPICS/RTEMS IOC based on the Altera System-on-Chip (SoC) FPGA is being designed at Jefferson Lab. The Altera SoC FPGA integrates a dual ARM Cortex-A9 Hard Processor System (HPS) consisting of processor, peripherals and memory interfaces tied seamlessly with the FPGA fabric using a high-bandwidth interconnect backbone. The embedded Altera SoC IOC has features of remote network boot via U-Boot from SD card or QSPI Flash, 1Gig Ethernet, 1GB DDR3 SDRAM on HPS, UART serial ports, and ISA bus interface. RTEMS for the ARM processor BSP were built with CEXP shell, which will dynamically load the EPICS applications at runtime. U-Boot is the primary bootloader to remotely load the kernel image into local memory from a DHCP/TFTP server over Ethernet, and automatically run RTEMS and EPICS. The first design of the SoC IOC will be compatible with Jefferson Lab’s current PC104 IOCs, which have been running in CEBAF 10 years. The next design would be mounting in a chassis and connected to a daughter card via standard HSMC connectors. This standard SoC IOC will become the next generation of low-level IOC for the accelerator controls at Jefferson Lab.

Yan, Jianxun↗

ENHANCED UNCERTAINTY ANALYSIS FOR SRS COMPOSITE ANALYSIS

The Composite Analysis (CA) performed for the Savannah River Site (SRS) in 2009 (SRS CA 2009) included a simplified uncertainty analysis. The uncertainty analysis in the CA (Smith et al. 2009b) was limited to considering at most five sources in a separate uncertainty calculation performed for each POA. To perform the uncertainty calculations in a reasonable amount of time, the analysis was limited to using 400 realizations, 2,000 years of simulated transport time, and the time steps used for the uncertainty analysis were increased from what was used in the CA base case analysis. As part of the CA maintenance plan, the Savannah River National Laboratory (SRNL) committed to improving the CA uncertainty/sensitivity analysis. The previous uncertainty analysis was constrained by the standard GoldSim licensing which limits the user to running at most four Monte Carlo uncertainty calculations (also called realizations) simultaneously. Some of the limitations on the number of realizations that could be practically run and the simulation time steps were removed by building a cluster of three HP Proliant windows servers with a total of 36 64-bit processors and by licensing the GoldSim DP-Plus distributed processing software. This allowed running as many as 35 realizations simultaneously (one processor is reserved as a master process that controls running the realizations). These enhancements to SRNL computing capabilities made uncertainty analysis: using 1000 realizations, using the time steps employed in the base case CA calculations, with more sources, and simulating radionuclide transport for 10,000 years feasible. In addition, an importance screening analysis was performed to identify the class of stochastic variables that have the most significant impact on model uncertainty. This analysis ran the uncertainty model separately testing the response to variations in the following five sets of model parameters: (a) K{sub d} values (72 parameters for the 36 CA elements in sand and clay), (b) Dose Parameters (34 parameters), (c) Material Properties (20 parameters), (d) Surface Water Flows (6 parameters), and (e) Vadose and Aquifer Flow (4 parameters). Results provided an assessment of which group of parameters is most significant in the dose uncertainty. It was found that K{sub d} and the vadose/aquifer flow parameters, both of which impact transport timing, had the greatest impact on dose uncertainty. Dose parameters had an intermediate level of impact while material properties and surface water flows had little impact on dose uncertainty. Results of the importance analysis are discussed further in Section 7 of this report. The objectives of this work were to address comments received during the CA review on the uncertainty analysis and to demonstrate an improved methodology for CA uncertainty calculations as part of CA maintenance. This report partially addresses the LFRG Review Team issue of producing an enhanced CA sensitivity and uncertainty analysis. This is described in Table 1-1 which provides specific responses to pertinent CA maintenance items extracted from Section 11 of the SRS CA (2009). As noted above, the original uncertainty analysis looked at each POA separately and only included the effects from at most five sources giving the highest peak doses at each POA. Only 17 of the 152 CA sources were used in the original uncertainty analysis and the simulation time was reduced from 10,000 to 2,000 years. A major constraint on the original uncertainty analysis was the limitation of only being able to use at most four distributed processes. This work expanded the analysis to 10,000 years using 39 of the CA sources, included cumulative dose effects at downstream POAs, with more realizations (1,000) and finer time steps. This was accomplished by using the GoldSim DP-Plus module and the 36 processors available on a new windows cluster. The last part of the work looked at the contribution to overall uncertainty from the main categories of uncertainty variables: K{sub d}s, dose parameters, flow parameters, and material properties. This was not intended to be a detailed sensitivity analysis but only to see in very general terms what broad category of parameters contributes most significantly to overall uncertainty in the CA dose. This analysis was not intended to be a final CA uncertainty calculation and there was no intent to revise the stochastic distributions as part of this work.

54 ENVIRONMENTAL SCIENCES↗

NMIS with Imaging and Gamma Ray Spectrometry for Pu, HEU, HE and Other Materials

The Nuclear Material Identification System (NMIS) has been under development at ORNL and the National Nuclear Security Administration (NNSA) Y-12 National Security Complex since 1984. In the mid-1990s, what is now the US Department of Energy (DOE) Office of Nuclear Verification (ONV) realized that it was a useful technology for future arms control treaty applications and supported further development of the system. In 2004, fast-neutron imaging was incorporated into the system. In 2007, the ONV decided to develop a fieldable version of the system, designated as FNMIS, for potential use in future treaties. The FNMIS is being developed to be compatible with the eventual incorporation of gamma-ray spectrometry and an information barrier. This report addresses how and what attributes could be determined by the FNMIS system with gamma-ray spectrometry. The NMIS is a time-dependent coincidence system that incorporates tomographic imaging (including mapping of the fission sites) and gamma-ray spectrometry. It utilizes a small, lightweight (30 lb), portable deuterium-tritium (DT) neutron (14.1 MeV) generator (4 x 10 7 neutrons/second) for active interrogation and can also perform passive interrogation. A high-purity germanium (HPGe) gamma-ray detector with multichannel analysis can be utilized in conjunction with the source for active interrogation or passively. The system uses proton recoil scintillators: 32 small 2.5 x 2.5 x 10.2-cm-thick plastic scintillators for imaging and at least two 2 x 2 arrays of 27 x 27 x 10-cm-thick plastic scintillators that detect induced fission radiation. The DT generator contains an alpha detector that time and directionally tags a fan beam of some of the neutrons emitted and subdivides it into pixels. A fast (1 GHz) time correlation processor measures the time-dependent coincidence among all detectors in the system. A computer-controlled scanner moves the small detectors and the source appropriately for scanning a target object for imaging. The system is based on detection of transmitted 14.1 MeV neutrons, fission neutrons, and gamma rays from spontaneous, inherent source fission of the target, fission neutrons and gamma rays induced by the external DT source, gamma rays from natural emissions of uranium and plutonium, and induced gamma-ray emission by the interaction of the 14.1 MeV neutrons from the DT source. The NMIS can and has been used with a time-tagged californium spontaneous fission source. It has also been used with pulsed interrogation sources such as LINACs, DT, and deuterium-deuterium (DD) sources. This system is uniquely suited for detection of shielded highly enriched uranium (HEU), plutonium, and other special nuclear materials and detection of high explosives (HE) and chemical agents. The NMIS will be adapted to utilize a trusted processor that incorporates information barrier and authentication techniques using open software and then be useful in some international applications for materials whose characteristics may be classified. The proposed information barrier version of the NMIS system would consist of detectors and cables, the red (classified side) computer system, which processes the data, and the black (unclassified side) computer, which handles the computer interface. The system could use the 'IB wrapper' concept proposed by Los Alamos National Laboratory and the software integrity (digital signatures) system proposed by Sandia. Since it is based entirely on commercially available components, the entire system, including NMIS data acquisition boards, can be built with commercial off-the-shelf components. This system is being developed into a fieldable system (FNMIS) for potential arms control treaties by the ONV. The system will be modularly constructed with the RF shielded modules connected to the processor by appropriate control and signal cable in metal conduit. The FNMIS is presently being designed for eventual incorporation of gamma-ray spectrometry and an information barrier to protect classified information. The system hardware and software can be configured to obtain the following: plutonium presence, plutonium mass, Pu-240/239 ratio, plutonium geometry, plutonium metal vs non-metallic (absence of metal), time (age) since processing for plutonium (or last purification), uranium presence, uranium mass, uranium enrichment, uranium geometry, uranium metal vs non-metallic compound (absence of metal), beryllium presence and mass, tritium and deuterium gas bottle presence, HE, and chemical weapons. A matrix of the quantities determined, the method of determination, whether active (external neutron source) or passive, and the measurement equipment involved is given in the Tables 1-4. Some of these attributes can be obtained by multiple data analysis methods. The gamma-ray spectrometry methods for HEU, plutonium, and HE have been developed by other laboratories, are well known, and will be incorporated.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗

A Programming Framework for Scientific Applications on CPU-GPU Systems

At a high level, my research interests center around designing, programming, and evaluating computer systems that use new approaches to solve interesting problems. The rapid change of technology allows a variety of different architectural approaches to computationally difficult problems, and a constantly shifting set of constraints and trends makes the solutions to these problems both challenging and interesting. One of the most important recent trends in computing has been a move to commodity parallel architectures. This sea change is motivated by the industry’s inability to continue to profitably increase performance on a single processor and instead to move to multiple parallel processors. In the period of review, my most significant work has been leading a research group looking at the use of the graphics processing unit (GPU) as a general-purpose processor. GPUs can potentially deliver superior performance on a broad range of problems than their CPU counterparts, but effectively mapping complex applications to a parallel programming model with an emerging programming environment is a significant and important research problem.

99 GENERAL AND MISCELLANEOUS↗

Integrating Fault-Tolerance into the Monte Carlo Application Toolkit

High-performance computer systems have passed the petascale mark, and the next decade will witness a drive towards exascale systems. Extrapolating from current hardware trends it is evident that exascale class systems will require a significantly larger number of cores than current petascale systems, and advanced architectures which include heterogeneous processors are gaining popularity. Such an increase in component count will decrease system mean time to interrupt (MTTI), resulting in greater emphasis on application fault tolerance. The additional memory requirements of exascale systems combined with the slow increase in hard drive write bandwidth will exclude current checkpoint/restart (C/R) techniques as a reliable means of fault tolerance. The current C/R approach to handling failures typically aborts the entire application and requires restarting the application, which is clearly an inefficient solution when utilizing 1000+ processors. Unless improvements are made in resilience and fault tolerance, it is predicted that the work throughput on the largest supercomputers could drop below 10% by 2015. As part of the ”Application Enablement for Next Generation Platforms” milestone, we addressed the shortcomings and limitations of C/R by developing a fault-tolerant component which would be integrated into the Monte Carlo Application Toolkit (MCATK) and demonstrated the ability to continue static K-effective calculations even in the presence of multiple processor failures. In section 2, we provide details about the milestone tasks. Section 3 describes the fault-tolerant framework, its integration into MCATK and its failure recovery scheme. We also describe FT-MPI, a fault-tolerant MPI implementation which provides the capabilities needed for an MPI-based application to survive and continue past failures. Section 4 reports future work that we hope to build upon the current accomplishments.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Detection, Classification and Estimation of Radioactive Contraband from Uncertain Low-Count Measurements

In this final report we discuss the development of the Statistical Radiation Detection System (SRaDS), the next-generation radiation detection software system capable of extracting all available physics information, photon-by-photon, employing Bayesian model-based sequential statistical processing techniques and capable of making a decision when statistically justified. It is a system of computational algorithms consisting of a simple photoelectron processor for the basic system with a combined photoelectron/downscatter processor for the advanced system. Both algorithms have been demonstrated on laboratory data and are available for integration into standard radiation detectors and acquisition systems as well as specialized embedded processing hardware for real-time operations. This report incorporates the basic research performed leading to the development of SRaDS in the form of attached publications discussing the theory, development and validation of the processor and subsequent designs for this powerful solution to the detection problem plaguing the radiation area for a long time.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Multigrid Smoothers for Ultra-Parallel Computing: Additional Theory and Discussion

This paper investigates the properties of smoothers in the context of algebraic multigrid (AMG) running on parallel computers with potentially millions of processors. The development of multigrid smoothers in this case is challenging, because some of the best relaxation schemes, such as the Gauss-Seidel (GS) algorithm, are inherently sequential. Based on the sharp two-grid multigrid theory from we characterize the smoothing properties of a number of practical candidates for parallel smoothers, including several C-F, polynomial, and hybrid schemes. We show, in particular, that the popular hybrid GS algorithm has multigrid smoothing properties which are independent of the number of processors in many practical applications, provided that the problem size per processor is large enough. This is encouraging news for the scalability of AMG on ultra-parallel computers. We also introduce the more robust `1 smoothers, which are always convergent and have already proven essential for the parallel solution of some electromagnetic problems.

97 MATHEMATICS AND COMPUTING↗

Implementation and Optimization of miniGMG - a Compact Geometric Multigrid Benchmark

Multigrid methods are widely used to accelerate the convergence of iterative solvers for linear systems used in a number of different application areas. In this report, we describe miniGMG, our compact geometric multigrid benchmark designed to proxy the multigrid solves found in AMR applications. We explore optimization techniques for geometric multigrid on existing and emerging multicore systems including the Opteron-based Cray XE6, Intel Sandy Bridge and Nehalem-based Infiniband clusters, as well as manycore-based architectures including NVIDIA's Fermi and Kepler GPUs and Intel's Knights Corner (KNC) co-processor. This report examines a variety of novel techniques including communication-aggregation, threaded wavefront-based DRAM communication-avoiding, dynamic threading decisions, SIMDization, and fusion of operators. We quantify performance through each phase of the V-cycle for both single-node and distributed-memory experiments and provide detailed analysis for each class of optimization. Results show our optimizations yield significant speedups across a variety of subdomain sizes while simultaneously demonstrating the potential of multi- and manycore processors to dramatically accelerate single-node performance. However, our analysis also indicates that improvements in networks and communication will be essential to reap the potential of manycore processors in large-scale multigrid calculations.

97 MATHEMATICS AND COMPUTING↗

Raexplore: Enabling Rapid, Automated Architecture Exploration for Full Applications

We present Raexplore, a performance modeling framework for architecture exploration. Raexplore enables rapid, automated, and systematic search of architecture design space by combining hardware counter-based performance characterization and analytical performance modeling. We demonstrate Raexplore for two recent manycore processors IBM Blue- Gene/Q compute chip and Intel Xeon Phi, targeting a set of scientific applications. Our framework is able to capture complex interactions between architectural components including instruction pipeline, cache, and memory, and to achieve a 3–22% error for same-architecture and cross-architecture performance predictions. Furthermore, we apply our framework to assess the two processors, and discover and evaluate a list of architectural scaling options for future processor designs.

96 KNOWLEDGE MANAGEMENT AND PRESERVATION↗

Load Balancing Scientific Applications

The largest supercomputers have millions of independent processors, and concurrency levels are rapidly increasing. For ideal efficiency, developers of the simulations that run on these machines must ensure that computational work is evenly balanced among processors. Assigning work evenly is challenging because many large modern parallel codes simulate behavior of physical systems that evolve over time, and their workloads change over time. Furthermore, the cost of imbalanced load increases with scale because most large-scale scientific simulations today use a Single Program Multiple Data (SPMD) parallel programming model, and an increasing number of processors will wait for the slowest one at the synchronization points. To address load imbalance, many large-scale parallel applications use dynamic load balance algorithms to redistribute work evenly. The research objective of this dissertation is to develop methods to decide when and how to load balance the application, and to balance it effectively and affordably. We measure and evaluate the computational load of the application, and develop strategies to decide when and how to correct the imbalance. Depending on the simulation, a fast, local load balance algorithm may be suitable, or a more sophisticated and expensive algorithm may be required. We developed a model for comparison of load balance algorithms for a specific state of the simulation that enables the selection of a balancing algorithm that will minimize overall runtime.

97 MATHEMATICS AND COMPUTING↗