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 811 records · Page 45

Multinode Multi-GPU Two-Electron Integrals: Code Generation Using the Regent Language

The computation of two-electron repulsion integrals (ERIs) is often the most expensive step of integral-direct self-consistent field methods. Formally it scales as O(N 4 ), where N is the number of Gaussian basis functions used to represent the molecular wave function. In practice, this scaling can be reduced to O(N 2 ) or less by neglecting small integrals with screening methods. The contributions of the ERIs to the Fock matrix are of Coulomb (J) and exchange (K) type and require separate algorithms to compute matrix elements efficiently. We previously implemented highly efficient GPU-accelerated J-matrix and K-matrix algorithms in the electronic structure code TeraChem. Although these implementations supported the use of multiple GPUs on a node, they did not support the use of multiple nodes. This presents a key bottleneck to cutting-edge ab initio simulations of large systems, e.g., excited state dynamics of photoactive proteins. We present our implementation of multinode multi-GPU J- and K-matrix algorithms in TeraChem using the Regent programming language. Regent directly supports distributed computation in a task-based model and can generate code for a variety of architectures, including NVIDIA GPUs. We demonstrate multinode scaling up to 45 GPUs (3 nodes) and benchmark against hand-coded TeraChem integral code. Finally, we also outline our metaprogrammed Regent implementation, which enables flexible code generation for integrals of different angular momenta.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Introducing the embedded random phase approximation: H 2 dissociative adsorption on Cu(111) as an exemplar

The random phase approximation (RPA) as a means of treating electron correlation recently has been shown to outperform standard density functional theory (DFT) approximations in a variety of cases. However, the computational cost of the RPA is substantially more than DFT, especially when aiming to study extended surfaces. Properly accounting for sufficient surface ensemble size, Brillouin zone sampling, and vacuum separation of periodic images in standard periodic-planewave-based DFT code raises the cost to achieve converged results. Here, we show that sub-system embedding schemes enable use of the RPA for modeling heterogeneous reactions at reduced computational cost. Further, we explore two different embedded RPA (emb-RPA) approaches, periodic emb-RPA and cluster emb-RPA. We use the (experimentally and theoretically) well-studied H 2 dissociative adsorption on Cu(111) as our exemplar, and first perform full periodic RPA calculations as a benchmark. The full RPA results match well the semi-empirical barrier fit to experimental observables and others derived from high-level computations, e.g., from recent embedded n-electron valence second order perturbation theory [Zhao et al., J. Chem. Theory Comput. 16(11), 7078–7088 (2020)] and quantum Monte Carlo [Doblhoff-Dier et al., J. Chem. Theory Comput. 13(7), 3208–3219 (2017)] simulations. Among the two emb-RPA approaches tested, the cluster emb-RPA accurately reproduces the energy profile (maximum error of 50 meV along the reaction pathway) while reducing the computational cost by approximately two orders of magnitude. We therefore expect that the embedded cluster approach will enable wider RPA implementation in heterogeneous catalysis.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

SPIKANs: separable physics-informed Kolmogorov–Arnold networks

Physics-Informed Neural Networks (PINNs) have emerged as a promising method for solving partial differential equations (PDEs) in scientific computing. While PINNs typically use multilayer perceptrons (MLPs) as their underlying architecture, recent advancements have explored alternative neural network structures. One such innovation is the Kolmogorov–Arnold Network (KAN), which has demonstrated benefits over traditional MLPs, including faster neural scaling and better interpretability. The application of KANs to physics-informed learning has led to the development of Physics-Informed KANs (PIKANs), enabling the use of KANs to solve PDEs. However, despite their advantages, KANs often suffer from slower training speeds, particularly in higher-dimensional problems where the number of collocation points grows exponentially with the dimensionality of the system. To address this challenge, we introduce Separable Physics-Informed Kolmogorov–Arnold Networks (SPIKANs). This novel architecture applies the principle of separation of variables to PIKANs, decomposing the problem such that each dimension is handled by an individual KAN. This approach drastically reduces the computational complexity of training without sacrificing accuracy, facilitating their application to higher-dimensional PDEs. Through a series of benchmark problems, we demonstrate the effectiveness of SPIKANs, showcasing their superior scalability and performance compared to PIKANs and highlighting their potential for solving complex, high-dimensional PDEs in scientific computing.

Kolmogorov-Arnold networks↗

A method for modeling laterally asymmetric proton beamlets resulting from collimation

Purpose: To introduce a method to model the 3D dose distribution of laterally asymmetric proton beamlets resulting from collimation. The model enables rapid beamlet calculation for spot scanning (SS) delivery using a novel penumbra-reducing dynamic collimation system (DCS) with two pairs of trimmers oriented perpendicular to each other. Methods: Trimmed beamlet dose distributions in water were simulated with MCNPX and the collimating effects noted in the simulations were validated by experimental measurement. The simulated beamlets were modeled analytically using integral depth dose curves along with an asymmetric Gaussian function to represent fluence in the beam’s eye view (BEV). The BEV parameters consisted of Gaussian standard deviations (sigmas) along each primary axis (σ{sub x1},σ{sub x2},σ{sub y1},σ{sub y2}) together with the spatial location of the maximum dose (μ{sub x},μ{sub y}). Percent depth dose variation with trimmer position was accounted for with a depth-dependent correction function. Beamlet growth with depth was accounted for by combining the in-air divergence with Hong’s fit of the Highland approximation along each axis in the BEV. Results: The beamlet model showed excellent agreement with the Monte Carlo simulation data used as a benchmark. The overall passing rate for a 3D gamma test with 3%/3 mm passing criteria was 96.1% between the analytical model and Monte Carlo data in an example treatment plan. Conclusions: The analytical model is capable of accurately representing individual asymmetric beamlets resulting from use of the DCS. This method enables integration of the DCS into a treatment planning system to perform dose computation in patient datasets. The method could be generalized for use with any SS collimation system in which blades, leaves, or trimmers are used to laterally sharpen beamlets.

07 ISOTOPE AND RADIATION SOURCES↗

CityBES v2021

City Buildings, Energy, and Sustainability (CityBES) is a web-based data and computing platform, focusing on energy modeling and analysis of a city's building stock to support district or city-scale building energy efficiency programs. CityBES uses an international open data standard, CityGML, to represent and exchange 3D city models. CityBES employs EnergyPlus to simulate building energy use and savings from energy efficient retrofits. Other CityBES features include energy benchmarking, district heating and cooling system modeling, rooftop PV analysis, building performance visualization, heat resilience modeling, as well as urban scale mapping of microclimate and heat vulnerability at census tract level. Different from other tools, CityBES uses integrated open and standard 3D city building data and models each individual building using EnergyPlus. CityBES can be used by urban planners, city energy managers, building owners, utilities, energy consultants and researchers.

Hong, Tianzhen↗

CLOMP: Accurately Characterizing OpenMP Application Overheads

Despite its ease of use, OpenMP has failed to gain widespread use on large scale systems, largely due to its failure to deliver sufficient performance. Our experience indicates that the cost of initiating OpenMP regions is simply too high for the desired OpenMP usage scenario of many applications. In this paper, we introduce CLOMP, a new benchmark to characterize this aspect of OpenMP implementations accurately. CLOMP complements the existing EPCC benchmark suite to provide simple, easy to understand measurements of OpenMP overheads in the context of application usage scenarios. Our results for several OpenMP implementations demonstrate that CLOMP identifies the amount of work required to compensate for the overheads observed with EPCC.We also show that CLOMP also captures limitations for OpenMP parallelization on SMT and NUMA systems. Finally, CLOMPI, our MPI extension of CLOMP, demonstrates which aspects of OpenMP interact poorly with MPI when MPI helper threads cannot run on the NIC.

99 GENERAL AND MISCELLANEOUS↗

QTRAJ 1.0: A Lindblad equation solver for heavy-quarkonium dynamics

Here we introduce an open-source package called QTraj that solves the Lindblad equation for heavy-quarkonium dynamics using the quantum trajectories algorithm. The package allows users to simulate the suppression of heavy-quarkonium states using externally-supplied input from 3+1D hydrodynamics simulations. The code uses a split-step pseudo-spectral method for updating the wave-function between jumps, which is implemented using the open-source multi-threaded FFTW3 package. This allows one to have manifestly unitary evolution when using real-valued potentials. In this paper, we provide detailed documentation of QTraj 1.0, installation instructions, and present various tests and benchmarks of the code.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Localized Radial Basis Functions for Radiation Transport in Slab Geometry

Meshing the spatial domain for the neutron transport equation can be computationally expensive and labor-intensive. Mesh-free methods, such as radial basis function (RBF) approximations, may alleviate these concerns by using a pointwise solution of the transport equation instead of the standard volume-integrated solution using cells or elements. This allows for the use of any geometric description that can provide the material at solution points and the boundary normal directions for boundary points. RBFs give spectral convergence when used to solve differential equations, but only if they are infinitely smooth. This occurs when the RBFs are effectively flat, creating an ill-conditioned, dense system of equations. One simple way to alleviate this issue is to use localized RBFs. Global RBFs have been applied previously to the monoenergetic radiative transport equation in X-Y geometry using source iteration. In this summary, localized RBFs are applied to the multigroup neutron transport equation in slab geometry using Krylov iteration. The results show agreement with benchmark solutions, provided the coefficients for the RBF basis are properly chosen. (authors)

97 MATHEMATICS AND COMPUTING↗

Performance of fully-coupled algebraic multigrid preconditioners for large-scale VMS resistive MHD

Here, we explore the current performance and scaling of a fully-implicit stabilized unstructured finite element (FE) variational multiscale (VMS) capability for large-scale simulations of 3D incompressible resistive magnetohydrodynamics (MHD). The large-scale linear systems that are generated by a Newton nonlinear solver approach are iteratively solved by preconditioned Krylov subspace methods. The efficiency of this approach is critically dependent on the scalability and performance of the algebraic multigrid preconditioner. Our study considers the performance of the numerical methods as recently implemented in the second-generation Trilinos implementation that is 64-bit compliant and is not limited by the 32-bit global identifiers of the original Epetra-based Trilinos. The study presents representative results for a Poisson problem on 1.6 million cores of an IBM Blue Gene/Q platform to demonstrate very large-scale parallel execution. Additionally, results for a more challenging steady-state MHD generator and a transient solution of a benchmark MHD turbulence calculation for the full resistive MHD system are also presented. These results are obtained on up to 131,000 cores of a Cray XC40 and one million cores of a BG/Q system.

97 MATHEMATICS AND COMPUTING↗

Bringing HPE Slingshot 11 support to Open MPI

The Cray HPE Slingshot 11 network is used on the new exascale systems arriving at the U.S. Department of Energy (DoE) laboratories (e.g., Frontier, Aurora, Perlmutter). As such, the support of this network is an important capability to meet the needs of exascale applications. Here, this article highlights recent work to develop supporting infrastructure to enable Open MPI to efficiently support these new platforms. A key component of this effort involves development of a new Open Fabrics Interface (OFI) provider, LinkX. We discuss the design and development of enhancements that take advantage of the new Slingshot 11 network and AMD GPUs. We include performance data from tests on the Frontier supercomputer using synthetic communication benchmarks, and the vendor provided MPI as a baseline for comparison. The tests demonstrate full functionality of Open MPI on the system and initial results show favorable performance when compared to the highly tuned vendor implementation.

97 MATHEMATICS AND COMPUTING↗

A data-driven network optimisation approach to coordinated control of distributed photovoltaic systems and smart buildings in distribution systems

The increasing integration of distributed energy resources, including demand-side resources and distributed photovoltaics (PVs), into distribution systems has resulted in more complicated power system operation. A data-driven network optimisation approach is proposed to coordinate the control of distributed PVs and smart buildings in distribution networks considering the uncertainties of solar power, outdoor temperature and heat gain associated with building thermal dynamics. These uncertain parameters have a significant impact on the operation and control of distributed PVs and smart buildings, bringing challenges to the distribution system operation. In the proposed data-driven distributionally robust optimisation (DRO) approach, the Wasserstein ball is used to construct an ambiguity set for the uncertain parameters, which does not require the probability distributions to be known. Furthermore, a conditional value-at-risk is incorporated into the Wasserstein-based DRO model and converted into a computationally tractable mixed-integer convex optimisation problem. Benchmarked with robust optimisation and chance-constrained programming, the proposed data-driven model can give a less conservative robust solution.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Numerical calculation of neoclassical distribution functions and current profiles in low collisionality, axisymmetric plasmas

A new code, the Neoclassical Ion-Electron Solver (NIES), has been written to solve for stationary, axisymmetric distribution functions (f) in the conventional banana regime for both ions and electrons using a set of drift-kinetic equations (DKEs) with linearized Fokker-Planck-Landau collision operators. Solvability conditions on the DKEs determine the relevant non-adiabatic pieces of f (called h). We work in a 4D phase space in which {psi} defines a flux surface, {theta} is the poloidal angle, v is the magnitude of the velocity referenced to the mean flow velocity, and {lambda} is the dimensionless magnetic moment parameter. We expand h in finite elements in both v and {lambda}. The Rosenbluth potentials, {Phi} and {Psi}, which define the integral part of the collision operator, are expanded in Legendre series in cos{chi}, where {chi} is the pitch angle, Fourier series in cos{theta}, and finite elements in v. At each {psi}, we solve a block tridiagonal system for h{sub i} (independent of f{sub e}), then solve another block tridiagonal system for h{sub e} (dependent on f{sub i}). We demonstrate that such a formulation can be accurately and efficiently solved. NIES is coupled to the MHD equilibrium code JSOLVER [J. DeLucia et al., J. Comput. Phys. 37, 183-204 (1980)] allowing us to work with realistic magnetic geometries. The bootstrap current is calculated as a simple moment of the distribution function. Results are benchmarked against the Sauter analytic formulas and can be used as a kinetic closure for an MHD code (e.g., M3D-C{sup 1}[S. C. Jardin et al., Comput. Sci. Discovery 5, 014002 (2012)]).

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Inclusion of nuclear quantum effects for simulations of nonlinear spectroscopy

The computation and interpretation of nonlinear vibrational spectroscopy is of vital importance for understanding a wide range of dynamical processes in molecular systems. Here, we introduce an approach to evaluate multi-time response functions in terms of multi-time double symmetrized Kubo transformed thermal correlation functions. Furthermore, we introduce a multi-time extension of ring polymer molecular dynamics to evaluate these Kubo transforms. Benchmark calculations show that the approximations are useful for short times even for nonlinear operators, providing a consistent improvement over classical simulations of multi-time correlation functions. In conclusion, the introduced methodology thus provides a practical way of including nuclear quantum effects in multi-time response functions of non-linear optical spectroscopy.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Automatic and Transparent Resource Contention Mitigation for Improving Large-Scale Parallel File System Performance

Proportional to the scale increases in HPC systems, many scientific applications are becoming increasingly data intensive, and parallel I/O has become one of the dominant factors impacting the large-scale HPC application performance. On a typical large-scale HPC system, we have observed that the lack of a global workload coordination coupled with the shared nature of storage systems cause load imbalance and resource contention over the end-to-end I/O paths resulting in severe performance degradation. I/O load imbalance on HPC systems is generally a self-inflicted wound and mostly occurs between the I/O paths and resources consumed by each individual job. In this paper, we introduce TAPP-IO, a dynamic, shared load balancing framework for mitigating resource contention. TAPP-IO extends our previous work and solves two major limitations: First, it transparently intercepts file creation calls during runtime to balance the workload over all available storage targets. The usage of TAPP-IO requires no application source code modifications and is independent from any I/O middleware. The framework can be applied to almost any HPC platform and is suitable for systems that lack a centralized file system resource manager. Second, the framework proposes a new placement strategy to support not only file-per-process I/O, but also single shared file I/O. This opens the door to a new class of scientific applications that can leverage the placement library for improved performance. We demonstrate the effectiveness of our integration on the Titan system at the Oak Ridge National Laboratory. Our experiments with a synthetic benchmark and real-world HPC workload show that, even in a noisy production environment, TAPP-IO can improve large-scale application performance significantly.

Neuwirth, Sarah↗

On the use of term rewriting for performance optimization of legacy HPC applications

Preparing codes for next generation supercomputer systems is anticipated to require significant changes to the optimization strategies employed in established HPC applications. In this paper, we present our experience in applying term-rewriting transformations for the optimization of such applications. We have designed application-specific term-rewriting transformations to improve the scalability, enhance locality and reduce the communication overhead of the Self-Consistent Field (SCF) computational chemistry benchmark. We present the rationale for the use of term-rewriting in this manner, the design of our transformations, and the much enhanced performance of the resulting code.

Global Arrays↗

SharP: Towards Programming Extreme-Scale Systems with Hierarchical Heterogeneous Memory

The pre-exascale systems are expected to have a significant amount of hierarchical and heterogeneous on-node memory, and this trend of system architecture in extreme-scale systems is expected to continue into the exascale era. Along with hierarchical-heterogeneous memory, the system typically has a high-performing network and a compute accelerator. This system architecture is not only effective for running traditional High Performance Computing (HPC) applications (Big-Compute), but also running data-intensive HPC applications and Big-Data applications. As a consequence, there is a growing desire to have a single system serve the needs of both Big-Compute and Big-Data applications. Though the system architecture supports the convergence of the Big-Compute and Big-Data, the programming models have yet to evolve to support either hierarchical-heterogeneous memory systems or the convergence. In this work, we propose and develop the programming abstraction called SHARed data-structure centric Programming abstraction (SharP) to address both of these goals, i.e., provide (1) a simple, usable, and portable abstraction for hierarchical-heterogeneous memory and (2) a unified programming abstraction for Big-Compute and Big-Data applications. To evaluate SharP, we implement a Stencil benchmark using SharP, port QMCPack, a petascale-capable application, and adapt Memcached ecosystem, a popular Big-Data framework, to use SharP, and quantify the performance and productivity advantages. Additionally, we demonstrate the simplicity of using SharP on different memories including DRAM, High-bandwidth Memory (HBM), and non-volatile random access memory (NVRAM).

97 MATHEMATICS AND COMPUTING↗

A Deep Generative Model for Non-Intrusive Identification of EV Charging Profiles

The proliferation of electric vehicles (EVs) brings environmental benefits and technical challenges to power grids. An identification algorithm which can accurately extract individual EV charging profiles out of widely available smart meter measurements has attracted great interests. This paper proposes a non-intrusive identification framework for EV charging profile extraction, which is driven by deep generative models (DGM). First, the proposed DGM is designed as a representation layer embedded into the Markov process and used to model the joint probability distribution of available time-series data. A novel contribution is to approximate posterior distributions by neural networks whose parameters are obtained by variational inference and supervised learning. Second, the EV charging status is inferred from the DGM via dynamic programming. Lastly, the desired EV charging profile can be reconstructed by the rated power of EV models and inferred status. Compared with the benchmark Hidden Markov Models, the proposed framework can better handle noise in data with less computational complexity and better overall accuracy performances with smaller recall. The proposed framework is validated by numerical experiments on the Pecan Street dataset.

33 ADVANCED PROPULSION SYSTEMS↗

gdess: A framework for evaluating simulated atmospheric CO 2 in Earth System Models

Atmospheric carbon dioxide (CO 2 ) plays a key role in the global carbon cycle and global warming. Climate-carbon feedbacks are often studied and estimated using Earth System Models (ESMs), which couple together multiple model components—including the atmosphere, ocean, terrestrial biosphere, and cryosphere—to jointly simulate mass and energy exchanges within and between these components. Despite tremendous advances, model intercomparisons and benchmarking are aspects of ESMs that warrant further improvement (Fer et al., 2021; Smith et al., 2014). Such benchmarking is critical because comparing the value of state variables in these simulations against observed values provides evidence for appropriately refining model components; moreover, researchers can learn much about Earth system dynamics in the process (Randall et al., 2019). We introduce `gdess` (a.k.a., Greenhouse gas Diagnostics for Earth System Simulations), which parses observational datasets and ESM simulation output, combines them to be in a consistent structure, computes statistical metrics, and generates diagnostic visualizations. In its current incarnation, `gdess` facilitates evaluating a model's ability to reproduce observed temporal and spatial variations of atmospheric CO 2 . The diagnostics implemented modularly in `gdess` support more rapid assessment and improvement of model-simulated global CO 2 sources and sinks associated with land and ocean ecosystem processes. We intend for this set of automated diagnostics to form an extensible, open source framework for future comparisons of simulated and observed concentrations of various greenhouse gases across Earth system models.

97 MATHEMATICS AND COMPUTING↗