Search NASA⌕ Search

SEARCH · Search NASA

Results for “partitioned operating system”

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 73 records · Page 4
All sourcesNASA NTRSDOE OSTI
DOE OSTI2021

Optimization of Asynchronous Communication Operations through Eager Notifications

UPC++ is a C++ library implementing the Asynchronous Partitioned Global Address Space (APGAS) model. We propose an enhancement to the completion mechanisms of UPC++ used to synchronize communication operations that is designed to reduce overhead for on-node operations. Our enhancement permits eager delivery of completion notification in cases where the data transfer semantics of an operation happen to complete synchronously, for example due to the use of shared-memory bypass. This semantic relaxation allows removing significant overhead from the critical path of the implementation in such cases. We evaluate our results on three different representative systems using a combination of microbenchmarks and five variations of the the HPCChallenge RandomAccess benchmark implemented in UPC++ and run on a single node to accentuate the impact of locality. We find that in RMA versions of the benchmark written in a straightforward manner (without manually optimizing for locality), the new eager notification mode can provide up to a 25% speedup when synchronizing with promises and up to a 13.5x speedup when synchronizing with conjoined futures. We also evaluate our results using a graph matching application written with UPC++ RMA communication, where we measure overall speedups of as much as 11% in single-node runs of the unmodified application code, due to our transparent enhancements.

Kamil, Amir↗
DOE OSTI2024

Distributed Multi-GPU Community Detection on Exascale Computing Platforms

Community detection is a fundamental operation in graph mining, and by uncovering hidden structures and patterns within complex systems it helps solve fundamental problems pertaining to social networks, such as information diffusion, epidemics, and recommender systems. Scaling graph algorithms for massive networks becomes challenging on modern distributed-memory multi-GPU (Graphics Processing Unit) systems due to limitations such as irregular memory access patterns, load imbalances, higher communication-computation ratios, and cross-platform support. We present a novel algorithm HiPDPL-GPU (distributed parallel Louvain) to address these challenges. We conduct experiments involving different partitioning techniques to achieve optimized performance of HiPDPL-GPU on the two largest supercomputers: Frontier and Summit. Remarkably, HiPDPL-GPU processes a graph with 4.2 billion edges in less than 3 minutes using 1024 GPUs. Qualitatively performance of HiPDPL-GPU is similar or better compared to other state-of-the-art CPU- and GPU-based implementations. While prior GPU implementations have predominantly employed CUDA, our first-of-its-kind implementation for community detection is cross-platform, accommodating both AMD and NVIDIA GPUs.

graph algorithms, high performance comptuing↗
DOE OSTI2023

Efficient Network Partitioning: Application for Decentralized State Estimation in Power Distribution Grids

Increase in the proliferation of distributed energy resources require real-time situational awareness for efficient grid operations. State estimation plays an important role for the real-time control and management of the power grid. As the sensing infrastructure grows, aggregating and handling high volumes of data at a centralized location is extremely difficult. To address this challenge, this paper first proposes a novel and efficient hier-archical spectral clustering-based network partitioning algorithm followed by a decentralized compressive sensing (DCS)-based state estimation. The applicability of the proposed network partitioning algorithm is tested on an IEEE 123-bus network, an IEEE 8,500-node system, and a 6,000+ node distribution network. The results shows that the proposed approach efficiently divides the network into multiple sub-networks with the minimum number of edge connections among the neighbors. Then, we perform DCS-based state estimation on the 6,000+ node distribution network after dividing the network into 18 optimal partitions. Simulation results show that the DCS-based state estimation recovers the system states with high accuracy and low complexity.

alternating direction method of multipliers↗
DOE OSTI2024

Distributed Multi-GPU Community Detection on Exascale Computing Platforms

Community detection is a fundamental operation in graph mining, and by uncovering hidden structures and patterns within complex systems it helps solve fundamental problems pertaining to social networks, such as information diffusion, epidemics, and recommender systems. Scaling graph algorithms for massive networks becomes challenging on modern distributed-memory multi-GPU (Graphics Processing Unit) systems due to limitations such as irregular memory access patterns, load imbalances, higher communication-computation ratios, and cross-platform support. We present a novel algorithm HiPDPL-GPU (Distributed Parallel Louvain) to address these challenges. We conduct experiments involving different partitioning techniques to achieve an optimized performance of HiPDPL-GPU on the two largest supercomputers: Frontier and Summit. Remarkably, HiPDPL-GPU processes a graph with 4.2 billion edges in less than 3 minutes using 1024 GPUs. Qualitatively, the performance of HiPDPL-GPU is similar or better compared to other state-of-the-art CPU- and GPU-based implementations. While prior GPU implementations have predominantly employed CUDA, our first-of-its-kind implementation for community detection is cross-platform, accommodating both AMD and NVIDIA GPUs.

Sattar, Naw Safrin↗
DOE OSTI2026

Lindblad many-body scars

Quantum many-body scars have received much recent attention for being both intriguing nonergodic states in otherwise quantum chaotic systems and promising candidates to encode quantum information efficiently. So far, these studies have mostly been restricted to Hermitian systems. Here, we study many-body scars in many-body quantum chaotic systems coupled to a Markovian bath, which we term Lindblad many-body scars. They are defined as simultaneous eigenvectors of the Hamiltonian and dissipative parts of the vectorized Liouvillian. Importantly, because their eigenvalues are purely real, they are not related to revivals. The number and nature of the scars depend on both the symmetry of the Hamiltonian and the choice of jump operators. For a dissipative four-body Sachdev-Ye-Kitaev (SYK) model with 𝑁 fermions, either Majorana or complex, we construct analytically some of these Lindblad scars while others could only be obtained numerically. As an example of the former, we identify 𝑁/2+1 scars for complex fermions due to the 𝑈⁡(1) symmetry of the model and two scars for Majorana fermions as a consequence of the parity symmetry. Similar results are obtained for a dissipative XXZ spin chain. We also characterize the physical properties of Lindblad scars. First, the operator size is independent of the disorder realization and has a vanishing variance. By contrast, the operator size for nonscarred states, believed to be quantum chaotic, is well described by a distribution centered around a specific size and a finite variance, which could be relevant for a precise definition of the eigenstate thermalization hypothesis in dissipative quantum chaos. Moreover, the entanglement entropy of these scars has distinct features such as a strong dependence on the partition choice and, in certain cases, a large entanglement.

Eigenstate thermalization↗
DOE OSTI2023

Distributed PV Hosting Capacity Evaluation Considering Equitable PV Accommodation

Distributed photovoltaic (DPV) hosting capacity is widely used by distribution system operators to evaluate how much DPV the distribution networks can accommodate without violating operational constraints. If the proliferation of DPV deployment is appropriately managed, it offers an excellent opportunity to mitigate the energy inequity issue in electric grids. Hosting capacity improvement plans that neglect energy equity can exacerbate unequal DPV access. To consider energy equity in distribution network planning to enhance DPV accommodation, hosting capacity evaluations should be improved to consider equity constraints endogenously. This paper proposes a DPV hosting capacity evaluation method that considers equitable DPV access among sub-regions of a distribution feeder partitioned by socioeconomic indicators, such as income level. The results demonstrate that considering equity in DPV hosting capacity evaluations can provide a valuable reference in the equitable energy transition. This work provides guidance for future distribution network upgrade planning for equitable DPV deployment.

distributed PV↗
DOE OSTI2024

OPER: Optimality-Guided Embedding Table Parallelization for Large-scale Recommendation Model

With the sharp increasing volume of user data, Deep Learning Recommendation Model (DLRM) becomes an indispensable infrastructure in large technology companies. However, large-scale DLRM on the multi-GPU platform is still inefficient due to unbalanced workload partitioning and intensive inter-GPU communication. To this end, we propose OPER, an OPtimality guided Embedding table placement for large-scale Recommendation model training and inference. OPER explores the potential of mitigating remote memory access latency in DLRM through fine-grained embedding table placement. Specifically, OPER proposes a theoretical modeling that builds up the relationship between EMT placement and the embedding communication latency in both training and inference. OPER proves the NP hardness of finding the optimal embedding table placement and proposes a heuristic algorithm that yields near optimal placement. OPER implements a SHMEM-based embedding table training system and a unified embedding index mapping to support fine-grained embedding table sharding and placement. Comprehensive experiments reveal that OPER achieves on average 3.4× and 5.1× speedup on training and inference respectively over state-of-the-art DLRM frameworks.

Wang, Zheng↗
DOE OSTI2020

Dominant Wave Energy Systems and Conditional Wave Resource Characterization for Coastal Waters of the United States

Opportunities and constraints for wave energy conversion technologies and projects are evaluated by identifying and characterizing the dominant wave energy systems for United States (US) coastal waters using marginal and joint distributions of the wave energy in terms of the peak period, wave direction, and month. These distributions are computed using partitioned wave parameters generated from a 30 year WaveWatch III model hindcast, and regionally averaged to identify the dominant wave systems contributing to the total annual available energy ( A A E ) for eleven distinct US wave energy climate regions. These dominant wave systems are linked to the wind systems driving their generation and propagation. In addition, conditional resource parameters characterizing peak period spread, directional spread, and seasonal variability, which consider dependencies of the peak period, direction, and month, are introduced to augment characterization methods recommended by international standards. These conditional resource parameters reveal information that supports project planning, conceptual design, and operation and maintenance. The present study shows that wave energy resources for the United States are dominated by long-period North Pacific swells (Alaska, West Coast, Hawaii), short-period trade winds and nor’easter swells (East Coast, Puerto Rico), and wind seas (Gulf of Mexico). Seasonality, peak period spread, and directional spread of these dominant wave systems are characterized to assess regional opportunities and constraints for wave energy conversion technologies targeting the dominant wave systems.

16 TIDAL AND WAVE POWER↗
DOE OSTI2020

Characteristics of the sub-micron ash aerosol generated during oxy-coal combustion at atmospheric and elevated pressures

This paper is concerned with the effect of pressure on the particle size distribution and the size-segregated composition of the sub-micron ash aerosol created during oxy-coal combustion under near practical self-sustaining combustion conditions. The problem is important because pressurized oxy-coal combustion has been proposed as one promising technology to minimize CO 2 emissions. Sub-micron ash plays a major role in ash deposition mechanisms, which, in turn, can control boiler performance. In this work, the same bituminous coal was burned at pressures of 1, 8 and 15 bar in O 2 /CO 2 environments. Here, tests employed a 100 kW (rated) oxy-fuel combustor (OFC) operated at atmospheric pressure (1 bar) and a 300 kW (rated) entrained-flow pressurized reactor (EFPR) at elevated pressures (8 and 15 bar). Although these tests were conducted under near practical combustion conditions, confounding effects of peak flame temperature variations were minimized for the 1 bar and 15 bar tests, allowing the role of elevated pressure to be isolated. For the EFPR tests, a specially designed sampling system was used to sample sub-micron ash aerosols from the pressurized combustor and is described in detail. Results showed that at the same peak temperature but higher pressure, the fractions of ash aerosol partitioned into the PM 0.6 and PM 1 size fractions were greatly diminished. Moreover, elevated pressures caused significant changes in the composition of the (size-segregated) sub-micron aerosol, especially in its alkali content, which increased significantly. Examination of fractions of each element that ended up in the sub-micron fume suggested that, at constant temperature, the effect of pressure on vaporization of semi-volatile metals was very different from that on the release into the gas phase of non-volatile metals and could not be explained by equilibrium.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2020

Iodine Speciation Effects in LAW Feeds

Tests were conducted on a DM10 vitrification system with prototypical off-gas treatment system, which includes an SBS, WESP, and evaporator based on the design of the evaporator to be installed at the WTP for concentration of the liquid effluents. This allowed determination of the partitioning and speciation of iodine between the melter and the off gas system components, including liquid effluents from the SBS and WESP as well as in the evaporator condensate and concentrate. Evaporator testing was decoupled from melter testing to permit filtration and pH adjustments of process solutions prior to evaporation. In addition, a slip stream was pulled from the exhaust downstream of the WESP and High-Efficiency Particulate Air (HEPA) filters and passed through a caustic scrubber simulating the WTP caustic scrubber to assess iodine partitioning to the caustic scrubber liquid effluent. The tests were conducted using simulated Hanford LAW from tank AP-107, which is the planned first feed for Direct Feed LAW (DFLAW) operations, at a nominal feed concentration of 5.6 molar Na over about 120 hours of testing.

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗
DOE OSTI2023

Distributed PV Hosting Capacity Evaluation Considering Equitable PV Accommodation: Preprint

In distribution systems, distributed photovoltaic (DPV) hosting capacity is widely used by system operators to evaluate how much DPV the distribution networks can accommodate without violating operational constraints. With the proliferation of DPV deployment, if managed properly, it offers a great opportunity to mitigate the persistent energy inequity issue in electric grid. Otherwise, energy inequity can be exacerbated with unequal DPV access when hosting capacity improvement plans ignore the equity factor. To consider energy equity in distribution network planning for enhancing DPV accommodation, the hosting capacity evaluation should be improved to consider equity constraints endogenously. In this paper, a DPV hosting capacity evaluation method is proposed considering equitable DPV access among sub-regions of a distribution feeder partitioned by social-economic indicators such as the income levels. The results demonstrate that considering equity in DPV hosting capacity evaluation can provide valuable reference in the equitable energy transition. This work provides a valuable guidance for future distribution network upgrade planning for equitable DPV deployment.

distributed PV↗
DOE OSTI2024

Jacobian-scaled K-means clustering for physics-informed segmentation of reacting flows

This work introduces Jacobian-scaled K-means (JSK-means) clustering, which is a physicsinformed clustering strategy centered on the K-means framework. The method allows for the injection of underlying physical knowledge into the clustering procedure through a distance function modification: instead of leveraging conventional Euclidean distance vectors, the JSKmeans procedure operates on distance vectors scaled by matrices obtained from dynamical system Jacobians evaluated at the cluster centroids. The goal of this work is to show how the JSKmeans algorithm - without modifying the input dataset - produces clusters that capture regions of dynamical similarity, in that the clusters are redistributed towards high-sensitivity regions in phase space and are described by similarity in the source terms of samples instead of the samples themselves. The algorithm is demonstrated on a complex reacting flow simulation dataset (a channel detonation configuration), where the dynamics in the thermochemical composition space are known through the highly nonlinear and stiff Arrhenius-based chemical source terms. Interpretations of cluster partitions in both physical space and composition space reveal how JSK-means shifts clusters produced by standard K-means towards regions of high chemical sensitivity (e.g., towards regions of peak heat release rate near the detonation reaction zone). Furthermore, the findings presented here illustrate the benefits of utilizing Jacobian-scaled distances in clustering techniques, and the JSK-means method in particular displays promising potential for improving former partition-based modeling strategies in reacting flow (and other multi-physics) applications.

Clustering↗
DOE OSTI2025

Screening green solvents for multilayer plastic film recycling processes

Multilayer (ML) plastic films are essential packaging materials that help protect products from diverse external factors; however, only 5% of all ML films are recycled in the United States. Solvent-based technologies are a promising alternative for recycling ML films because they enable recovery of constituent polymer resins. For example, the Solvent Targeted Recovery and Precipitation (STRAPTM) process sequentially dissolves and separates polymer components using a series of targeted solvent washes. A crucial design aspect of this process is the impact of selected solvents on human health and on the environment. Here, this work introduces a computational framework that integrates molecular modeling, process modeling, techno-economic analysis (TEA), and life-cycle analysis (LCA) to quickly screen green solvents for solvent-based ML recycling processes. Initial screening for solvents based on selectivity is performed by estimating temperature-dependent solubilities using molecular-scale models. Subsequent screening uses basic estimates of energy use and octanol-water partition coefficients (logP) as key measures of health, safety, and environmental hazards. Detailed process modeling, TEA, and LCA are used on a reduced set of promising solvents identified in early screening steps to more accurately determine how solvent selection and associated operating conditions impact overall economics and environmental impacts. The framework is used for the identification of green solvents (from a database of 1,000 solvents) that separate an industrial ML film composed of polyethylene (PE), ethylene vinyl alcohol (EVOH), and polyethylene terephthalate (PET). Our analysis shows the effectiveness of the framework and reveals fundamental trade-offs between solvent greenness, solubility, and economics. Our work emphasizes the importance of taking a holistic systems view during solvent design and aims to inform the development of new processes for ML film recycling and the identification of new ML films that are easier to recycle.

economics↗
DOE OSTI2023

A Fine-grained Asynchronous Bulk Synchronous parallelism model for PGAS applications

The Partitioned Global Address Space (PGAS) model is well suited for executing irregular applications on cluster-based systems, due to its efficient support for short, one-sided messages. Separately, the actor model has been gaining popularity as a productive asynchronous message-passing approach for distributed objects in enterprise and cloud computing platforms, typically implemented in languages such as Erlang, Scala or Rust. To the best of our knowledge, there has been no past work on using the actor model to deliver both productivity and scalability to irregular PGAS applications with large number of small messages. In this paper, we introduce a new programming system for PGAS applications, in which point-to-point remote operations can be expressed as fine-grained asynchronous actor messages. In our approach, the programmer does not need to worry about programming complexities related to message aggregation and termination detection. Our approach can be viewed as extending the classical Bulk Synchronous Parallelism model with fine-grained asynchronous communications within a phase or superstep. Here, we believe that our approach offers a desirable point in the productivity-performance space for PGAS applications, with more scalable performance and higher productivity relative to past approaches. Specifically, for seven irregular mini-applications from the Bale Kernels and three graph kernels executed using 2048 cores in the NERSC Cori system, our approach shows geometric mean performance improvements of ≥ 20X relative to standard PGAS versions (UPC and OpenSHMEM) while maintaining comparable productivity to those versions.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2023

Quantum simulations of SO(5) many-fermion systems using qudits

The structure and dynamics of many-body systems are the result of a delicate interplay between underlying interactions. Fermionic pairing, for example, plays a central role in various physical systems, ranging from condensed matter to nuclear systems, where it can lead to collective phenomena such as superconductivity and superfluidity. In atomic nuclei, the interplay between pairing and particle-hole interactions leads to a high degree of complexity and intricate entanglement structures. Despite this apparent complexity, symmetries emerge and manifest themselves in observable regular patterns. These symmetries and their breakings have long been used to determine relevant degrees of freedom and simplify classical descriptions of many-body systems. Here, this work explores the potential utility of quantum computers with arrays of qudits in simulating interacting fermionic systems, when the qudits can naturally map the relevant degrees of freedom determined by an underlying symmetry group. The Agassi model of fermions interacting via particle-hole and pairing interactions is based on an underlying so(5) algebra. Such systems can intuitively be partitioned into pairs of modes with five basis states, which thus naturally map to arrays of d = 5 qudits (qu5its). Classical noiseless simulations of the time evolution of systems with up to twelve qu5its are performed, by implementing quantum circuits that are developed herein, using PYTHON codes invoking Google's CIRQ software. The resource requirements of the qu5it circuits are analyzed and compared with two different mappings to qubit systems: a physics-aware Jordan-Wigner mapping requiring four qubits per mode pair and a state-to-state mapping requiring three qubits per mode pair. While the dimensionality of Hilbert spaces in mappings to qu5it systems are less than those for the corresponding qubit systems, the number of entangling operations, depending on the available hardware, can either be greater or smaller than for the physics-aware Jordan-Wigner mapping. The state-to-state mapping, while having a smaller Hilbert space than Jordan-Wigner mappings, appears to be the least efficient in gate counts. Further, a previously unknown sign problem has been identified from Trotterization errors in time evolving high-energy excitations. There appear to be advantages in employing quantum computers with arrays of qudits to perform simulations of many-body dynamics that exploit the role of underlying symmetries, specifically in lowering the required quantum resources and in reducing anticipated errors that take the simulation out of the physical space. If the necessary entangling gates are not directly supported by the hardware, physics-aware mappings to qubits may, however, be advantageous for other aspects.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
DOE OSTI2021

Modeling Data Movement Performance on Heterogeneous Architectures

The cost of data movement on parallel systems varies greatly with machine architecture, job partition, and nearby jobs. Performance models that accurately capture the cost of data movement provide a tool for analysis, allowing for communication bottlenecks to be pinpointed. Modern heterogeneous architectures yield increased variance in data movement as there are a number of viable paths for inter-GPU communication. In this paper, we present performance models for the various paths of inter-node communication on modern heterogeneous architectures, including the trade-off between GPUDirect communication and copying to CPUs. Furthermore, we present a novel optimization for inter-node communication based on these models, utilizing all available CPU cores per node. Finally, we show associated performance improvements for MPI collective operations.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

BCSR on GPU: A Way Forward Extreme-scale Graph Processing on Accelerator-enabled Frontier Supercomputer

Handling large graphs in a distributed environment requires effective partitioning across processors and efficient management of local partitions. In 2D partitioning, local graphs often become too sparse, making memory-efficient data structures crucial. Using the Compressed Sparse Row (CSR) format wastes space, especially for > 83% of vertices with empty edges for the sparse graphs. This study explores bit-CSR (BCSR), a modified CSR representation, on GPUs to reduce memory usage in graph computations. We achieved 16.67% memory savings on a sparse rmat dataset with 268 million vertices and 357 million edges, without performance degradation, supported by both theoretical and experimental storage savings of 33%. However, we observed a 1.7× slowdown in degree lookup times due to bitwise operations on AMD CPUs. This analysis highlights the potential of BCSR on GPUs for improving Graph500 benchmark performance on GPU-accelerated systems, such as the Frontier supercomputer.

Sattar, Naw Safrin↗
DOE OSTI2021

Scalable Approaches to Selecting Key Entities in Large Networked Infrastructure Systems

This work aims at bringing advances in discrete optimization algorithms to solving practical engineering problems at scale. Often times, in many engineering design problems, there is a need to select a small set of influential or representative elements from a large ground set of entities in an optimal fashion. Submodular optimization provides for a formal way to solve such problems. Common examples with infrastructure systems involve sensor placement and identification of key entities with certain objectives. However, scaling these approaches to large infrastructure systems can be challenging because of the high computational complexity of the overall framework that include the optimization algorithms as well as high-complexity compute-oracles that provide the necessary objective function values. In this work, we explore a well-studied and widely-applicable paradigm, namely leader-selection in a multi-agent networked setting in the context of scalable methodologies. We demonstrate novel frameworks that utilize variations of accelerated submodular optimization algorithms along with linear-algebraic methods that can help accelerate the oracle computations. We further explore this combination in conjunction with graph partitioning paradigms to take advantage of the accelerated algorithms in a distributed setting. Finally we demonstrate the key findings on a practical problem in an operational setting. For this, we leverage an example road network with approximately 18k nodes and 27k edges in a traffic control application, where we seek a limited number of k=200 key intersections. This problem can be solved in a serial setting in just under 5 hours providing more than 2 orders of magnitude speed-up over methods that do not consider acceleration techniques.

Visweswara Sathanur, Arun↗