Search NASA⌕ Search

SEARCH · Search NASA

Results for “distributed memory”

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 631 records · Page 35

Enhanced deep neural networks with transfer learning for distribution LMP considering load and PV uncertainties

As the flexibility of generation and demand increases in distribution systems, the residential loads are emerging as a promising means to participate in demand response and the transactive energy market. Market pricing is an instrumental mechanism for the distribution system operator to exploit the full potential of the flexible resources. The distribution locational marginal price (DLMP) can be used to guide the residential load consumption. This type of market signal helps the distribution system operator to optimize the scheduling of all resources while satisfying related network constraints through a day-ahead market. However, solving the optimization problem for large-scale systems can be computationally expensive. To address the scalability and practicability limitations of the DLMP framework, a learning-based approach is proposed in this paper to complement the day-ahead distribution market framework. Here, the proposed approach combines long short-term memory and transfer learning to develop deep neural network that can capture the spatial–temporal correlation of the input data. The model can determine the optimal DLMP for each node in a distribution system without the system parameters required to formulate the optimization problem. Testing results on IEEE 33-bus and 123-bus systems show that the proposed approach can generate a comparable DLMP against the optimization solutions.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Modular chip-integrated photonic control of artificial atoms in diamond waveguides

A central goal in creating long-distance quantum networks and distributed quantum computing is the development of interconnected and individually controlled qubit nodes. Atom-like emitters in diamond have emerged as a leading system for optically networked quantum memories, motivating the development of visible-spectrum, multi-channel photonic integrated circuit (PIC) systems for scalable atom control. However, it has remained an open challenge to realize optical programmability with a qubit layer that can achieve high optical detection probability over many optical channels. Here, we address this problem by introducing a modular architecture of piezoelectrically actuated atom-control PICs (APICs) and artificial atoms embedded in diamond nanostructures designed for high-efficiency free-space collection. The high-speed four-channel APIC is based on a splitting tree mesh with triple-phase shifter Mach–Zehnder interferometers. This design simultaneously achieves optically broadband operation at visible wavelengths, high-fidelity switching (>40dB) at low voltages, submicrosecond modulation timescales (>30MHz), and minimal channel-to-channel crosstalk for repeatable optical pulse carving. Via a reconfigurable free-space interconnect, we use the APIC to address single silicon vacancy color centers in individual diamond waveguides with inverse tapered couplers, achieving efficient single photon detection probabilities (∼15%) and second-order autocorrelation measurements g (2) (0)<0.14 for all channels. The modularity of this distributed APIC–quantum memory system simplifies the quantum control problem, potentially enabling further scaling to thousands of channels.

47 OTHER INSTRUMENTATION↗

Dense LU Factorization on Multicore Supercomputer Nodes

Dense LU factorization is a prominent benchmark used to rank the performance of supercomputers. Many implementations, including the reference code HPL, use block-cyclic distributions of matrix blocks onto a two-dimensional process grid. The process grid dimensions drive a trade-off between communication and computation and are architecture- and implementation-sensitive. We show how the critical panel factorization steps can be made less communication-bound by overlapping asynchronous collectives for pivot identification and exchange with the computation of rank-k updates. By shifting this trade-off, a modified block-cyclic distribution can beneficially exploit more available parallelism on the critical path, and reduce panel factorization's memory hierarchy contention on now-ubiquitous multi-core architectures. The missed parallelism in traditional block-cyclic distributions arises because active panel factorization, triangular solves, and subsequent broadcasts are spread over single process columns or rows (respectively) of the process grid. Increasing one dimension of the process grid decreases the number of distinct processes in the other dimension. To increase parallelism in both dimensions, periodic 'rotation' is applied to the process grid to recover the row-parallelism lost by a tall process grid. During active panel factorization, rank-1 updates stream through memory with minimal reuse. In a column-major process grid, the performance of this access pattern degrades as too many streaming processors contend for access to memory. A block-cyclic mapping in the more popular row-major order does not encounter this problem, but consequently sacrifices node and network locality in the critical pivoting steps. We introduce 'striding' to vary between the two extremes of row- and column-major process grids. As a test-bed for further mapping experiments, we describe a dense LU implementation that allows a block distribution to be defined as a general function of block to processor. Other mappings can be tested with only small, local changes to the code.

99 GENERAL AND MISCELLANEOUS↗

Temporal Forecasting of Distributed Temperature Sensing in a Thermal Hydraulic System With Machine Learning and Statistical Models

We benchmark performance of long-short term memory (LSTM) network machine learning model and autoregressive integrated moving average (ARIMA) statistical model in temporal forecasting of distributed temperature sensing (DTS). Data in this study consists of fluid temperature transient measured with two co-located Rayleigh scattering fiber optic sensors (FOS) in a forced convection mixing zone of a thermal tee. We treat each gauge of a FOS as an independent temperature sensor. We first study prediction of DTS time series using Vanilla LSTM and ARIMA models trained on prior history of the same FOS that is used for testing. The results yield maximum absolute percentage error (MaxAPE) and root mean squared percentage error (RMSPE) of 1.58% and 0.06% for ARIMA, and 3.14% and 0.44% for LSTM, respectively. Next, we investigate zero-shot forecasting (ZSF) with LSTM and ARIMA trained on history of the co-located FOS only, which is advantageous when limited training data is available. The ZSF MaxAPE and RMSPE values for ARIMA are comparable to those of the Vanilla use case, while the error values for LSTM increase. We show that in ZSF, performance of LSTM network can be improved by training on most correlated gauges between the two FOS, which are identified by calculating the Pearson correlation coefficient. The improved ZSF MaxAPE and RMSPE for LSTM are 4.4% and 0.33%, respectively. Performance of ZSF LSTM can be further enhanced through transfer learning (TL), where LSTM is re-trained on a subset of the FOS that is the target of forecasting. We show that LSTM pre-trained on correlated dataset and re-trained on 30% of testing target dataset achieves MaxAPE and RMSPE values of 2.32% and 0.28%, respectively.

ARIMA↗

In-Place Zero-Space Memory Protection for CNN

Convolutional Neural Networks (CNN) are being actively explored for safetycritical applications such as autonomous vehicles and aerospace, where it is essential to ensure the reliability of inference results in the presence of possible memory faults. Traditional methods such as error correction codes (ECC) and Triple Modular Redundancy (TMR) are CNN-oblivious and incur substantial memory overhead and energy cost. This paper introduces in-place zero-space ECC assisted with a new training scheme weight distribution-oriented training. The new method provides the first known zero space cost memory protection for CNNs without compromising the reliability offered by traditional ECC.

Guan, Hui↗

Computationally efficient implementation of combustion chemistry in parallel PDF calculations

In parallel calculations of combustion processes with realistic chemistry, the serial in situ adaptive tabulation (ISAT) algorithm [S.B. Pope, Computationally efficient implementation of combustion chemistry using in situ adaptive tabulation, Combustion Theory and Modelling, 1 (1997) 41-63; L. Lu, S.B. Pope, An improved algorithm for in situ adaptive tabulation, Journal of Computational Physics 228 (2009) 361-386] substantially speeds up the chemistry calculations on each processor. To improve the parallel efficiency of large ensembles of such calculations in parallel computations, in this work, the ISAT algorithm is extended to the multi-processor environment, with the aim of minimizing the wall clock time required for the whole ensemble. Parallel ISAT strategies are developed by combining the existing serial ISAT algorithm with different distribution strategies, namely purely local processing (PLP), uniformly random distribution (URAN), and preferential distribution (PREF). The distribution strategies enable the queued load redistribution of chemistry calculations among processors using message passing. They are implemented in the software x2f{sub m}pi, which is a Fortran 95 library for facilitating many parallel evaluations of a general vector function. The relative performance of the parallel ISAT strategies is investigated in different computational regimes via the PDF calculations of multiple partially stirred reactors burning methane/air mixtures. The results show that the performance of ISAT with a fixed distribution strategy strongly depends on certain computational regimes, based on how much memory is available and how much overlap exists between tabulated information on different processors. No one fixed strategy consistently achieves good performance in all the regimes. Therefore, an adaptive distribution strategy, which blends PLP, URAN and PREF, is devised and implemented. It yields consistently good performance in all regimes. In the adaptive parallel ISAT strategy, the type and extent of redistribution is determined 'on the fly' based on the prediction of future simulation time. Compared to the PLP/ISAT strategy where chemistry calculations are essentially serial, a speed-up factor of up to 30 is achieved. The study also demonstrates that the adaptive strategy has acceptable parallel scalability.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Ionic Tuning of Cobaltites at the Nanoscale

Control of materials through custom design of ionic distributions is one of the most promising approaches for developing future technologies ranging from spintronic logic and memory devices to energy storage. Perovskites have shown particular promise for ionic-devices due to their high ion mobility and sensitivity to chemical stoichiometry. In this work, we demonstrate control of ion distributions in (La,Sr)CoO3. Commensurate with control of the oxygen stoichiometry, the Co valence and saturation magnetization show a smooth continuous variation. In contrast, magnetoresistance measurements show a sharp, step-wise transition in the magnetic anisotropy and resistivity over the same range of oxygen stoichiometry. These results suggest significant phase separation, with conductive magnetic regions and oxygen-deficient, insulating, non-magnetic regions, forming percolated networks. Indeed, X-ray diffraction identifies oxygen vacancy ordering, including the emergence of Brownmillerite ordering. The unusual appearance of the Brownmillerite phase under ambient temperature is further confirmed by high resolution scanning transmission electron microscopy. This work demonstrates room-temperature ionic control of magnetism, electrical conductivity and crystalline structure, in a 40 nm film, presenting new opportunities for ionic devices that leverage multiple material properties.

thin films, magnetism, xmcd, pnr, STEM, defects↗

TAZeR: Hiding the Cost of Remote I/O in Distributed Scientific Workflows

A perennial bottleneck in distributed workflow analytics is long access latencies for remote data. We ask the question: assuming that data must be accessed remotely, can latencies be hidden? We present TAZeR, a framework that reduces data access latency while increasing data reuse. TAZeR transparently converts POSIX I/O into operations that interleave application work with data transfer, i.e. read, prefetching, and write stage-out. TAZeR ensures read data moves directly to application memory without synchronous intervention (soft zero-copy). TAZeR uses distributed bandwidth-aware staging to exploit reuse across application tasks and to manage the capacity constraints of fast hierarchical storage. We evaluate TAZeR on a High Energy Physics workflow that requests remote data at 48 Gb/s (over two 1 Gb/s WAN links) using complex access patterns. TAZeR is 12× and 22× faster than XRootD (state-of-the-art) and file copies (current approach), respectively; and within 7% of optimal. We discuss conditions when TAZeR can hide I/O accesses; and evaluate performance as effective staging sizes change.

Tazer, remote I/O, data staging↗

Efficient Heterogeneous Execution on Large Multicore and Accelerator Platforms: Case Study Using a Block Tridiagonal Solver

The algorithmic and implementation principles are explored in gainfully exploiting GPU accelerators in conjunction with multicore processors on high-end systems with large numbers of compute nodes, and evaluated in an implementation of a scalable block tridiagonal solver. The accelerator of each compute node is exploited in combination with multicore processors of that node in performing block-level linear algebra operations in the overall, distributed solver algorithm. Optimizations incorporated include: (1) an efficient memory mapping and synchronization interface to minimize data movement, (2) multi-process sharing of the accelerator within a node to obtain balanced load with multicore processors, and (3) an automatic memory management system to efficiently utilize accelerator memory when sub-matrices spill over the limits of device memory. Results are reported from our novel implementation that uses MAGMA and CUBLAS accelerator software systems simultaneously with ACML for multithreaded execution on processors. Overall, using 940 nVidia Tesla X2090 accelerators and 15,040 cores, the best heterogeneous execution delivers a 10.9-fold reduction in run time relative to an already efficient parallel multicore-only baseline implementation that is highly optimized with intra-node and inter-node concurrency and computation-communication overlap. Detailed quantitative results are presented to explain all critical runtime components contributing to hybrid performance.

Park, Alfred J↗

Reconfigurable perovskite nickelate electronics for artificial intelligence

Reconfigurable devices offer the ability to program electronic circuits on demand. Here, in this work, we demonstrated on-demand creation of artificial neurons, synapses, and memory capacitors in post-fabricated perovskite NdNiO 3 devices that can be simply reconfigured for a specific purpose by single-shot electric pulses. The sensitivity of electronic properties of perovskite nickelates to the local distribution of hydrogen ions enabled these results. With experimental data from our memory capacitors, simulation results of a reservoir computing framework showed excellent performance for tasks such as digit recognition and classification of electrocardiogram heartbeat activity. Using our reconfigurable artificial neurons and synapses, simulated dynamic networks outperformed static networks for incremental learning scenarios. The ability to fashion the building blocks of brain-inspired computers on demand opens up new directions in adaptive networks.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

Progressive CPU Volume Rendering with Sample Accumulation

Here, we present a new method for progressive volume rendering by accumulating object-space samples over successively rendered frames. Existing methods for progressive refinement either use image space methods or average pixels over frames, which can blur features or integrate incorrectly with respect to depth. Our approach stores samples along each ray, accumulates new samples each frame into a buffer, and progressively interleaves and integrates these samples. Though this process requires additional memory, it ensures interactivity and is well suited for CPU architectures with large memory and cache. This approach also extends well to distributed rendering in cluster environments. We implement this technique in Intel’s open source OSPRay CPU ray tracing framework and demonstrate that it is particularly useful for rendering volumetric data with costly sampling functions.

97 MATHEMATICS AND COMPUTING↗

Some Problems of Distributed and Boundary Control for Systems with Integral Aftereffect

We consider the problem of exact control for a system described by an equation with integral “memory.” It is shown that, under certain conditions, this system can be brought to rest in finite time by distributed control bounded in absolute value, and, in a special one-dimensional case, by control applied to an end-point of the interval. We consider different types of kernels in the integral term of the equation and describe some relationships between problems of controllability of some hyperbolic and parabolic systems.

97 MATHEMATICS AND COMPUTING↗

Role of atomistic structure in the stochastic nature of conductivity in substoichiometric tantalum pentoxide

First-principles calculations of electrical conductivity (σ{sub o}) are revisited to determine the atomistic origin of its stochasticity in a distribution generated from sampling 14 ab-initio molecular dynamics configurations from 10 independently quenched models (n = 140) of substoichiometric amorphous Ta{sub 2}O{sub 5}, where each structure contains a neutral O monovacancy (V{sub O}{sup 0}). Structural analysis revealed a distinct minimum Ta-Ta separation (dimer/trimer) corresponding to each V{sub O}{sup 0} location. Bader charge decomposition using a commonality analysis approach based on the σ{sub o} distribution extremes revealed nanostructural signatures indicating that both the magnitude and distribution of cationic charge on the Ta subnetwork have a profound influence on σ{sub o}. Furthermore, visualization of local defect structures and their electron densities reinforces these conclusions and suggests σ{sub o} in the amorphous oxide is best suppressed by a highly charged, compact Ta cation shell that effectively screens and minimizes localized V{sub O}{sup 0} interaction with the a-Ta{sub 2}O{sub 5} network; conversely, delocalization of V{sub O}{sup 0} corresponds to metallic character and high σ{sub o}. The random network of a-Ta{sub 2}O{sub 5} provides countless variations of an ionic configuration scaffold in which small perturbations affect the electronic charge distribution and result in a fixed-stoichiometry distribution of σ{sub o}; consequently, precisely controlled and highly repeatable oxide fabrication processes are likely paramount for advancement of resistive memory technologies.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

User-level File Systems Specialized for HPC Workloads Year-End Report

High-performance computing (HPC) clusters attract the attention of Deep Learning (DL) training users due to the clusters' powerful computation capabilities. While there have been many existing efforts to enable deep neural works to leverage the powerful CPU and GPU processors from leadership high-performance computing (HPC) systems, large-scale deep learning with larger datasets requires efficient I/O support from the underlying file and storage systems. As some current and upcoming HPC clusters have large on-node memory or are equipped with NVMe SSD on compute nodes, more distributed DL trainings are considered to leverage those storages to store dataset for efficient dataset access. Our project goal is to design a specialized DL-oriented file system to improve datasets loading performance of any DL training applications. Over the past year, the team at Florida State University has performed research activities in three different aspects, including 1) completing a specialized memory-based I/O framework (DeepIO) for improving dataset loading performance on DL applications; 2) proposing a more generalized file system (DLFS) for DL applications on node-local SSDs; 3) completing the BeeGFS performance evaluation project for Deep Neural Networks.

97 MATHEMATICS AND COMPUTING↗

Set statistics in conductive bridge random access memory device with Cu/HfO{sub 2}/Pt structure

The switching parameter variation of resistive switching memory is one of the most important challenges in its application. In this letter, we have studied the set statistics of conductive bridge random access memory with a Cu/HfO{sub 2}/Pt structure. The experimental distributions of the set parameters in several off resistance ranges are shown to nicely fit a Weibull model. The Weibull slopes of the set voltage and current increase and decrease logarithmically with off resistance, respectively. This experimental behavior is perfectly captured by a Monte Carlo simulator based on the cell-based set voltage statistics model and the Quantum Point Contact electron transport model. Our work provides indications for the improvement of the switching uniformity.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Minimizing CGYRO HPC Communication Costs in Ensembles with XGYRO by Sharing the Collisional Constant Tensor Structure

First-principles fusion plasma simulations are both compute and memory intensive, and CGYRO is no exception. The use of many HPC nodes to fit the problem in the available memory thus results in significant communication overhead, which is hard to avoid for any single simulation. That said, most fusion studies are composed of ensembles of simulations, so we developed a new tool, named XGYRO, that executes a whole ensemble of CGYRO simulations as a single HPC job. By treating the ensemble as a unit, XGYRO can alter the global buffer distribution logic and apply optimizations that are not feasible on any single simulation, but only on the ensemble as a whole. The main saving comes from the sharing of the collisional constant tensor structure, since its values are typically identical between parameter-sweep simulations. This data structure dominates the memory consumption of CGYRO simulations, so distributing it among the whole ensemble results in drastic memory savings for each simulation, which in turn results in overall lower communication overhead.

CGYRO↗

Classical non-Markovian Boltzmann equation

The modeling of particle transport involves anomalous diffusion, (x²(t) ) ∝ t{sup α} with α ≠ 1, with subdiffusive transport corresponding to 0 < α < 1 and superdiffusive transport to α > 1. These anomalies give rise to fractional advection-dispersion equations with memory in space and time. The usual Boltzmann equation, with only isolated binary collisions, is Markovian and, in particular, the contributions of the three-particle distribution function are neglected. We show that the inclusion of higher-order distribution functions give rise to an exact, non-Markovian Boltzmann equation with resulting transport equations for mass, momentum, and kinetic energy with memory in both time and space. The two- and the three-particle distribution functions are considered under the assumption that the two- and the three-particle correlation functions are translationally invariant that allows us to obtain advection-dispersion equations for modeling transport in terms of spatial and temporal fractional derivatives.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Techniques for storing data to enhance recovery and detection of data corruption errors

Often there are errors when reading data from computer memory. To detect and correct these errors, there are multiple types of error correction codes. Disclosed is an error correction architecture that creates a codeword having a data portion and an error correction code portion. Swizzling rearranges the order of bits and distributes the bits among different codewords. Because the data is redistributed, a potential memory error of up to N contiguous bits, where N for example equals 2 times the number of codewords swizzled together, only affects up to, at most, two bits per swizzled codeword. This keeps the error within the error detecting capabilities of the error correction architecture. Furthermore, this can allow improved error correction and detection without requiring a change to error correcting code generators and checkers.

Mills, Peter↗