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 577 records · Page 32

High Level Synthesis of RDF Queries for Graph Analytics

In this paper we present a set of techniques that enable the synthesis of efficient custom accelerators for memory intensive, irregular applications. To address irregular applications challenges (large memory footprints, unpredictable fine- grained data accesses, and high synchronization intensity), and exploit their opportunities (thread level parallelism, memory level parallelism), we propose a novel accelerator design which take advantage of an adaptive and Distributed Controller (DC) architecture, and a Memory Interface (MI) that supports parallel memory subsystems. Among the multitude of algorithms that may benefit from our solution, we focus on the acceleration of graph analytics applications, and in particular, on the syn- thesis of SPARQL queries on Resource Description Framework (RDF) databases. We achieve this objective by incorporating the synthesis techniques into Bambu, an Open Source high- level synthesis tools, and interfacing the system with GEMS, the Graph database Engine for Multithreaded Systems. The front- end of GEMS generates optimized C implementations of the input queries, modeled as pattern matching routines, which are then automatically synthesized by Bambu. We validate our approach synthesizing several SPARQL queries from the Lehigh University Benchmark (LUBM).

HLS, FPGA, Convey, RDF, GEMS, Memory-intensive,↗

Ambipolar organic thin-film transistor-based nano-floating-gate nonvolatile memory

An ambipolar organic thin-film transistor-based nano-floating-gate nonvolatile memory was demonstrated, with discrete distributed gold nanoparticles, tetratetracontane (TTC), pentacene as the floating-gate layer, tunneling layer, and active layer, respectively. The electron traps at the TTC/pentacene interface were significantly suppressed, which resulted in an ambipolar operation in present memory. As both electrons and holes were supplied in the channel and trapped in the floating-gate by programming/erasing operations, respectively, i.e., one type of charge carriers was used to overwrite the other, trapped, one, a large memory window, extending on both sides of the initial threshold voltage, was realized.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Fast and Scalable Sparse Triangular Solver for Multi-GPU Based HPC Architectures

Designing efficient and scalable sparse linear algebra kernels on modern multi-GPU based HPC systems is a daunting task due to significant irregular memory references and workload imbalance across the GPUs. This is particularly the case for \textit{Sparse Triangular Solver (SpTRSV)} which introduces additional two-dimensional computation dependencies among subsequent computation steps. Dependency information is exchanged and shared among GPUs, thus warrant for efficient memory allocation, data partitioning, and workload distribution as well as fine-grained communication and synchronization support. In this work, we demonstrate that directly adopting unified memory can adversely affect the performance of SpTRSV on multi-GPU architectures, despite linking via fast interconnect like NVLinks and NVSwitches. Alternatively, we employ the latest NVSHMEM technology based on Partitioned Global Address Space programming model to enable efficient fine-grained communication and drastic synchronization overhead reduction. Furthermore, to handle workload imbalance, we propose a malleable task-pool execution model which can further enhance the utilization of GPUs. By applying these techniques, our experiments on the NVIDIA multi-GPU supernode V100-DGX-1 and DGX-2 systems demonstrate that our design can achieve on average 3.53x (up to 9.86x) speedup on a DGX-1 system and 3.66x (up to 9.64x) speedup on a DGX-2 system with 4-GPUs over the Unified-Memory design. The comprehensive sensitivity and scalability studies also show that the proposed zero-copy SpTRSV is able to fully utilize the computing and communication resources of the multi-GPU system.

Xie, Chenhao↗

GAP CLEARING BY PLANETS IN A COLLISIONAL DEBRIS DISK

We apply our 3D debris disk model, SMACK, to simulate a planet on a circular orbit near a ring of planetesimals that are experiencing destructive collisions. Previous simulations of a planet opening a gap in a collisionless debris disk have found that the width of the gap scales as the planet mass to the 2/7th power (α = 2/7). We find that gap sizes in a collisional disk still obey a power law scaling with planet mass, but that the index α of the power law depends on the age of the system t relative to the collisional timescale t {sub coll} of the disk by α = 0.32(t/t {sub coll}){sup –0.04}, with inferred planet masses up to five times smaller than those predicted by the classical gap law. The increased gap sizes likely stem from the interaction between collisions and the mean motion resonances near the chaotic zone. We investigate the effects of the initial eccentricity distribution of the disk particles and find a negligible effect on the gap size at Jovian planet masses, since collisions tend to erase memory of the initial particle eccentricity distributions. Finally, we find that the presence of Trojan analogs is a potentially powerful diagnostic of planets in the mass range ∼1-10 M {sub Jup}. We apply our model to place new upper limits on planets around Fomalhaut, HR 4796 A, HD 202628, HD 181327, and β Pictoris.

79 ASTRONOMY AND ASTROPHYSICS↗

Efficient sampling for spatial uncertainty qualification in multibody system dynamics applications.

We present two methods for efficiently sampling the response (trajectory space) of multibody systems operating under spatial uncertainty, when the latter is assumed to be representable with Gaussian processes. In this case, the dynamics (time evolution) of the multibody systems depends on spatially indexed uncertain parameters that span infinite-dimensional spaces. This places a heavy computational burden on existing methodologies, an issue addressed herein with two new conditional sampling approaches. When a single instance of the uncertainty is needed in the entire domain, we use a fast Fourier transform technique. When the initial conditions are fixed and the path distribution of the dynamical system is relatively narrow, we use an incremental sampling approach that is fast and has a small memory footprint. Both methods produce the same distributions as the widely used Cholesky-based approaches. We illustrate this convergence at a smaller computational effort and memory cost for a simple non-linear vehicle model.

97 MATHEMATICS AND COMPUTING↗

Support for non-locking parallel reception of packets belonging to a single memory reception FIFO

A method and apparatus for distributed parallel messaging in a parallel computing system. A plurality of DMA engine units are configured in a multiprocessor system to operate in parallel, one DMA engine unit for transferring a current packet received at a network reception queue to a memory location in a memory FIFO (rmFIFO) region of a memory. A control unit implements logic to determine whether any prior received packet destined for that rmFIFO is still in a process of being stored in the associated memory by another DMA engine unit of the plurality, and prevent the one DMA engine unit from indicating completion of storing the current received packet in the reception memory FIFO (rmFIFO) until all prior received packets destined for that rmFIFO are completely stored by the other DMA engine units. Thus, there is provided non-locking support so that multiple packets destined for a single rmFIFO are transferred and stored in parallel to predetermined locations in a memory.

97 MATHEMATICS AND COMPUTING↗

Implementing High-Performance Geometric Multigrid Solver with Naturally Grained Messages

Structured-grid linear solvers often require manually packing and unpacking of communication data to achieve high performance.Orchestrating this process efficiently is challenging, labor-intensive, and potentially error-prone.In this paper, we explore an alternative approach that communicates the data with naturally grained messagesizes without manual packing and unpacking. This approach is the distributed analogue of shared-memory programming, taking advantage of the global addressspace in PGAS languages to provide substantial programming ease. However, its performance may suffer from the large number of small messages. We investigate theruntime support required in the UPC ++ library for this naturally grained version to close the performance gap between the two approaches and attain comparable performance at scale using the High-Performance Geometric Multgrid (HPGMG-FV) benchmark as a driver.

97 MATHEMATICS AND COMPUTING↗

Automatic Parallelization of Numerical Python Applications using the Global Arrays Toolkit

Global Arrays is a software system from Pacific Northwest National Laboratory that enables an efficient, portable, and parallel shared-memory programming interface to manipulate distributed dense arrays. The NumPy module is the de facto standard for numerical calculation in the Python programming language, a language whose use is growing rapidly in the scientific and engineering communities. NumPy provides a powerful N-dimensional array class as well as other scientific computing capabilities. However, like the majority of the core Python modules, NumPy is inherently serial. Using a combination of Global Arrays and NumPy, we have reimplemented NumPy as a distributed drop-in replacement called Global Arrays in NumPy (GAiN). Serial NumPy applications can become parallel, scalable GAiN applications with only minor source code changes. Scalability studies of several different GAiN applications will be presented showing the utility of developing serial NumPy codes which can later run on more capable clusters or supercomputers.

99 GENERAL AND MISCELLANEOUS↗

Optimizing the Accelerated Recursive Doubling Algorithm for Block Tridiagonal Systems of Equations

The need to solve block tridiagonal systems with hundreds or thousands of right-hand sides for the same block tridiagonal matrix is common in a variety of disciplines. To meet this need, the Accelerated Recursive Doubling Algorithm was developed. After a right-hand side independent phase, the algorithm allows for the quick, online calculation of solutions for different right-hand sides. In this work, we present methods to optimize the Accelerated Recursive Doubling Algorithm in memory usage and computation time in a hybrid parallelization model. The right-hand side independent phase of the naïve implementation takes ≥ 11/3 the amount of memory required to store the tridiagonal matrix, while our implementation reduces the fraction to ≈ 5/3 . The right-hand side dependent phase of the naïve implementation takes ≥ 6 times the amount of memory required to store the right-hand side, while our implementation reduces the fraction to ≈ 3. The computation time for the independent phase is reduced to ≈ 2/3 times that of the naïve implementation, while the computation time for the dependent phase is reduced to ≈ 5/9 . With increasing numbers of shared-memory threads q on every distributed processing element, we have O(q) theoretical speedup.

97 MATHEMATICS AND COMPUTING↗

Charged particle tracking in real-time using a full-mesh data delivery architecture and associative memory techniques

We present a flexible and scalable approach to address the challenges of charged particle track reconstruction in real-time event filters (Level-1 triggers) in collider physics experiments. The method described here is based on a full-mesh architecture for data distribution and relies on the Associative Memory approach to implement a pattern recognition algorithm that quickly identifies and organizes hits associated to trajectories of particles originating from particle collisions. We describe a successful implementation of a demonstration system composed of several innovative hardware and algorithmic elements. The implementation of a full-size system relies on the assumption that an Associative Memory device with the sufficient pattern density becomes available in the future, either through a dedicated ASIC or a modern FPGA. We demonstrate excellent performance in terms of track reconstruction efficiency, purity, momentum resolution, and processing time measured with data from a simulated LHC-like tracking detector.

47 OTHER INSTRUMENTATION↗

The Nature of Grand Minima and Maxima from Fully Nonlinear Flux Transport Dynamos

We aim to investigate the nature and occurrence characteristics of grand solar minimum and maximum periods, which are observed in the solar proxy records such as {sup 10}Be and {sup 14}C, using a fully nonlinear Babcock–Leighton type flux transport dynamo including momentum and entropy equations. The differential rotation and meridional circulation are generated from the effect of turbulent Reynolds stress and are subjected to back-reaction from the magnetic field. To generate grand minimum- and maximum-like periods in our simulations, we used random fluctuations in the angular momentum transport process, namely the Λ-mechanism, and in the Babcock–Leighton mechanism. To characterize the nature and occurrences of the identified grand minima and maxima in our simulations, we used the waiting time distribution analyses, which reflect whether the underlying distribution arises from a random or a memory-bearing process. The results show that, in the majority of the cases, the distributions of grand minima and maxima reveal that the nature of these events originates from memoryless processes. We also found that in our simulations the meridional circulation speed tends to be smaller during grand maximum, while it is faster during grand minimum periods. The radial differential rotation tends to be larger during grand maxima, while it is smaller during grand minima. The latitudinal differential rotation, on the other hand, is found to be larger during grand minima.

79 ASTRONOMY AND ASTROPHYSICS↗

Accurate, Fast and Scalable Kernel Ridge Regression on Parallel and Distributed Systems

Kernel Ridge Regression (KRR) is a fundamental method in machine learning. Given an n-by-d data matrix as input, a traditional implementation requires Θ(n2) memory to form an n-by-n kernel matrix and Θ(n3) flops to compute the final model. These time and storage costs prohibit KRR from scaling up to large datasets. For example, even on a relatively small dataset (a 520k-by-90 input requiring 357 MB), KRR requires 2 TB memory just to store the kernel matrix. Additionally, the reason is that n usually is much larger than d for real-world applications. On the other hand, weak scaling becomes a problem: if we keep d and n/p fixed as p grows (p is # machines), the memory needed grows as Θ(p) per processor and the flops as Θ(p2) per processor. In the perfect weak scaling situation, both the memory needed and the flops grow as Θ(1) per processor (i.e. memory and flops are constant). The traditional Distributed KRR implementation (DKRR) only achieved 0.32% weak scaling efficiency from 96 to 1536 processors. In this work, we propose two new methods to address these problems: the Balanced KRR (BKRR) and K-means KRR (KKRR). These methods consider alternative ways to partition the input dataset into p different parts, generating p different models, and then selecting the best model among them. Compared to a conventional implementation, KKRR2 (optimized version of KKRR) improves the weak scaling efficiency from 0.32% to 38% and achieves a 591x speedup for getting the same accuracy by using the same data and the same hardware (1536 processors). BKRR2 (optimized version of BKRR) achieves a higher accuracy than the current fastest method using less training time for a variety of datasets. For the applications requiring only approximate solutions, BKRR2 improves the weak scaling efficiency to 92% and achieves 3505x speedup (theoretical speedup: 4096x).

97 MATHEMATICS AND COMPUTING↗

Implementing efficient data compression and encryption in a persistent key-value store for HPC

Recently, persistent data structures, like key-value stores (KVSs), which are stored in a high-performance computing (HPC) system’s nonvolatile memory, provide an attractive solution for a number of emerging challenges like limited I/O performance. Data compression and encryption are two well-known techniques for improving several properties of such data-oriented systems. This article investigates how to efficiently integrate data compression and encryption into persistent KVSs for HPC with the ultimate goal of hiding their costs and complexity in terms of performance and ease of use. Our compression technique exploits deep memory hierarchy in an HPC system to achieve both storage reduction and performance improvement. Our encryption technique provides a practical level of security and enables sharing of sensitive data securely in complex scientific workflows with nearly imperceptible cost. We implement the proposed techniques on top of a distributed embedded KVS to evaluate the benefits and costs of incorporating these capabilities along different points in the dataflow path, illustrating differences in effective bandwidth, latency, and additional computational expense on Swiss National Supercomputing Centre’s Grand Tavé and National Energy Research Scientific Computing Center’s Cori.

97 MATHEMATICS AND COMPUTING↗

Concurrent Monte Carlo transport and fluence optimization with fluence adjusting scalable transport Monte Carlo

Purpose: The future of radiation therapy will require advanced inverse planning solutions to support single-arc, multiple-arc, and “4π” delivery modes, which present unique challenges in finding an optimal treatment plan over a vast search space, while still preserving dosimetric accuracy. The successful clinical implementation of such methods would benefit from Monte Carlo (MC) based dose calculation methods, which can offer improvements in dosimetric accuracy when compared to deterministic methods. The standard method for MC based treatment planning optimization leverages the accuracy of the MC dose calculation and efficiency of well-developed optimization methods, by precalculating the fluence to dose relationship within a patient with MC methods and subsequently optimizing the fluence weights. However, the sequential nature of this implementation is computationally time consuming and memory intensive. Methods to reduce the overhead of the MC precalculation have been explored in the past, demonstrating promising reductions of computational time overhead, but with limited impact on the memory overhead due to the sequential nature of the dose calculation and fluence optimization. The authors propose an entirely new form of “concurrent” Monte Carlo treat plan optimization: a platform which optimizes the fluence during the dose calculation, reduces wasted computation time being spent on beamlets that weakly contribute to the final dose distribution, and requires only a low memory footprint to function. In this initial investigation, the authors explore the key theoretical and practical considerations of optimizing fluence in such a manner. Methods: The authors present a novel derivation and implementation of a gradient descent algorithm that allows for optimization during MC particle transport, based on highly stochastic information generated through particle transport of very few histories. A gradient rescaling and renormalization algorithm, and the concept of momentum from stochastic gradient descent were used to address obstacles unique to performing gradient descent fluence optimization during MC particle transport. The authors have applied their method to two simple geometrical phantoms, and one clinical patient geometry to examine the capability of this platform to generate conformal plans as well as assess its computational scaling and efficiency, respectively. Results: The authors obtain a reduction of at least 50% in total histories transported in their investigation compared to a theoretical unweighted beamlet calculation and subsequent fluence optimization method, and observe a roughly fixed optimization time overhead consisting of ∼10% of the total computation time in all cases. Finally, the authors demonstrate a negligible increase in memory overhead of ∼7–8 MB to allow for optimization of a clinical patient geometry surrounded by 36 beams using their platform. Conclusions: This study demonstrates a fluence optimization approach, which could significantly improve the development of next generation radiation therapy solutions while incurring minimal additional computational overhead.

60 APPLIED LIFE SCIENCES↗

Modular Spiking Neural Circuits for Mapping Long Short-Term Memory on a Neurosynaptic Processor

Due to the distributed and asynchronous nature of neural computation through low-energy spikes, brain-inspired hardware systems offer high energy efficiency and massive parallelism. One such platform is the IBM TrueNorth neurosynaptic system. Recently, TrueNorth compatible representation learning algorithms have emerged, achieving close to the state-of-the-art performance in various data sets. However, its application in temporal sequence processing models, such as recurrent neural networks (RNNs), is still only at the proof of concept level. There is an inherent difficulty in capturing temporal dynamics of an RNN using spiking neurons, which is only exasperated by the hardware constraints in connectivity and synaptic weight resolution. This paper presents a design flow that overcomes these difficulties and maps a special case of recurrent networks called long short-term memory (LSTM) onto a spike-based platform. The framework utilizes various approximation techniques, such as activation discretization, weight quantization, and scaling and rounding, spiking neural circuits that implement the complex gating mechanisms, and a store-and-release technique to enable neuron synchronization and faithful storage. While the presented techniques can be applied to map LSTM to any spiking neural network (SNN) simulator/emulator, here we choose the TrueNorth chip as the target platform by adhering to its hardware constraints. Three LSTM applications, parity check, extended Reber grammar, and question classification, are evaluated. The tradeoffs among accuracy, performance, and energy tradeoffs achieved on TrueNorth are demonstrated. Finally, this is compared with the performance on an SNN platform without hardware constraints, which represents the upper bound of the achievable accuracy.

97 MATHEMATICS AND COMPUTING↗

A generalized massively parallel ultra-high order FFT-based Maxwell solver

Dispersion-free ultra-high order FFT-based Maxwell solvers have recently proven to be paramount to a large range of applications, including the high-fidelity modeling of high-intensity laser–matter interactions with Particle-In-Cell (PIC) codes. To enable a massively parallel scaling of these solvers, a novel parallelization technique was recently proposed, which consists in splitting the simulation domain into several processor sub-domains, with guard regions appended at each sub-domain boundary. Maxwell's equations are advanced independently on each sub-domain using local shared-memory FFTs (instead of a single distributed global FFT). This implies small truncation errors at sub-domain boundaries, the amplitude of which depends on guard regions sizes and order of the Maxwell solver. For moderate guard region sizes, this ’local’ technique proved to be highly scalable on up to a million cores and notably enabled the 3D modeling of so-called plasma mirrors, for which 8 guard cells only were enough to prevent truncation error growth. Yet, for other applications, the required number of guard cells might be much higher, which would severely limit the parallel efficiency of this technique due to the large volume of guard cells to be exchanged between sub-domains. In this context, we propose a novel parallelization technique that ensures very good scaling of FFT-based solvers with an arbitrarily high number of guard cells. Our ’hybrid’ technique consists in performing distributed FFTs on local groups of processors with guard regions now appended to boundaries of each group of processors. It uses a dual domain decomposition method for the Maxwell solver and other parts of the PIC cycle to keep the simulation load-balanced. This ’hybrid’ technique was implemented in the open source exascale library PICSAR. Benchmarks show that for a large number of guard cells (>16), the ’hybrid’ technique offers up to ×3 speed-up and ×8 memory savings compared to the ’local’ one.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Distribution of nanoscale nuclei in the amorphous dome of a phase change random access memory

The nanoscale crystal nuclei in an amorphous Ge{sub 2}Sb{sub 2}Te{sub 5} bit in a phase change memory device were evaluated by fluctuation transmission electron microscopy. The quench time in the device (∼10 ns) afforded more and larger nuclei in the melt-quenched state than in the as-deposited state. However, nuclei were even more numerous and larger in a test structure with a longer quench time (∼100 ns), verifying the prediction of nucleation theory that slower cooling produces more nuclei. It also demonstrates that the thermal design of devices will strongly influence the population of nuclei, and thus the speed and data retention characteristics.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

A simple and efficient random walk solution of multi-rate mobile/immobile mass transport equations

We extend the particle-tracking method to simulate general multi-rate mass transfer (MRMT) equations. Previous methods for single-rate equations used two-state Markov chains and found that the time a par- ticle spends in the mobile state between waiting time epochs is random and exponentially distributed. Using Bochner's subordination technique for Markov processes, we find that the random mobile times are still exponential for the stochastic process that corresponds to the MRMT equations. The random times in the immobile phase have a distribution that is directly related to the memory function of the MRMT equation. This connection allows us to interpret the MRMT memory function as the rate at which particles of a certain age, measured by residence time in the immobile zone, exit to become mobile once again. Because the exact distributions of mobile and immobile times are known from the MRMT equa- tions, they can be simulated very simply and efficiently using random walks.

58 GEOSCIENCES↗