Search NASA⌕ Search

SEARCH · Search NASA

Results for “computer system benchmarking”

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 1,135 records · Page 63

Enabling Diverse Software Stacks on Supercomputers using High Performance Virtual Clusters.

While large-scale simulations have been the hallmark of the High Performance Computing (HPC) community for decades, Large Scale Data Analytics (LSDA) workloads are gaining attention within the scientific community not only as a processing component to large HPC simulations, but also as standalone scientific tools for knowledge discovery. With the path towards Exascale, new HPC runtime systems are also emerging in a way that differs from classical distributed com- puting models. However, system software for such capabilities on the latest extreme-scale DOE supercomputing needs to be enhanced to more appropriately support these types of emerging soft- ware ecosystems. In this paper, we propose the use of Virtual Clusters on advanced supercomputing resources to enable systems to support not only HPC workloads, but also emerging big data stacks. Specifi- cally, we have deployed the KVM hypervisor within Cray's Compute Node Linux on a XC-series supercomputer testbed. We also use libvirt and QEMU to manage and provision VMs directly on compute nodes, leveraging Ethernet-over-Aries network emulation. To our knowledge, this is the first known use of KVM on a true MPP supercomputer. We investigate the overhead our solution using HPC benchmarks, both evaluating single-node performance as well as weak scaling of a 32-node virtual cluster. Overall, we find single node performance of our solution using KVM on a Cray is very efficient with near-native performance. However overhead increases by up to 20% as virtual cluster size increases, due to limitations of the Ethernet-over-Aries bridged network. Furthermore, we deploy Apache Spark with large data analysis workloads in a Virtual Cluster, ef- fectively demonstrating how diverse software ecosystems can be supported by High Performance Virtual Clusters.

97 MATHEMATICS AND COMPUTING↗

Cyclus fuel cycle simulation capabilities with the CYDER disposal system model

An algorithm and supporting database for rapid thermal repository loading calculation was implemented in CYDER. This algorithm employs a Specific Temperature Change (STC) method and has resulted from combining detailed spent nuclear fuel composition data with a detailed thermal repository performance analysis tool from Lawrence Livermore National Laboratory (LLNL), Argonne National Laboratory (ANL), and the Used Fuel Disposition (UFD) campaign. By abstraction of and benchmarking against these detailed thermal models, CYDER captures the dominant physics of thermal phenomena affecting repository capacity in various geologic media and as a function of spent fuel composition. Abstraction based on detailed computational thermal repository performance calculations with the LLNL semi-analytic model has resulted in implementation of the STC estimation algorithm and a supporting reference dataset. This method is capable of rapid estimation of temperature increase near emplacement tunnels as a function of waste composition, limiting radius, waste package spacing, near field thermal conductivity and near field thermal diffusivity. The sensitivity analyses and validation efforts conducted in this work demonstrate the capability of the CYDER tool to provide repository capacity and performance metrics in the context of dynamic fuel cycle.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗

Solving Inverse Detection Problems Using Passive Radiation Signatures [Slides]

The ability to reconstruct an unknown radioactive object based on its passive gamma-ray and neutron signatures is very important in homeland security applications. Often in the analysis of unknown radioactive objects, for simplicity or speed or because there is no other information, they are modeled as spherically symmetric regardless of their actual geometry. In these presentation we discuss the accuracy and implications of this approximation for decay gamma rays and for neutron-induced gamma rays. We discuss an extension of spherical raytracing (for uncollided fluxes) that allows it to be used when the exterior shielding is flat or cylindrical. We revisit some early results in boundary perturbation theory, showing that the Roussopolos estimate is the correct one to use when the quantity of interest is the flux or leakage on the boundary. We apply boundary perturbation theory to problems in which spherically symmetric systems are perturbed in asymmetric nonspherical ways. We apply mesh adaptive direct search (MADS) algorithms to object reconstructions. We present a benchmark test set that may be used to quantitatively evaluate inverse detection methods.

42 ENGINEERING↗

Respecting causality for training physics-informed neural networks

While the popularity of physics-informed neural networks (PINNs) is steadily rising, to this date PINNs have not been successful in simulating dynamical systems whose solution exhibits multi-scale, chaotic or turbulent behavior. In this work we attribute this shortcoming to the inability of existing PINNs formulations to respect the spatio-temporal causal structure that is inherent to the evolution of physical systems. We argue that this is a fundamental limitation and a key source of error that can ultimately steer PINN models to converge towards erroneous solutions. We address this pathology by proposing a simple re-formulation of PINNs loss functions that can explicitly account for physical causality during model training. We demonstrate that this simple modification alone is enough to introduce significant accuracy improvements, as well as a practical quantitative mechanism for assessing the convergence of a PINNs model. Here we provide state-of-the-art numerical results across a series of benchmarks for which existing PINNs formulations fail, including the chaotic Lorenz system, the Kuramoto–Sivashinsky equation in the chaotic regime, and the Navier–Stokes equations. To the best of our knowledge, this is the first time that PINNs have been successful in simulating such systems, introducing new opportunities for their applicability to problems of industrial complexity.

97 MATHEMATICS AND COMPUTING↗

On the Scalability of Data Reduction Techniques in Current and Upcoming HPC Systems from an Application Perspective

We implement and benchmark parallel I/O methods for the fully-manycore driven particle-in-cell code PIConGPU. Identifying throughput and overall I/O size as a major challenge for applications on today’s and future HPC systems, we present a scaling law characterizing performance bottlenecks in state-of-the-art approaches for data reduction. Consequently, we propose, implement and verify multi-threaded data-transformations for the I/O library ADIOS as a feasible way to trade underutilized host-side compute potential on heterogeneous systems for reduced I/O latency.

Huebl, Axel↗

Role of Dynamic Polarization Interactions in the Electrical Double Layer at Calcite (104) Interfaces with Aqueous Solutions

Reactions at mineral interfaces with aqueous solutions control many geochemical and biogeochemical processes in the Earth’s critical zone. At the molecular level, insights into important properties such as the structure of the electrical double layer (EDL) at specific mineral interfaces continue to improve due to the increasing fidelity of laboratory instrumentation and computational approaches. However, molecular simulation approaches suffer from limited reach into relevant scales of time, length, and system complexity. To span this gap, a novel hybrid approach that couples first principle plane-wave density functional theory (DFT) with classical DFT (cDFT) is demonstrated and applied to calcite (104) interfaces with various electrolytes. In this approach, a region of interest described using DFT interacts with the surrounding medium described using cDFT to arrive at a self-consistent ground state. Benchmarking against experimental observations and entirely first principle DFT simulations demonstrates that this hybrid model efficiently encompasses the key short-range and collective interactions in the EDL. Simulations of calcite (104)/solution interfaces reveal the key static and dynamic polarization interactions that give rise to structuring of ions and water. Ion hydration interactions have the strongest effect on the depth of the first minimum in the density distribution of counterions at the surface, and the position and width of the first density peak is largely determined by the strength of ion-correlation forces. Finer details of ion distributions are controlled by mutual polarization of the calcite surface and interfacial electrolyte. Finally, this new ability to efficiently and rigorously predict EDL structure at mineral surfaces in contact with complex solutions paves the way to accurately modeling sorption, nucleation, dissolution, and growth in realistic systems.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

How strongly do hydrogen and water molecules stick to carbon nanomaterials?

The interaction strength of molecular hydrogen and water to carbon nanomaterials is relevant to, among many applications, hydrogen storage, water treatment, and water flow. Yet, accurate interaction energies for hydrogen and water with carbon nanotubes (CNTs) remain scarce despite the importance of having reliable benchmark data to inform experiments and to validate computational models. Here, benchmark fixed-node diffusion Monte Carlo (DMC) interaction energies are provided for hydrogen and water monomers inside and outside a typical zigzag CNT. The DMC interaction energies provide valuable insight into molecular interactions with CNTs in general and are also expected to be particularly relevant to gas uptake studies on CNTs. In addition, a selection of density functional theory (DFT) exchange-correlation (xc) functionals and force field potentials that ought to be suitable for these systems is compared. An unexpected variation is found in the performance of DFT van der Waals (vdW) models in particular. An analysis of the peculiar discrepancy between different vdW models indicates that medium-range correlation (at circa 3 to 5 Å) plays a key role inside CNTs and is poorly predicted by some vdW models. Using accurate reference information, this work reveals which xc functionals and force fields perform well for molecules interacting with CNTs. The results will be valuable to future work on these and related systems that involve molecules interacting with low-dimensional systems.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Molecular NMR shieldings, J -couplings, and magnetizabilities from numeric atom-centered orbital based density-functional calculations

This paper reports and benchmarks a new implementation of nuclear magnetic resonance shieldings, magnetizabilities, and J-couplings for molecules within semilocal density functional theory, based on numeric atom-centered orbital (NAO) basis sets. NAO basis sets are attractive for the calculation of these nuclear magnetic resonance (NMR) parameters because NAOs provide accurate atomic orbital representations especially near the nucleus, enabling high-quality results at modest computational cost. Moreover, NAOs are readily adaptable for linear scaling methods, enabling efficient calculations of large systems. Here, the paper has five main parts: (1) It reviews the formalism of density functional calculations of NMR parameters in one comprehensive text to make the mathematical background available in a self-contained way. (2) The paper quantifies the attainable precision of NAO basis sets for shieldings in comparison to specialized Gaussian basis sets, showing similar performance for similar basis set size. (3) The paper quantifies the precision of calculated magnetizabilities, where the NAO basis sets appear to outperform several established Gaussian basis sets of similar size. (4) The paper quantifies the precision of computed J-couplings, for which a group of customized NAO basis sets achieves precision of ~Hz for smaller basis set sizes than some established Gaussian basis sets. (5) The paper demonstrates that the implementation is applicable to systems beyond 1000 atoms in size.

74 ATOMIC AND MOLECULAR PHYSICS↗

Performance Characteristics of the BlueField-2 SmartNIC

High-performance computing (HPC) researchers have long envisioned scenarios where application workflows could be improved through the use of programmable processing elements embedded in the network fabric. Recently, vendors have introduced programmable Smart Network Interface Cards (SmartNICs) that enable computations to be offloaded to the edge of the network. There is great interest in both the HPC and high-performance data analytics (HPDA) communities in understanding the roles these devices may play in the data paths of upcoming systems. This paper focuses on characterizing both the networking and computing aspects of NVIDIA’s new BlueField-2 SmartNIC when used in a 100Gb/s Ethernet environment. For the networking evaluation we conducted multiple transfer experiments between processors located at the host, the SmartNIC, and a remote host. These tests illuminate how much effort is required to saturate the network and help estimate the processing headroom available on the SmartNIC during transfers. For the computing evaluation we used the stress-ng benchmark to compare the BlueField-2 to other servers and place realistic bounds on the types of offload operations that are appropriate for the hardware. Our findings from this work indicate that while the BlueField-2 provides a flexible means of processing data at the network’s edge, great care must be taken to not overwhelm the hardware. While the host can easily saturate the network link, the SmartNIC’s embedded processors may not have enough computing resources to sustain more than half the expected bandwidth when using kernel-space packet processing. From a computational perspective, encryption operations, memory operations under contention, and on-card IPC operations on the SmartNIC perform significantly better than the general-purpose servers used for comparisons in our experiments. Therefore, applications that mainly focus on these operations may be good candidates for offloading to the SmartNIC.

97 MATHEMATICS AND COMPUTING↗

Evaluation of Cache-based Superscalar and Cacheless Vector Architectures for Scientific Computations

The growing gap between sustained and peak performance for scientific applications has become a well-known problem in high performance computing. The recent development of parallel vector systems offers the potential to bridge this gap for a significant number of computational science codes and deliver a substantial increase in computing capabilities. This paper examines the intranode performance of the NEC SX6 vector processor and the cache-based IBM Power3/4 superscalar architectures across a number of key scientific computing areas. First, we present the performance of a microbenchmark suite that examines a full spectrum of low-level machine characteristics. Next, we study the behavior of the NAS Parallel Benchmarks using some simple optimizations. Finally, we evaluate the perfor- mance of several numerical codes from key scientific computing domains. Overall results demonstrate that the SX6 achieves high performance on a large fraction of our application suite and in many cases significantly outperforms the RISC-based architectures. However, certain classes of applications are not easily amenable to vectorization and would likely require extensive reengineering of both algorithm and implementation to utilize the SX6 effectively.

Oliker, Leonid↗

A History of High-Performance Computing

Faster than most speedy computers. More powerful than its NASA data-processing predecessors. Able to leap large, mission-related computational problems in a single bound. Clearly, it s neither a bird nor a plane, nor does it need to don a red cape, because it s super in its own way. It's Columbia, NASA s newest supercomputer and one of the world s most powerful production/processing units. Named Columbia to honor the STS-107 Space Shuttle Columbia crewmembers, the new supercomputer is making it possible for NASA to achieve breakthroughs in science and engineering, fulfilling the Agency s missions, and, ultimately, the Vision for Space Exploration. Shortly after being built in 2004, Columbia achieved a benchmark rating of 51.9 teraflop/s on 10,240 processors, making it the world s fastest operational computer at the time of completion. Putting this speed into perspective, 20 years ago, the most powerful computer at NASA s Ames Research Center, home of the NASA Advanced Supercomputing Division (NAS), ran at a speed of about 1 gigaflop (one billion calculations per second). The Columbia supercomputer is 50,000 times faster than this computer and offers a tenfold increase in capacity over the prior system housed at Ames. What s more, Columbia is considered the world s largest Linux-based, shared-memory system. The system is offering immeasurable benefits to society and is the zenith of years of NASA/private industry collaboration that has spawned new generations of commercial, high-speed computing systems.

Source record↗

ConnectX-2 InfiniBand Management Queues: First Investigation of the New Support for Network Offloaded Collective Operations

This paper introduces the newly developed Infini-Band (IB) Management Queue capability, used by the Host Channel Adapter (HCA) to manage network task data flow dependancies, and progress the communications associated with such flows. These tasks include sends, receives, and the newly supported wait task, and are scheduled by the HCA based on a data dependency description provided by the user. This functionality is supported by the ConnectX-2 HCA, and provides the means for delegating collective communication management and progress to the HCA, also known as collective communication offload. This provides a means for overlapping collective communications managed by the HCA and computation on the Central Processing Unit (CPU), thus making it possible to reduce the impact of system noise on parallel applications using collective operations. This paper further describes how this new capability can be used to implement scalable Message Passing Interface (MPI) collective operations, describing the high level details of how this new capability is used to implement the MPI Barrier collective operation, focusing on the latency sensitive performance aspects of this new capability. This paper concludes with small scale benchmark experiments comparing implementations of the barrier collective operation, using the new network offload capabilities, with established point-to-point based implementations of these same algorithms, which manage the data flow using the central processing unit. These early results demonstrate the promise this new capability provides to improve the scalability of high performance applications using collective communications. The latency of the HCA based implementation of the barrier is similar to that of the best performing point-to-point based implementation managed by the central processing unit, starting to outperform these as the number of processes involved in the collective operation increases.

Graham, Richard L↗

ConnectX2 In niBand Management Queues: New support for Network Of oaded

This paper introduces the newly developed InfiniBand (IB) Management Queue capability, used by the Host Channel Adapter (HCA) to manage network task data flow dependancies, and progress the communications associated with such flows. These tasks include sends, receives, and the newly supported wait task, and are scheduled by the HCA based on a data dependency description provided by the user. This functionality is supported by the ConnectX-2 HCA, and provides the means for delegating collective communication management and progress to the HCA, also known as collective communication offload. This provides a means for overlapping collective communications managed by the HCA and computation on the Central Processing Unit (CPU), thus making it possible to reduce the impact of system noise on parallel applications using collective operations. This paper further describes how this new capability can be used to implement scalable Message Passing Interface (MPI) collective operations, describing the high level details of how this new capability is used to implement the MPI Barrier collective operation, focusing on the latency sensitive performance aspects of this new capability. This paper concludes with small scale benchmark experiments comparing implementations of the barrier collective operation, using the new network offload capabilities, with established point-to-point based implementations of these same algorithms, which manage the data flow using the central processing unit. These early results demonstrate the promise this new capability provides to improve the scalability of high-performance applications using collective communications. The latency of the HCA based implementation of the barrier is similar to that of the best performing point-to-point based implementation managed by the central processing unit, starting to outperform these as the number of processes involved in the collective operation increases.

99 GENERAL AND MISCELLANEOUS↗

Plutonium Critical Mass Curve Comparison to Mass at Upper Subcritical Limit (USL) Using Whisper

Whisper is computational software designed to assist the nuclear criticality safety analyst with validation studies with the MCNP ® Monte Carlo radiation transport package. Standard approaches to validation rely on the selection of benchmarks based upon expert judgment. Whisper uses sensitivity/uncertainty (S/U) methods to select relevant benchmarks to a particular application or set of applications being analyzed. Using these benchmarks, Whisper computes a calculational margin. Whisper attempts to quantify the margin of subcriticality (MOS) from errors in software and uncertainties in nuclear data. The combination of the Whisper-derived calculational margin and MOS comprise the baseline upper subcritical limit (USL), to which an additional margin may be applied by the nuclear criticality safety analyst as appropriate to ensure subcriticality. A series of critical mass curves for plutonium, similar to those found in Figure 31 of LA-10860-MS, have been generated using MCNP6.1.1 and the iterative parameter study software, WORM_Solver. The baseline USL for each of the data points of the curves was then computed using Whisper 1.1. The USL was then used to determine the equivalent mass for plutonium metal-water system. ANSI/ANS-8.1 states that it is acceptable to use handbook data, such as the data directly from the LA-10860-MS, as it is already considered validated (Section 4.3 4) “Use of subcritical limit data provided in ANSI/ANS standards or accepted reference publications does not require further validation.”). This paper attempts to take a novel approach to visualize traditional critical mass curves and allows comparison with the amount of mass for which the k eff is equal to the USL (calculational margin + margin of subcriticality). However, the intent is to plot the critical mass data along with USL, not to suggest that already accepted handbook data should have new and more rigorous requirements for validation.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗

Validation for 233U-Fueled Systems in KENO V.a in SCALE 6.2

Validation of the KENO V.a Monte Carlo code has been performed in recent years using benchmark experiments contained in the International Criticality Safety Benchmark Evaluation Project (ICSBEP) Handbook. The models used in these recent validation efforts have been drawn from the Verified, Archived Library of Inputs and Data (VALID). Models must be generated and reviewed by qualified individuals to be added to the VALID library, which leads to high confidence in the models and reduces the probability that apparent validation issues are a result of modeling errors. The contents of the VALID library are somewhat limited because of the effort involved in generating and reviewing models, so no 233U-fueled experiments are included in VALID. A large number of models of 233U experiments have been modeled in the past in support of the criticality safety of 233U in storage at Oak Ridge National Laboratory (ORNL). In June 2016, two midshipmen from the United States Naval Academy were assigned to ORNL, and they reviewed and updated these models for execution in SCALE 6.2. The validation of 233U is of interest because some discrepancies in SCALE results were identified by a beta tester at the Nuclear Energy Agency modeling these systems. These discrepancies were related to a data thinning procedure in the continuous-energy nuclear data process, and had been resolved prior to the release of SCALE 6.2

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗

Flux in a 1D-Reactor Cell by Response Matrix

Legacy computational reactor physics software tools and protocols currently used in support of the Advanced Test Reactor (ATR) core fuel management and safety assurance, and to some extent, experiment management, are inconsistent with the state of modern nuclear engineering practice. Thus, they are difficult, if not impossible, to verify and validate (V and V) according to modern standards. Furthermore, the legacy staff knowledge required for effective application of their application and protocols from the 1960's and 1970's is rapidly being lost due to staff turnover and retirements. In late 2009, the Idaho National Laboratory (INL) initiated a focused effort, the ATR Core Modeling Update Project, to address this situation through the introduction of modern high-fidelity transport theory based computational software and models to replace the legacy methods, with appropriate (V and V) protocols according to applicable national standards. The new computational tools include: - The 2-dimensional HELIOS deterministic transport code for integrated ATR core fuel and experiment management and Core Safety Analysis Package (CSAP) physics data generation. The MCNP5 3-dimensional stochastic simulation (Monte Carlo) code for verification and validation applications and for specialized ATR core analysis along with its longstanding traditional role in experiment support - The 2-dimensional deterministic NEWT transport code and the 3-dimensional KENO6 stochastic simulation code, with their SCALE-based support infrastructure for use in V and V applications and as supporting modules for sensitivity-uncertainty analysis via the TSUNAMI system. The ATTILA 3-dimensional discrete-ordinates transport code for specialized experiment support and possibly additional applications. Regardless of application, every neutron transport code, new or legacy, requires verification. The aim of this presentation is to provide a neutron transport benchmark for a simple 1D-reactor cell in plane geometry. The benchmark finds use as a standard of comparison for the codes listed above intended for the ATR upgrade. The method developed here is in contrast to the finite Fourier transform solution found in Ref. 1, which is limited to isotropic scattering and particularly difficult to numerically implement unless the total cross section is constant over the cell.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗

Principal component analysis of computed emission lines from protostellar jets

A very important issue concerning protostellar jets is the mechanism behind their formation. Obtaining information on the region at the base of a jet can shed light on the subject, and some years ago this was done through a search for a rotational signature in the jet line spectrum. The existence of such signatures, however, remains controversial. In order to contribute to the clarification of this issue, in this paper we show that principal component analysis (PCA) can potentially help to distinguish between rotation and precession effects in protostellar jet images. This method reduces the dimensions of the data, facilitating the efficient extraction of information from large data sets such as those arising from integral field spectroscopy. PCA transforms the system of correlated coordinates into a system of uncorrelated coordinates, the eigenvectors, ordered by principal components of decreasing variance. The projection of the data on these coordinates produces images called tomograms, while eigenvectors can be displayed as eigenspectra. The combined analysis of both can allow the identification of patterns correlated to a particular physical property that would otherwise remain hidden, and can help to separate the effects of physically uncorrelated phenomena in the data. These are, for example, rotation and precession in the kinematics of a stellar jet. In order to show the potential of PCA analysis, we apply it to synthetic spectro-imaging datacubes generated as an output of numerical simulations of protostellar jets. In this way we generate a benchmark with which a PCA diagnostics of real observations can be confronted. Using the computed emission line profiles for [O i]λ6300 and [S ii]λ6716, we recover and analyze the effects of rotation and precession in tomograms generated by PCA. We show that different combinations of the eigenvectors can be used to enhance and to identify the rotation features present in the data. Our results indicate that PCA can be useful for disentangling rotation from precession in jets with an inclination of the jet with respect to the plane of the sky as high as 45°. We have been able to recover the initially imposed rotation jet profile for models at a moderate inclination angle (ϕ≪15°) and without precession.

79 ASTRONOMY AND ASTROPHYSICS↗

Automatic Qubit Characterization and Gate Optimization with QubiC

As the size and complexity of a quantum computer increases, quantum bit (qubit) characterization and gate optimization become complex and time-consuming tasks. Current calibration techniques require complicated and verbose measurements to tune up qubits and gates, which cannot easily expand to the large-scale quantum systems. We develop a concise and automatic calibration protocol to characterize qubits and optimize gates using QubiC, which is an open source FPGA (field-programmable gate array) based control and measurement system for superconducting quantum information processors. We propose multi-dimensional loss-based optimization of single-qubit gates and full XY-plane measurement method for the two-qubit CNOT gate calibration. We demonstrate the QubiC automatic calibration protocols are capable of delivering high-fidelity gates on the state-of-the-art transmon-type processor operating at the Advanced Quantum Testbed at Lawrence Berkeley National Laboratory. Finally, the single-qubit and two-qubit Clifford gate infidelities measured by randomized benchmarking are of 4.9(1.1) × 10 -4 and 1.4(3) × 10 -2 , respectively.

97 MATHEMATICS AND COMPUTING↗