Search NASA⌕ Search

SEARCH · Search NASA

Results for “Parallelization efficiency”

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 1,531 records · Page 85

BALANCING THE LOAD: A VORONOI BASED SCHEME FOR PARALLEL COMPUTATIONS

One of the key issues when running a simulation on multiple CPUs is maintaining a proper load balance throughout the run and minimizing communications between CPUs. We propose a novel method of utilizing a Voronoi diagram to achieve a nearly perfect load balance without the need of any global redistributions of data. As a show case, we implement our method in RICH, a two-dimensional moving mesh hydrodynamical code, but it can be extended trivially to other codes in two or three dimensions. Our tests show that this method is indeed efficient and can be used in a large variety of existing hydrodynamical codes.

79 ASTRONOMY AND ASTROPHYSICS↗

Parallelized telecom quantum networking with an ytterbium-171 atom array

The integration of quantum computers and sensors into a quantum network enables new capabilities in quantum information science. Most networks with atom-like qubits operate at visible or near-ultraviolet wavelengths and require conversion to the telecom band for long-distance communication, which reduces efficiency and potentially introduces noise. In this article we report high-fidelity entanglement between ytterbium-171 atoms and optical photons generated directly in the telecommunication band, where fibre loss is low. The nuclear spin of the atom is entangled with a single photon in the time-bin basis, yielding a high atom-measurement-corrected atom–photon Bell state fidelity. This can be further improved by addressing photon measurement errors. By imaging the atom array onto an optical fibre array, we also implement a parallelized networking protocol that can increase the remote entanglement rate proportionately with the number of channels. We also preserve coherence on a memory qubit during operations on communication qubits. These results support the integration of atomic systems into scalable quantum networks.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Additive manufactured bipolar plate for high-efficiency hydrogen production in proton exchange membrane electrolyzer cells

Additive manufacturing (AM) technology is capable of fast and low-cost prototyping from complex 3D digital models. To take advantage of this technology, a stainless steel (SS) plate with parallel flow field served as a combination of a cathode bipolar plate and a current distributor; it was fabricated using selective laser melting (SLM) techniques and investigated in a proton exchange membrane electrolyzer cell (PEMEC) in-situ for the first time. The experimental results show that the PEMEC with an AM SS cathode bipolar plate can achieve an excellent performance for hydrogen production for a voltage of 1.779 V and a current density of 2.0 A/cm 2 . The AM SS cathode bipolar plate was also characterized by SEM and EDS, and the results show a uniform elemental distribution across the plate with very limited oxidization. As a result, this research demonstrates that AM method could be a route to aid cost-effective and rapid development of PEMECs.

08 HYDROGEN↗

FOX: A Fault Oblivious Extreme-Scale Execution Environment

The FOX project, funded under the ASCR X-stack I program, developed systems software and runtime libraries for a new approach to the data and work distribution for massively parallel, fault oblivious application execution. Our work was motivated by the premise that exascale computing systems will provide a thousand-fold increase in parallelism and a proportional increase in failure rate relative to today’s machines. To deliver the capability of exascale hardware, the systems software must provide the infrastructure to support existing applications while simultaneously enabling efficient execution of new programming models that naturally express dynamic, adaptive, irregular computation; coupled simulations; and massive data analysis in a highly unreliable hardware environment with billions of threads of execution. Our OS research has prototyped new methods to provide efficient resource sharing, synchronization, and protection in a many-core compute node. We have experimented with alternative task/dataflow programming models and shown scalability in some cases to hundreds of thousands of cores. Much of our software is in active development through open source projects. Concepts from FOX are being pursued in next generation exascale operating systems. Our OS work focused on adaptive, application tailored OS services optimized for multi → many core processors. We developed a new operating system NIX that supports role-based allocation of cores to processes which was released to open source. We contributed to the IBM FusedOS project, which promoted the concept of latency-optimized and throughput-optimized cores. We built a task queue library based on distributed, fault tolerant key-value store and identified scaling issues. A second fault tolerant task parallel library was developed, based on the Linda tuple space model, that used low level interconnect primitives for optimized communication. We designed fault tolerance mechanisms for task parallel computations employing work stealing for load balancing that scaled to the largest existing supercomputers. Finally, we implemented the Elastic Building Blocks runtime, a library to manage object-oriented distributed software components. To support the research, we won two INCITE awards for time on Intrepid (BG/P) and Mira (BG/Q). Much of our work has had impact in the OS and runtime community through the ASCR Exascale OS/R workshop and report, leading to the research agenda of the Exascale OS/R program. Our project was, however, also affected by attrition of multiple PIs. While the PIs continued to participate and offer guidance as time permitted, losing these key individuals was unfortunate both for the project and for the DOE HPC community.

97 MATHEMATICS AND COMPUTING↗

DyG-DPCD: A Distributed Parallel Community Detection Algorithm for Large-Scale Dynamic Graphs

Dynamic (Temporal) graphs capture the valuable evolution of real-world systems, from the continuously evolving patterns of social interactions and genetic pathways to the dynamic fluctuations of economic forces. Detecting communities for such evolving networks poses unique challenges. Detecting and analyzing the evolution of communities within dynamic graphs unlocks valuable insights into the underlying structural and temporal patterns of real-world systems. However, the sheer volume of modern graph data and the inherent complexity of the temporal dimension pose significant challenges to scalable community detection algorithms. Addressing this gap, our work explores the limited landscape of scalable distributed-memory parallel methods specifically designed for dynamic network community detection. We propose a novel parallel algorithm, DyG-DPCD (Dynamic Graph Distributed Parallel Community Detection), to detect communities in dynamic networks using the Message Passing Interface (MPI) framework. We present a vertex-centric approach, allowing us to detect communities through local optimization. Furthermore, we enhance our baseline algorithm by incorporating three heuristics, which improve the algorithm’s performance significantly while maintaining the quality of the solutions. We demonstrate the efficiency of our algorithm by experimenting on several real-world large-scale networks with hundreds of millions of edges spanning diverse domains. Notably, DyG-DPCD achieves speedups between 25× and 30× for large networks that we experimented on using NERSC compute nodes. In conclusion, our algorithm outperforms the STINGER parallel re-agglomeration algorithm by 30×.

97 MATHEMATICS AND COMPUTING↗

Large Scale Caching and Streaming of Training Data for Online Deep Learning

The training of deep neural network models on large data remains a difficult problem, despite progress towards scalable techniques. In particular, there is a mismatch between the random but predetermined order in which AI flows select training samples and the streaming I/O patterns for which traditional HPC data storage (e.g., parallel file systems) are designed. In addition, as more data are obtained, it is feasible neither simply to train learning models incrementally, due to catastrophic forgetting (i.e., bias towards new samples), nor to train frequently from scratch, due to prohibitive time and/or resource constraints. In this paper, we study data management techniques that combine caching and streaming with rehearsal support in order to enable efficient access to training samples in both offline training and continual learning. We revisit state-of-art streaming approaches based on data pipelines that transparently handle prefetching, caching, shuffling, and data augmentation, and discuss the challenges and opportunities that arise when combining these methods with data-parallel training techniques. We also report on preliminary experiments that evaluate the I/O overheads involved in accessing the training samples from a parallel file system (PFS) under several concurrency scenarios, highlighting the impact of the PFS on the design of the data pipelines.

data pipelines↗

Multiplexed high resolution soft x-ray RIXS

High-resolution Resonance Inelastic X-ray Scattering (RIXS) is a technique that allows us to probe the electronic excitations of complex materials with unprecedented precision. However, the RIXS process has a low cross section, compounded by the fact that the optical spectrometers used to analyze the scattered photons can only collect a small solid angle and overall have a small efficiency. Here we present a method to significantly increase the throughput of RIXS systems, by energy multiplexing, so that a complete RIXS map of scattered intensity versus photon energy in and photon energy out can be recorded simultaneously{sup 1}. This parallel acquisition scheme should provide a gain in throughput of over 100.. A system based on this principle, QERLIN, is under construction at the Advanced Light Source (ALS).

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Parallel exponential time differencing methods for geophysical flow simulations

Two ocean models are considered for geophysical flow simulations: the multilayer shallow water equations and the multilayer primitive equations. For the former, we investigate the parallel performance of exponential time differencing (ETD) methods, including exponential Rosenbrock–Euler, ETD2wave, and B-ETD2wave. For the latter, we take advantage of the splitting of barotropic and baroclinic modes and propose a new two-level method in which an ETD method is applied to solve the fast barotropic mode. Furthermore, these methods could improve the computational efficiency of numerical simulations because ETD methods allow for much larger time step sizes than traditional explicit time-stepping techniques that are commonly used in existing computational ocean models. Several standard benchmark tests for ocean modeling are performed and comparison of the numerical results demonstrates a great potential of applying the parallel ETD methods for simulating real-world geophysical flows.

54 ENVIRONMENTAL SCIENCES↗

RELATIVISTIC RECONNECTION: AN EFFICIENT SOURCE OF NON-THERMAL PARTICLES

In magnetized astrophysical outflows, the dissipation of field energy into particle energy via magnetic reconnection is often invoked to explain the observed non-thermal signatures. By means of two- and three-dimensional particle-in-cell simulations, we investigate anti-parallel reconnection in magnetically dominated electron-positron plasmas. Our simulations extend to unprecedentedly long temporal and spatial scales, so we can capture the asymptotic state of the system beyond the initial transients, and without any artificial limitation by the boundary conditions. At late times, the reconnection layer is organized into a chain of large magnetic islands connected by thin X-lines. The plasmoid instability further fragments each X-line into a series of smaller islands, separated by X-points. At the X-points, the particles become unmagnetized and they get accelerated along the reconnection electric field. We provide definitive evidence that the late-time particle spectrum integrated over the whole reconnection region is a power law whose slope is harder than –2 for magnetizations σ ≳ 10. Efficient particle acceleration to non-thermal energies is a generic by-product of the long-term evolution of relativistic reconnection in both two and three dimensions. In three dimensions, the drift-kink mode corrugates the reconnection layer at early times, but the long-term evolution is controlled by the plasmoid instability which facilitates efficient particle acceleration, analogous to the two-dimensional physics. Our findings have important implications for the generation of hard photon spectra in pulsar winds and relativistic astrophysical jets.

79 ASTRONOMY AND ASTROPHYSICS↗

The influence of preexisting host rock damage on fault network localization

The transition from stable to unstable fracture propagation occurs when fractures begin to interact and link. Thus, fracture network coalescence controls how rocks and engineered structures fail. Here, to constrain the factors that influence localization in shear zones under brittle conditions, we build discrete element method models with a rough fault embedded in a shear zone. We add varying numbers of diffuse, randomly-placed weaknesses to examine the influence of diffuse damage on fracture network localization. The number of weaknesses controls the localization behavior of the fault network and the final fault geometry. We quantify localization using the Gini coefficient of the fracture volume, which measures the nonuniformity in a population. Each model generally increases in localization toward failure. However, models with more diffuse damage experience delocalization phases that are superimposed on the overall trend of increasing localization. The observed link between delocalization and host rock damage may help explain the varying localization of low magnitude seismicity in southern California. Models with more diffuse damage produce more complex fault geometries comprised of several parallel strands of wing cracks. The propagation of these wing cracks reduces the shear stress acting on the model boundaries, indicating that this fracture development increases the mechanical efficiency of the system.

58 GEOSCIENCES↗

Petascale Orbital-Free Density Functional Theory Enabled by Small-Box Algorithms

Orbital-free density functional theory (OFDFT) is a quantum-mechanics-based method that utilizes electron density as its sole variable. The main computational cost in OFDFT is the ubiquitous use of the fast Fourier transform (FFT), which is mainly adopted to evaluate the kinetic energy density functional (KEDF) and electron–electron Coulomb interaction terms. We design and implement a small-box FFT (SBFFT) algorithm to overcome the parallelization limitations of conventional FFT algorithms. We also propose real-space truncation of the nonlocal Wang–Teter KEDF kernel. The scalability of the SBFFT is demonstrated by efficiently simulating one full optimization step (electron density, energies, forces, and stresses) of 1,024,000 lithium (Li) atoms on up to 65,536 cores. We perform other tests using Li as a test material, including calculations of physical properties of different phases of bulk Li, geometry optimizations of nanocrystalline Li, and molecular dynamics simulations of liquid Li. All of the tests yield excellent agreement with the original OFDFT results, suggesting that the OFDFT-SBFFT algorithm opens the door to efficient first-principles simulations of materials containing millions of atoms.

Chemistry↗

Large Scale GW Calculations

We present GW calculations of molecules, ordered and disordered solids and interfaces, which employ an efficient contour deformation technique for frequency integration and do not require the explicit evaluation of virtual electronic states nor the inversion of dielectric matrices. We also present a parallel implementation of the algorithm which takes advantage of separable expressions of both the single particle Green's function and the screened Coulomb interaction. The method can be used starting from density functional theory calculations performed with semilocal or hybrid functionals. We applied the newly developed technique to GW calculations of systems of unprecedented size, including water/semiconductor interfaces with thousands of electrons.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Investigation of lower hybrid physics through power modulation experiments on Alcator C-Mod

Lower hybrid current drive (LHCD) is an attractive tool for off-axis current profile control in magnetically confined tokamak plasmas and burning plasmas (ITER), because of its high current drive efficiency. The LHCD system on Alcator C-Mod operates at 4.6 GHz, with {approx} 1 MW of coupled power, and can produce a wide range of launched parallel refractive index (n) spectra. A 32 chord, perpendicularly viewing hard x-ray camera has been used to measure the spatial and energy distribution of fast electrons generated by lower hybrid (LH) waves. Square-wave modulation of LH power on a time scale much faster than the current relaxation time does not significantly alter the poloidal magnetic field inside the plasma and thus allows for realistic modeling and consistent plasma conditions for different n spectra. Inverted hard x-ray profiles show clear changes in LH-driven fast electron location with differing n. Boxcar binning of hard x-rays during LH power modulation allows for {approx} 1 ms time resolution which is sufficient to resolve the build-up, steady-state, and slowing-down phases of fast electrons. Ray-tracing/Fokker-Planck modeling in combination with a synthetic hard x-ray diagnostic shows quantitative agreement with the x-ray data for high n cases. The time histories of hollow x-ray profiles have been used to measure off-axis fast electron transport in the outer half of the plasma, which is found to be small on a slowing down time scale.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Phoebe: a high-performance framework for solving phonon and electron Boltzmann transport equations

Understanding the electrical and thermal transport properties of materials is critical to the design of electronics, sensors, and energy conversion devices. Computational modeling can accurately predict material properties but, in order to be reliable, requires accurate descriptions of electron and phonon states and their interactions. While first-principles methods are capable of describing the energy spectrum of each carrier, using them to compute transport properties is still a formidable task, both computationally demanding and memory intensive, requiring integration of fine microscopic scattering details for estimation of macroscopic transport properties. To address this challenge, we present Phoebe—a newly developed software package that includes the effects of electron–phonon, phonon–phonon, boundary, and isotope scattering in computations of electrical and thermal transport properties of materials with a variety of available methods and approximations. This open source C++ code combines MPI-OpenMP hybrid parallelization with GPU acceleration and distributed memory structures to manage computational cost, allowing Phoebe to effectively take advantage of contemporary computing infrastructures. We demonstrate that Phoebe accurately and efficiently predicts a wide range of transport properties, opening avenues for accelerated computational analysis of complex crystals.

36 MATERIALS SCIENCE↗

Geometric GNNs for charged particle tracking at GlueX

Nuclear physics experiments are aimed at uncovering the fundamental building blocks of matter. The experiments involve high-energy collisions that produce complex events with many particle trajectories. Tracking charged particles resulting from collisions in the presence of a strong magnetic field is critical to enable the reconstruction of particle trajectories and precise determination of interactions. It is traditionally achieved through combinatorial approaches that scale worse than linearly as the number of hits grows. Since particle hit data naturally form a point cloud and can be structured as graphs, graph neural networks (GNNs) emerge as an intuitive and effective choice for this task. In this study, we evaluate the GNN model for track finding on the data from the GlueX experiment at Jefferson Lab. We use simulation data to train the model and test on both simulation and real GlueX measurements. We demonstrate that GNN-based track finding outperforms the currently used traditional method at GlueX in terms of segment-based efficiency at a fixed purity while providing faster inferences. We show that the GNN model can achieve significant speedup by processing multiple events in batches, which exploits the parallel computation capability of graphical processing units (GPUs). Finally, we compare the GNN implementation on GPU and field-programmable gate array and describe the trade-off.

batched GNN pipeline↗

Computationally efficient zero-noise extrapolation for quantum-gate-error mitigation

Zero noise extrapolation (ZNE) is a widely used technique for gate error mitigation on near term quantum computers because it can be implemented in software and does not require knowledge of the quantum computer noise parameters. Traditional ZNE requires a significant resource overhead in terms of quantum operations. A recent proposal using a targeted (or random) instead of fixed identity insertion method (riim versus fiim) requires significantly fewer quantum gates for the same formal precision. We start by showing that riim can allow for ZNE to be deployed on deeper circuits than fiim but requires many more measurements to maintain the same statistical uncertainty. We develop two extensions to fiim and riim. The List Identity Insertion Method (liim) allows to mitigate the error from certain cnot gates, typically those with the largest error. Set Identity Insertion Method (siim) naturally interpolates between the measurement-efficient fiim and the gate-efficient riim allowing to trade off fewer cnot gates for more measurements. Finally, we investigate a way to boost the number of measurements, namely to run ZNE in parallel, utilizing as many quantum devices as are available. We explore the performance of riim in a parallel setting where there is a non-trivial spread in noise across sets of qubits within or across quantum computers.

97 MATHEMATICS AND COMPUTING↗

Sparse Binary Matrix-Vector Multiplication on Neuromorphic Computers

Neuromorphic computers offer the opportunity for low-power, efficient computation. Though they have been primarily applied to neural network tasks, there is also the opportunity to leverage the inherent characteristics of neuromorphic computers (low power, massive parallelism, collocated processing and memory) to perform non-neural network tasks. Here, we demonstrate how an approach for performing sparse binary matrix-vector multiplication on neuromorphic computers. We describe the approach, which relies on the connection between binary matrix-vector multiplication and breadth first search, and we introduce the algorithm for performing this calculation in a neuromorphic way. We validate the approach in simulation. Finally, we provide a discussion of the runtime of this algorithm and discuss where neuromorphic computers in the future may have a computational advantage when performing this computation.

Schuman, Catherine↗

Meta Parallel Coordinates For Visualizing Features in Large, High-Dimensional, Time-Varying Data

Managing computational complexity and designing effective visual representations are two important challenges for the visualization of large, complex, high-dimensional datasets. Parallel coordinates are an effective technique for visualizing high- dimensional data, but do not scale well to very large datasets. The addition of the temporal dimension leads to more uncertainty due to clutter on screen. Consequently, this poses a significant challenge for visually finding trends and patterns that maximize insight about the underlying time-varying properties of the data. To address these problems, we present meta parallel coordinates, a parallel coordinates display that is guided by perceptually mo- tivated visual metrics. These metrics describe the visual structures typically found in parallel coordinates and thus aid the user’s analysis by providing meaningful views of the data. Since they are computed in screen space, our metrics are computationally more efficient than data-based metrics. Our choice of metrics is driven by the different analytical tasks that a user typically wants to perform with time-varying multivariate data. In particular, we have worked with domain scientists who performed simulations of bioremediation experiments, and use their data and results to demonstrate the usefulness of our approach

Dasgupta, Aritra↗